From 2a64b546a0efd4bb7f5907db9a02a03c2bb045ed Mon Sep 17 00:00:00 2001 From: V3n3RiX Date: Sun, 4 Nov 2018 02:42:59 +0000 Subject: sys-kernel/linux-{image,sources}-redcore-lts : * version bump to v4.14.75 * adjust MuQSS v0.162 CPU scheduler to linux-hardened and enable it by default * add BFQ-SQ v8r12 I/O scheduler and enable it by default * lower HZ to 100 (recommended value for MuQSS enabled kernels) * disable dynamic ticks (MuQSS itself is a tickless scheduler, so no longer required) --- .../files/0001-BFQ-v8r12-20171108.patch | 25199 +++++++++++++++++++ ...ersion-0.162-CPU-scheduler-linux-hardened.patch | 9571 +++++++ .../files/0002-BFQ-v8r12-20180404.patch | 4611 ++++ .../0002-Make-preemptible-kernel-default.patch | 733 + ...-Expose-vmsplit-for-our-poor-32-bit-users.patch | 48 + ...res-timeout-variants-of-schedule_timeout-.patch | 153 + ...e-calls-of-schedule_timeout-1-to-use-the-.patch | 50 + ...onvert-msleep-to-use-hrtimers-when-active.patch | 54 + ...-schedule-timeout-1-with-schedule_min_hrt.patch | 529 + ...-calls-to-schedule_timeout_interruptible-.patch | 311 + ...-calls-to-schedule_timeout_uninterruptibl.patch | 160 + ...rtimer-overlay-when-pm_freezing-since-som.patch | 69 + ...r-granularity-and-minimum-hrtimeout-confi.patch | 136 + ...efault-Hz-of-100-in-combination-with-MuQS.patch | 81 + ...ed-IRQs-optionally-the-default-which-can-.patch | 61 + .../files/0014-Swap-sucks.patch | 25 + .../0015-Enable-BFQ-io-scheduler-by-default.patch | 38 - ...SS.c-needs-irq_regs.h-to-use-get_irq_regs.patch | 19 + .../0016-unfuck-MuQSS-on-linux-4_14_15+.patch | 48 + .../files/redcore-lts-amd64.config | 41 +- .../files/uksm-for-linux-hardened.patch | 6919 ----- .../files/uksm-linux-hardened.patch | 6919 +++++ 22 files changed, 48798 insertions(+), 6977 deletions(-) create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0001-BFQ-v8r12-20171108.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0001-MuQSS-version-0.162-CPU-scheduler-linux-hardened.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0002-BFQ-v8r12-20180404.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0002-Make-preemptible-kernel-default.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0003-Expose-vmsplit-for-our-poor-32-bit-users.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0004-Create-highres-timeout-variants-of-schedule_timeout-.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0005-Special-case-calls-of-schedule_timeout-1-to-use-the-.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0006-Convert-msleep-to-use-hrtimers-when-active.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0007-Replace-all-schedule-timeout-1-with-schedule_min_hrt.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0008-Replace-all-calls-to-schedule_timeout_interruptible-.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0009-Replace-all-calls-to-schedule_timeout_uninterruptibl.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0010-Don-t-use-hrtimer-overlay-when-pm_freezing-since-som.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0011-Make-hrtimer-granularity-and-minimum-hrtimeout-confi.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0012-Reinstate-default-Hz-of-100-in-combination-with-MuQS.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0013-Make-threaded-IRQs-optionally-the-default-which-can-.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0014-Swap-sucks.patch delete mode 100644 sys-kernel/linux-sources-redcore-lts/files/0015-Enable-BFQ-io-scheduler-by-default.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0015-MuQSS.c-needs-irq_regs.h-to-use-get_irq_regs.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/0016-unfuck-MuQSS-on-linux-4_14_15+.patch delete mode 100644 sys-kernel/linux-sources-redcore-lts/files/uksm-for-linux-hardened.patch create mode 100644 sys-kernel/linux-sources-redcore-lts/files/uksm-linux-hardened.patch (limited to 'sys-kernel/linux-sources-redcore-lts/files') diff --git a/sys-kernel/linux-sources-redcore-lts/files/0001-BFQ-v8r12-20171108.patch b/sys-kernel/linux-sources-redcore-lts/files/0001-BFQ-v8r12-20171108.patch new file mode 100644 index 00000000..db7d064b --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0001-BFQ-v8r12-20171108.patch @@ -0,0 +1,25199 @@ +From c21f53f17430230dab50df29b8ea1b71f99d09d6 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Tue, 7 Apr 2015 13:39:12 +0200 +Subject: [PATCH 01/51] Add BFQ-v8r12 + +This commit is the result of the following operations. + +1. The squash of all the commits between "block: cgroups, kconfig, +build bits for BFQ-v7r11-4.5.0" and BFQ-v8r12 in the branch +bfq-mq-v8-v4.11 + +2. The renaming of two files (block/bfq-cgroup.c -> +block/bfq-cgroup-included.c and block/bfq-iosched.c -> +block/bfq-sq-iosched.c) and of one option (CONFIG_BFQ_GROUP_IOSCHED -> +CONFIG_BFQ_SQ_GROUP_IOSCHED), to avoid name clashes. These name +clashes are due to the presence of bfq in mainline from 4.12. + +3. The modification of block/Makefile and block/Kconfig.iosched to +comply with the above renaming. + +Signed-off-by: Mauro Andreolini +Signed-off-by: Arianna Avanzini +Signed-off-by: Linus Walleij +Signed-off-by: Paolo Valente +--- + Makefile | 2 +- + block/Kconfig.iosched | 31 + + block/bfq-cgroup-included.c | 1190 ++++++++++ + block/bfq-ioc.c | 36 + + block/bfq-sched.c | 2002 ++++++++++++++++ + block/bfq-sq-iosched.c | 5379 +++++++++++++++++++++++++++++++++++++++++++ + block/bfq.h | 948 ++++++++ + include/linux/blkdev.h | 2 +- + 9 files changed, 9589 insertions(+), 2 deletions(-) + create mode 100644 block/bfq-cgroup-included.c + create mode 100644 block/bfq-ioc.c + create mode 100644 block/bfq-sched.c + create mode 100644 block/bfq-sq-iosched.c + create mode 100644 block/bfq.h + +diff --git a/block/Kconfig.iosched b/block/Kconfig.iosched +index a4a8914bf7a4..9e3f4c2f7390 100644 +--- a/block/Kconfig.iosched ++++ b/block/Kconfig.iosched +@@ -40,6 +40,26 @@ config CFQ_GROUP_IOSCHED + ---help--- + Enable group IO scheduling in CFQ. + ++config IOSCHED_BFQ_SQ ++ tristate "BFQ-SQ I/O scheduler" ++ default n ++ ---help--- ++ The BFQ-SQ I/O scheduler (for legacy blk: SQ stands for ++ SingleQueue) distributes bandwidth among all processes ++ according to their weights, regardless of the device ++ parameters and with any workload. It also guarantees a low ++ latency to interactive and soft real-time applications. ++ Details in Documentation/block/bfq-iosched.txt ++ ++config BFQ_SQ_GROUP_IOSCHED ++ bool "BFQ-SQ hierarchical scheduling support" ++ depends on IOSCHED_BFQ_SQ && BLK_CGROUP ++ default n ++ ---help--- ++ ++ Enable hierarchical scheduling in BFQ-SQ, using the blkio ++ (cgroups-v1) or io (cgroups-v2) controller. ++ + choice + + prompt "Default I/O scheduler" +@@ -54,6 +74,16 @@ choice + config DEFAULT_CFQ + bool "CFQ" if IOSCHED_CFQ=y + ++ config DEFAULT_BFQ_SQ ++ bool "BFQ-SQ" if IOSCHED_BFQ_SQ=y ++ help ++ Selects BFQ-SQ as the default I/O scheduler which will be ++ used by default for all block devices. ++ The BFQ-SQ I/O scheduler aims at distributing the bandwidth ++ as desired, independently of the disk parameters and with ++ any workload. It also tries to guarantee low latency to ++ interactive and soft real-time applications. ++ + config DEFAULT_NOOP + bool "No-op" + +@@ -63,6 +93,7 @@ config DEFAULT_IOSCHED + string + default "deadline" if DEFAULT_DEADLINE + default "cfq" if DEFAULT_CFQ ++ default "bfq-sq" if DEFAULT_BFQ_SQ + default "noop" if DEFAULT_NOOP + + config MQ_IOSCHED_DEADLINE +diff --git a/block/Makefile b/block/Makefile +index 6a56303b9925..59026b425791 100644 +--- a/block/Makefile ++++ b/block/Makefile +@@ -24,6 +24,7 @@ obj-$(CONFIG_MQ_IOSCHED_DEADLINE) += mq-deadline.o + obj-$(CONFIG_MQ_IOSCHED_KYBER) += kyber-iosched.o + bfq-y := bfq-iosched.o bfq-wf2q.o bfq-cgroup.o + obj-$(CONFIG_IOSCHED_BFQ) += bfq.o ++obj-$(CONFIG_IOSCHED_BFQ_SQ) += bfq-sq-iosched.o + + obj-$(CONFIG_BLOCK_COMPAT) += compat_ioctl.o + obj-$(CONFIG_BLK_CMDLINE_PARSER) += cmdline-parser.o +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +new file mode 100644 +index 000000000000..af7c216a3540 +--- /dev/null ++++ b/block/bfq-cgroup-included.c +@@ -0,0 +1,1190 @@ ++/* ++ * BFQ: CGROUPS support. ++ * ++ * Based on ideas and code from CFQ: ++ * Copyright (C) 2003 Jens Axboe ++ * ++ * Copyright (C) 2008 Fabio Checconi ++ * Paolo Valente ++ * ++ * Copyright (C) 2015 Paolo Valente ++ * ++ * Copyright (C) 2016 Paolo Valente ++ * ++ * Licensed under the GPL-2 as detailed in the accompanying COPYING.BFQ ++ * file. ++ */ ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ ++/* bfqg stats flags */ ++enum bfqg_stats_flags { ++ BFQG_stats_waiting = 0, ++ BFQG_stats_idling, ++ BFQG_stats_empty, ++}; ++ ++#define BFQG_FLAG_FNS(name) \ ++static void bfqg_stats_mark_##name(struct bfqg_stats *stats) \ ++{ \ ++ stats->flags |= (1 << BFQG_stats_##name); \ ++} \ ++static void bfqg_stats_clear_##name(struct bfqg_stats *stats) \ ++{ \ ++ stats->flags &= ~(1 << BFQG_stats_##name); \ ++} \ ++static int bfqg_stats_##name(struct bfqg_stats *stats) \ ++{ \ ++ return (stats->flags & (1 << BFQG_stats_##name)) != 0; \ ++} \ ++ ++BFQG_FLAG_FNS(waiting) ++BFQG_FLAG_FNS(idling) ++BFQG_FLAG_FNS(empty) ++#undef BFQG_FLAG_FNS ++ ++/* This should be called with the queue_lock held. */ ++static void bfqg_stats_update_group_wait_time(struct bfqg_stats *stats) ++{ ++ unsigned long long now; ++ ++ if (!bfqg_stats_waiting(stats)) ++ return; ++ ++ now = sched_clock(); ++ if (time_after64(now, stats->start_group_wait_time)) ++ blkg_stat_add(&stats->group_wait_time, ++ now - stats->start_group_wait_time); ++ bfqg_stats_clear_waiting(stats); ++} ++ ++/* This should be called with the queue_lock held. */ ++static void bfqg_stats_set_start_group_wait_time(struct bfq_group *bfqg, ++ struct bfq_group *curr_bfqg) ++{ ++ struct bfqg_stats *stats = &bfqg->stats; ++ ++ if (bfqg_stats_waiting(stats)) ++ return; ++ if (bfqg == curr_bfqg) ++ return; ++ stats->start_group_wait_time = sched_clock(); ++ bfqg_stats_mark_waiting(stats); ++} ++ ++/* This should be called with the queue_lock held. */ ++static void bfqg_stats_end_empty_time(struct bfqg_stats *stats) ++{ ++ unsigned long long now; ++ ++ if (!bfqg_stats_empty(stats)) ++ return; ++ ++ now = sched_clock(); ++ if (time_after64(now, stats->start_empty_time)) ++ blkg_stat_add(&stats->empty_time, ++ now - stats->start_empty_time); ++ bfqg_stats_clear_empty(stats); ++} ++ ++static void bfqg_stats_update_dequeue(struct bfq_group *bfqg) ++{ ++ blkg_stat_add(&bfqg->stats.dequeue, 1); ++} ++ ++static void bfqg_stats_set_start_empty_time(struct bfq_group *bfqg) ++{ ++ struct bfqg_stats *stats = &bfqg->stats; ++ ++ if (blkg_rwstat_total(&stats->queued)) ++ return; ++ ++ /* ++ * group is already marked empty. This can happen if bfqq got new ++ * request in parent group and moved to this group while being added ++ * to service tree. Just ignore the event and move on. ++ */ ++ if (bfqg_stats_empty(stats)) ++ return; ++ ++ stats->start_empty_time = sched_clock(); ++ bfqg_stats_mark_empty(stats); ++} ++ ++static void bfqg_stats_update_idle_time(struct bfq_group *bfqg) ++{ ++ struct bfqg_stats *stats = &bfqg->stats; ++ ++ if (bfqg_stats_idling(stats)) { ++ unsigned long long now = sched_clock(); ++ ++ if (time_after64(now, stats->start_idle_time)) ++ blkg_stat_add(&stats->idle_time, ++ now - stats->start_idle_time); ++ bfqg_stats_clear_idling(stats); ++ } ++} ++ ++static void bfqg_stats_set_start_idle_time(struct bfq_group *bfqg) ++{ ++ struct bfqg_stats *stats = &bfqg->stats; ++ ++ stats->start_idle_time = sched_clock(); ++ bfqg_stats_mark_idling(stats); ++} ++ ++static void bfqg_stats_update_avg_queue_size(struct bfq_group *bfqg) ++{ ++ struct bfqg_stats *stats = &bfqg->stats; ++ ++ blkg_stat_add(&stats->avg_queue_size_sum, ++ blkg_rwstat_total(&stats->queued)); ++ blkg_stat_add(&stats->avg_queue_size_samples, 1); ++ bfqg_stats_update_group_wait_time(stats); ++} ++ ++static struct blkcg_policy blkcg_policy_bfq; ++ ++/* ++ * blk-cgroup policy-related handlers ++ * The following functions help in converting between blk-cgroup ++ * internal structures and BFQ-specific structures. ++ */ ++ ++static struct bfq_group *pd_to_bfqg(struct blkg_policy_data *pd) ++{ ++ return pd ? container_of(pd, struct bfq_group, pd) : NULL; ++} ++ ++static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg) ++{ ++ return pd_to_blkg(&bfqg->pd); ++} ++ ++static struct bfq_group *blkg_to_bfqg(struct blkcg_gq *blkg) ++{ ++ struct blkg_policy_data *pd = blkg_to_pd(blkg, &blkcg_policy_bfq); ++ ++ return pd_to_bfqg(pd); ++} ++ ++/* ++ * bfq_group handlers ++ * The following functions help in navigating the bfq_group hierarchy ++ * by allowing to find the parent of a bfq_group or the bfq_group ++ * associated to a bfq_queue. ++ */ ++ ++static struct bfq_group *bfqg_parent(struct bfq_group *bfqg) ++{ ++ struct blkcg_gq *pblkg = bfqg_to_blkg(bfqg)->parent; ++ ++ return pblkg ? blkg_to_bfqg(pblkg) : NULL; ++} ++ ++static struct bfq_group *bfqq_group(struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *group_entity = bfqq->entity.parent; ++ ++ return group_entity ? container_of(group_entity, struct bfq_group, ++ entity) : ++ bfqq->bfqd->root_group; ++} ++ ++/* ++ * The following two functions handle get and put of a bfq_group by ++ * wrapping the related blk-cgroup hooks. ++ */ ++ ++static void bfqg_get(struct bfq_group *bfqg) ++{ ++ return blkg_get(bfqg_to_blkg(bfqg)); ++} ++ ++static void bfqg_put(struct bfq_group *bfqg) ++{ ++ return blkg_put(bfqg_to_blkg(bfqg)); ++} ++ ++static void bfqg_stats_update_io_add(struct bfq_group *bfqg, ++ struct bfq_queue *bfqq, ++ unsigned int op) ++{ ++ blkg_rwstat_add(&bfqg->stats.queued, op, 1); ++ bfqg_stats_end_empty_time(&bfqg->stats); ++ if (!(bfqq == ((struct bfq_data *)bfqg->bfqd)->in_service_queue)) ++ bfqg_stats_set_start_group_wait_time(bfqg, bfqq_group(bfqq)); ++} ++ ++static void bfqg_stats_update_io_remove(struct bfq_group *bfqg, unsigned int op) ++{ ++ blkg_rwstat_add(&bfqg->stats.queued, op, -1); ++} ++ ++static void bfqg_stats_update_io_merged(struct bfq_group *bfqg, unsigned int op) ++{ ++ blkg_rwstat_add(&bfqg->stats.merged, op, 1); ++} ++ ++static void bfqg_stats_update_completion(struct bfq_group *bfqg, ++ uint64_t start_time, uint64_t io_start_time, ++ unsigned int op) ++{ ++ struct bfqg_stats *stats = &bfqg->stats; ++ unsigned long long now = sched_clock(); ++ ++ if (time_after64(now, io_start_time)) ++ blkg_rwstat_add(&stats->service_time, op, ++ now - io_start_time); ++ if (time_after64(io_start_time, start_time)) ++ blkg_rwstat_add(&stats->wait_time, op, ++ io_start_time - start_time); ++} ++ ++/* @stats = 0 */ ++static void bfqg_stats_reset(struct bfqg_stats *stats) ++{ ++ /* queued stats shouldn't be cleared */ ++ blkg_rwstat_reset(&stats->merged); ++ blkg_rwstat_reset(&stats->service_time); ++ blkg_rwstat_reset(&stats->wait_time); ++ blkg_stat_reset(&stats->time); ++ blkg_stat_reset(&stats->avg_queue_size_sum); ++ blkg_stat_reset(&stats->avg_queue_size_samples); ++ blkg_stat_reset(&stats->dequeue); ++ blkg_stat_reset(&stats->group_wait_time); ++ blkg_stat_reset(&stats->idle_time); ++ blkg_stat_reset(&stats->empty_time); ++} ++ ++/* @to += @from */ ++static void bfqg_stats_add_aux(struct bfqg_stats *to, struct bfqg_stats *from) ++{ ++ if (!to || !from) ++ return; ++ ++ /* queued stats shouldn't be cleared */ ++ blkg_rwstat_add_aux(&to->merged, &from->merged); ++ blkg_rwstat_add_aux(&to->service_time, &from->service_time); ++ blkg_rwstat_add_aux(&to->wait_time, &from->wait_time); ++ blkg_stat_add_aux(&from->time, &from->time); ++ blkg_stat_add_aux(&to->avg_queue_size_sum, &from->avg_queue_size_sum); ++ blkg_stat_add_aux(&to->avg_queue_size_samples, ++ &from->avg_queue_size_samples); ++ blkg_stat_add_aux(&to->dequeue, &from->dequeue); ++ blkg_stat_add_aux(&to->group_wait_time, &from->group_wait_time); ++ blkg_stat_add_aux(&to->idle_time, &from->idle_time); ++ blkg_stat_add_aux(&to->empty_time, &from->empty_time); ++} ++ ++/* ++ * Transfer @bfqg's stats to its parent's dead_stats so that the ancestors' ++ * recursive stats can still account for the amount used by this bfqg after ++ * it's gone. ++ */ ++static void bfqg_stats_xfer_dead(struct bfq_group *bfqg) ++{ ++ struct bfq_group *parent; ++ ++ if (!bfqg) /* root_group */ ++ return; ++ ++ parent = bfqg_parent(bfqg); ++ ++ lockdep_assert_held(bfqg_to_blkg(bfqg)->q->queue_lock); ++ ++ if (unlikely(!parent)) ++ return; ++ ++ bfqg_stats_add_aux(&parent->stats, &bfqg->stats); ++ bfqg_stats_reset(&bfqg->stats); ++} ++ ++static void bfq_init_entity(struct bfq_entity *entity, ++ struct bfq_group *bfqg) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ ++ entity->weight = entity->new_weight; ++ entity->orig_weight = entity->new_weight; ++ if (bfqq) { ++ bfqq->ioprio = bfqq->new_ioprio; ++ bfqq->ioprio_class = bfqq->new_ioprio_class; ++ bfqg_get(bfqg); ++ } ++ entity->parent = bfqg->my_entity; /* NULL for root group */ ++ entity->sched_data = &bfqg->sched_data; ++} ++ ++static void bfqg_stats_exit(struct bfqg_stats *stats) ++{ ++ blkg_rwstat_exit(&stats->merged); ++ blkg_rwstat_exit(&stats->service_time); ++ blkg_rwstat_exit(&stats->wait_time); ++ blkg_rwstat_exit(&stats->queued); ++ blkg_stat_exit(&stats->time); ++ blkg_stat_exit(&stats->avg_queue_size_sum); ++ blkg_stat_exit(&stats->avg_queue_size_samples); ++ blkg_stat_exit(&stats->dequeue); ++ blkg_stat_exit(&stats->group_wait_time); ++ blkg_stat_exit(&stats->idle_time); ++ blkg_stat_exit(&stats->empty_time); ++} ++ ++static int bfqg_stats_init(struct bfqg_stats *stats, gfp_t gfp) ++{ ++ if (blkg_rwstat_init(&stats->merged, gfp) || ++ blkg_rwstat_init(&stats->service_time, gfp) || ++ blkg_rwstat_init(&stats->wait_time, gfp) || ++ blkg_rwstat_init(&stats->queued, gfp) || ++ blkg_stat_init(&stats->time, gfp) || ++ blkg_stat_init(&stats->avg_queue_size_sum, gfp) || ++ blkg_stat_init(&stats->avg_queue_size_samples, gfp) || ++ blkg_stat_init(&stats->dequeue, gfp) || ++ blkg_stat_init(&stats->group_wait_time, gfp) || ++ blkg_stat_init(&stats->idle_time, gfp) || ++ blkg_stat_init(&stats->empty_time, gfp)) { ++ bfqg_stats_exit(stats); ++ return -ENOMEM; ++ } ++ ++ return 0; ++} ++ ++static struct bfq_group_data *cpd_to_bfqgd(struct blkcg_policy_data *cpd) ++{ ++ return cpd ? container_of(cpd, struct bfq_group_data, pd) : NULL; ++} ++ ++static struct bfq_group_data *blkcg_to_bfqgd(struct blkcg *blkcg) ++{ ++ return cpd_to_bfqgd(blkcg_to_cpd(blkcg, &blkcg_policy_bfq)); ++} ++ ++static struct blkcg_policy_data *bfq_cpd_alloc(gfp_t gfp) ++{ ++ struct bfq_group_data *bgd; ++ ++ bgd = kzalloc(sizeof(*bgd), gfp); ++ if (!bgd) ++ return NULL; ++ return &bgd->pd; ++} ++ ++static void bfq_cpd_init(struct blkcg_policy_data *cpd) ++{ ++ struct bfq_group_data *d = cpd_to_bfqgd(cpd); ++ ++ d->weight = cgroup_subsys_on_dfl(io_cgrp_subsys) ? ++ CGROUP_WEIGHT_DFL : BFQ_WEIGHT_LEGACY_DFL; ++} ++ ++static void bfq_cpd_free(struct blkcg_policy_data *cpd) ++{ ++ kfree(cpd_to_bfqgd(cpd)); ++} ++ ++static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, int node) ++{ ++ struct bfq_group *bfqg; ++ ++ bfqg = kzalloc_node(sizeof(*bfqg), gfp, node); ++ if (!bfqg) ++ return NULL; ++ ++ if (bfqg_stats_init(&bfqg->stats, gfp)) { ++ kfree(bfqg); ++ return NULL; ++ } ++ ++ return &bfqg->pd; ++} ++ ++static void bfq_pd_init(struct blkg_policy_data *pd) ++{ ++ struct blkcg_gq *blkg; ++ struct bfq_group *bfqg; ++ struct bfq_data *bfqd; ++ struct bfq_entity *entity; ++ struct bfq_group_data *d; ++ ++ blkg = pd_to_blkg(pd); ++ BUG_ON(!blkg); ++ bfqg = blkg_to_bfqg(blkg); ++ bfqd = blkg->q->elevator->elevator_data; ++ entity = &bfqg->entity; ++ d = blkcg_to_bfqgd(blkg->blkcg); ++ ++ entity->orig_weight = entity->weight = entity->new_weight = d->weight; ++ entity->my_sched_data = &bfqg->sched_data; ++ bfqg->my_entity = entity; /* ++ * the root_group's will be set to NULL ++ * in bfq_init_queue() ++ */ ++ bfqg->bfqd = bfqd; ++ bfqg->active_entities = 0; ++ bfqg->rq_pos_tree = RB_ROOT; ++} ++ ++static void bfq_pd_free(struct blkg_policy_data *pd) ++{ ++ struct bfq_group *bfqg = pd_to_bfqg(pd); ++ ++ bfqg_stats_exit(&bfqg->stats); ++ return kfree(bfqg); ++} ++ ++static void bfq_pd_reset_stats(struct blkg_policy_data *pd) ++{ ++ struct bfq_group *bfqg = pd_to_bfqg(pd); ++ ++ bfqg_stats_reset(&bfqg->stats); ++} ++ ++static void bfq_group_set_parent(struct bfq_group *bfqg, ++ struct bfq_group *parent) ++{ ++ struct bfq_entity *entity; ++ ++ BUG_ON(!parent); ++ BUG_ON(!bfqg); ++ BUG_ON(bfqg == parent); ++ ++ entity = &bfqg->entity; ++ entity->parent = parent->my_entity; ++ entity->sched_data = &parent->sched_data; ++} ++ ++static struct bfq_group *bfq_lookup_bfqg(struct bfq_data *bfqd, ++ struct blkcg *blkcg) ++{ ++ struct blkcg_gq *blkg; ++ ++ blkg = blkg_lookup(blkcg, bfqd->queue); ++ if (likely(blkg)) ++ return blkg_to_bfqg(blkg); ++ return NULL; ++} ++ ++static struct bfq_group *bfq_find_set_group(struct bfq_data *bfqd, ++ struct blkcg *blkcg) ++{ ++ struct bfq_group *bfqg, *parent; ++ struct bfq_entity *entity; ++ ++ assert_spin_locked(bfqd->queue->queue_lock); ++ ++ bfqg = bfq_lookup_bfqg(bfqd, blkcg); ++ ++ if (unlikely(!bfqg)) ++ return NULL; ++ ++ /* ++ * Update chain of bfq_groups as we might be handling a leaf group ++ * which, along with some of its relatives, has not been hooked yet ++ * to the private hierarchy of BFQ. ++ */ ++ entity = &bfqg->entity; ++ for_each_entity(entity) { ++ bfqg = container_of(entity, struct bfq_group, entity); ++ BUG_ON(!bfqg); ++ if (bfqg != bfqd->root_group) { ++ parent = bfqg_parent(bfqg); ++ if (!parent) ++ parent = bfqd->root_group; ++ BUG_ON(!parent); ++ bfq_group_set_parent(bfqg, parent); ++ } ++ } ++ ++ return bfqg; ++} ++ ++static void bfq_pos_tree_add_move(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq); ++ ++static void bfq_bfqq_expire(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ bool compensate, ++ enum bfqq_expiration reason); ++ ++/** ++ * bfq_bfqq_move - migrate @bfqq to @bfqg. ++ * @bfqd: queue descriptor. ++ * @bfqq: the queue to move. ++ * @bfqg: the group to move to. ++ * ++ * Move @bfqq to @bfqg, deactivating it from its old group and reactivating ++ * it on the new one. Avoid putting the entity on the old group idle tree. ++ * ++ * Must be called under the queue lock; the cgroup owning @bfqg must ++ * not disappear (by now this just means that we are called under ++ * rcu_read_lock()). ++ */ ++static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ struct bfq_group *bfqg) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ ++ BUG_ON(!bfq_bfqq_busy(bfqq) && !RB_EMPTY_ROOT(&bfqq->sort_list)); ++ BUG_ON(!RB_EMPTY_ROOT(&bfqq->sort_list) && !entity->on_st); ++ BUG_ON(bfq_bfqq_busy(bfqq) && RB_EMPTY_ROOT(&bfqq->sort_list) ++ && entity->on_st && ++ bfqq != bfqd->in_service_queue); ++ BUG_ON(!bfq_bfqq_busy(bfqq) && bfqq == bfqd->in_service_queue); ++ ++ /* If bfqq is empty, then bfq_bfqq_expire also invokes ++ * bfq_del_bfqq_busy, thereby removing bfqq and its entity ++ * from data structures related to current group. Otherwise we ++ * need to remove bfqq explicitly with bfq_deactivate_bfqq, as ++ * we do below. ++ */ ++ if (bfqq == bfqd->in_service_queue) ++ bfq_bfqq_expire(bfqd, bfqd->in_service_queue, ++ false, BFQ_BFQQ_PREEMPTED); ++ ++ BUG_ON(entity->on_st && !bfq_bfqq_busy(bfqq) ++ && &bfq_entity_service_tree(entity)->idle != ++ entity->tree); ++ ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list) && bfq_bfqq_busy(bfqq)); ++ ++ if (bfq_bfqq_busy(bfqq)) ++ bfq_deactivate_bfqq(bfqd, bfqq, false, false); ++ else if (entity->on_st) { ++ BUG_ON(&bfq_entity_service_tree(entity)->idle != ++ entity->tree); ++ bfq_put_idle_entity(bfq_entity_service_tree(entity), entity); ++ } ++ bfqg_put(bfqq_group(bfqq)); ++ ++ /* ++ * Here we use a reference to bfqg. We don't need a refcounter ++ * as the cgroup reference will not be dropped, so that its ++ * destroy() callback will not be invoked. ++ */ ++ entity->parent = bfqg->my_entity; ++ entity->sched_data = &bfqg->sched_data; ++ bfqg_get(bfqg); ++ ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list) && bfq_bfqq_busy(bfqq)); ++ if (bfq_bfqq_busy(bfqq)) { ++ bfq_pos_tree_add_move(bfqd, bfqq); ++ bfq_activate_bfqq(bfqd, bfqq); ++ } ++ ++ if (!bfqd->in_service_queue && !bfqd->rq_in_driver) ++ bfq_schedule_dispatch(bfqd); ++ BUG_ON(entity->on_st && !bfq_bfqq_busy(bfqq) ++ && &bfq_entity_service_tree(entity)->idle != ++ entity->tree); ++} ++ ++/** ++ * __bfq_bic_change_cgroup - move @bic to @cgroup. ++ * @bfqd: the queue descriptor. ++ * @bic: the bic to move. ++ * @blkcg: the blk-cgroup to move to. ++ * ++ * Move bic to blkcg, assuming that bfqd->queue is locked; the caller ++ * has to make sure that the reference to cgroup is valid across the call. ++ * ++ * NOTE: an alternative approach might have been to store the current ++ * cgroup in bfqq and getting a reference to it, reducing the lookup ++ * time here, at the price of slightly more complex code. ++ */ ++static struct bfq_group *__bfq_bic_change_cgroup(struct bfq_data *bfqd, ++ struct bfq_io_cq *bic, ++ struct blkcg *blkcg) ++{ ++ struct bfq_queue *async_bfqq = bic_to_bfqq(bic, 0); ++ struct bfq_queue *sync_bfqq = bic_to_bfqq(bic, 1); ++ struct bfq_group *bfqg; ++ struct bfq_entity *entity; ++ ++ lockdep_assert_held(bfqd->queue->queue_lock); ++ ++ bfqg = bfq_find_set_group(bfqd, blkcg); ++ ++ if (unlikely(!bfqg)) ++ bfqg = bfqd->root_group; ++ ++ if (async_bfqq) { ++ entity = &async_bfqq->entity; ++ ++ if (entity->sched_data != &bfqg->sched_data) { ++ bic_set_bfqq(bic, NULL, 0); ++ bfq_log_bfqq(bfqd, async_bfqq, ++ "bic_change_group: %p %d", ++ async_bfqq, ++ async_bfqq->ref); ++ bfq_put_queue(async_bfqq); ++ } ++ } ++ ++ if (sync_bfqq) { ++ entity = &sync_bfqq->entity; ++ if (entity->sched_data != &bfqg->sched_data) ++ bfq_bfqq_move(bfqd, sync_bfqq, bfqg); ++ } ++ ++ return bfqg; ++} ++ ++static void bfq_bic_update_cgroup(struct bfq_io_cq *bic, struct bio *bio) ++{ ++ struct bfq_data *bfqd = bic_to_bfqd(bic); ++ struct bfq_group *bfqg = NULL; ++ uint64_t serial_nr; ++ ++ rcu_read_lock(); ++ serial_nr = bio_blkcg(bio)->css.serial_nr; ++ ++ /* ++ * Check whether blkcg has changed. The condition may trigger ++ * spuriously on a newly created cic but there's no harm. ++ */ ++ if (unlikely(!bfqd) || likely(bic->blkcg_serial_nr == serial_nr)) ++ goto out; ++ ++ bfqg = __bfq_bic_change_cgroup(bfqd, bic, bio_blkcg(bio)); ++ bic->blkcg_serial_nr = serial_nr; ++out: ++ rcu_read_unlock(); ++} ++ ++/** ++ * bfq_flush_idle_tree - deactivate any entity on the idle tree of @st. ++ * @st: the service tree being flushed. ++ */ ++static void bfq_flush_idle_tree(struct bfq_service_tree *st) ++{ ++ struct bfq_entity *entity = st->first_idle; ++ ++ for (; entity ; entity = st->first_idle) ++ __bfq_deactivate_entity(entity, false); ++} ++ ++/** ++ * bfq_reparent_leaf_entity - move leaf entity to the root_group. ++ * @bfqd: the device data structure with the root group. ++ * @entity: the entity to move. ++ */ ++static void bfq_reparent_leaf_entity(struct bfq_data *bfqd, ++ struct bfq_entity *entity) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ ++ BUG_ON(!bfqq); ++ bfq_bfqq_move(bfqd, bfqq, bfqd->root_group); ++} ++ ++/** ++ * bfq_reparent_active_entities - move to the root group all active ++ * entities. ++ * @bfqd: the device data structure with the root group. ++ * @bfqg: the group to move from. ++ * @st: the service tree with the entities. ++ * ++ * Needs queue_lock to be taken and reference to be valid over the call. ++ */ ++static void bfq_reparent_active_entities(struct bfq_data *bfqd, ++ struct bfq_group *bfqg, ++ struct bfq_service_tree *st) ++{ ++ struct rb_root *active = &st->active; ++ struct bfq_entity *entity = NULL; ++ ++ if (!RB_EMPTY_ROOT(&st->active)) ++ entity = bfq_entity_of(rb_first(active)); ++ ++ for (; entity ; entity = bfq_entity_of(rb_first(active))) ++ bfq_reparent_leaf_entity(bfqd, entity); ++ ++ if (bfqg->sched_data.in_service_entity) ++ bfq_reparent_leaf_entity(bfqd, ++ bfqg->sched_data.in_service_entity); ++} ++ ++/** ++ * bfq_pd_offline - deactivate the entity associated with @pd, ++ * and reparent its children entities. ++ * @pd: descriptor of the policy going offline. ++ * ++ * blkio already grabs the queue_lock for us, so no need to use ++ * RCU-based magic ++ */ ++static void bfq_pd_offline(struct blkg_policy_data *pd) ++{ ++ struct bfq_service_tree *st; ++ struct bfq_group *bfqg; ++ struct bfq_data *bfqd; ++ struct bfq_entity *entity; ++ int i; ++ ++ BUG_ON(!pd); ++ bfqg = pd_to_bfqg(pd); ++ BUG_ON(!bfqg); ++ bfqd = bfqg->bfqd; ++ BUG_ON(bfqd && !bfqd->root_group); ++ ++ entity = bfqg->my_entity; ++ ++ if (!entity) /* root group */ ++ return; ++ ++ /* ++ * Empty all service_trees belonging to this group before ++ * deactivating the group itself. ++ */ ++ for (i = 0; i < BFQ_IOPRIO_CLASSES; i++) { ++ BUG_ON(!bfqg->sched_data.service_tree); ++ st = bfqg->sched_data.service_tree + i; ++ /* ++ * The idle tree may still contain bfq_queues belonging ++ * to exited task because they never migrated to a different ++ * cgroup from the one being destroyed now. No one else ++ * can access them so it's safe to act without any lock. ++ */ ++ bfq_flush_idle_tree(st); ++ ++ /* ++ * It may happen that some queues are still active ++ * (busy) upon group destruction (if the corresponding ++ * processes have been forced to terminate). We move ++ * all the leaf entities corresponding to these queues ++ * to the root_group. ++ * Also, it may happen that the group has an entity ++ * in service, which is disconnected from the active ++ * tree: it must be moved, too. ++ * There is no need to put the sync queues, as the ++ * scheduler has taken no reference. ++ */ ++ bfq_reparent_active_entities(bfqd, bfqg, st); ++ BUG_ON(!RB_EMPTY_ROOT(&st->active)); ++ BUG_ON(!RB_EMPTY_ROOT(&st->idle)); ++ } ++ BUG_ON(bfqg->sched_data.next_in_service); ++ BUG_ON(bfqg->sched_data.in_service_entity); ++ ++ __bfq_deactivate_entity(entity, false); ++ bfq_put_async_queues(bfqd, bfqg); ++ ++ /* ++ * @blkg is going offline and will be ignored by ++ * blkg_[rw]stat_recursive_sum(). Transfer stats to the parent so ++ * that they don't get lost. If IOs complete after this point, the ++ * stats for them will be lost. Oh well... ++ */ ++ bfqg_stats_xfer_dead(bfqg); ++} ++ ++static void bfq_end_wr_async(struct bfq_data *bfqd) ++{ ++ struct blkcg_gq *blkg; ++ ++ list_for_each_entry(blkg, &bfqd->queue->blkg_list, q_node) { ++ struct bfq_group *bfqg = blkg_to_bfqg(blkg); ++ BUG_ON(!bfqg); ++ ++ bfq_end_wr_async_queues(bfqd, bfqg); ++ } ++ bfq_end_wr_async_queues(bfqd, bfqd->root_group); ++} ++ ++static int bfq_io_show_weight(struct seq_file *sf, void *v) ++{ ++ struct blkcg *blkcg = css_to_blkcg(seq_css(sf)); ++ struct bfq_group_data *bfqgd = blkcg_to_bfqgd(blkcg); ++ unsigned int val = 0; ++ ++ if (bfqgd) ++ val = bfqgd->weight; ++ ++ seq_printf(sf, "%u\n", val); ++ ++ return 0; ++} ++ ++static int bfq_io_set_weight_legacy(struct cgroup_subsys_state *css, ++ struct cftype *cftype, ++ u64 val) ++{ ++ struct blkcg *blkcg = css_to_blkcg(css); ++ struct bfq_group_data *bfqgd = blkcg_to_bfqgd(blkcg); ++ struct blkcg_gq *blkg; ++ int ret = -ERANGE; ++ ++ if (val < BFQ_MIN_WEIGHT || val > BFQ_MAX_WEIGHT) ++ return ret; ++ ++ ret = 0; ++ spin_lock_irq(&blkcg->lock); ++ bfqgd->weight = (unsigned short)val; ++ hlist_for_each_entry(blkg, &blkcg->blkg_list, blkcg_node) { ++ struct bfq_group *bfqg = blkg_to_bfqg(blkg); ++ ++ if (!bfqg) ++ continue; ++ /* ++ * Setting the prio_changed flag of the entity ++ * to 1 with new_weight == weight would re-set ++ * the value of the weight to its ioprio mapping. ++ * Set the flag only if necessary. ++ */ ++ if ((unsigned short)val != bfqg->entity.new_weight) { ++ bfqg->entity.new_weight = (unsigned short)val; ++ /* ++ * Make sure that the above new value has been ++ * stored in bfqg->entity.new_weight before ++ * setting the prio_changed flag. In fact, ++ * this flag may be read asynchronously (in ++ * critical sections protected by a different ++ * lock than that held here), and finding this ++ * flag set may cause the execution of the code ++ * for updating parameters whose value may ++ * depend also on bfqg->entity.new_weight (in ++ * __bfq_entity_update_weight_prio). ++ * This barrier makes sure that the new value ++ * of bfqg->entity.new_weight is correctly ++ * seen in that code. ++ */ ++ smp_wmb(); ++ bfqg->entity.prio_changed = 1; ++ } ++ } ++ spin_unlock_irq(&blkcg->lock); ++ ++ return ret; ++} ++ ++static ssize_t bfq_io_set_weight(struct kernfs_open_file *of, ++ char *buf, size_t nbytes, ++ loff_t off) ++{ ++ u64 weight; ++ /* First unsigned long found in the file is used */ ++ int ret = kstrtoull(strim(buf), 0, &weight); ++ ++ if (ret) ++ return ret; ++ ++ return bfq_io_set_weight_legacy(of_css(of), NULL, weight); ++} ++ ++static int bfqg_print_stat(struct seq_file *sf, void *v) ++{ ++ blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), blkg_prfill_stat, ++ &blkcg_policy_bfq, seq_cft(sf)->private, false); ++ return 0; ++} ++ ++static int bfqg_print_rwstat(struct seq_file *sf, void *v) ++{ ++ blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), blkg_prfill_rwstat, ++ &blkcg_policy_bfq, seq_cft(sf)->private, true); ++ return 0; ++} ++ ++static u64 bfqg_prfill_stat_recursive(struct seq_file *sf, ++ struct blkg_policy_data *pd, int off) ++{ ++ u64 sum = blkg_stat_recursive_sum(pd_to_blkg(pd), ++ &blkcg_policy_bfq, off); ++ return __blkg_prfill_u64(sf, pd, sum); ++} ++ ++static u64 bfqg_prfill_rwstat_recursive(struct seq_file *sf, ++ struct blkg_policy_data *pd, int off) ++{ ++ struct blkg_rwstat sum = blkg_rwstat_recursive_sum(pd_to_blkg(pd), ++ &blkcg_policy_bfq, ++ off); ++ return __blkg_prfill_rwstat(sf, pd, &sum); ++} ++ ++static int bfqg_print_stat_recursive(struct seq_file *sf, void *v) ++{ ++ blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), ++ bfqg_prfill_stat_recursive, &blkcg_policy_bfq, ++ seq_cft(sf)->private, false); ++ return 0; ++} ++ ++static int bfqg_print_rwstat_recursive(struct seq_file *sf, void *v) ++{ ++ blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), ++ bfqg_prfill_rwstat_recursive, &blkcg_policy_bfq, ++ seq_cft(sf)->private, true); ++ return 0; ++} ++ ++static u64 bfqg_prfill_sectors(struct seq_file *sf, struct blkg_policy_data *pd, ++ int off) ++{ ++ u64 sum = blkg_rwstat_total(&pd->blkg->stat_bytes); ++ ++ return __blkg_prfill_u64(sf, pd, sum >> 9); ++} ++ ++static int bfqg_print_stat_sectors(struct seq_file *sf, void *v) ++{ ++ blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), ++ bfqg_prfill_sectors, &blkcg_policy_bfq, 0, false); ++ return 0; ++} ++ ++static u64 bfqg_prfill_sectors_recursive(struct seq_file *sf, ++ struct blkg_policy_data *pd, int off) ++{ ++ struct blkg_rwstat tmp = blkg_rwstat_recursive_sum(pd->blkg, NULL, ++ offsetof(struct blkcg_gq, stat_bytes)); ++ u64 sum = atomic64_read(&tmp.aux_cnt[BLKG_RWSTAT_READ]) + ++ atomic64_read(&tmp.aux_cnt[BLKG_RWSTAT_WRITE]); ++ ++ return __blkg_prfill_u64(sf, pd, sum >> 9); ++} ++ ++static int bfqg_print_stat_sectors_recursive(struct seq_file *sf, void *v) ++{ ++ blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), ++ bfqg_prfill_sectors_recursive, &blkcg_policy_bfq, 0, ++ false); ++ return 0; ++} ++ ++ ++static u64 bfqg_prfill_avg_queue_size(struct seq_file *sf, ++ struct blkg_policy_data *pd, int off) ++{ ++ struct bfq_group *bfqg = pd_to_bfqg(pd); ++ u64 samples = blkg_stat_read(&bfqg->stats.avg_queue_size_samples); ++ u64 v = 0; ++ ++ if (samples) { ++ v = blkg_stat_read(&bfqg->stats.avg_queue_size_sum); ++ v = div64_u64(v, samples); ++ } ++ __blkg_prfill_u64(sf, pd, v); ++ return 0; ++} ++ ++/* print avg_queue_size */ ++static int bfqg_print_avg_queue_size(struct seq_file *sf, void *v) ++{ ++ blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), ++ bfqg_prfill_avg_queue_size, &blkcg_policy_bfq, ++ 0, false); ++ return 0; ++} ++ ++static struct bfq_group * ++bfq_create_group_hierarchy(struct bfq_data *bfqd, int node) ++{ ++ int ret; ++ ++ ret = blkcg_activate_policy(bfqd->queue, &blkcg_policy_bfq); ++ if (ret) ++ return NULL; ++ ++ return blkg_to_bfqg(bfqd->queue->root_blkg); ++} ++ ++static struct cftype bfq_blkcg_legacy_files[] = { ++ { ++ .name = "bfq.weight", ++ .flags = CFTYPE_NOT_ON_ROOT, ++ .seq_show = bfq_io_show_weight, ++ .write_u64 = bfq_io_set_weight_legacy, ++ }, ++ ++ /* statistics, covers only the tasks in the bfqg */ ++ { ++ .name = "bfq.time", ++ .private = offsetof(struct bfq_group, stats.time), ++ .seq_show = bfqg_print_stat, ++ }, ++ { ++ .name = "bfq.sectors", ++ .seq_show = bfqg_print_stat_sectors, ++ }, ++ { ++ .name = "bfq.io_service_bytes", ++ .private = (unsigned long)&blkcg_policy_bfq, ++ .seq_show = blkg_print_stat_bytes, ++ }, ++ { ++ .name = "bfq.io_serviced", ++ .private = (unsigned long)&blkcg_policy_bfq, ++ .seq_show = blkg_print_stat_ios, ++ }, ++ { ++ .name = "bfq.io_service_time", ++ .private = offsetof(struct bfq_group, stats.service_time), ++ .seq_show = bfqg_print_rwstat, ++ }, ++ { ++ .name = "bfq.io_wait_time", ++ .private = offsetof(struct bfq_group, stats.wait_time), ++ .seq_show = bfqg_print_rwstat, ++ }, ++ { ++ .name = "bfq.io_merged", ++ .private = offsetof(struct bfq_group, stats.merged), ++ .seq_show = bfqg_print_rwstat, ++ }, ++ { ++ .name = "bfq.io_queued", ++ .private = offsetof(struct bfq_group, stats.queued), ++ .seq_show = bfqg_print_rwstat, ++ }, ++ ++ /* the same statictics which cover the bfqg and its descendants */ ++ { ++ .name = "bfq.time_recursive", ++ .private = offsetof(struct bfq_group, stats.time), ++ .seq_show = bfqg_print_stat_recursive, ++ }, ++ { ++ .name = "bfq.sectors_recursive", ++ .seq_show = bfqg_print_stat_sectors_recursive, ++ }, ++ { ++ .name = "bfq.io_service_bytes_recursive", ++ .private = (unsigned long)&blkcg_policy_bfq, ++ .seq_show = blkg_print_stat_bytes_recursive, ++ }, ++ { ++ .name = "bfq.io_serviced_recursive", ++ .private = (unsigned long)&blkcg_policy_bfq, ++ .seq_show = blkg_print_stat_ios_recursive, ++ }, ++ { ++ .name = "bfq.io_service_time_recursive", ++ .private = offsetof(struct bfq_group, stats.service_time), ++ .seq_show = bfqg_print_rwstat_recursive, ++ }, ++ { ++ .name = "bfq.io_wait_time_recursive", ++ .private = offsetof(struct bfq_group, stats.wait_time), ++ .seq_show = bfqg_print_rwstat_recursive, ++ }, ++ { ++ .name = "bfq.io_merged_recursive", ++ .private = offsetof(struct bfq_group, stats.merged), ++ .seq_show = bfqg_print_rwstat_recursive, ++ }, ++ { ++ .name = "bfq.io_queued_recursive", ++ .private = offsetof(struct bfq_group, stats.queued), ++ .seq_show = bfqg_print_rwstat_recursive, ++ }, ++ { ++ .name = "bfq.avg_queue_size", ++ .seq_show = bfqg_print_avg_queue_size, ++ }, ++ { ++ .name = "bfq.group_wait_time", ++ .private = offsetof(struct bfq_group, stats.group_wait_time), ++ .seq_show = bfqg_print_stat, ++ }, ++ { ++ .name = "bfq.idle_time", ++ .private = offsetof(struct bfq_group, stats.idle_time), ++ .seq_show = bfqg_print_stat, ++ }, ++ { ++ .name = "bfq.empty_time", ++ .private = offsetof(struct bfq_group, stats.empty_time), ++ .seq_show = bfqg_print_stat, ++ }, ++ { ++ .name = "bfq.dequeue", ++ .private = offsetof(struct bfq_group, stats.dequeue), ++ .seq_show = bfqg_print_stat, ++ }, ++ { } /* terminate */ ++}; ++ ++static struct cftype bfq_blkg_files[] = { ++ { ++ .name = "bfq.weight", ++ .flags = CFTYPE_NOT_ON_ROOT, ++ .seq_show = bfq_io_show_weight, ++ .write = bfq_io_set_weight, ++ }, ++ {} /* terminate */ ++}; ++ ++#else /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++ ++static inline void bfqg_stats_update_io_add(struct bfq_group *bfqg, ++ struct bfq_queue *bfqq, unsigned int op) { } ++static inline void ++bfqg_stats_update_io_remove(struct bfq_group *bfqg, unsigned int op) { } ++static inline void ++bfqg_stats_update_io_merged(struct bfq_group *bfqg, unsigned int op) { } ++static inline void bfqg_stats_update_completion(struct bfq_group *bfqg, ++ uint64_t start_time, uint64_t io_start_time, ++ unsigned int op) { } ++static inline void ++bfqg_stats_set_start_group_wait_time(struct bfq_group *bfqg, ++ struct bfq_group *curr_bfqg) { } ++static inline void bfqg_stats_end_empty_time(struct bfqg_stats *stats) { } ++static inline void bfqg_stats_update_dequeue(struct bfq_group *bfqg) { } ++static inline void bfqg_stats_set_start_empty_time(struct bfq_group *bfqg) { } ++static inline void bfqg_stats_update_idle_time(struct bfq_group *bfqg) { } ++static inline void bfqg_stats_set_start_idle_time(struct bfq_group *bfqg) { } ++static inline void bfqg_stats_update_avg_queue_size(struct bfq_group *bfqg) { } ++ ++static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ struct bfq_group *bfqg) {} ++ ++static void bfq_init_entity(struct bfq_entity *entity, ++ struct bfq_group *bfqg) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ ++ entity->weight = entity->new_weight; ++ entity->orig_weight = entity->new_weight; ++ if (bfqq) { ++ bfqq->ioprio = bfqq->new_ioprio; ++ bfqq->ioprio_class = bfqq->new_ioprio_class; ++ } ++ entity->sched_data = &bfqg->sched_data; ++} ++ ++static void bfq_bic_update_cgroup(struct bfq_io_cq *bic, struct bio *bio) {} ++ ++static void bfq_end_wr_async(struct bfq_data *bfqd) ++{ ++ bfq_end_wr_async_queues(bfqd, bfqd->root_group); ++} ++ ++static struct bfq_group *bfq_find_set_group(struct bfq_data *bfqd, ++ struct blkcg *blkcg) ++{ ++ return bfqd->root_group; ++} ++ ++static struct bfq_group *bfqq_group(struct bfq_queue *bfqq) ++{ ++ return bfqq->bfqd->root_group; ++} ++ ++static struct bfq_group * ++bfq_create_group_hierarchy(struct bfq_data *bfqd, int node) ++{ ++ struct bfq_group *bfqg; ++ int i; ++ ++ bfqg = kmalloc_node(sizeof(*bfqg), GFP_KERNEL | __GFP_ZERO, node); ++ if (!bfqg) ++ return NULL; ++ ++ for (i = 0; i < BFQ_IOPRIO_CLASSES; i++) ++ bfqg->sched_data.service_tree[i] = BFQ_SERVICE_TREE_INIT; ++ ++ return bfqg; ++} ++#endif +diff --git a/block/bfq-ioc.c b/block/bfq-ioc.c +new file mode 100644 +index 000000000000..fb7bb8f08b75 +--- /dev/null ++++ b/block/bfq-ioc.c +@@ -0,0 +1,36 @@ ++/* ++ * BFQ: I/O context handling. ++ * ++ * Based on ideas and code from CFQ: ++ * Copyright (C) 2003 Jens Axboe ++ * ++ * Copyright (C) 2008 Fabio Checconi ++ * Paolo Valente ++ * ++ * Copyright (C) 2010 Paolo Valente ++ */ ++ ++/** ++ * icq_to_bic - convert iocontext queue structure to bfq_io_cq. ++ * @icq: the iocontext queue. ++ */ ++static struct bfq_io_cq *icq_to_bic(struct io_cq *icq) ++{ ++ /* bic->icq is the first member, %NULL will convert to %NULL */ ++ return container_of(icq, struct bfq_io_cq, icq); ++} ++ ++/** ++ * bfq_bic_lookup - search into @ioc a bic associated to @bfqd. ++ * @bfqd: the lookup key. ++ * @ioc: the io_context of the process doing I/O. ++ * ++ * Queue lock must be held. ++ */ ++static struct bfq_io_cq *bfq_bic_lookup(struct bfq_data *bfqd, ++ struct io_context *ioc) ++{ ++ if (ioc) ++ return icq_to_bic(ioc_lookup_icq(ioc, bfqd->queue)); ++ return NULL; ++} +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +new file mode 100644 +index 000000000000..ac8991bca9fa +--- /dev/null ++++ b/block/bfq-sched.c +@@ -0,0 +1,2002 @@ ++/* ++ * BFQ: Hierarchical B-WF2Q+ scheduler. ++ * ++ * Based on ideas and code from CFQ: ++ * Copyright (C) 2003 Jens Axboe ++ * ++ * Copyright (C) 2008 Fabio Checconi ++ * Paolo Valente ++ * ++ * Copyright (C) 2015 Paolo Valente ++ * ++ * Copyright (C) 2016 Paolo Valente ++ */ ++ ++static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); ++ ++/** ++ * bfq_gt - compare two timestamps. ++ * @a: first ts. ++ * @b: second ts. ++ * ++ * Return @a > @b, dealing with wrapping correctly. ++ */ ++static int bfq_gt(u64 a, u64 b) ++{ ++ return (s64)(a - b) > 0; ++} ++ ++static struct bfq_entity *bfq_root_active_entity(struct rb_root *tree) ++{ ++ struct rb_node *node = tree->rb_node; ++ ++ return rb_entry(node, struct bfq_entity, rb_node); ++} ++ ++static struct bfq_entity *bfq_lookup_next_entity(struct bfq_sched_data *sd); ++ ++static bool bfq_update_parent_budget(struct bfq_entity *next_in_service); ++ ++/** ++ * bfq_update_next_in_service - update sd->next_in_service ++ * @sd: sched_data for which to perform the update. ++ * @new_entity: if not NULL, pointer to the entity whose activation, ++ * requeueing or repositionig triggered the invocation of ++ * this function. ++ * ++ * This function is called to update sd->next_in_service, which, in ++ * its turn, may change as a consequence of the insertion or ++ * extraction of an entity into/from one of the active trees of ++ * sd. These insertions/extractions occur as a consequence of ++ * activations/deactivations of entities, with some activations being ++ * 'true' activations, and other activations being requeueings (i.e., ++ * implementing the second, requeueing phase of the mechanism used to ++ * reposition an entity in its active tree; see comments on ++ * __bfq_activate_entity and __bfq_requeue_entity for details). In ++ * both the last two activation sub-cases, new_entity points to the ++ * just activated or requeued entity. ++ * ++ * Returns true if sd->next_in_service changes in such a way that ++ * entity->parent may become the next_in_service for its parent ++ * entity. ++ */ ++static bool bfq_update_next_in_service(struct bfq_sched_data *sd, ++ struct bfq_entity *new_entity) ++{ ++ struct bfq_entity *next_in_service = sd->next_in_service; ++ struct bfq_queue *bfqq; ++ bool parent_sched_may_change = false; ++ ++ /* ++ * If this update is triggered by the activation, requeueing ++ * or repositiong of an entity that does not coincide with ++ * sd->next_in_service, then a full lookup in the active tree ++ * can be avoided. In fact, it is enough to check whether the ++ * just-modified entity has a higher priority than ++ * sd->next_in_service, or, even if it has the same priority ++ * as sd->next_in_service, is eligible and has a lower virtual ++ * finish time than sd->next_in_service. If this compound ++ * condition holds, then the new entity becomes the new ++ * next_in_service. Otherwise no change is needed. ++ */ ++ if (new_entity && new_entity != sd->next_in_service) { ++ /* ++ * Flag used to decide whether to replace ++ * sd->next_in_service with new_entity. Tentatively ++ * set to true, and left as true if ++ * sd->next_in_service is NULL. ++ */ ++ bool replace_next = true; ++ ++ /* ++ * If there is already a next_in_service candidate ++ * entity, then compare class priorities or timestamps ++ * to decide whether to replace sd->service_tree with ++ * new_entity. ++ */ ++ if (next_in_service) { ++ unsigned int new_entity_class_idx = ++ bfq_class_idx(new_entity); ++ struct bfq_service_tree *st = ++ sd->service_tree + new_entity_class_idx; ++ ++ /* ++ * For efficiency, evaluate the most likely ++ * sub-condition first. ++ */ ++ replace_next = ++ (new_entity_class_idx == ++ bfq_class_idx(next_in_service) ++ && ++ !bfq_gt(new_entity->start, st->vtime) ++ && ++ bfq_gt(next_in_service->finish, ++ new_entity->finish)) ++ || ++ new_entity_class_idx < ++ bfq_class_idx(next_in_service); ++ } ++ ++ if (replace_next) ++ next_in_service = new_entity; ++ } else /* invoked because of a deactivation: lookup needed */ ++ next_in_service = bfq_lookup_next_entity(sd); ++ ++ if (next_in_service) { ++ parent_sched_may_change = !sd->next_in_service || ++ bfq_update_parent_budget(next_in_service); ++ } ++ ++ sd->next_in_service = next_in_service; ++ ++ if (!next_in_service) ++ return parent_sched_may_change; ++ ++ bfqq = bfq_entity_to_bfqq(next_in_service); ++ if (bfqq) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "update_next_in_service: chosen this queue"); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { ++ struct bfq_group *bfqg = ++ container_of(next_in_service, ++ struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "update_next_in_service: chosen this entity"); ++ } ++#endif ++ return parent_sched_may_change; ++} ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++/* both next loops stop at one of the child entities of the root group */ ++#define for_each_entity(entity) \ ++ for (; entity ; entity = entity->parent) ++ ++/* ++ * For each iteration, compute parent in advance, so as to be safe if ++ * entity is deallocated during the iteration. Such a deallocation may ++ * happen as a consequence of a bfq_put_queue that frees the bfq_queue ++ * containing entity. ++ */ ++#define for_each_entity_safe(entity, parent) \ ++ for (; entity && ({ parent = entity->parent; 1; }); entity = parent) ++ ++/* ++ * Returns true if this budget changes may let next_in_service->parent ++ * become the next_in_service entity for its parent entity. ++ */ ++static bool bfq_update_parent_budget(struct bfq_entity *next_in_service) ++{ ++ struct bfq_entity *bfqg_entity; ++ struct bfq_group *bfqg; ++ struct bfq_sched_data *group_sd; ++ bool ret = false; ++ ++ BUG_ON(!next_in_service); ++ ++ group_sd = next_in_service->sched_data; ++ ++ bfqg = container_of(group_sd, struct bfq_group, sched_data); ++ /* ++ * bfq_group's my_entity field is not NULL only if the group ++ * is not the root group. We must not touch the root entity ++ * as it must never become an in-service entity. ++ */ ++ bfqg_entity = bfqg->my_entity; ++ if (bfqg_entity) { ++ if (bfqg_entity->budget > next_in_service->budget) ++ ret = true; ++ bfqg_entity->budget = next_in_service->budget; ++ } ++ ++ return ret; ++} ++ ++/* ++ * This function tells whether entity stops being a candidate for next ++ * service, according to the following logic. ++ * ++ * This function is invoked for an entity that is about to be set in ++ * service. If such an entity is a queue, then the entity is no longer ++ * a candidate for next service (i.e, a candidate entity to serve ++ * after the in-service entity is expired). The function then returns ++ * true. ++ * ++ * In contrast, the entity could stil be a candidate for next service ++ * if it is not a queue, and has more than one child. In fact, even if ++ * one of its children is about to be set in service, other children ++ * may still be the next to serve. As a consequence, a non-queue ++ * entity is not a candidate for next-service only if it has only one ++ * child. And only if this condition holds, then the function returns ++ * true for a non-queue entity. ++ */ ++static bool bfq_no_longer_next_in_service(struct bfq_entity *entity) ++{ ++ struct bfq_group *bfqg; ++ ++ if (bfq_entity_to_bfqq(entity)) ++ return true; ++ ++ bfqg = container_of(entity, struct bfq_group, entity); ++ ++ BUG_ON(bfqg == ((struct bfq_data *)(bfqg->bfqd))->root_group); ++ BUG_ON(bfqg->active_entities == 0); ++ if (bfqg->active_entities == 1) ++ return true; ++ ++ return false; ++} ++ ++#else /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++#define for_each_entity(entity) \ ++ for (; entity ; entity = NULL) ++ ++#define for_each_entity_safe(entity, parent) \ ++ for (parent = NULL; entity ; entity = parent) ++ ++static bool bfq_update_parent_budget(struct bfq_entity *next_in_service) ++{ ++ return false; ++} ++ ++static bool bfq_no_longer_next_in_service(struct bfq_entity *entity) ++{ ++ return true; ++} ++ ++#endif /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++ ++/* ++ * Shift for timestamp calculations. This actually limits the maximum ++ * service allowed in one timestamp delta (small shift values increase it), ++ * the maximum total weight that can be used for the queues in the system ++ * (big shift values increase it), and the period of virtual time ++ * wraparounds. ++ */ ++#define WFQ_SERVICE_SHIFT 22 ++ ++static struct bfq_queue *bfq_entity_to_bfqq(struct bfq_entity *entity) ++{ ++ struct bfq_queue *bfqq = NULL; ++ ++ BUG_ON(!entity); ++ ++ if (!entity->my_sched_data) ++ bfqq = container_of(entity, struct bfq_queue, entity); ++ ++ return bfqq; ++} ++ ++ ++/** ++ * bfq_delta - map service into the virtual time domain. ++ * @service: amount of service. ++ * @weight: scale factor (weight of an entity or weight sum). ++ */ ++static u64 bfq_delta(unsigned long service, unsigned long weight) ++{ ++ u64 d = (u64)service << WFQ_SERVICE_SHIFT; ++ ++ do_div(d, weight); ++ return d; ++} ++ ++/** ++ * bfq_calc_finish - assign the finish time to an entity. ++ * @entity: the entity to act upon. ++ * @service: the service to be charged to the entity. ++ */ ++static void bfq_calc_finish(struct bfq_entity *entity, unsigned long service) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ unsigned long long start, finish, delta; ++ ++ BUG_ON(entity->weight == 0); ++ ++ entity->finish = entity->start + ++ bfq_delta(service, entity->weight); ++ ++ start = ((entity->start>>10)*1000)>>12; ++ finish = ((entity->finish>>10)*1000)>>12; ++ delta = ((bfq_delta(service, entity->weight)>>10)*1000)>>12; ++ ++ if (bfqq) { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "calc_finish: serv %lu, w %d", ++ service, entity->weight); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "calc_finish: start %llu, finish %llu, delta %llu", ++ start, finish, delta); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ } else { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "calc_finish group: serv %lu, w %d", ++ service, entity->weight); ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "calc_finish group: start %llu, finish %llu, delta %llu", ++ start, finish, delta); ++#endif ++ } ++} ++ ++/** ++ * bfq_entity_of - get an entity from a node. ++ * @node: the node field of the entity. ++ * ++ * Convert a node pointer to the relative entity. This is used only ++ * to simplify the logic of some functions and not as the generic ++ * conversion mechanism because, e.g., in the tree walking functions, ++ * the check for a %NULL value would be redundant. ++ */ ++static struct bfq_entity *bfq_entity_of(struct rb_node *node) ++{ ++ struct bfq_entity *entity = NULL; ++ ++ if (node) ++ entity = rb_entry(node, struct bfq_entity, rb_node); ++ ++ return entity; ++} ++ ++/** ++ * bfq_extract - remove an entity from a tree. ++ * @root: the tree root. ++ * @entity: the entity to remove. ++ */ ++static void bfq_extract(struct rb_root *root, struct bfq_entity *entity) ++{ ++ BUG_ON(entity->tree != root); ++ ++ entity->tree = NULL; ++ rb_erase(&entity->rb_node, root); ++} ++ ++/** ++ * bfq_idle_extract - extract an entity from the idle tree. ++ * @st: the service tree of the owning @entity. ++ * @entity: the entity being removed. ++ */ ++static void bfq_idle_extract(struct bfq_service_tree *st, ++ struct bfq_entity *entity) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ struct rb_node *next; ++ ++ BUG_ON(entity->tree != &st->idle); ++ ++ if (entity == st->first_idle) { ++ next = rb_next(&entity->rb_node); ++ st->first_idle = bfq_entity_of(next); ++ } ++ ++ if (entity == st->last_idle) { ++ next = rb_prev(&entity->rb_node); ++ st->last_idle = bfq_entity_of(next); ++ } ++ ++ bfq_extract(&st->idle, entity); ++ ++ if (bfqq) ++ list_del(&bfqq->bfqq_list); ++} ++ ++/** ++ * bfq_insert - generic tree insertion. ++ * @root: tree root. ++ * @entity: entity to insert. ++ * ++ * This is used for the idle and the active tree, since they are both ++ * ordered by finish time. ++ */ ++static void bfq_insert(struct rb_root *root, struct bfq_entity *entity) ++{ ++ struct bfq_entity *entry; ++ struct rb_node **node = &root->rb_node; ++ struct rb_node *parent = NULL; ++ ++ BUG_ON(entity->tree); ++ ++ while (*node) { ++ parent = *node; ++ entry = rb_entry(parent, struct bfq_entity, rb_node); ++ ++ if (bfq_gt(entry->finish, entity->finish)) ++ node = &parent->rb_left; ++ else ++ node = &parent->rb_right; ++ } ++ ++ rb_link_node(&entity->rb_node, parent, node); ++ rb_insert_color(&entity->rb_node, root); ++ ++ entity->tree = root; ++} ++ ++/** ++ * bfq_update_min - update the min_start field of a entity. ++ * @entity: the entity to update. ++ * @node: one of its children. ++ * ++ * This function is called when @entity may store an invalid value for ++ * min_start due to updates to the active tree. The function assumes ++ * that the subtree rooted at @node (which may be its left or its right ++ * child) has a valid min_start value. ++ */ ++static void bfq_update_min(struct bfq_entity *entity, struct rb_node *node) ++{ ++ struct bfq_entity *child; ++ ++ if (node) { ++ child = rb_entry(node, struct bfq_entity, rb_node); ++ if (bfq_gt(entity->min_start, child->min_start)) ++ entity->min_start = child->min_start; ++ } ++} ++ ++/** ++ * bfq_update_active_node - recalculate min_start. ++ * @node: the node to update. ++ * ++ * @node may have changed position or one of its children may have moved, ++ * this function updates its min_start value. The left and right subtrees ++ * are assumed to hold a correct min_start value. ++ */ ++static void bfq_update_active_node(struct rb_node *node) ++{ ++ struct bfq_entity *entity = rb_entry(node, struct bfq_entity, rb_node); ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ ++ entity->min_start = entity->start; ++ bfq_update_min(entity, node->rb_right); ++ bfq_update_min(entity, node->rb_left); ++ ++ if (bfqq) { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "update_active_node: new min_start %llu", ++ ((entity->min_start>>10)*1000)>>12); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ } else { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "update_active_node: new min_start %llu", ++ ((entity->min_start>>10)*1000)>>12); ++#endif ++ } ++} ++ ++/** ++ * bfq_update_active_tree - update min_start for the whole active tree. ++ * @node: the starting node. ++ * ++ * @node must be the deepest modified node after an update. This function ++ * updates its min_start using the values held by its children, assuming ++ * that they did not change, and then updates all the nodes that may have ++ * changed in the path to the root. The only nodes that may have changed ++ * are the ones in the path or their siblings. ++ */ ++static void bfq_update_active_tree(struct rb_node *node) ++{ ++ struct rb_node *parent; ++ ++up: ++ bfq_update_active_node(node); ++ ++ parent = rb_parent(node); ++ if (!parent) ++ return; ++ ++ if (node == parent->rb_left && parent->rb_right) ++ bfq_update_active_node(parent->rb_right); ++ else if (parent->rb_left) ++ bfq_update_active_node(parent->rb_left); ++ ++ node = parent; ++ goto up; ++} ++ ++static void bfq_weights_tree_add(struct bfq_data *bfqd, ++ struct bfq_entity *entity, ++ struct rb_root *root); ++ ++static void bfq_weights_tree_remove(struct bfq_data *bfqd, ++ struct bfq_entity *entity, ++ struct rb_root *root); ++ ++ ++/** ++ * bfq_active_insert - insert an entity in the active tree of its ++ * group/device. ++ * @st: the service tree of the entity. ++ * @entity: the entity being inserted. ++ * ++ * The active tree is ordered by finish time, but an extra key is kept ++ * per each node, containing the minimum value for the start times of ++ * its children (and the node itself), so it's possible to search for ++ * the eligible node with the lowest finish time in logarithmic time. ++ */ ++static void bfq_active_insert(struct bfq_service_tree *st, ++ struct bfq_entity *entity) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ struct rb_node *node = &entity->rb_node; ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ struct bfq_sched_data *sd = NULL; ++ struct bfq_group *bfqg = NULL; ++ struct bfq_data *bfqd = NULL; ++#endif ++ ++ bfq_insert(&st->active, entity); ++ ++ if (node->rb_left) ++ node = node->rb_left; ++ else if (node->rb_right) ++ node = node->rb_right; ++ ++ bfq_update_active_tree(node); ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ sd = entity->sched_data; ++ bfqg = container_of(sd, struct bfq_group, sched_data); ++ BUG_ON(!bfqg); ++ bfqd = (struct bfq_data *)bfqg->bfqd; ++#endif ++ if (bfqq) ++ list_add(&bfqq->bfqq_list, &bfqq->bfqd->active_list); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { /* bfq_group */ ++ BUG_ON(!bfqd); ++ bfq_weights_tree_add(bfqd, entity, &bfqd->group_weights_tree); ++ } ++ if (bfqg != bfqd->root_group) { ++ BUG_ON(!bfqg); ++ BUG_ON(!bfqd); ++ bfqg->active_entities++; ++ } ++#endif ++} ++ ++/** ++ * bfq_ioprio_to_weight - calc a weight from an ioprio. ++ * @ioprio: the ioprio value to convert. ++ */ ++static unsigned short bfq_ioprio_to_weight(int ioprio) ++{ ++ BUG_ON(ioprio < 0 || ioprio >= IOPRIO_BE_NR); ++ return (IOPRIO_BE_NR - ioprio) * BFQ_WEIGHT_CONVERSION_COEFF; ++} ++ ++/** ++ * bfq_weight_to_ioprio - calc an ioprio from a weight. ++ * @weight: the weight value to convert. ++ * ++ * To preserve as much as possible the old only-ioprio user interface, ++ * 0 is used as an escape ioprio value for weights (numerically) equal or ++ * larger than IOPRIO_BE_NR * BFQ_WEIGHT_CONVERSION_COEFF. ++ */ ++static unsigned short bfq_weight_to_ioprio(int weight) ++{ ++ BUG_ON(weight < BFQ_MIN_WEIGHT || weight > BFQ_MAX_WEIGHT); ++ return IOPRIO_BE_NR * BFQ_WEIGHT_CONVERSION_COEFF - weight < 0 ? ++ 0 : IOPRIO_BE_NR * BFQ_WEIGHT_CONVERSION_COEFF - weight; ++} ++ ++static void bfq_get_entity(struct bfq_entity *entity) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ ++ if (bfqq) { ++ bfqq->ref++; ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "get_entity: %p %d", ++ bfqq, bfqq->ref); ++ } ++} ++ ++/** ++ * bfq_find_deepest - find the deepest node that an extraction can modify. ++ * @node: the node being removed. ++ * ++ * Do the first step of an extraction in an rb tree, looking for the ++ * node that will replace @node, and returning the deepest node that ++ * the following modifications to the tree can touch. If @node is the ++ * last node in the tree return %NULL. ++ */ ++static struct rb_node *bfq_find_deepest(struct rb_node *node) ++{ ++ struct rb_node *deepest; ++ ++ if (!node->rb_right && !node->rb_left) ++ deepest = rb_parent(node); ++ else if (!node->rb_right) ++ deepest = node->rb_left; ++ else if (!node->rb_left) ++ deepest = node->rb_right; ++ else { ++ deepest = rb_next(node); ++ if (deepest->rb_right) ++ deepest = deepest->rb_right; ++ else if (rb_parent(deepest) != node) ++ deepest = rb_parent(deepest); ++ } ++ ++ return deepest; ++} ++ ++/** ++ * bfq_active_extract - remove an entity from the active tree. ++ * @st: the service_tree containing the tree. ++ * @entity: the entity being removed. ++ */ ++static void bfq_active_extract(struct bfq_service_tree *st, ++ struct bfq_entity *entity) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ struct rb_node *node; ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ struct bfq_sched_data *sd = NULL; ++ struct bfq_group *bfqg = NULL; ++ struct bfq_data *bfqd = NULL; ++#endif ++ ++ node = bfq_find_deepest(&entity->rb_node); ++ bfq_extract(&st->active, entity); ++ ++ if (node) ++ bfq_update_active_tree(node); ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ sd = entity->sched_data; ++ bfqg = container_of(sd, struct bfq_group, sched_data); ++ BUG_ON(!bfqg); ++ bfqd = (struct bfq_data *)bfqg->bfqd; ++#endif ++ if (bfqq) ++ list_del(&bfqq->bfqq_list); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { /* bfq_group */ ++ BUG_ON(!bfqd); ++ bfq_weights_tree_remove(bfqd, entity, ++ &bfqd->group_weights_tree); ++ } ++ if (bfqg != bfqd->root_group) { ++ BUG_ON(!bfqg); ++ BUG_ON(!bfqd); ++ BUG_ON(!bfqg->active_entities); ++ bfqg->active_entities--; ++ } ++#endif ++} ++ ++/** ++ * bfq_idle_insert - insert an entity into the idle tree. ++ * @st: the service tree containing the tree. ++ * @entity: the entity to insert. ++ */ ++static void bfq_idle_insert(struct bfq_service_tree *st, ++ struct bfq_entity *entity) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ struct bfq_entity *first_idle = st->first_idle; ++ struct bfq_entity *last_idle = st->last_idle; ++ ++ if (!first_idle || bfq_gt(first_idle->finish, entity->finish)) ++ st->first_idle = entity; ++ if (!last_idle || bfq_gt(entity->finish, last_idle->finish)) ++ st->last_idle = entity; ++ ++ bfq_insert(&st->idle, entity); ++ ++ if (bfqq) ++ list_add(&bfqq->bfqq_list, &bfqq->bfqd->idle_list); ++} ++ ++/** ++ * bfq_forget_entity - do not consider entity any longer for scheduling ++ * @st: the service tree. ++ * @entity: the entity being removed. ++ * @is_in_service: true if entity is currently the in-service entity. ++ * ++ * Forget everything about @entity. In addition, if entity represents ++ * a queue, and the latter is not in service, then release the service ++ * reference to the queue (the one taken through bfq_get_entity). In ++ * fact, in this case, there is really no more service reference to ++ * the queue, as the latter is also outside any service tree. If, ++ * instead, the queue is in service, then __bfq_bfqd_reset_in_service ++ * will take care of putting the reference when the queue finally ++ * stops being served. ++ */ ++static void bfq_forget_entity(struct bfq_service_tree *st, ++ struct bfq_entity *entity, ++ bool is_in_service) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ BUG_ON(!entity->on_st); ++ ++ entity->on_st = false; ++ st->wsum -= entity->weight; ++ if (bfqq && !is_in_service) { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "forget_entity (before): %p %d", ++ bfqq, bfqq->ref); ++ bfq_put_queue(bfqq); ++ } ++} ++ ++/** ++ * bfq_put_idle_entity - release the idle tree ref of an entity. ++ * @st: service tree for the entity. ++ * @entity: the entity being released. ++ */ ++static void bfq_put_idle_entity(struct bfq_service_tree *st, ++ struct bfq_entity *entity) ++{ ++ bfq_idle_extract(st, entity); ++ bfq_forget_entity(st, entity, ++ entity == entity->sched_data->in_service_entity); ++} ++ ++/** ++ * bfq_forget_idle - update the idle tree if necessary. ++ * @st: the service tree to act upon. ++ * ++ * To preserve the global O(log N) complexity we only remove one entry here; ++ * as the idle tree will not grow indefinitely this can be done safely. ++ */ ++static void bfq_forget_idle(struct bfq_service_tree *st) ++{ ++ struct bfq_entity *first_idle = st->first_idle; ++ struct bfq_entity *last_idle = st->last_idle; ++ ++ if (RB_EMPTY_ROOT(&st->active) && last_idle && ++ !bfq_gt(last_idle->finish, st->vtime)) { ++ /* ++ * Forget the whole idle tree, increasing the vtime past ++ * the last finish time of idle entities. ++ */ ++ st->vtime = last_idle->finish; ++ } ++ ++ if (first_idle && !bfq_gt(first_idle->finish, st->vtime)) ++ bfq_put_idle_entity(st, first_idle); ++} ++ ++/* ++ * Update weight and priority of entity. If update_class_too is true, ++ * then update the ioprio_class of entity too. ++ * ++ * The reason why the update of ioprio_class is controlled through the ++ * last parameter is as follows. Changing the ioprio class of an ++ * entity implies changing the destination service trees for that ++ * entity. If such a change occurred when the entity is already on one ++ * of the service trees for its previous class, then the state of the ++ * entity would become more complex: none of the new possible service ++ * trees for the entity, according to bfq_entity_service_tree(), would ++ * match any of the possible service trees on which the entity ++ * is. Complex operations involving these trees, such as entity ++ * activations and deactivations, should take into account this ++ * additional complexity. To avoid this issue, this function is ++ * invoked with update_class_too unset in the points in the code where ++ * entity may happen to be on some tree. ++ */ ++static struct bfq_service_tree * ++__bfq_entity_update_weight_prio(struct bfq_service_tree *old_st, ++ struct bfq_entity *entity, ++ bool update_class_too) ++{ ++ struct bfq_service_tree *new_st = old_st; ++ ++ if (entity->prio_changed) { ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ unsigned int prev_weight, new_weight; ++ struct bfq_data *bfqd = NULL; ++ struct rb_root *root; ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ struct bfq_sched_data *sd; ++ struct bfq_group *bfqg; ++#endif ++ ++ if (bfqq) ++ bfqd = bfqq->bfqd; ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { ++ sd = entity->my_sched_data; ++ bfqg = container_of(sd, struct bfq_group, sched_data); ++ BUG_ON(!bfqg); ++ bfqd = (struct bfq_data *)bfqg->bfqd; ++ BUG_ON(!bfqd); ++ } ++#endif ++ ++ BUG_ON(old_st->wsum < entity->weight); ++ old_st->wsum -= entity->weight; ++ ++ if (entity->new_weight != entity->orig_weight) { ++ if (entity->new_weight < BFQ_MIN_WEIGHT || ++ entity->new_weight > BFQ_MAX_WEIGHT) { ++ pr_crit("update_weight_prio: new_weight %d\n", ++ entity->new_weight); ++ if (entity->new_weight < BFQ_MIN_WEIGHT) ++ entity->new_weight = BFQ_MIN_WEIGHT; ++ else ++ entity->new_weight = BFQ_MAX_WEIGHT; ++ } ++ entity->orig_weight = entity->new_weight; ++ if (bfqq) ++ bfqq->ioprio = ++ bfq_weight_to_ioprio(entity->orig_weight); ++ } ++ ++ if (bfqq && update_class_too) ++ bfqq->ioprio_class = bfqq->new_ioprio_class; ++ ++ /* ++ * Reset prio_changed only if the ioprio_class change ++ * is not pending any longer. ++ */ ++ if (!bfqq || bfqq->ioprio_class == bfqq->new_ioprio_class) ++ entity->prio_changed = 0; ++ ++ /* ++ * NOTE: here we may be changing the weight too early, ++ * this will cause unfairness. The correct approach ++ * would have required additional complexity to defer ++ * weight changes to the proper time instants (i.e., ++ * when entity->finish <= old_st->vtime). ++ */ ++ new_st = bfq_entity_service_tree(entity); ++ ++ prev_weight = entity->weight; ++ new_weight = entity->orig_weight * ++ (bfqq ? bfqq->wr_coeff : 1); ++ /* ++ * If the weight of the entity changes, remove the entity ++ * from its old weight counter (if there is a counter ++ * associated with the entity), and add it to the counter ++ * associated with its new weight. ++ */ ++ if (prev_weight != new_weight) { ++ if (bfqq) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "weight changed %d %d(%d %d)", ++ prev_weight, new_weight, ++ entity->orig_weight, ++ bfqq->wr_coeff); ++ ++ root = bfqq ? &bfqd->queue_weights_tree : ++ &bfqd->group_weights_tree; ++ bfq_weights_tree_remove(bfqd, entity, root); ++ } ++ entity->weight = new_weight; ++ /* ++ * Add the entity to its weights tree only if it is ++ * not associated with a weight-raised queue. ++ */ ++ if (prev_weight != new_weight && ++ (bfqq ? bfqq->wr_coeff == 1 : 1)) ++ /* If we get here, root has been initialized. */ ++ bfq_weights_tree_add(bfqd, entity, root); ++ ++ new_st->wsum += entity->weight; ++ ++ if (new_st != old_st) ++ entity->start = new_st->vtime; ++ } ++ ++ return new_st; ++} ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++static void bfqg_stats_set_start_empty_time(struct bfq_group *bfqg); ++#endif ++ ++/** ++ * bfq_bfqq_served - update the scheduler status after selection for ++ * service. ++ * @bfqq: the queue being served. ++ * @served: bytes to transfer. ++ * ++ * NOTE: this can be optimized, as the timestamps of upper level entities ++ * are synchronized every time a new bfqq is selected for service. By now, ++ * we keep it to better check consistency. ++ */ ++static void bfq_bfqq_served(struct bfq_queue *bfqq, int served) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ struct bfq_service_tree *st; ++ ++ for_each_entity(entity) { ++ st = bfq_entity_service_tree(entity); ++ ++ entity->service += served; ++ ++ BUG_ON(st->wsum == 0); ++ ++ st->vtime += bfq_delta(served, st->wsum); ++ bfq_forget_idle(st); ++ } ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ bfqg_stats_set_start_empty_time(bfqq_group(bfqq)); ++#endif ++ st = bfq_entity_service_tree(&bfqq->entity); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "bfqq_served %d secs, vtime %llu on %p", ++ served, ((st->vtime>>10)*1000)>>12, st); ++} ++ ++/** ++ * bfq_bfqq_charge_time - charge an amount of service equivalent to the length ++ * of the time interval during which bfqq has been in ++ * service. ++ * @bfqd: the device ++ * @bfqq: the queue that needs a service update. ++ * @time_ms: the amount of time during which the queue has received service ++ * ++ * If a queue does not consume its budget fast enough, then providing ++ * the queue with service fairness may impair throughput, more or less ++ * severely. For this reason, queues that consume their budget slowly ++ * are provided with time fairness instead of service fairness. This ++ * goal is achieved through the BFQ scheduling engine, even if such an ++ * engine works in the service, and not in the time domain. The trick ++ * is charging these queues with an inflated amount of service, equal ++ * to the amount of service that they would have received during their ++ * service slot if they had been fast, i.e., if their requests had ++ * been dispatched at a rate equal to the estimated peak rate. ++ * ++ * It is worth noting that time fairness can cause important ++ * distortions in terms of bandwidth distribution, on devices with ++ * internal queueing. The reason is that I/O requests dispatched ++ * during the service slot of a queue may be served after that service ++ * slot is finished, and may have a total processing time loosely ++ * correlated with the duration of the service slot. This is ++ * especially true for short service slots. ++ */ ++static void bfq_bfqq_charge_time(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ unsigned long time_ms) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ int tot_serv_to_charge = entity->service; ++ unsigned int timeout_ms = jiffies_to_msecs(bfq_timeout); ++ ++ if (time_ms > 0 && time_ms < timeout_ms) ++ tot_serv_to_charge = ++ (bfqd->bfq_max_budget * time_ms) / timeout_ms; ++ ++ if (tot_serv_to_charge < entity->service) ++ tot_serv_to_charge = entity->service; ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "charge_time: %lu/%u ms, %d/%d/%d sectors", ++ time_ms, timeout_ms, entity->service, ++ tot_serv_to_charge, entity->budget); ++ ++ /* Increase budget to avoid inconsistencies */ ++ if (tot_serv_to_charge > entity->budget) ++ entity->budget = tot_serv_to_charge; ++ ++ bfq_bfqq_served(bfqq, ++ max_t(int, 0, tot_serv_to_charge - entity->service)); ++} ++ ++static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, ++ struct bfq_service_tree *st, ++ bool backshifted) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ struct bfq_sched_data *sd = entity->sched_data; ++ ++ /* ++ * When this function is invoked, entity is not in any service ++ * tree, then it is safe to invoke next function with the last ++ * parameter set (see the comments on the function). ++ */ ++ st = __bfq_entity_update_weight_prio(st, entity, true); ++ bfq_calc_finish(entity, entity->budget); ++ ++ /* ++ * If some queues enjoy backshifting for a while, then their ++ * (virtual) finish timestamps may happen to become lower and ++ * lower than the system virtual time. In particular, if ++ * these queues often happen to be idle for short time ++ * periods, and during such time periods other queues with ++ * higher timestamps happen to be busy, then the backshifted ++ * timestamps of the former queues can become much lower than ++ * the system virtual time. In fact, to serve the queues with ++ * higher timestamps while the ones with lower timestamps are ++ * idle, the system virtual time may be pushed-up to much ++ * higher values than the finish timestamps of the idle ++ * queues. As a consequence, the finish timestamps of all new ++ * or newly activated queues may end up being much larger than ++ * those of lucky queues with backshifted timestamps. The ++ * latter queues may then monopolize the device for a lot of ++ * time. This would simply break service guarantees. ++ * ++ * To reduce this problem, push up a little bit the ++ * backshifted timestamps of the queue associated with this ++ * entity (only a queue can happen to have the backshifted ++ * flag set): just enough to let the finish timestamp of the ++ * queue be equal to the current value of the system virtual ++ * time. This may introduce a little unfairness among queues ++ * with backshifted timestamps, but it does not break ++ * worst-case fairness guarantees. ++ * ++ * As a special case, if bfqq is weight-raised, push up ++ * timestamps much less, to keep very low the probability that ++ * this push up causes the backshifted finish timestamps of ++ * weight-raised queues to become higher than the backshifted ++ * finish timestamps of non weight-raised queues. ++ */ ++ if (backshifted && bfq_gt(st->vtime, entity->finish)) { ++ unsigned long delta = st->vtime - entity->finish; ++ ++ if (bfqq) ++ delta /= bfqq->wr_coeff; ++ ++ entity->start += delta; ++ entity->finish += delta; ++ ++ if (bfqq) { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "__activate_entity: new queue finish %llu", ++ ((entity->finish>>10)*1000)>>12); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ } else { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "__activate_entity: new group finish %llu", ++ ((entity->finish>>10)*1000)>>12); ++#endif ++ } ++ } ++ ++ bfq_active_insert(st, entity); ++ ++ if (bfqq) { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "__activate_entity: queue %seligible in st %p", ++ entity->start <= st->vtime ? "" : "non ", st); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ } else { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "__activate_entity: group %seligible in st %p", ++ entity->start <= st->vtime ? "" : "non ", st); ++#endif ++ } ++ BUG_ON(RB_EMPTY_ROOT(&st->active)); ++ BUG_ON(&st->active != &sd->service_tree->active && ++ &st->active != &(sd->service_tree+1)->active && ++ &st->active != &(sd->service_tree+2)->active); ++} ++ ++/** ++ * __bfq_activate_entity - handle activation of entity. ++ * @entity: the entity being activated. ++ * @non_blocking_wait_rq: true if entity was waiting for a request ++ * ++ * Called for a 'true' activation, i.e., if entity is not active and ++ * one of its children receives a new request. ++ * ++ * Basically, this function updates the timestamps of entity and ++ * inserts entity into its active tree, ater possible extracting it ++ * from its idle tree. ++ */ ++static void __bfq_activate_entity(struct bfq_entity *entity, ++ bool non_blocking_wait_rq) ++{ ++ struct bfq_sched_data *sd = entity->sched_data; ++ struct bfq_service_tree *st = bfq_entity_service_tree(entity); ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ bool backshifted = false; ++ unsigned long long min_vstart; ++ ++ BUG_ON(!sd); ++ BUG_ON(!st); ++ ++ /* See comments on bfq_fqq_update_budg_for_activation */ ++ if (non_blocking_wait_rq && bfq_gt(st->vtime, entity->finish)) { ++ backshifted = true; ++ min_vstart = entity->finish; ++ } else ++ min_vstart = st->vtime; ++ ++ if (entity->tree == &st->idle) { ++ /* ++ * Must be on the idle tree, bfq_idle_extract() will ++ * check for that. ++ */ ++ bfq_idle_extract(st, entity); ++ entity->start = bfq_gt(min_vstart, entity->finish) ? ++ min_vstart : entity->finish; ++ } else { ++ /* ++ * The finish time of the entity may be invalid, and ++ * it is in the past for sure, otherwise the queue ++ * would have been on the idle tree. ++ */ ++ entity->start = min_vstart; ++ st->wsum += entity->weight; ++ /* ++ * entity is about to be inserted into a service tree, ++ * and then set in service: get a reference to make ++ * sure entity does not disappear until it is no ++ * longer in service or scheduled for service. ++ */ ++ bfq_get_entity(entity); ++ ++ BUG_ON(entity->on_st && bfqq); ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ if (entity->on_st && !bfqq) { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, ++ entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, ++ bfqg, ++ "activate bug, class %d in_service %p", ++ bfq_class_idx(entity), sd->in_service_entity); ++ } ++#endif ++ BUG_ON(entity->on_st && !bfqq); ++ entity->on_st = true; ++ } ++ ++ bfq_update_fin_time_enqueue(entity, st, backshifted); ++} ++ ++/** ++ * __bfq_requeue_entity - handle requeueing or repositioning of an entity. ++ * @entity: the entity being requeued or repositioned. ++ * ++ * Requeueing is needed if this entity stops being served, which ++ * happens if a leaf descendant entity has expired. On the other hand, ++ * repositioning is needed if the next_inservice_entity for the child ++ * entity has changed. See the comments inside the function for ++ * details. ++ * ++ * Basically, this function: 1) removes entity from its active tree if ++ * present there, 2) updates the timestamps of entity and 3) inserts ++ * entity back into its active tree (in the new, right position for ++ * the new values of the timestamps). ++ */ ++static void __bfq_requeue_entity(struct bfq_entity *entity) ++{ ++ struct bfq_sched_data *sd = entity->sched_data; ++ struct bfq_service_tree *st = bfq_entity_service_tree(entity); ++ ++ BUG_ON(!sd); ++ BUG_ON(!st); ++ ++ BUG_ON(entity != sd->in_service_entity && ++ entity->tree != &st->active); ++ ++ if (entity == sd->in_service_entity) { ++ /* ++ * We are requeueing the current in-service entity, ++ * which may have to be done for one of the following ++ * reasons: ++ * - entity represents the in-service queue, and the ++ * in-service queue is being requeued after an ++ * expiration; ++ * - entity represents a group, and its budget has ++ * changed because one of its child entities has ++ * just been either activated or requeued for some ++ * reason; the timestamps of the entity need then to ++ * be updated, and the entity needs to be enqueued ++ * or repositioned accordingly. ++ * ++ * In particular, before requeueing, the start time of ++ * the entity must be moved forward to account for the ++ * service that the entity has received while in ++ * service. This is done by the next instructions. The ++ * finish time will then be updated according to this ++ * new value of the start time, and to the budget of ++ * the entity. ++ */ ++ bfq_calc_finish(entity, entity->service); ++ entity->start = entity->finish; ++ BUG_ON(entity->tree && entity->tree != &st->active); ++ /* ++ * In addition, if the entity had more than one child ++ * when set in service, then was not extracted from ++ * the active tree. This implies that the position of ++ * the entity in the active tree may need to be ++ * changed now, because we have just updated the start ++ * time of the entity, and we will update its finish ++ * time in a moment (the requeueing is then, more ++ * precisely, a repositioning in this case). To ++ * implement this repositioning, we: 1) dequeue the ++ * entity here, 2) update the finish time and ++ * requeue the entity according to the new ++ * timestamps below. ++ */ ++ if (entity->tree) ++ bfq_active_extract(st, entity); ++ } else { /* The entity is already active, and not in service */ ++ /* ++ * In this case, this function gets called only if the ++ * next_in_service entity below this entity has ++ * changed, and this change has caused the budget of ++ * this entity to change, which, finally implies that ++ * the finish time of this entity must be ++ * updated. Such an update may cause the scheduling, ++ * i.e., the position in the active tree, of this ++ * entity to change. We handle this change by: 1) ++ * dequeueing the entity here, 2) updating the finish ++ * time and requeueing the entity according to the new ++ * timestamps below. This is the same approach as the ++ * non-extracted-entity sub-case above. ++ */ ++ bfq_active_extract(st, entity); ++ } ++ ++ bfq_update_fin_time_enqueue(entity, st, false); ++} ++ ++static void __bfq_activate_requeue_entity(struct bfq_entity *entity, ++ struct bfq_sched_data *sd, ++ bool non_blocking_wait_rq) ++{ ++ struct bfq_service_tree *st = bfq_entity_service_tree(entity); ++ ++ if (sd->in_service_entity == entity || entity->tree == &st->active) ++ /* ++ * in service or already queued on the active tree, ++ * requeue or reposition ++ */ ++ __bfq_requeue_entity(entity); ++ else ++ /* ++ * Not in service and not queued on its active tree: ++ * the activity is idle and this is a true activation. ++ */ ++ __bfq_activate_entity(entity, non_blocking_wait_rq); ++} ++ ++ ++/** ++ * bfq_activate_entity - activate or requeue an entity representing a bfq_queue, ++ * and activate, requeue or reposition all ancestors ++ * for which such an update becomes necessary. ++ * @entity: the entity to activate. ++ * @non_blocking_wait_rq: true if this entity was waiting for a request ++ * @requeue: true if this is a requeue, which implies that bfqq is ++ * being expired; thus ALL its ancestors stop being served and must ++ * therefore be requeued ++ */ ++static void bfq_activate_requeue_entity(struct bfq_entity *entity, ++ bool non_blocking_wait_rq, ++ bool requeue) ++{ ++ struct bfq_sched_data *sd; ++ ++ for_each_entity(entity) { ++ BUG_ON(!entity); ++ sd = entity->sched_data; ++ __bfq_activate_requeue_entity(entity, sd, non_blocking_wait_rq); ++ ++ BUG_ON(RB_EMPTY_ROOT(&sd->service_tree->active) && ++ RB_EMPTY_ROOT(&(sd->service_tree+1)->active) && ++ RB_EMPTY_ROOT(&(sd->service_tree+2)->active)); ++ ++ if (!bfq_update_next_in_service(sd, entity) && !requeue) { ++ BUG_ON(!sd->next_in_service); ++ break; ++ } ++ BUG_ON(!sd->next_in_service); ++ } ++} ++ ++/** ++ * __bfq_deactivate_entity - deactivate an entity from its service tree. ++ * @entity: the entity to deactivate. ++ * @ins_into_idle_tree: if false, the entity will not be put into the ++ * idle tree. ++ * ++ * Deactivates an entity, independently from its previous state. Must ++ * be invoked only if entity is on a service tree. Extracts the entity ++ * from that tree, and if necessary and allowed, puts it on the idle ++ * tree. ++ */ ++static bool __bfq_deactivate_entity(struct bfq_entity *entity, ++ bool ins_into_idle_tree) ++{ ++ struct bfq_sched_data *sd = entity->sched_data; ++ struct bfq_service_tree *st; ++ bool is_in_service; ++ ++ if (!entity->on_st) { /* entity never activated, or already inactive */ ++ BUG_ON(sd && entity == sd->in_service_entity); ++ return false; ++ } ++ ++ /* ++ * If we get here, then entity is active, which implies that ++ * bfq_group_set_parent has already been invoked for the group ++ * represented by entity. Therefore, the field ++ * entity->sched_data has been set, and we can safely use it. ++ */ ++ st = bfq_entity_service_tree(entity); ++ is_in_service = entity == sd->in_service_entity; ++ ++ BUG_ON(is_in_service && entity->tree && entity->tree != &st->active); ++ ++ if (is_in_service) ++ bfq_calc_finish(entity, entity->service); ++ ++ if (entity->tree == &st->active) ++ bfq_active_extract(st, entity); ++ else if (!is_in_service && entity->tree == &st->idle) ++ bfq_idle_extract(st, entity); ++ else if (entity->tree) ++ BUG(); ++ ++ if (!ins_into_idle_tree || !bfq_gt(entity->finish, st->vtime)) ++ bfq_forget_entity(st, entity, is_in_service); ++ else ++ bfq_idle_insert(st, entity); ++ ++ return true; ++} ++ ++/** ++ * bfq_deactivate_entity - deactivate an entity representing a bfq_queue. ++ * @entity: the entity to deactivate. ++ * @ins_into_idle_tree: true if the entity can be put on the idle tree ++ */ ++static void bfq_deactivate_entity(struct bfq_entity *entity, ++ bool ins_into_idle_tree, ++ bool expiration) ++{ ++ struct bfq_sched_data *sd; ++ struct bfq_entity *parent = NULL; ++ ++ for_each_entity_safe(entity, parent) { ++ sd = entity->sched_data; ++ ++ BUG_ON(sd == NULL); /* ++ * It would mean that this is the ++ * root group. ++ */ ++ ++ BUG_ON(expiration && entity != sd->in_service_entity); ++ ++ BUG_ON(entity != sd->in_service_entity && ++ entity->tree == ++ &bfq_entity_service_tree(entity)->active && ++ !sd->next_in_service); ++ ++ if (!__bfq_deactivate_entity(entity, ins_into_idle_tree)) { ++ /* ++ * entity is not in any tree any more, so ++ * this deactivation is a no-op, and there is ++ * nothing to change for upper-level entities ++ * (in case of expiration, this can never ++ * happen). ++ */ ++ BUG_ON(expiration); /* ++ * entity cannot be already out of ++ * any tree ++ */ ++ return; ++ } ++ ++ if (sd->next_in_service == entity) ++ /* ++ * entity was the next_in_service entity, ++ * then, since entity has just been ++ * deactivated, a new one must be found. ++ */ ++ bfq_update_next_in_service(sd, NULL); ++ ++ if (sd->next_in_service) { ++ /* ++ * The parent entity is still backlogged, ++ * because next_in_service is not NULL. So, no ++ * further upwards deactivation must be ++ * performed. Yet, next_in_service has ++ * changed. Then the schedule does need to be ++ * updated upwards. ++ */ ++ BUG_ON(sd->next_in_service == entity); ++ break; ++ } ++ ++ /* ++ * If we get here, then the parent is no more ++ * backlogged and we need to propagate the ++ * deactivation upwards. Thus let the loop go on. ++ */ ++ ++ /* ++ * Also let parent be queued into the idle tree on ++ * deactivation, to preserve service guarantees, and ++ * assuming that who invoked this function does not ++ * need parent entities too to be removed completely. ++ */ ++ ins_into_idle_tree = true; ++ } ++ ++ /* ++ * If the deactivation loop is fully executed, then there are ++ * no more entities to touch and next loop is not executed at ++ * all. Otherwise, requeue remaining entities if they are ++ * about to stop receiving service, or reposition them if this ++ * is not the case. ++ */ ++ entity = parent; ++ for_each_entity(entity) { ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ ++ /* ++ * Invoke __bfq_requeue_entity on entity, even if ++ * already active, to requeue/reposition it in the ++ * active tree (because sd->next_in_service has ++ * changed) ++ */ ++ __bfq_requeue_entity(entity); ++ ++ sd = entity->sched_data; ++ BUG_ON(expiration && sd->in_service_entity != entity); ++ ++ if (bfqq) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "invoking udpdate_next for this queue"); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { ++ struct bfq_group *bfqg = ++ container_of(entity, ++ struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "invoking udpdate_next for this entity"); ++ } ++#endif ++ if (!bfq_update_next_in_service(sd, entity) && ++ !expiration) ++ /* ++ * next_in_service unchanged or not causing ++ * any change in entity->parent->sd, and no ++ * requeueing needed for expiration: stop ++ * here. ++ */ ++ break; ++ } ++} ++ ++/** ++ * bfq_calc_vtime_jump - compute the value to which the vtime should jump, ++ * if needed, to have at least one entity eligible. ++ * @st: the service tree to act upon. ++ * ++ * Assumes that st is not empty. ++ */ ++static u64 bfq_calc_vtime_jump(struct bfq_service_tree *st) ++{ ++ struct bfq_entity *root_entity = bfq_root_active_entity(&st->active); ++ ++ if (bfq_gt(root_entity->min_start, st->vtime)) { ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(root_entity); ++ ++ if (bfqq) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "calc_vtime_jump: new value %llu", ++ root_entity->min_start); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { ++ struct bfq_group *bfqg = ++ container_of(root_entity, struct bfq_group, ++ entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "calc_vtime_jump: new value %llu", ++ root_entity->min_start); ++ } ++#endif ++ return root_entity->min_start; ++ } ++ return st->vtime; ++} ++ ++static void bfq_update_vtime(struct bfq_service_tree *st, u64 new_value) ++{ ++ if (new_value > st->vtime) { ++ st->vtime = new_value; ++ bfq_forget_idle(st); ++ } ++} ++ ++/** ++ * bfq_first_active_entity - find the eligible entity with ++ * the smallest finish time ++ * @st: the service tree to select from. ++ * @vtime: the system virtual to use as a reference for eligibility ++ * ++ * This function searches the first schedulable entity, starting from the ++ * root of the tree and going on the left every time on this side there is ++ * a subtree with at least one eligible (start >= vtime) entity. The path on ++ * the right is followed only if a) the left subtree contains no eligible ++ * entities and b) no eligible entity has been found yet. ++ */ ++static struct bfq_entity *bfq_first_active_entity(struct bfq_service_tree *st, ++ u64 vtime) ++{ ++ struct bfq_entity *entry, *first = NULL; ++ struct rb_node *node = st->active.rb_node; ++ ++ while (node) { ++ entry = rb_entry(node, struct bfq_entity, rb_node); ++left: ++ if (!bfq_gt(entry->start, vtime)) ++ first = entry; ++ ++ BUG_ON(bfq_gt(entry->min_start, vtime)); ++ ++ if (node->rb_left) { ++ entry = rb_entry(node->rb_left, ++ struct bfq_entity, rb_node); ++ if (!bfq_gt(entry->min_start, vtime)) { ++ node = node->rb_left; ++ goto left; ++ } ++ } ++ if (first) ++ break; ++ node = node->rb_right; ++ } ++ ++ BUG_ON(!first && !RB_EMPTY_ROOT(&st->active)); ++ return first; ++} ++ ++/** ++ * __bfq_lookup_next_entity - return the first eligible entity in @st. ++ * @st: the service tree. ++ * ++ * If there is no in-service entity for the sched_data st belongs to, ++ * then return the entity that will be set in service if: ++ * 1) the parent entity this st belongs to is set in service; ++ * 2) no entity belonging to such parent entity undergoes a state change ++ * that would influence the timestamps of the entity (e.g., becomes idle, ++ * becomes backlogged, changes its budget, ...). ++ * ++ * In this first case, update the virtual time in @st too (see the ++ * comments on this update inside the function). ++ * ++ * In constrast, if there is an in-service entity, then return the ++ * entity that would be set in service if not only the above ++ * conditions, but also the next one held true: the currently ++ * in-service entity, on expiration, ++ * 1) gets a finish time equal to the current one, or ++ * 2) is not eligible any more, or ++ * 3) is idle. ++ */ ++static struct bfq_entity * ++__bfq_lookup_next_entity(struct bfq_service_tree *st, bool in_service ++#if 0 ++ , bool force ++#endif ++ ) ++{ ++ struct bfq_entity *entity ++#if 0 ++ , *new_next_in_service = NULL ++#endif ++ ; ++ u64 new_vtime; ++ struct bfq_queue *bfqq; ++ ++ if (RB_EMPTY_ROOT(&st->active)) ++ return NULL; ++ ++ /* ++ * Get the value of the system virtual time for which at ++ * least one entity is eligible. ++ */ ++ new_vtime = bfq_calc_vtime_jump(st); ++ ++ /* ++ * If there is no in-service entity for the sched_data this ++ * active tree belongs to, then push the system virtual time ++ * up to the value that guarantees that at least one entity is ++ * eligible. If, instead, there is an in-service entity, then ++ * do not make any such update, because there is already an ++ * eligible entity, namely the in-service one (even if the ++ * entity is not on st, because it was extracted when set in ++ * service). ++ */ ++ if (!in_service) ++ bfq_update_vtime(st, new_vtime); ++ ++ entity = bfq_first_active_entity(st, new_vtime); ++ BUG_ON(bfq_gt(entity->start, new_vtime)); ++ ++ /* Log some information */ ++ bfqq = bfq_entity_to_bfqq(entity); ++ if (bfqq) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "__lookup_next: start %llu vtime %llu st %p", ++ ((entity->start>>10)*1000)>>12, ++ ((new_vtime>>10)*1000)>>12, st); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "__lookup_next: start %llu vtime %llu st %p", ++ ((entity->start>>10)*1000)>>12, ++ ((new_vtime>>10)*1000)>>12, st); ++ } ++#endif ++ ++ BUG_ON(!entity); ++ ++ return entity; ++} ++ ++/** ++ * bfq_lookup_next_entity - return the first eligible entity in @sd. ++ * @sd: the sched_data. ++ * ++ * This function is invoked when there has been a change in the trees ++ * for sd, and we need know what is the new next entity after this ++ * change. ++ */ ++static struct bfq_entity *bfq_lookup_next_entity(struct bfq_sched_data *sd) ++{ ++ struct bfq_service_tree *st = sd->service_tree; ++ struct bfq_service_tree *idle_class_st = st + (BFQ_IOPRIO_CLASSES - 1); ++ struct bfq_entity *entity = NULL; ++ struct bfq_queue *bfqq; ++ int class_idx = 0; ++ ++ BUG_ON(!sd); ++ BUG_ON(!st); ++ /* ++ * Choose from idle class, if needed to guarantee a minimum ++ * bandwidth to this class (and if there is some active entity ++ * in idle class). This should also mitigate ++ * priority-inversion problems in case a low priority task is ++ * holding file system resources. ++ */ ++ if (time_is_before_jiffies(sd->bfq_class_idle_last_service + ++ BFQ_CL_IDLE_TIMEOUT)) { ++ if (!RB_EMPTY_ROOT(&idle_class_st->active)) ++ class_idx = BFQ_IOPRIO_CLASSES - 1; ++ /* About to be served if backlogged, or not yet backlogged */ ++ sd->bfq_class_idle_last_service = jiffies; ++ } ++ ++ /* ++ * Find the next entity to serve for the highest-priority ++ * class, unless the idle class needs to be served. ++ */ ++ for (; class_idx < BFQ_IOPRIO_CLASSES; class_idx++) { ++ entity = __bfq_lookup_next_entity(st + class_idx, ++ sd->in_service_entity); ++ ++ if (entity) ++ break; ++ } ++ ++ BUG_ON(!entity && ++ (!RB_EMPTY_ROOT(&st->active) || !RB_EMPTY_ROOT(&(st+1)->active) || ++ !RB_EMPTY_ROOT(&(st+2)->active))); ++ ++ if (!entity) ++ return NULL; ++ ++ /* Log some information */ ++ bfqq = bfq_entity_to_bfqq(entity); ++ if (bfqq) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "chosen from st %p %d", ++ st + class_idx, class_idx); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "chosen from st %p %d", ++ st + class_idx, class_idx); ++ } ++#endif ++ ++ return entity; ++} ++ ++static bool next_queue_may_preempt(struct bfq_data *bfqd) ++{ ++ struct bfq_sched_data *sd = &bfqd->root_group->sched_data; ++ ++ return sd->next_in_service != sd->in_service_entity; ++} ++ ++/* ++ * Get next queue for service. ++ */ ++static struct bfq_queue *bfq_get_next_queue(struct bfq_data *bfqd) ++{ ++ struct bfq_entity *entity = NULL; ++ struct bfq_sched_data *sd; ++ struct bfq_queue *bfqq; ++ ++ BUG_ON(bfqd->in_service_queue); ++ ++ if (bfqd->busy_queues == 0) ++ return NULL; ++ ++ /* ++ * Traverse the path from the root to the leaf entity to ++ * serve. Set in service all the entities visited along the ++ * way. ++ */ ++ sd = &bfqd->root_group->sched_data; ++ for (; sd ; sd = entity->my_sched_data) { ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ if (entity) { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg(bfqd, bfqg, ++ "get_next_queue: lookup in this group"); ++ if (!sd->next_in_service) ++ pr_crit("get_next_queue: lookup in this group"); ++ } else { ++ bfq_log_bfqg(bfqd, bfqd->root_group, ++ "get_next_queue: lookup in root group"); ++ if (!sd->next_in_service) ++ pr_crit("get_next_queue: lookup in root group"); ++ } ++#endif ++ ++ BUG_ON(!sd->next_in_service); ++ ++ /* ++ * WARNING. We are about to set the in-service entity ++ * to sd->next_in_service, i.e., to the (cached) value ++ * returned by bfq_lookup_next_entity(sd) the last ++ * time it was invoked, i.e., the last time when the ++ * service order in sd changed as a consequence of the ++ * activation or deactivation of an entity. In this ++ * respect, if we execute bfq_lookup_next_entity(sd) ++ * in this very moment, it may, although with low ++ * probability, yield a different entity than that ++ * pointed to by sd->next_in_service. This rare event ++ * happens in case there was no CLASS_IDLE entity to ++ * serve for sd when bfq_lookup_next_entity(sd) was ++ * invoked for the last time, while there is now one ++ * such entity. ++ * ++ * If the above event happens, then the scheduling of ++ * such entity in CLASS_IDLE is postponed until the ++ * service of the sd->next_in_service entity ++ * finishes. In fact, when the latter is expired, ++ * bfq_lookup_next_entity(sd) gets called again, ++ * exactly to update sd->next_in_service. ++ */ ++ ++ /* Make next_in_service entity become in_service_entity */ ++ entity = sd->next_in_service; ++ sd->in_service_entity = entity; ++ ++ /* ++ * Reset the accumulator of the amount of service that ++ * the entity is about to receive. ++ */ ++ entity->service = 0; ++ ++ /* ++ * If entity is no longer a candidate for next ++ * service, then we extract it from its active tree, ++ * for the following reason. To further boost the ++ * throughput in some special case, BFQ needs to know ++ * which is the next candidate entity to serve, while ++ * there is already an entity in service. In this ++ * respect, to make it easy to compute/update the next ++ * candidate entity to serve after the current ++ * candidate has been set in service, there is a case ++ * where it is necessary to extract the current ++ * candidate from its service tree. Such a case is ++ * when the entity just set in service cannot be also ++ * a candidate for next service. Details about when ++ * this conditions holds are reported in the comments ++ * on the function bfq_no_longer_next_in_service() ++ * invoked below. ++ */ ++ if (bfq_no_longer_next_in_service(entity)) ++ bfq_active_extract(bfq_entity_service_tree(entity), ++ entity); ++ ++ /* ++ * For the same reason why we may have just extracted ++ * entity from its active tree, we may need to update ++ * next_in_service for the sched_data of entity too, ++ * regardless of whether entity has been extracted. ++ * In fact, even if entity has not been extracted, a ++ * descendant entity may get extracted. Such an event ++ * would cause a change in next_in_service for the ++ * level of the descendant entity, and thus possibly ++ * back to upper levels. ++ * ++ * We cannot perform the resulting needed update ++ * before the end of this loop, because, to know which ++ * is the correct next-to-serve candidate entity for ++ * each level, we need first to find the leaf entity ++ * to set in service. In fact, only after we know ++ * which is the next-to-serve leaf entity, we can ++ * discover whether the parent entity of the leaf ++ * entity becomes the next-to-serve, and so on. ++ */ ++ ++ /* Log some information */ ++ bfqq = bfq_entity_to_bfqq(entity); ++ if (bfqq) ++ bfq_log_bfqq(bfqd, bfqq, ++ "get_next_queue: this queue, finish %llu", ++ (((entity->finish>>10)*1000)>>10)>>2); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg(bfqd, bfqg, ++ "get_next_queue: this entity, finish %llu", ++ (((entity->finish>>10)*1000)>>10)>>2); ++ } ++#endif ++ ++ } ++ ++ BUG_ON(!entity); ++ bfqq = bfq_entity_to_bfqq(entity); ++ BUG_ON(!bfqq); ++ ++ /* ++ * We can finally update all next-to-serve entities along the ++ * path from the leaf entity just set in service to the root. ++ */ ++ for_each_entity(entity) { ++ struct bfq_sched_data *sd = entity->sched_data; ++ ++ if(!bfq_update_next_in_service(sd, NULL)) ++ break; ++ } ++ ++ return bfqq; ++} ++ ++static void __bfq_bfqd_reset_in_service(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *in_serv_bfqq = bfqd->in_service_queue; ++ struct bfq_entity *in_serv_entity = &in_serv_bfqq->entity; ++ struct bfq_entity *entity = in_serv_entity; ++ ++ if (bfqd->in_service_bic) { ++ put_io_context(bfqd->in_service_bic->icq.ioc); ++ bfqd->in_service_bic = NULL; ++ } ++ ++ bfq_clear_bfqq_wait_request(in_serv_bfqq); ++ hrtimer_try_to_cancel(&bfqd->idle_slice_timer); ++ bfqd->in_service_queue = NULL; ++ ++ /* ++ * When this function is called, all in-service entities have ++ * been properly deactivated or requeued, so we can safely ++ * execute the final step: reset in_service_entity along the ++ * path from entity to the root. ++ */ ++ for_each_entity(entity) ++ entity->sched_data->in_service_entity = NULL; ++ ++ /* ++ * in_serv_entity is no longer in service, so, if it is in no ++ * service tree either, then release the service reference to ++ * the queue it represents (taken with bfq_get_entity). ++ */ ++ if (!in_serv_entity->on_st) ++ bfq_put_queue(in_serv_bfqq); ++} ++ ++static void bfq_deactivate_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ bool ins_into_idle_tree, bool expiration) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ ++ bfq_deactivate_entity(entity, ins_into_idle_tree, expiration); ++} ++ ++static void bfq_activate_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ struct bfq_service_tree *st = bfq_entity_service_tree(entity); ++ ++ BUG_ON(bfqq == bfqd->in_service_queue); ++ BUG_ON(entity->tree != &st->active && entity->tree != &st->idle && ++ entity->on_st); ++ ++ bfq_activate_requeue_entity(entity, bfq_bfqq_non_blocking_wait_rq(bfqq), ++ false); ++ bfq_clear_bfqq_non_blocking_wait_rq(bfqq); ++} ++ ++static void bfq_requeue_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ ++ bfq_activate_requeue_entity(entity, false, ++ bfqq == bfqd->in_service_queue); ++} ++ ++static void bfqg_stats_update_dequeue(struct bfq_group *bfqg); ++ ++/* ++ * Called when the bfqq no longer has requests pending, remove it from ++ * the service tree. As a special case, it can be invoked during an ++ * expiration. ++ */ ++static void bfq_del_bfqq_busy(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ bool expiration) ++{ ++ BUG_ON(!bfq_bfqq_busy(bfqq)); ++ BUG_ON(!RB_EMPTY_ROOT(&bfqq->sort_list)); ++ ++ bfq_log_bfqq(bfqd, bfqq, "del from busy"); ++ ++ bfq_clear_bfqq_busy(bfqq); ++ ++ BUG_ON(bfqd->busy_queues == 0); ++ bfqd->busy_queues--; ++ ++ if (!bfqq->dispatched) ++ bfq_weights_tree_remove(bfqd, &bfqq->entity, ++ &bfqd->queue_weights_tree); ++ ++ if (bfqq->wr_coeff > 1) { ++ bfqd->wr_busy_queues--; ++ BUG_ON(bfqd->wr_busy_queues < 0); ++ } ++ ++ bfqg_stats_update_dequeue(bfqq_group(bfqq)); ++ ++ BUG_ON(bfqq->entity.budget < 0); ++ ++ bfq_deactivate_bfqq(bfqd, bfqq, true, expiration); ++} ++ ++/* ++ * Called when an inactive queue receives a new request. ++ */ ++static void bfq_add_bfqq_busy(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ BUG_ON(bfq_bfqq_busy(bfqq)); ++ BUG_ON(bfqq == bfqd->in_service_queue); ++ ++ bfq_log_bfqq(bfqd, bfqq, "add to busy"); ++ ++ bfq_activate_bfqq(bfqd, bfqq); ++ ++ bfq_mark_bfqq_busy(bfqq); ++ bfqd->busy_queues++; ++ ++ if (!bfqq->dispatched) ++ if (bfqq->wr_coeff == 1) ++ bfq_weights_tree_add(bfqd, &bfqq->entity, ++ &bfqd->queue_weights_tree); ++ ++ if (bfqq->wr_coeff > 1) { ++ bfqd->wr_busy_queues++; ++ BUG_ON(bfqd->wr_busy_queues > bfqd->busy_queues); ++ } ++ ++} +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +new file mode 100644 +index 000000000000..65e7c7e77f3c +--- /dev/null ++++ b/block/bfq-sq-iosched.c +@@ -0,0 +1,5379 @@ ++/* ++ * Budget Fair Queueing (BFQ) I/O scheduler. ++ * ++ * Based on ideas and code from CFQ: ++ * Copyright (C) 2003 Jens Axboe ++ * ++ * Copyright (C) 2008 Fabio Checconi ++ * Paolo Valente ++ * ++ * Copyright (C) 2015 Paolo Valente ++ * ++ * Copyright (C) 2017 Paolo Valente ++ * ++ * Licensed under the GPL-2 as detailed in the accompanying COPYING.BFQ ++ * file. ++ * ++ * BFQ is a proportional-share I/O scheduler, with some extra ++ * low-latency capabilities. BFQ also supports full hierarchical ++ * scheduling through cgroups. Next paragraphs provide an introduction ++ * on BFQ inner workings. Details on BFQ benefits and usage can be ++ * found in Documentation/block/bfq-iosched.txt. ++ * ++ * BFQ is a proportional-share storage-I/O scheduling algorithm based ++ * on the slice-by-slice service scheme of CFQ. But BFQ assigns ++ * budgets, measured in number of sectors, to processes instead of ++ * time slices. The device is not granted to the in-service process ++ * for a given time slice, but until it has exhausted its assigned ++ * budget. This change from the time to the service domain enables BFQ ++ * to distribute the device throughput among processes as desired, ++ * without any distortion due to throughput fluctuations, or to device ++ * internal queueing. BFQ uses an ad hoc internal scheduler, called ++ * B-WF2Q+, to schedule processes according to their budgets. More ++ * precisely, BFQ schedules queues associated with processes. Thanks to ++ * the accurate policy of B-WF2Q+, BFQ can afford to assign high ++ * budgets to I/O-bound processes issuing sequential requests (to ++ * boost the throughput), and yet guarantee a low latency to ++ * interactive and soft real-time applications. ++ * ++ * NOTE: if the main or only goal, with a given device, is to achieve ++ * the maximum-possible throughput at all times, then do switch off ++ * all low-latency heuristics for that device, by setting low_latency ++ * to 0. ++ * ++ * BFQ is described in [1], where also a reference to the initial, more ++ * theoretical paper on BFQ can be found. The interested reader can find ++ * in the latter paper full details on the main algorithm, as well as ++ * formulas of the guarantees and formal proofs of all the properties. ++ * With respect to the version of BFQ presented in these papers, this ++ * implementation adds a few more heuristics, such as the one that ++ * guarantees a low latency to soft real-time applications, and a ++ * hierarchical extension based on H-WF2Q+. ++ * ++ * B-WF2Q+ is based on WF2Q+, that is described in [2], together with ++ * H-WF2Q+, while the augmented tree used to implement B-WF2Q+ with O(log N) ++ * complexity derives from the one introduced with EEVDF in [3]. ++ * ++ * [1] P. Valente, A. Avanzini, "Evolution of the BFQ Storage I/O ++ * Scheduler", Proceedings of the First Workshop on Mobile System ++ * Technologies (MST-2015), May 2015. ++ * http://algogroup.unimore.it/people/paolo/disk_sched/mst-2015.pdf ++ * ++ * http://algogroup.unimo.it/people/paolo/disk_sched/bf1-v1-suite-results.pdf ++ * ++ * [2] Jon C.R. Bennett and H. Zhang, ``Hierarchical Packet Fair Queueing ++ * Algorithms,'' IEEE/ACM Transactions on Networking, 5(5):675-689, ++ * Oct 1997. ++ * ++ * http://www.cs.cmu.edu/~hzhang/papers/TON-97-Oct.ps.gz ++ * ++ * [3] I. Stoica and H. Abdel-Wahab, ``Earliest Eligible Virtual Deadline ++ * First: A Flexible and Accurate Mechanism for Proportional Share ++ * Resource Allocation,'' technical report. ++ * ++ * http://www.cs.berkeley.edu/~istoica/papers/eevdf-tr-95.pdf ++ */ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include "blk.h" ++#include "bfq.h" ++ ++/* Expiration time of sync (0) and async (1) requests, in ns. */ ++static const u64 bfq_fifo_expire[2] = { NSEC_PER_SEC / 4, NSEC_PER_SEC / 8 }; ++ ++/* Maximum backwards seek, in KiB. */ ++static const int bfq_back_max = (16 * 1024); ++ ++/* Penalty of a backwards seek, in number of sectors. */ ++static const int bfq_back_penalty = 2; ++ ++/* Idling period duration, in ns. */ ++static u32 bfq_slice_idle = (NSEC_PER_SEC / 125); ++ ++/* Minimum number of assigned budgets for which stats are safe to compute. */ ++static const int bfq_stats_min_budgets = 194; ++ ++/* Default maximum budget values, in sectors and number of requests. */ ++static const int bfq_default_max_budget = (16 * 1024); ++ ++/* ++ * Async to sync throughput distribution is controlled as follows: ++ * when an async request is served, the entity is charged the number ++ * of sectors of the request, multiplied by the factor below ++ */ ++static const int bfq_async_charge_factor = 10; ++ ++/* Default timeout values, in jiffies, approximating CFQ defaults. */ ++static const int bfq_timeout = (HZ / 8); ++ ++static struct kmem_cache *bfq_pool; ++ ++/* Below this threshold (in ns), we consider thinktime immediate. */ ++#define BFQ_MIN_TT (2 * NSEC_PER_MSEC) ++ ++/* hw_tag detection: parallel requests threshold and min samples needed. */ ++#define BFQ_HW_QUEUE_THRESHOLD 4 ++#define BFQ_HW_QUEUE_SAMPLES 32 ++ ++#define BFQQ_SEEK_THR (sector_t)(8 * 100) ++#define BFQQ_SECT_THR_NONROT (sector_t)(2 * 32) ++#define BFQQ_CLOSE_THR (sector_t)(8 * 1024) ++#define BFQQ_SEEKY(bfqq) (hweight32(bfqq->seek_history) > 32/8) ++ ++/* Min number of samples required to perform peak-rate update */ ++#define BFQ_RATE_MIN_SAMPLES 32 ++/* Min observation time interval required to perform a peak-rate update (ns) */ ++#define BFQ_RATE_MIN_INTERVAL (300*NSEC_PER_MSEC) ++/* Target observation time interval for a peak-rate update (ns) */ ++#define BFQ_RATE_REF_INTERVAL NSEC_PER_SEC ++ ++/* Shift used for peak rate fixed precision calculations. */ ++#define BFQ_RATE_SHIFT 16 ++ ++/* ++ * By default, BFQ computes the duration of the weight raising for ++ * interactive applications automatically, using the following formula: ++ * duration = (R / r) * T, where r is the peak rate of the device, and ++ * R and T are two reference parameters. ++ * In particular, R is the peak rate of the reference device (see below), ++ * and T is a reference time: given the systems that are likely to be ++ * installed on the reference device according to its speed class, T is ++ * about the maximum time needed, under BFQ and while reading two files in ++ * parallel, to load typical large applications on these systems. ++ * In practice, the slower/faster the device at hand is, the more/less it ++ * takes to load applications with respect to the reference device. ++ * Accordingly, the longer/shorter BFQ grants weight raising to interactive ++ * applications. ++ * ++ * BFQ uses four different reference pairs (R, T), depending on: ++ * . whether the device is rotational or non-rotational; ++ * . whether the device is slow, such as old or portable HDDs, as well as ++ * SD cards, or fast, such as newer HDDs and SSDs. ++ * ++ * The device's speed class is dynamically (re)detected in ++ * bfq_update_peak_rate() every time the estimated peak rate is updated. ++ * ++ * In the following definitions, R_slow[0]/R_fast[0] and ++ * T_slow[0]/T_fast[0] are the reference values for a slow/fast ++ * rotational device, whereas R_slow[1]/R_fast[1] and ++ * T_slow[1]/T_fast[1] are the reference values for a slow/fast ++ * non-rotational device. Finally, device_speed_thresh are the ++ * thresholds used to switch between speed classes. The reference ++ * rates are not the actual peak rates of the devices used as a ++ * reference, but slightly lower values. The reason for using these ++ * slightly lower values is that the peak-rate estimator tends to ++ * yield slightly lower values than the actual peak rate (it can yield ++ * the actual peak rate only if there is only one process doing I/O, ++ * and the process does sequential I/O). ++ * ++ * Both the reference peak rates and the thresholds are measured in ++ * sectors/usec, left-shifted by BFQ_RATE_SHIFT. ++ */ ++static int R_slow[2] = {1000, 10700}; ++static int R_fast[2] = {14000, 33000}; ++/* ++ * To improve readability, a conversion function is used to initialize the ++ * following arrays, which entails that they can be initialized only in a ++ * function. ++ */ ++static int T_slow[2]; ++static int T_fast[2]; ++static int device_speed_thresh[2]; ++ ++#define BFQ_SERVICE_TREE_INIT ((struct bfq_service_tree) \ ++ { RB_ROOT, RB_ROOT, NULL, NULL, 0, 0 }) ++ ++#define RQ_BIC(rq) ((struct bfq_io_cq *) (rq)->elv.priv[0]) ++#define RQ_BFQQ(rq) ((rq)->elv.priv[1]) ++ ++static void bfq_schedule_dispatch(struct bfq_data *bfqd); ++ ++#include "bfq-ioc.c" ++#include "bfq-sched.c" ++#include "bfq-cgroup-included.c" ++ ++#define bfq_class_idle(bfqq) ((bfqq)->ioprio_class == IOPRIO_CLASS_IDLE) ++#define bfq_class_rt(bfqq) ((bfqq)->ioprio_class == IOPRIO_CLASS_RT) ++ ++#define bfq_sample_valid(samples) ((samples) > 80) ++ ++/* ++ * Scheduler run of queue, if there are requests pending and no one in the ++ * driver that will restart queueing. ++ */ ++static void bfq_schedule_dispatch(struct bfq_data *bfqd) ++{ ++ if (bfqd->queued != 0) { ++ bfq_log(bfqd, "schedule dispatch"); ++ kblockd_schedule_work(&bfqd->unplug_work); ++ } ++} ++ ++/* ++ * Lifted from AS - choose which of rq1 and rq2 that is best served now. ++ * We choose the request that is closesr to the head right now. Distance ++ * behind the head is penalized and only allowed to a certain extent. ++ */ ++static struct request *bfq_choose_req(struct bfq_data *bfqd, ++ struct request *rq1, ++ struct request *rq2, ++ sector_t last) ++{ ++ sector_t s1, s2, d1 = 0, d2 = 0; ++ unsigned long back_max; ++#define BFQ_RQ1_WRAP 0x01 /* request 1 wraps */ ++#define BFQ_RQ2_WRAP 0x02 /* request 2 wraps */ ++ unsigned int wrap = 0; /* bit mask: requests behind the disk head? */ ++ ++ if (!rq1 || rq1 == rq2) ++ return rq2; ++ if (!rq2) ++ return rq1; ++ ++ if (rq_is_sync(rq1) && !rq_is_sync(rq2)) ++ return rq1; ++ else if (rq_is_sync(rq2) && !rq_is_sync(rq1)) ++ return rq2; ++ if ((rq1->cmd_flags & REQ_META) && !(rq2->cmd_flags & REQ_META)) ++ return rq1; ++ else if ((rq2->cmd_flags & REQ_META) && !(rq1->cmd_flags & REQ_META)) ++ return rq2; ++ ++ s1 = blk_rq_pos(rq1); ++ s2 = blk_rq_pos(rq2); ++ ++ /* ++ * By definition, 1KiB is 2 sectors. ++ */ ++ back_max = bfqd->bfq_back_max * 2; ++ ++ /* ++ * Strict one way elevator _except_ in the case where we allow ++ * short backward seeks which are biased as twice the cost of a ++ * similar forward seek. ++ */ ++ if (s1 >= last) ++ d1 = s1 - last; ++ else if (s1 + back_max >= last) ++ d1 = (last - s1) * bfqd->bfq_back_penalty; ++ else ++ wrap |= BFQ_RQ1_WRAP; ++ ++ if (s2 >= last) ++ d2 = s2 - last; ++ else if (s2 + back_max >= last) ++ d2 = (last - s2) * bfqd->bfq_back_penalty; ++ else ++ wrap |= BFQ_RQ2_WRAP; ++ ++ /* Found required data */ ++ ++ /* ++ * By doing switch() on the bit mask "wrap" we avoid having to ++ * check two variables for all permutations: --> faster! ++ */ ++ switch (wrap) { ++ case 0: /* common case for CFQ: rq1 and rq2 not wrapped */ ++ if (d1 < d2) ++ return rq1; ++ else if (d2 < d1) ++ return rq2; ++ ++ if (s1 >= s2) ++ return rq1; ++ else ++ return rq2; ++ ++ case BFQ_RQ2_WRAP: ++ return rq1; ++ case BFQ_RQ1_WRAP: ++ return rq2; ++ case (BFQ_RQ1_WRAP|BFQ_RQ2_WRAP): /* both rqs wrapped */ ++ default: ++ /* ++ * Since both rqs are wrapped, ++ * start with the one that's further behind head ++ * (--> only *one* back seek required), ++ * since back seek takes more time than forward. ++ */ ++ if (s1 <= s2) ++ return rq1; ++ else ++ return rq2; ++ } ++} ++ ++static struct bfq_queue * ++bfq_rq_pos_tree_lookup(struct bfq_data *bfqd, struct rb_root *root, ++ sector_t sector, struct rb_node **ret_parent, ++ struct rb_node ***rb_link) ++{ ++ struct rb_node **p, *parent; ++ struct bfq_queue *bfqq = NULL; ++ ++ parent = NULL; ++ p = &root->rb_node; ++ while (*p) { ++ struct rb_node **n; ++ ++ parent = *p; ++ bfqq = rb_entry(parent, struct bfq_queue, pos_node); ++ ++ /* ++ * Sort strictly based on sector. Smallest to the left, ++ * largest to the right. ++ */ ++ if (sector > blk_rq_pos(bfqq->next_rq)) ++ n = &(*p)->rb_right; ++ else if (sector < blk_rq_pos(bfqq->next_rq)) ++ n = &(*p)->rb_left; ++ else ++ break; ++ p = n; ++ bfqq = NULL; ++ } ++ ++ *ret_parent = parent; ++ if (rb_link) ++ *rb_link = p; ++ ++ bfq_log(bfqd, "rq_pos_tree_lookup %llu: returning %d", ++ (unsigned long long) sector, ++ bfqq ? bfqq->pid : 0); ++ ++ return bfqq; ++} ++ ++static void bfq_pos_tree_add_move(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ struct rb_node **p, *parent; ++ struct bfq_queue *__bfqq; ++ ++ if (bfqq->pos_root) { ++ rb_erase(&bfqq->pos_node, bfqq->pos_root); ++ bfqq->pos_root = NULL; ++ } ++ ++ if (bfq_class_idle(bfqq)) ++ return; ++ if (!bfqq->next_rq) ++ return; ++ ++ bfqq->pos_root = &bfq_bfqq_to_bfqg(bfqq)->rq_pos_tree; ++ __bfqq = bfq_rq_pos_tree_lookup(bfqd, bfqq->pos_root, ++ blk_rq_pos(bfqq->next_rq), &parent, &p); ++ if (!__bfqq) { ++ rb_link_node(&bfqq->pos_node, parent, p); ++ rb_insert_color(&bfqq->pos_node, bfqq->pos_root); ++ } else ++ bfqq->pos_root = NULL; ++} ++ ++/* ++ * Tell whether there are active queues or groups with differentiated weights. ++ */ ++static bool bfq_differentiated_weights(struct bfq_data *bfqd) ++{ ++ /* ++ * For weights to differ, at least one of the trees must contain ++ * at least two nodes. ++ */ ++ return (!RB_EMPTY_ROOT(&bfqd->queue_weights_tree) && ++ (bfqd->queue_weights_tree.rb_node->rb_left || ++ bfqd->queue_weights_tree.rb_node->rb_right) ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ ) || ++ (!RB_EMPTY_ROOT(&bfqd->group_weights_tree) && ++ (bfqd->group_weights_tree.rb_node->rb_left || ++ bfqd->group_weights_tree.rb_node->rb_right) ++#endif ++ ); ++} ++ ++/* ++ * The following function returns true if every queue must receive the ++ * same share of the throughput (this condition is used when deciding ++ * whether idling may be disabled, see the comments in the function ++ * bfq_bfqq_may_idle()). ++ * ++ * Such a scenario occurs when: ++ * 1) all active queues have the same weight, ++ * 2) all active groups at the same level in the groups tree have the same ++ * weight, ++ * 3) all active groups at the same level in the groups tree have the same ++ * number of children. ++ * ++ * Unfortunately, keeping the necessary state for evaluating exactly the ++ * above symmetry conditions would be quite complex and time-consuming. ++ * Therefore this function evaluates, instead, the following stronger ++ * sub-conditions, for which it is much easier to maintain the needed ++ * state: ++ * 1) all active queues have the same weight, ++ * 2) all active groups have the same weight, ++ * 3) all active groups have at most one active child each. ++ * In particular, the last two conditions are always true if hierarchical ++ * support and the cgroups interface are not enabled, thus no state needs ++ * to be maintained in this case. ++ */ ++static bool bfq_symmetric_scenario(struct bfq_data *bfqd) ++{ ++ return !bfq_differentiated_weights(bfqd); ++} ++ ++/* ++ * If the weight-counter tree passed as input contains no counter for ++ * the weight of the input entity, then add that counter; otherwise just ++ * increment the existing counter. ++ * ++ * Note that weight-counter trees contain few nodes in mostly symmetric ++ * scenarios. For example, if all queues have the same weight, then the ++ * weight-counter tree for the queues may contain at most one node. ++ * This holds even if low_latency is on, because weight-raised queues ++ * are not inserted in the tree. ++ * In most scenarios, the rate at which nodes are created/destroyed ++ * should be low too. ++ */ ++static void bfq_weights_tree_add(struct bfq_data *bfqd, ++ struct bfq_entity *entity, ++ struct rb_root *root) ++{ ++ struct rb_node **new = &(root->rb_node), *parent = NULL; ++ ++ /* ++ * Do not insert if the entity is already associated with a ++ * counter, which happens if: ++ * 1) the entity is associated with a queue, ++ * 2) a request arrival has caused the queue to become both ++ * non-weight-raised, and hence change its weight, and ++ * backlogged; in this respect, each of the two events ++ * causes an invocation of this function, ++ * 3) this is the invocation of this function caused by the ++ * second event. This second invocation is actually useless, ++ * and we handle this fact by exiting immediately. More ++ * efficient or clearer solutions might possibly be adopted. ++ */ ++ if (entity->weight_counter) ++ return; ++ ++ while (*new) { ++ struct bfq_weight_counter *__counter = container_of(*new, ++ struct bfq_weight_counter, ++ weights_node); ++ parent = *new; ++ ++ if (entity->weight == __counter->weight) { ++ entity->weight_counter = __counter; ++ goto inc_counter; ++ } ++ if (entity->weight < __counter->weight) ++ new = &((*new)->rb_left); ++ else ++ new = &((*new)->rb_right); ++ } ++ ++ entity->weight_counter = kzalloc(sizeof(struct bfq_weight_counter), ++ GFP_ATOMIC); ++ ++ /* ++ * In the unlucky event of an allocation failure, we just ++ * exit. This will cause the weight of entity to not be ++ * considered in bfq_differentiated_weights, which, in its ++ * turn, causes the scenario to be deemed wrongly symmetric in ++ * case entity's weight would have been the only weight making ++ * the scenario asymmetric. On the bright side, no unbalance ++ * will however occur when entity becomes inactive again (the ++ * invocation of this function is triggered by an activation ++ * of entity). In fact, bfq_weights_tree_remove does nothing ++ * if !entity->weight_counter. ++ */ ++ if (unlikely(!entity->weight_counter)) ++ return; ++ ++ entity->weight_counter->weight = entity->weight; ++ rb_link_node(&entity->weight_counter->weights_node, parent, new); ++ rb_insert_color(&entity->weight_counter->weights_node, root); ++ ++inc_counter: ++ entity->weight_counter->num_active++; ++} ++ ++/* ++ * Decrement the weight counter associated with the entity, and, if the ++ * counter reaches 0, remove the counter from the tree. ++ * See the comments to the function bfq_weights_tree_add() for considerations ++ * about overhead. ++ */ ++static void bfq_weights_tree_remove(struct bfq_data *bfqd, ++ struct bfq_entity *entity, ++ struct rb_root *root) ++{ ++ if (!entity->weight_counter) ++ return; ++ ++ BUG_ON(RB_EMPTY_ROOT(root)); ++ BUG_ON(entity->weight_counter->weight != entity->weight); ++ ++ BUG_ON(!entity->weight_counter->num_active); ++ entity->weight_counter->num_active--; ++ if (entity->weight_counter->num_active > 0) ++ goto reset_entity_pointer; ++ ++ rb_erase(&entity->weight_counter->weights_node, root); ++ kfree(entity->weight_counter); ++ ++reset_entity_pointer: ++ entity->weight_counter = NULL; ++} ++ ++/* ++ * Return expired entry, or NULL to just start from scratch in rbtree. ++ */ ++static struct request *bfq_check_fifo(struct bfq_queue *bfqq, ++ struct request *last) ++{ ++ struct request *rq; ++ ++ if (bfq_bfqq_fifo_expire(bfqq)) ++ return NULL; ++ ++ bfq_mark_bfqq_fifo_expire(bfqq); ++ ++ rq = rq_entry_fifo(bfqq->fifo.next); ++ ++ if (rq == last || ktime_get_ns() < rq->fifo_time) ++ return NULL; ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "check_fifo: returned %p", rq); ++ BUG_ON(RB_EMPTY_NODE(&rq->rb_node)); ++ return rq; ++} ++ ++static struct request *bfq_find_next_rq(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ struct request *last) ++{ ++ struct rb_node *rbnext = rb_next(&last->rb_node); ++ struct rb_node *rbprev = rb_prev(&last->rb_node); ++ struct request *next, *prev = NULL; ++ ++ BUG_ON(list_empty(&bfqq->fifo)); ++ ++ /* Follow expired path, else get first next available. */ ++ next = bfq_check_fifo(bfqq, last); ++ if (next) { ++ BUG_ON(next == last); ++ return next; ++ } ++ ++ BUG_ON(RB_EMPTY_NODE(&last->rb_node)); ++ ++ if (rbprev) ++ prev = rb_entry_rq(rbprev); ++ ++ if (rbnext) ++ next = rb_entry_rq(rbnext); ++ else { ++ rbnext = rb_first(&bfqq->sort_list); ++ if (rbnext && rbnext != &last->rb_node) ++ next = rb_entry_rq(rbnext); ++ } ++ ++ return bfq_choose_req(bfqd, next, prev, blk_rq_pos(last)); ++} ++ ++/* see the definition of bfq_async_charge_factor for details */ ++static unsigned long bfq_serv_to_charge(struct request *rq, ++ struct bfq_queue *bfqq) ++{ ++ if (bfq_bfqq_sync(bfqq) || bfqq->wr_coeff > 1) ++ return blk_rq_sectors(rq); ++ ++ /* ++ * If there are no weight-raised queues, then amplify service ++ * by just the async charge factor; otherwise amplify service ++ * by twice the async charge factor, to further reduce latency ++ * for weight-raised queues. ++ */ ++ if (bfqq->bfqd->wr_busy_queues == 0) ++ return blk_rq_sectors(rq) * bfq_async_charge_factor; ++ ++ return blk_rq_sectors(rq) * 2 * bfq_async_charge_factor; ++} ++ ++/** ++ * bfq_updated_next_req - update the queue after a new next_rq selection. ++ * @bfqd: the device data the queue belongs to. ++ * @bfqq: the queue to update. ++ * ++ * If the first request of a queue changes we make sure that the queue ++ * has enough budget to serve at least its first request (if the ++ * request has grown). We do this because if the queue has not enough ++ * budget for its first request, it has to go through two dispatch ++ * rounds to actually get it dispatched. ++ */ ++static void bfq_updated_next_req(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ struct bfq_service_tree *st = bfq_entity_service_tree(entity); ++ struct request *next_rq = bfqq->next_rq; ++ unsigned long new_budget; ++ ++ if (!next_rq) ++ return; ++ ++ if (bfqq == bfqd->in_service_queue) ++ /* ++ * In order not to break guarantees, budgets cannot be ++ * changed after an entity has been selected. ++ */ ++ return; ++ ++ BUG_ON(entity->tree != &st->active); ++ BUG_ON(entity == entity->sched_data->in_service_entity); ++ ++ new_budget = max_t(unsigned long, bfqq->max_budget, ++ bfq_serv_to_charge(next_rq, bfqq)); ++ if (entity->budget != new_budget) { ++ entity->budget = new_budget; ++ bfq_log_bfqq(bfqd, bfqq, "updated next rq: new budget %lu", ++ new_budget); ++ bfq_requeue_bfqq(bfqd, bfqq); ++ } ++} ++ ++static unsigned int bfq_wr_duration(struct bfq_data *bfqd) ++{ ++ u64 dur; ++ ++ if (bfqd->bfq_wr_max_time > 0) ++ return bfqd->bfq_wr_max_time; ++ ++ dur = bfqd->RT_prod; ++ do_div(dur, bfqd->peak_rate); ++ ++ /* ++ * Limit duration between 3 and 13 seconds. Tests show that ++ * higher values than 13 seconds often yield the opposite of ++ * the desired result, i.e., worsen responsiveness by letting ++ * non-interactive and non-soft-real-time applications ++ * preserve weight raising for a too long time interval. ++ * ++ * On the other end, lower values than 3 seconds make it ++ * difficult for most interactive tasks to complete their jobs ++ * before weight-raising finishes. ++ */ ++ if (dur > msecs_to_jiffies(13000)) ++ dur = msecs_to_jiffies(13000); ++ else if (dur < msecs_to_jiffies(3000)) ++ dur = msecs_to_jiffies(3000); ++ ++ return dur; ++} ++ ++static void ++bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, ++ struct bfq_io_cq *bic, bool bfq_already_existing) ++{ ++ unsigned int old_wr_coeff; ++ bool busy = bfq_already_existing && bfq_bfqq_busy(bfqq); ++ ++ if (bic->saved_idle_window) ++ bfq_mark_bfqq_idle_window(bfqq); ++ else ++ bfq_clear_bfqq_idle_window(bfqq); ++ ++ if (bic->saved_IO_bound) ++ bfq_mark_bfqq_IO_bound(bfqq); ++ else ++ bfq_clear_bfqq_IO_bound(bfqq); ++ ++ if (unlikely(busy)) ++ old_wr_coeff = bfqq->wr_coeff; ++ ++ bfqq->wr_coeff = bic->saved_wr_coeff; ++ bfqq->wr_start_at_switch_to_srt = bic->saved_wr_start_at_switch_to_srt; ++ BUG_ON(time_is_after_jiffies(bfqq->wr_start_at_switch_to_srt)); ++ bfqq->last_wr_start_finish = bic->saved_last_wr_start_finish; ++ bfqq->wr_cur_max_time = bic->saved_wr_cur_max_time; ++ BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); ++ ++ if (bfqq->wr_coeff > 1 && (bfq_bfqq_in_large_burst(bfqq) || ++ time_is_before_jiffies(bfqq->last_wr_start_finish + ++ bfqq->wr_cur_max_time))) { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "resume state: switching off wr (%lu + %lu < %lu)", ++ bfqq->last_wr_start_finish, bfqq->wr_cur_max_time, ++ jiffies); ++ ++ bfqq->wr_coeff = 1; ++ } ++ ++ /* make sure weight will be updated, however we got here */ ++ bfqq->entity.prio_changed = 1; ++ ++ if (likely(!busy)) ++ return; ++ ++ if (old_wr_coeff == 1 && bfqq->wr_coeff > 1) { ++ bfqd->wr_busy_queues++; ++ BUG_ON(bfqd->wr_busy_queues > bfqd->busy_queues); ++ } else if (old_wr_coeff > 1 && bfqq->wr_coeff == 1) { ++ bfqd->wr_busy_queues--; ++ BUG_ON(bfqd->wr_busy_queues < 0); ++ } ++} ++ ++static int bfqq_process_refs(struct bfq_queue *bfqq) ++{ ++ int process_refs, io_refs; ++ ++ lockdep_assert_held(bfqq->bfqd->queue->queue_lock); ++ ++ io_refs = bfqq->allocated[READ] + bfqq->allocated[WRITE]; ++ process_refs = bfqq->ref - io_refs - bfqq->entity.on_st; ++ BUG_ON(process_refs < 0); ++ return process_refs; ++} ++ ++/* Empty burst list and add just bfqq (see comments to bfq_handle_burst) */ ++static void bfq_reset_burst_list(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ struct bfq_queue *item; ++ struct hlist_node *n; ++ ++ hlist_for_each_entry_safe(item, n, &bfqd->burst_list, burst_list_node) ++ hlist_del_init(&item->burst_list_node); ++ hlist_add_head(&bfqq->burst_list_node, &bfqd->burst_list); ++ bfqd->burst_size = 1; ++ bfqd->burst_parent_entity = bfqq->entity.parent; ++} ++ ++/* Add bfqq to the list of queues in current burst (see bfq_handle_burst) */ ++static void bfq_add_to_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ /* Increment burst size to take into account also bfqq */ ++ bfqd->burst_size++; ++ ++ bfq_log_bfqq(bfqd, bfqq, "add_to_burst %d", bfqd->burst_size); ++ ++ BUG_ON(bfqd->burst_size > bfqd->bfq_large_burst_thresh); ++ ++ if (bfqd->burst_size == bfqd->bfq_large_burst_thresh) { ++ struct bfq_queue *pos, *bfqq_item; ++ struct hlist_node *n; ++ ++ /* ++ * Enough queues have been activated shortly after each ++ * other to consider this burst as large. ++ */ ++ bfqd->large_burst = true; ++ bfq_log_bfqq(bfqd, bfqq, "add_to_burst: large burst started"); ++ ++ /* ++ * We can now mark all queues in the burst list as ++ * belonging to a large burst. ++ */ ++ hlist_for_each_entry(bfqq_item, &bfqd->burst_list, ++ burst_list_node) { ++ bfq_mark_bfqq_in_large_burst(bfqq_item); ++ bfq_log_bfqq(bfqd, bfqq_item, "marked in large burst"); ++ } ++ bfq_mark_bfqq_in_large_burst(bfqq); ++ bfq_log_bfqq(bfqd, bfqq, "marked in large burst"); ++ ++ /* ++ * From now on, and until the current burst finishes, any ++ * new queue being activated shortly after the last queue ++ * was inserted in the burst can be immediately marked as ++ * belonging to a large burst. So the burst list is not ++ * needed any more. Remove it. ++ */ ++ hlist_for_each_entry_safe(pos, n, &bfqd->burst_list, ++ burst_list_node) ++ hlist_del_init(&pos->burst_list_node); ++ } else /* ++ * Burst not yet large: add bfqq to the burst list. Do ++ * not increment the ref counter for bfqq, because bfqq ++ * is removed from the burst list before freeing bfqq ++ * in put_queue. ++ */ ++ hlist_add_head(&bfqq->burst_list_node, &bfqd->burst_list); ++} ++ ++/* ++ * If many queues belonging to the same group happen to be created ++ * shortly after each other, then the processes associated with these ++ * queues have typically a common goal. In particular, bursts of queue ++ * creations are usually caused by services or applications that spawn ++ * many parallel threads/processes. Examples are systemd during boot, ++ * or git grep. To help these processes get their job done as soon as ++ * possible, it is usually better to not grant either weight-raising ++ * or device idling to their queues. ++ * ++ * In this comment we describe, firstly, the reasons why this fact ++ * holds, and, secondly, the next function, which implements the main ++ * steps needed to properly mark these queues so that they can then be ++ * treated in a different way. ++ * ++ * The above services or applications benefit mostly from a high ++ * throughput: the quicker the requests of the activated queues are ++ * cumulatively served, the sooner the target job of these queues gets ++ * completed. As a consequence, weight-raising any of these queues, ++ * which also implies idling the device for it, is almost always ++ * counterproductive. In most cases it just lowers throughput. ++ * ++ * On the other hand, a burst of queue creations may be caused also by ++ * the start of an application that does not consist of a lot of ++ * parallel I/O-bound threads. In fact, with a complex application, ++ * several short processes may need to be executed to start-up the ++ * application. In this respect, to start an application as quickly as ++ * possible, the best thing to do is in any case to privilege the I/O ++ * related to the application with respect to all other ++ * I/O. Therefore, the best strategy to start as quickly as possible ++ * an application that causes a burst of queue creations is to ++ * weight-raise all the queues created during the burst. This is the ++ * exact opposite of the best strategy for the other type of bursts. ++ * ++ * In the end, to take the best action for each of the two cases, the ++ * two types of bursts need to be distinguished. Fortunately, this ++ * seems relatively easy, by looking at the sizes of the bursts. In ++ * particular, we found a threshold such that only bursts with a ++ * larger size than that threshold are apparently caused by ++ * services or commands such as systemd or git grep. For brevity, ++ * hereafter we call just 'large' these bursts. BFQ *does not* ++ * weight-raise queues whose creation occurs in a large burst. In ++ * addition, for each of these queues BFQ performs or does not perform ++ * idling depending on which choice boosts the throughput more. The ++ * exact choice depends on the device and request pattern at ++ * hand. ++ * ++ * Unfortunately, false positives may occur while an interactive task ++ * is starting (e.g., an application is being started). The ++ * consequence is that the queues associated with the task do not ++ * enjoy weight raising as expected. Fortunately these false positives ++ * are very rare. They typically occur if some service happens to ++ * start doing I/O exactly when the interactive task starts. ++ * ++ * Turning back to the next function, it implements all the steps ++ * needed to detect the occurrence of a large burst and to properly ++ * mark all the queues belonging to it (so that they can then be ++ * treated in a different way). This goal is achieved by maintaining a ++ * "burst list" that holds, temporarily, the queues that belong to the ++ * burst in progress. The list is then used to mark these queues as ++ * belonging to a large burst if the burst does become large. The main ++ * steps are the following. ++ * ++ * . when the very first queue is created, the queue is inserted into the ++ * list (as it could be the first queue in a possible burst) ++ * ++ * . if the current burst has not yet become large, and a queue Q that does ++ * not yet belong to the burst is activated shortly after the last time ++ * at which a new queue entered the burst list, then the function appends ++ * Q to the burst list ++ * ++ * . if, as a consequence of the previous step, the burst size reaches ++ * the large-burst threshold, then ++ * ++ * . all the queues in the burst list are marked as belonging to a ++ * large burst ++ * ++ * . the burst list is deleted; in fact, the burst list already served ++ * its purpose (keeping temporarily track of the queues in a burst, ++ * so as to be able to mark them as belonging to a large burst in the ++ * previous sub-step), and now is not needed any more ++ * ++ * . the device enters a large-burst mode ++ * ++ * . if a queue Q that does not belong to the burst is created while ++ * the device is in large-burst mode and shortly after the last time ++ * at which a queue either entered the burst list or was marked as ++ * belonging to the current large burst, then Q is immediately marked ++ * as belonging to a large burst. ++ * ++ * . if a queue Q that does not belong to the burst is created a while ++ * later, i.e., not shortly after, than the last time at which a queue ++ * either entered the burst list or was marked as belonging to the ++ * current large burst, then the current burst is deemed as finished and: ++ * ++ * . the large-burst mode is reset if set ++ * ++ * . the burst list is emptied ++ * ++ * . Q is inserted in the burst list, as Q may be the first queue ++ * in a possible new burst (then the burst list contains just Q ++ * after this step). ++ */ ++static void bfq_handle_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ /* ++ * If bfqq is already in the burst list or is part of a large ++ * burst, or finally has just been split, then there is ++ * nothing else to do. ++ */ ++ if (!hlist_unhashed(&bfqq->burst_list_node) || ++ bfq_bfqq_in_large_burst(bfqq) || ++ time_is_after_eq_jiffies(bfqq->split_time + ++ msecs_to_jiffies(10))) ++ return; ++ ++ /* ++ * If bfqq's creation happens late enough, or bfqq belongs to ++ * a different group than the burst group, then the current ++ * burst is finished, and related data structures must be ++ * reset. ++ * ++ * In this respect, consider the special case where bfqq is ++ * the very first queue created after BFQ is selected for this ++ * device. In this case, last_ins_in_burst and ++ * burst_parent_entity are not yet significant when we get ++ * here. But it is easy to verify that, whether or not the ++ * following condition is true, bfqq will end up being ++ * inserted into the burst list. In particular the list will ++ * happen to contain only bfqq. And this is exactly what has ++ * to happen, as bfqq may be the first queue of the first ++ * burst. ++ */ ++ if (time_is_before_jiffies(bfqd->last_ins_in_burst + ++ bfqd->bfq_burst_interval) || ++ bfqq->entity.parent != bfqd->burst_parent_entity) { ++ bfqd->large_burst = false; ++ bfq_reset_burst_list(bfqd, bfqq); ++ bfq_log_bfqq(bfqd, bfqq, ++ "handle_burst: late activation or different group"); ++ goto end; ++ } ++ ++ /* ++ * If we get here, then bfqq is being activated shortly after the ++ * last queue. So, if the current burst is also large, we can mark ++ * bfqq as belonging to this large burst immediately. ++ */ ++ if (bfqd->large_burst) { ++ bfq_log_bfqq(bfqd, bfqq, "handle_burst: marked in burst"); ++ bfq_mark_bfqq_in_large_burst(bfqq); ++ goto end; ++ } ++ ++ /* ++ * If we get here, then a large-burst state has not yet been ++ * reached, but bfqq is being activated shortly after the last ++ * queue. Then we add bfqq to the burst. ++ */ ++ bfq_add_to_burst(bfqd, bfqq); ++end: ++ /* ++ * At this point, bfqq either has been added to the current ++ * burst or has caused the current burst to terminate and a ++ * possible new burst to start. In particular, in the second ++ * case, bfqq has become the first queue in the possible new ++ * burst. In both cases last_ins_in_burst needs to be moved ++ * forward. ++ */ ++ bfqd->last_ins_in_burst = jiffies; ++ ++} ++ ++static int bfq_bfqq_budget_left(struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ ++ return entity->budget - entity->service; ++} ++ ++/* ++ * If enough samples have been computed, return the current max budget ++ * stored in bfqd, which is dynamically updated according to the ++ * estimated disk peak rate; otherwise return the default max budget ++ */ ++static int bfq_max_budget(struct bfq_data *bfqd) ++{ ++ if (bfqd->budgets_assigned < bfq_stats_min_budgets) ++ return bfq_default_max_budget; ++ else ++ return bfqd->bfq_max_budget; ++} ++ ++/* ++ * Return min budget, which is a fraction of the current or default ++ * max budget (trying with 1/32) ++ */ ++static int bfq_min_budget(struct bfq_data *bfqd) ++{ ++ if (bfqd->budgets_assigned < bfq_stats_min_budgets) ++ return bfq_default_max_budget / 32; ++ else ++ return bfqd->bfq_max_budget / 32; ++} ++ ++static void bfq_bfqq_expire(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ bool compensate, ++ enum bfqq_expiration reason); ++ ++/* ++ * The next function, invoked after the input queue bfqq switches from ++ * idle to busy, updates the budget of bfqq. The function also tells ++ * whether the in-service queue should be expired, by returning ++ * true. The purpose of expiring the in-service queue is to give bfqq ++ * the chance to possibly preempt the in-service queue, and the reason ++ * for preempting the in-service queue is to achieve one of the two ++ * goals below. ++ * ++ * 1. Guarantee to bfqq its reserved bandwidth even if bfqq has ++ * expired because it has remained idle. In particular, bfqq may have ++ * expired for one of the following two reasons: ++ * ++ * - BFQ_BFQQ_NO_MORE_REQUEST bfqq did not enjoy any device idling and ++ * did not make it to issue a new request before its last request ++ * was served; ++ * ++ * - BFQ_BFQQ_TOO_IDLE bfqq did enjoy device idling, but did not issue ++ * a new request before the expiration of the idling-time. ++ * ++ * Even if bfqq has expired for one of the above reasons, the process ++ * associated with the queue may be however issuing requests greedily, ++ * and thus be sensitive to the bandwidth it receives (bfqq may have ++ * remained idle for other reasons: CPU high load, bfqq not enjoying ++ * idling, I/O throttling somewhere in the path from the process to ++ * the I/O scheduler, ...). But if, after every expiration for one of ++ * the above two reasons, bfqq has to wait for the service of at least ++ * one full budget of another queue before being served again, then ++ * bfqq is likely to get a much lower bandwidth or resource time than ++ * its reserved ones. To address this issue, two countermeasures need ++ * to be taken. ++ * ++ * First, the budget and the timestamps of bfqq need to be updated in ++ * a special way on bfqq reactivation: they need to be updated as if ++ * bfqq did not remain idle and did not expire. In fact, if they are ++ * computed as if bfqq expired and remained idle until reactivation, ++ * then the process associated with bfqq is treated as if, instead of ++ * being greedy, it stopped issuing requests when bfqq remained idle, ++ * and restarts issuing requests only on this reactivation. In other ++ * words, the scheduler does not help the process recover the "service ++ * hole" between bfqq expiration and reactivation. As a consequence, ++ * the process receives a lower bandwidth than its reserved one. In ++ * contrast, to recover this hole, the budget must be updated as if ++ * bfqq was not expired at all before this reactivation, i.e., it must ++ * be set to the value of the remaining budget when bfqq was ++ * expired. Along the same line, timestamps need to be assigned the ++ * value they had the last time bfqq was selected for service, i.e., ++ * before last expiration. Thus timestamps need to be back-shifted ++ * with respect to their normal computation (see [1] for more details ++ * on this tricky aspect). ++ * ++ * Secondly, to allow the process to recover the hole, the in-service ++ * queue must be expired too, to give bfqq the chance to preempt it ++ * immediately. In fact, if bfqq has to wait for a full budget of the ++ * in-service queue to be completed, then it may become impossible to ++ * let the process recover the hole, even if the back-shifted ++ * timestamps of bfqq are lower than those of the in-service queue. If ++ * this happens for most or all of the holes, then the process may not ++ * receive its reserved bandwidth. In this respect, it is worth noting ++ * that, being the service of outstanding requests unpreemptible, a ++ * little fraction of the holes may however be unrecoverable, thereby ++ * causing a little loss of bandwidth. ++ * ++ * The last important point is detecting whether bfqq does need this ++ * bandwidth recovery. In this respect, the next function deems the ++ * process associated with bfqq greedy, and thus allows it to recover ++ * the hole, if: 1) the process is waiting for the arrival of a new ++ * request (which implies that bfqq expired for one of the above two ++ * reasons), and 2) such a request has arrived soon. The first ++ * condition is controlled through the flag non_blocking_wait_rq, ++ * while the second through the flag arrived_in_time. If both ++ * conditions hold, then the function computes the budget in the ++ * above-described special way, and signals that the in-service queue ++ * should be expired. Timestamp back-shifting is done later in ++ * __bfq_activate_entity. ++ * ++ * 2. Reduce latency. Even if timestamps are not backshifted to let ++ * the process associated with bfqq recover a service hole, bfqq may ++ * however happen to have, after being (re)activated, a lower finish ++ * timestamp than the in-service queue. That is, the next budget of ++ * bfqq may have to be completed before the one of the in-service ++ * queue. If this is the case, then preempting the in-service queue ++ * allows this goal to be achieved, apart from the unpreemptible, ++ * outstanding requests mentioned above. ++ * ++ * Unfortunately, regardless of which of the above two goals one wants ++ * to achieve, service trees need first to be updated to know whether ++ * the in-service queue must be preempted. To have service trees ++ * correctly updated, the in-service queue must be expired and ++ * rescheduled, and bfqq must be scheduled too. This is one of the ++ * most costly operations (in future versions, the scheduling ++ * mechanism may be re-designed in such a way to make it possible to ++ * know whether preemption is needed without needing to update service ++ * trees). In addition, queue preemptions almost always cause random ++ * I/O, and thus loss of throughput. Because of these facts, the next ++ * function adopts the following simple scheme to avoid both costly ++ * operations and too frequent preemptions: it requests the expiration ++ * of the in-service queue (unconditionally) only for queues that need ++ * to recover a hole, or that either are weight-raised or deserve to ++ * be weight-raised. ++ */ ++static bool bfq_bfqq_update_budg_for_activation(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ bool arrived_in_time, ++ bool wr_or_deserves_wr) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ ++ if (bfq_bfqq_non_blocking_wait_rq(bfqq) && arrived_in_time) { ++ /* ++ * We do not clear the flag non_blocking_wait_rq here, as ++ * the latter is used in bfq_activate_bfqq to signal ++ * that timestamps need to be back-shifted (and is ++ * cleared right after). ++ */ ++ ++ /* ++ * In next assignment we rely on that either ++ * entity->service or entity->budget are not updated ++ * on expiration if bfqq is empty (see ++ * __bfq_bfqq_recalc_budget). Thus both quantities ++ * remain unchanged after such an expiration, and the ++ * following statement therefore assigns to ++ * entity->budget the remaining budget on such an ++ * expiration. For clarity, entity->service is not ++ * updated on expiration in any case, and, in normal ++ * operation, is reset only when bfqq is selected for ++ * service (see bfq_get_next_queue). ++ */ ++ BUG_ON(bfqq->max_budget < 0); ++ entity->budget = min_t(unsigned long, ++ bfq_bfqq_budget_left(bfqq), ++ bfqq->max_budget); ++ ++ BUG_ON(entity->budget < 0); ++ return true; ++ } ++ ++ BUG_ON(bfqq->max_budget < 0); ++ entity->budget = max_t(unsigned long, bfqq->max_budget, ++ bfq_serv_to_charge(bfqq->next_rq, bfqq)); ++ BUG_ON(entity->budget < 0); ++ ++ bfq_clear_bfqq_non_blocking_wait_rq(bfqq); ++ return wr_or_deserves_wr; ++} ++ ++static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ unsigned int old_wr_coeff, ++ bool wr_or_deserves_wr, ++ bool interactive, ++ bool in_burst, ++ bool soft_rt) ++{ ++ if (old_wr_coeff == 1 && wr_or_deserves_wr) { ++ /* start a weight-raising period */ ++ if (interactive) { ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ } else { ++ bfqq->wr_start_at_switch_to_srt = jiffies; ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff * ++ BFQ_SOFTRT_WEIGHT_FACTOR; ++ bfqq->wr_cur_max_time = ++ bfqd->bfq_wr_rt_max_time; ++ } ++ /* ++ * If needed, further reduce budget to make sure it is ++ * close to bfqq's backlog, so as to reduce the ++ * scheduling-error component due to a too large ++ * budget. Do not care about throughput consequences, ++ * but only about latency. Finally, do not assign a ++ * too small budget either, to avoid increasing ++ * latency by causing too frequent expirations. ++ */ ++ bfqq->entity.budget = min_t(unsigned long, ++ bfqq->entity.budget, ++ 2 * bfq_min_budget(bfqd)); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "wrais starting at %lu, rais_max_time %u", ++ jiffies, ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ } else if (old_wr_coeff > 1) { ++ if (interactive) { /* update wr coeff and duration */ ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ } else if (in_burst) { ++ bfqq->wr_coeff = 1; ++ bfq_log_bfqq(bfqd, bfqq, ++ "wrais ending at %lu, rais_max_time %u", ++ jiffies, ++ jiffies_to_msecs(bfqq-> ++ wr_cur_max_time)); ++ } else if (soft_rt) { ++ /* ++ * The application is now or still meeting the ++ * requirements for being deemed soft rt. We ++ * can then correctly and safely (re)charge ++ * the weight-raising duration for the ++ * application with the weight-raising ++ * duration for soft rt applications. ++ * ++ * In particular, doing this recharge now, i.e., ++ * before the weight-raising period for the ++ * application finishes, reduces the probability ++ * of the following negative scenario: ++ * 1) the weight of a soft rt application is ++ * raised at startup (as for any newly ++ * created application), ++ * 2) since the application is not interactive, ++ * at a certain time weight-raising is ++ * stopped for the application, ++ * 3) at that time the application happens to ++ * still have pending requests, and hence ++ * is destined to not have a chance to be ++ * deemed soft rt before these requests are ++ * completed (see the comments to the ++ * function bfq_bfqq_softrt_next_start() ++ * for details on soft rt detection), ++ * 4) these pending requests experience a high ++ * latency because the application is not ++ * weight-raised while they are pending. ++ */ ++ if (bfqq->wr_cur_max_time != ++ bfqd->bfq_wr_rt_max_time) { ++ bfqq->wr_start_at_switch_to_srt = ++ bfqq->last_wr_start_finish; ++ BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); ++ ++ bfqq->wr_cur_max_time = ++ bfqd->bfq_wr_rt_max_time; ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff * ++ BFQ_SOFTRT_WEIGHT_FACTOR; ++ bfq_log_bfqq(bfqd, bfqq, ++ "switching to soft_rt wr"); ++ } else ++ bfq_log_bfqq(bfqd, bfqq, ++ "moving forward soft_rt wr duration"); ++ bfqq->last_wr_start_finish = jiffies; ++ } ++ } ++} ++ ++static bool bfq_bfqq_idle_for_long_time(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ return bfqq->dispatched == 0 && ++ time_is_before_jiffies( ++ bfqq->budget_timeout + ++ bfqd->bfq_wr_min_idle_time); ++} ++ ++static void bfq_bfqq_handle_idle_busy_switch(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ int old_wr_coeff, ++ struct request *rq, ++ bool *interactive) ++{ ++ bool soft_rt, in_burst, wr_or_deserves_wr, ++ bfqq_wants_to_preempt, ++ idle_for_long_time = bfq_bfqq_idle_for_long_time(bfqd, bfqq), ++ /* ++ * See the comments on ++ * bfq_bfqq_update_budg_for_activation for ++ * details on the usage of the next variable. ++ */ ++ arrived_in_time = ktime_get_ns() <= ++ RQ_BIC(rq)->ttime.last_end_request + ++ bfqd->bfq_slice_idle * 3; ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "bfq_add_request non-busy: " ++ "jiffies %lu, in_time %d, idle_long %d busyw %d " ++ "wr_coeff %u", ++ jiffies, arrived_in_time, ++ idle_for_long_time, ++ bfq_bfqq_non_blocking_wait_rq(bfqq), ++ old_wr_coeff); ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ BUG_ON(bfqq == bfqd->in_service_queue); ++ bfqg_stats_update_io_add(bfqq_group(RQ_BFQQ(rq)), bfqq, rq->cmd_flags); ++ ++ /* ++ * bfqq deserves to be weight-raised if: ++ * - it is sync, ++ * - it does not belong to a large burst, ++ * - it has been idle for enough time or is soft real-time, ++ * - is linked to a bfq_io_cq (it is not shared in any sense) ++ */ ++ in_burst = bfq_bfqq_in_large_burst(bfqq); ++ soft_rt = bfqd->bfq_wr_max_softrt_rate > 0 && ++ !in_burst && ++ time_is_before_jiffies(bfqq->soft_rt_next_start); ++ *interactive = ++ !in_burst && ++ idle_for_long_time; ++ wr_or_deserves_wr = bfqd->low_latency && ++ (bfqq->wr_coeff > 1 || ++ (bfq_bfqq_sync(bfqq) && ++ bfqq->bic && (*interactive || soft_rt))); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "bfq_add_request: " ++ "in_burst %d, " ++ "soft_rt %d (next %lu), inter %d, bic %p", ++ bfq_bfqq_in_large_burst(bfqq), soft_rt, ++ bfqq->soft_rt_next_start, ++ *interactive, ++ bfqq->bic); ++ ++ /* ++ * Using the last flag, update budget and check whether bfqq ++ * may want to preempt the in-service queue. ++ */ ++ bfqq_wants_to_preempt = ++ bfq_bfqq_update_budg_for_activation(bfqd, bfqq, ++ arrived_in_time, ++ wr_or_deserves_wr); ++ ++ /* ++ * If bfqq happened to be activated in a burst, but has been ++ * idle for much more than an interactive queue, then we ++ * assume that, in the overall I/O initiated in the burst, the ++ * I/O associated with bfqq is finished. So bfqq does not need ++ * to be treated as a queue belonging to a burst ++ * anymore. Accordingly, we reset bfqq's in_large_burst flag ++ * if set, and remove bfqq from the burst list if it's ++ * there. We do not decrement burst_size, because the fact ++ * that bfqq does not need to belong to the burst list any ++ * more does not invalidate the fact that bfqq was created in ++ * a burst. ++ */ ++ if (likely(!bfq_bfqq_just_created(bfqq)) && ++ idle_for_long_time && ++ time_is_before_jiffies( ++ bfqq->budget_timeout + ++ msecs_to_jiffies(10000))) { ++ hlist_del_init(&bfqq->burst_list_node); ++ bfq_clear_bfqq_in_large_burst(bfqq); ++ } ++ ++ bfq_clear_bfqq_just_created(bfqq); ++ ++ if (!bfq_bfqq_IO_bound(bfqq)) { ++ if (arrived_in_time) { ++ bfqq->requests_within_timer++; ++ if (bfqq->requests_within_timer >= ++ bfqd->bfq_requests_within_timer) ++ bfq_mark_bfqq_IO_bound(bfqq); ++ } else ++ bfqq->requests_within_timer = 0; ++ bfq_log_bfqq(bfqd, bfqq, "requests in time %d", ++ bfqq->requests_within_timer); ++ } ++ ++ if (bfqd->low_latency) { ++ if (unlikely(time_is_after_jiffies(bfqq->split_time))) ++ /* wraparound */ ++ bfqq->split_time = ++ jiffies - bfqd->bfq_wr_min_idle_time - 1; ++ ++ if (time_is_before_jiffies(bfqq->split_time + ++ bfqd->bfq_wr_min_idle_time)) { ++ bfq_update_bfqq_wr_on_rq_arrival(bfqd, bfqq, ++ old_wr_coeff, ++ wr_or_deserves_wr, ++ *interactive, ++ in_burst, ++ soft_rt); ++ ++ if (old_wr_coeff != bfqq->wr_coeff) ++ bfqq->entity.prio_changed = 1; ++ } ++ } ++ ++ bfqq->last_idle_bklogged = jiffies; ++ bfqq->service_from_backlogged = 0; ++ bfq_clear_bfqq_softrt_update(bfqq); ++ ++ bfq_add_bfqq_busy(bfqd, bfqq); ++ ++ /* ++ * Expire in-service queue only if preemption may be needed ++ * for guarantees. In this respect, the function ++ * next_queue_may_preempt just checks a simple, necessary ++ * condition, and not a sufficient condition based on ++ * timestamps. In fact, for the latter condition to be ++ * evaluated, timestamps would need first to be updated, and ++ * this operation is quite costly (see the comments on the ++ * function bfq_bfqq_update_budg_for_activation). ++ */ ++ if (bfqd->in_service_queue && bfqq_wants_to_preempt && ++ bfqd->in_service_queue->wr_coeff < bfqq->wr_coeff && ++ next_queue_may_preempt(bfqd)) { ++ struct bfq_queue *in_serv = ++ bfqd->in_service_queue; ++ BUG_ON(in_serv == bfqq); ++ ++ bfq_bfqq_expire(bfqd, bfqd->in_service_queue, ++ false, BFQ_BFQQ_PREEMPTED); ++ } ++} ++ ++static void bfq_add_request(struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ struct bfq_data *bfqd = bfqq->bfqd; ++ struct request *next_rq, *prev; ++ unsigned int old_wr_coeff = bfqq->wr_coeff; ++ bool interactive = false; ++ ++ bfq_log_bfqq(bfqd, bfqq, "add_request: size %u %s", ++ blk_rq_sectors(rq), rq_is_sync(rq) ? "S" : "A"); ++ ++ if (bfqq->wr_coeff > 1) /* queue is being weight-raised */ ++ bfq_log_bfqq(bfqd, bfqq, ++ "raising period dur %u/%u msec, old coeff %u, w %d(%d)", ++ jiffies_to_msecs(jiffies - bfqq->last_wr_start_finish), ++ jiffies_to_msecs(bfqq->wr_cur_max_time), ++ bfqq->wr_coeff, ++ bfqq->entity.weight, bfqq->entity.orig_weight); ++ ++ bfqq->queued[rq_is_sync(rq)]++; ++ bfqd->queued++; ++ ++ elv_rb_add(&bfqq->sort_list, rq); ++ ++ /* ++ * Check if this request is a better next-to-serve candidate. ++ */ ++ prev = bfqq->next_rq; ++ next_rq = bfq_choose_req(bfqd, bfqq->next_rq, rq, bfqd->last_position); ++ BUG_ON(!next_rq); ++ bfqq->next_rq = next_rq; ++ ++ /* ++ * Adjust priority tree position, if next_rq changes. ++ */ ++ if (prev != bfqq->next_rq) ++ bfq_pos_tree_add_move(bfqd, bfqq); ++ ++ if (!bfq_bfqq_busy(bfqq)) /* switching to busy ... */ ++ bfq_bfqq_handle_idle_busy_switch(bfqd, bfqq, old_wr_coeff, ++ rq, &interactive); ++ else { ++ if (bfqd->low_latency && old_wr_coeff == 1 && !rq_is_sync(rq) && ++ time_is_before_jiffies( ++ bfqq->last_wr_start_finish + ++ bfqd->bfq_wr_min_inter_arr_async)) { ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ ++ bfqd->wr_busy_queues++; ++ BUG_ON(bfqd->wr_busy_queues > bfqd->busy_queues); ++ bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqd, bfqq, ++ "non-idle wrais starting, " ++ "wr_max_time %u wr_busy %d", ++ jiffies_to_msecs(bfqq->wr_cur_max_time), ++ bfqd->wr_busy_queues); ++ } ++ if (prev != bfqq->next_rq) ++ bfq_updated_next_req(bfqd, bfqq); ++ } ++ ++ /* ++ * Assign jiffies to last_wr_start_finish in the following ++ * cases: ++ * ++ * . if bfqq is not going to be weight-raised, because, for ++ * non weight-raised queues, last_wr_start_finish stores the ++ * arrival time of the last request; as of now, this piece ++ * of information is used only for deciding whether to ++ * weight-raise async queues ++ * ++ * . if bfqq is not weight-raised, because, if bfqq is now ++ * switching to weight-raised, then last_wr_start_finish ++ * stores the time when weight-raising starts ++ * ++ * . if bfqq is interactive, because, regardless of whether ++ * bfqq is currently weight-raised, the weight-raising ++ * period must start or restart (this case is considered ++ * separately because it is not detected by the above ++ * conditions, if bfqq is already weight-raised) ++ * ++ * last_wr_start_finish has to be updated also if bfqq is soft ++ * real-time, because the weight-raising period is constantly ++ * restarted on idle-to-busy transitions for these queues, but ++ * this is already done in bfq_bfqq_handle_idle_busy_switch if ++ * needed. ++ */ ++ if (bfqd->low_latency && ++ (old_wr_coeff == 1 || bfqq->wr_coeff == 1 || interactive)) ++ bfqq->last_wr_start_finish = jiffies; ++} ++ ++static struct request *bfq_find_rq_fmerge(struct bfq_data *bfqd, ++ struct bio *bio) ++{ ++ struct task_struct *tsk = current; ++ struct bfq_io_cq *bic; ++ struct bfq_queue *bfqq; ++ ++ bic = bfq_bic_lookup(bfqd, tsk->io_context); ++ if (!bic) ++ return NULL; ++ ++ bfqq = bic_to_bfqq(bic, op_is_sync(bio->bi_opf)); ++ if (bfqq) ++ return elv_rb_find(&bfqq->sort_list, bio_end_sector(bio)); ++ ++ return NULL; ++} ++ ++static sector_t get_sdist(sector_t last_pos, struct request *rq) ++{ ++ sector_t sdist = 0; ++ ++ if (last_pos) { ++ if (last_pos < blk_rq_pos(rq)) ++ sdist = blk_rq_pos(rq) - last_pos; ++ else ++ sdist = last_pos - blk_rq_pos(rq); ++ } ++ ++ return sdist; ++} ++ ++static void bfq_activate_request(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ bfqd->rq_in_driver++; ++} ++ ++static void bfq_deactivate_request(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ ++ BUG_ON(bfqd->rq_in_driver == 0); ++ bfqd->rq_in_driver--; ++} ++ ++static void bfq_remove_request(struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ struct bfq_data *bfqd = bfqq->bfqd; ++ const int sync = rq_is_sync(rq); ++ ++ BUG_ON(bfqq->entity.service > bfqq->entity.budget && ++ bfqq == bfqd->in_service_queue); ++ ++ if (bfqq->next_rq == rq) { ++ bfqq->next_rq = bfq_find_next_rq(bfqd, bfqq, rq); ++ bfq_updated_next_req(bfqd, bfqq); ++ } ++ ++ if (rq->queuelist.prev != &rq->queuelist) ++ list_del_init(&rq->queuelist); ++ BUG_ON(bfqq->queued[sync] == 0); ++ bfqq->queued[sync]--; ++ bfqd->queued--; ++ elv_rb_del(&bfqq->sort_list, rq); ++ ++ if (RB_EMPTY_ROOT(&bfqq->sort_list)) { ++ bfqq->next_rq = NULL; ++ ++ BUG_ON(bfqq->entity.budget < 0); ++ ++ if (bfq_bfqq_busy(bfqq) && bfqq != bfqd->in_service_queue) { ++ BUG_ON(bfqq->ref < 2); /* referred by rq and on tree */ ++ bfq_del_bfqq_busy(bfqd, bfqq, false); ++ /* ++ * bfqq emptied. In normal operation, when ++ * bfqq is empty, bfqq->entity.service and ++ * bfqq->entity.budget must contain, ++ * respectively, the service received and the ++ * budget used last time bfqq emptied. These ++ * facts do not hold in this case, as at least ++ * this last removal occurred while bfqq is ++ * not in service. To avoid inconsistencies, ++ * reset both bfqq->entity.service and ++ * bfqq->entity.budget, if bfqq has still a ++ * process that may issue I/O requests to it. ++ */ ++ bfqq->entity.budget = bfqq->entity.service = 0; ++ } ++ ++ /* ++ * Remove queue from request-position tree as it is empty. ++ */ ++ if (bfqq->pos_root) { ++ rb_erase(&bfqq->pos_node, bfqq->pos_root); ++ bfqq->pos_root = NULL; ++ } ++ } ++ ++ if (rq->cmd_flags & REQ_META) { ++ BUG_ON(bfqq->meta_pending == 0); ++ bfqq->meta_pending--; ++ } ++ bfqg_stats_update_io_remove(bfqq_group(bfqq), rq->cmd_flags); ++} ++ ++static enum elv_merge bfq_merge(struct request_queue *q, struct request **req, ++ struct bio *bio) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct request *__rq; ++ ++ __rq = bfq_find_rq_fmerge(bfqd, bio); ++ if (__rq && elv_bio_merge_ok(__rq, bio)) { ++ *req = __rq; ++ return ELEVATOR_FRONT_MERGE; ++ } ++ ++ return ELEVATOR_NO_MERGE; ++} ++ ++static void bfq_merged_request(struct request_queue *q, struct request *req, ++ enum elv_merge type) ++{ ++ if (type == ELEVATOR_FRONT_MERGE && ++ rb_prev(&req->rb_node) && ++ blk_rq_pos(req) < ++ blk_rq_pos(container_of(rb_prev(&req->rb_node), ++ struct request, rb_node))) { ++ struct bfq_queue *bfqq = RQ_BFQQ(req); ++ struct bfq_data *bfqd = bfqq->bfqd; ++ struct request *prev, *next_rq; ++ ++ /* Reposition request in its sort_list */ ++ elv_rb_del(&bfqq->sort_list, req); ++ elv_rb_add(&bfqq->sort_list, req); ++ /* Choose next request to be served for bfqq */ ++ prev = bfqq->next_rq; ++ next_rq = bfq_choose_req(bfqd, bfqq->next_rq, req, ++ bfqd->last_position); ++ BUG_ON(!next_rq); ++ bfqq->next_rq = next_rq; ++ /* ++ * If next_rq changes, update both the queue's budget to ++ * fit the new request and the queue's position in its ++ * rq_pos_tree. ++ */ ++ if (prev != bfqq->next_rq) { ++ bfq_updated_next_req(bfqd, bfqq); ++ bfq_pos_tree_add_move(bfqd, bfqq); ++ } ++ } ++} ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++static void bfq_bio_merged(struct request_queue *q, struct request *req, ++ struct bio *bio) ++{ ++ bfqg_stats_update_io_merged(bfqq_group(RQ_BFQQ(req)), bio->bi_opf); ++} ++#endif ++ ++static void bfq_merged_requests(struct request_queue *q, struct request *rq, ++ struct request *next) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq), *next_bfqq = RQ_BFQQ(next); ++ ++ /* ++ * If next and rq belong to the same bfq_queue and next is older ++ * than rq, then reposition rq in the fifo (by substituting next ++ * with rq). Otherwise, if next and rq belong to different ++ * bfq_queues, never reposition rq: in fact, we would have to ++ * reposition it with respect to next's position in its own fifo, ++ * which would most certainly be too expensive with respect to ++ * the benefits. ++ */ ++ if (bfqq == next_bfqq && ++ !list_empty(&rq->queuelist) && !list_empty(&next->queuelist) && ++ next->fifo_time < rq->fifo_time) { ++ list_del_init(&rq->queuelist); ++ list_replace_init(&next->queuelist, &rq->queuelist); ++ rq->fifo_time = next->fifo_time; ++ } ++ ++ if (bfqq->next_rq == next) ++ bfqq->next_rq = rq; ++ ++ bfq_remove_request(next); ++ bfqg_stats_update_io_merged(bfqq_group(bfqq), next->cmd_flags); ++} ++ ++/* Must be called with bfqq != NULL */ ++static void bfq_bfqq_end_wr(struct bfq_queue *bfqq) ++{ ++ BUG_ON(!bfqq); ++ ++ if (bfq_bfqq_busy(bfqq)) { ++ bfqq->bfqd->wr_busy_queues--; ++ BUG_ON(bfqq->bfqd->wr_busy_queues < 0); ++ } ++ bfqq->wr_coeff = 1; ++ bfqq->wr_cur_max_time = 0; ++ bfqq->last_wr_start_finish = jiffies; ++ /* ++ * Trigger a weight change on the next invocation of ++ * __bfq_entity_update_weight_prio. ++ */ ++ bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "end_wr: wrais ending at %lu, rais_max_time %u", ++ bfqq->last_wr_start_finish, ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "end_wr: wr_busy %d", ++ bfqq->bfqd->wr_busy_queues); ++} ++ ++static void bfq_end_wr_async_queues(struct bfq_data *bfqd, ++ struct bfq_group *bfqg) ++{ ++ int i, j; ++ ++ for (i = 0; i < 2; i++) ++ for (j = 0; j < IOPRIO_BE_NR; j++) ++ if (bfqg->async_bfqq[i][j]) ++ bfq_bfqq_end_wr(bfqg->async_bfqq[i][j]); ++ if (bfqg->async_idle_bfqq) ++ bfq_bfqq_end_wr(bfqg->async_idle_bfqq); ++} ++ ++static void bfq_end_wr(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq; ++ ++ spin_lock_irq(bfqd->queue->queue_lock); ++ ++ list_for_each_entry(bfqq, &bfqd->active_list, bfqq_list) ++ bfq_bfqq_end_wr(bfqq); ++ list_for_each_entry(bfqq, &bfqd->idle_list, bfqq_list) ++ bfq_bfqq_end_wr(bfqq); ++ bfq_end_wr_async(bfqd); ++ ++ spin_unlock_irq(bfqd->queue->queue_lock); ++} ++ ++static sector_t bfq_io_struct_pos(void *io_struct, bool request) ++{ ++ if (request) ++ return blk_rq_pos(io_struct); ++ else ++ return ((struct bio *)io_struct)->bi_iter.bi_sector; ++} ++ ++static int bfq_rq_close_to_sector(void *io_struct, bool request, ++ sector_t sector) ++{ ++ return abs(bfq_io_struct_pos(io_struct, request) - sector) <= ++ BFQQ_CLOSE_THR; ++} ++ ++static struct bfq_queue *bfqq_find_close(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ sector_t sector) ++{ ++ struct rb_root *root = &bfq_bfqq_to_bfqg(bfqq)->rq_pos_tree; ++ struct rb_node *parent, *node; ++ struct bfq_queue *__bfqq; ++ ++ if (RB_EMPTY_ROOT(root)) ++ return NULL; ++ ++ /* ++ * First, if we find a request starting at the end of the last ++ * request, choose it. ++ */ ++ __bfqq = bfq_rq_pos_tree_lookup(bfqd, root, sector, &parent, NULL); ++ if (__bfqq) ++ return __bfqq; ++ ++ /* ++ * If the exact sector wasn't found, the parent of the NULL leaf ++ * will contain the closest sector (rq_pos_tree sorted by ++ * next_request position). ++ */ ++ __bfqq = rb_entry(parent, struct bfq_queue, pos_node); ++ if (bfq_rq_close_to_sector(__bfqq->next_rq, true, sector)) ++ return __bfqq; ++ ++ if (blk_rq_pos(__bfqq->next_rq) < sector) ++ node = rb_next(&__bfqq->pos_node); ++ else ++ node = rb_prev(&__bfqq->pos_node); ++ if (!node) ++ return NULL; ++ ++ __bfqq = rb_entry(node, struct bfq_queue, pos_node); ++ if (bfq_rq_close_to_sector(__bfqq->next_rq, true, sector)) ++ return __bfqq; ++ ++ return NULL; ++} ++ ++static struct bfq_queue *bfq_find_close_cooperator(struct bfq_data *bfqd, ++ struct bfq_queue *cur_bfqq, ++ sector_t sector) ++{ ++ struct bfq_queue *bfqq; ++ ++ /* ++ * We shall notice if some of the queues are cooperating, ++ * e.g., working closely on the same area of the device. In ++ * that case, we can group them together and: 1) don't waste ++ * time idling, and 2) serve the union of their requests in ++ * the best possible order for throughput. ++ */ ++ bfqq = bfqq_find_close(bfqd, cur_bfqq, sector); ++ if (!bfqq || bfqq == cur_bfqq) ++ return NULL; ++ ++ return bfqq; ++} ++ ++static struct bfq_queue * ++bfq_setup_merge(struct bfq_queue *bfqq, struct bfq_queue *new_bfqq) ++{ ++ int process_refs, new_process_refs; ++ struct bfq_queue *__bfqq; ++ ++ /* ++ * If there are no process references on the new_bfqq, then it is ++ * unsafe to follow the ->new_bfqq chain as other bfqq's in the chain ++ * may have dropped their last reference (not just their last process ++ * reference). ++ */ ++ if (!bfqq_process_refs(new_bfqq)) ++ return NULL; ++ ++ /* Avoid a circular list and skip interim queue merges. */ ++ while ((__bfqq = new_bfqq->new_bfqq)) { ++ if (__bfqq == bfqq) ++ return NULL; ++ new_bfqq = __bfqq; ++ } ++ ++ process_refs = bfqq_process_refs(bfqq); ++ new_process_refs = bfqq_process_refs(new_bfqq); ++ /* ++ * If the process for the bfqq has gone away, there is no ++ * sense in merging the queues. ++ */ ++ if (process_refs == 0 || new_process_refs == 0) ++ return NULL; ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "scheduling merge with queue %d", ++ new_bfqq->pid); ++ ++ /* ++ * Merging is just a redirection: the requests of the process ++ * owning one of the two queues are redirected to the other queue. ++ * The latter queue, in its turn, is set as shared if this is the ++ * first time that the requests of some process are redirected to ++ * it. ++ * ++ * We redirect bfqq to new_bfqq and not the opposite, because we ++ * are in the context of the process owning bfqq, hence we have ++ * the io_cq of this process. So we can immediately configure this ++ * io_cq to redirect the requests of the process to new_bfqq. ++ * ++ * NOTE, even if new_bfqq coincides with the in-service queue, the ++ * io_cq of new_bfqq is not available, because, if the in-service ++ * queue is shared, bfqd->in_service_bic may not point to the ++ * io_cq of the in-service queue. ++ * Redirecting the requests of the process owning bfqq to the ++ * currently in-service queue is in any case the best option, as ++ * we feed the in-service queue with new requests close to the ++ * last request served and, by doing so, hopefully increase the ++ * throughput. ++ */ ++ bfqq->new_bfqq = new_bfqq; ++ new_bfqq->ref += process_refs; ++ return new_bfqq; ++} ++ ++static bool bfq_may_be_close_cooperator(struct bfq_queue *bfqq, ++ struct bfq_queue *new_bfqq) ++{ ++ if (bfq_class_idle(bfqq) || bfq_class_idle(new_bfqq) || ++ (bfqq->ioprio_class != new_bfqq->ioprio_class)) ++ return false; ++ ++ /* ++ * If either of the queues has already been detected as seeky, ++ * then merging it with the other queue is unlikely to lead to ++ * sequential I/O. ++ */ ++ if (BFQQ_SEEKY(bfqq) || BFQQ_SEEKY(new_bfqq)) ++ return false; ++ ++ /* ++ * Interleaved I/O is known to be done by (some) applications ++ * only for reads, so it does not make sense to merge async ++ * queues. ++ */ ++ if (!bfq_bfqq_sync(bfqq) || !bfq_bfqq_sync(new_bfqq)) ++ return false; ++ ++ return true; ++} ++ ++/* ++ * If this function returns true, then bfqq cannot be merged. The idea ++ * is that true cooperation happens very early after processes start ++ * to do I/O. Usually, late cooperations are just accidental false ++ * positives. In case bfqq is weight-raised, such false positives ++ * would evidently degrade latency guarantees for bfqq. ++ */ ++static bool wr_from_too_long(struct bfq_queue *bfqq) ++{ ++ return bfqq->wr_coeff > 1 && ++ time_is_before_jiffies(bfqq->last_wr_start_finish + ++ msecs_to_jiffies(100)); ++} ++ ++/* ++ * Attempt to schedule a merge of bfqq with the currently in-service ++ * queue or with a close queue among the scheduled queues. Return ++ * NULL if no merge was scheduled, a pointer to the shared bfq_queue ++ * structure otherwise. ++ * ++ * The OOM queue is not allowed to participate to cooperation: in fact, since ++ * the requests temporarily redirected to the OOM queue could be redirected ++ * again to dedicated queues at any time, the state needed to correctly ++ * handle merging with the OOM queue would be quite complex and expensive ++ * to maintain. Besides, in such a critical condition as an out of memory, ++ * the benefits of queue merging may be little relevant, or even negligible. ++ * ++ * Weight-raised queues can be merged only if their weight-raising ++ * period has just started. In fact cooperating processes are usually ++ * started together. Thus, with this filter we avoid false positives ++ * that would jeopardize low-latency guarantees. ++ * ++ * WARNING: queue merging may impair fairness among non-weight raised ++ * queues, for at least two reasons: 1) the original weight of a ++ * merged queue may change during the merged state, 2) even being the ++ * weight the same, a merged queue may be bloated with many more ++ * requests than the ones produced by its originally-associated ++ * process. ++ */ ++static struct bfq_queue * ++bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ void *io_struct, bool request) ++{ ++ struct bfq_queue *in_service_bfqq, *new_bfqq; ++ ++ if (bfqq->new_bfqq) ++ return bfqq->new_bfqq; ++ ++ if (io_struct && wr_from_too_long(bfqq) && ++ likely(bfqq != &bfqd->oom_bfqq)) ++ bfq_log_bfqq(bfqd, bfqq, ++ "would have looked for coop, but bfq%d wr", ++ bfqq->pid); ++ ++ if (!io_struct || ++ wr_from_too_long(bfqq) || ++ unlikely(bfqq == &bfqd->oom_bfqq)) ++ return NULL; ++ ++ /* If there is only one backlogged queue, don't search. */ ++ if (bfqd->busy_queues == 1) ++ return NULL; ++ ++ in_service_bfqq = bfqd->in_service_queue; ++ ++ if (in_service_bfqq && in_service_bfqq != bfqq && ++ bfqd->in_service_bic && wr_from_too_long(in_service_bfqq) ++ && likely(in_service_bfqq == &bfqd->oom_bfqq)) ++ bfq_log_bfqq(bfqd, bfqq, ++ "would have tried merge with in-service-queue, but wr"); ++ ++ if (!in_service_bfqq || in_service_bfqq == bfqq || ++ !bfqd->in_service_bic || wr_from_too_long(in_service_bfqq) || ++ unlikely(in_service_bfqq == &bfqd->oom_bfqq)) ++ goto check_scheduled; ++ ++ if (bfq_rq_close_to_sector(io_struct, request, bfqd->last_position) && ++ bfqq->entity.parent == in_service_bfqq->entity.parent && ++ bfq_may_be_close_cooperator(bfqq, in_service_bfqq)) { ++ new_bfqq = bfq_setup_merge(bfqq, in_service_bfqq); ++ if (new_bfqq) ++ return new_bfqq; ++ } ++ /* ++ * Check whether there is a cooperator among currently scheduled ++ * queues. The only thing we need is that the bio/request is not ++ * NULL, as we need it to establish whether a cooperator exists. ++ */ ++check_scheduled: ++ new_bfqq = bfq_find_close_cooperator(bfqd, bfqq, ++ bfq_io_struct_pos(io_struct, request)); ++ ++ BUG_ON(new_bfqq && bfqq->entity.parent != new_bfqq->entity.parent); ++ ++ if (new_bfqq && wr_from_too_long(new_bfqq) && ++ likely(new_bfqq != &bfqd->oom_bfqq) && ++ bfq_may_be_close_cooperator(bfqq, new_bfqq)) ++ bfq_log_bfqq(bfqd, bfqq, ++ "would have merged with bfq%d, but wr", ++ new_bfqq->pid); ++ ++ if (new_bfqq && !wr_from_too_long(new_bfqq) && ++ likely(new_bfqq != &bfqd->oom_bfqq) && ++ bfq_may_be_close_cooperator(bfqq, new_bfqq)) ++ return bfq_setup_merge(bfqq, new_bfqq); ++ ++ return NULL; ++} ++ ++static void bfq_bfqq_save_state(struct bfq_queue *bfqq) ++{ ++ struct bfq_io_cq *bic = bfqq->bic; ++ ++ /* ++ * If !bfqq->bic, the queue is already shared or its requests ++ * have already been redirected to a shared queue; both idle window ++ * and weight raising state have already been saved. Do nothing. ++ */ ++ if (!bic) ++ return; ++ ++ bic->saved_idle_window = bfq_bfqq_idle_window(bfqq); ++ bic->saved_IO_bound = bfq_bfqq_IO_bound(bfqq); ++ bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq); ++ bic->was_in_burst_list = !hlist_unhashed(&bfqq->burst_list_node); ++ bic->saved_wr_coeff = bfqq->wr_coeff; ++ bic->saved_wr_start_at_switch_to_srt = bfqq->wr_start_at_switch_to_srt; ++ bic->saved_last_wr_start_finish = bfqq->last_wr_start_finish; ++ bic->saved_wr_cur_max_time = bfqq->wr_cur_max_time; ++ BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); ++} ++ ++static void bfq_get_bic_reference(struct bfq_queue *bfqq) ++{ ++ /* ++ * If bfqq->bic has a non-NULL value, the bic to which it belongs ++ * is about to begin using a shared bfq_queue. ++ */ ++ if (bfqq->bic) ++ atomic_long_inc(&bfqq->bic->icq.ioc->refcount); ++} ++ ++static void ++bfq_merge_bfqqs(struct bfq_data *bfqd, struct bfq_io_cq *bic, ++ struct bfq_queue *bfqq, struct bfq_queue *new_bfqq) ++{ ++ bfq_log_bfqq(bfqd, bfqq, "merging with queue %lu", ++ (unsigned long) new_bfqq->pid); ++ /* Save weight raising and idle window of the merged queues */ ++ bfq_bfqq_save_state(bfqq); ++ bfq_bfqq_save_state(new_bfqq); ++ if (bfq_bfqq_IO_bound(bfqq)) ++ bfq_mark_bfqq_IO_bound(new_bfqq); ++ bfq_clear_bfqq_IO_bound(bfqq); ++ ++ /* ++ * If bfqq is weight-raised, then let new_bfqq inherit ++ * weight-raising. To reduce false positives, neglect the case ++ * where bfqq has just been created, but has not yet made it ++ * to be weight-raised (which may happen because EQM may merge ++ * bfqq even before bfq_add_request is executed for the first ++ * time for bfqq). Handling this case would however be very ++ * easy, thanks to the flag just_created. ++ */ ++ if (new_bfqq->wr_coeff == 1 && bfqq->wr_coeff > 1) { ++ new_bfqq->wr_coeff = bfqq->wr_coeff; ++ new_bfqq->wr_cur_max_time = bfqq->wr_cur_max_time; ++ new_bfqq->last_wr_start_finish = bfqq->last_wr_start_finish; ++ new_bfqq->wr_start_at_switch_to_srt = ++ bfqq->wr_start_at_switch_to_srt; ++ if (bfq_bfqq_busy(new_bfqq)) { ++ bfqd->wr_busy_queues++; ++ BUG_ON(bfqd->wr_busy_queues > bfqd->busy_queues); ++ } ++ ++ new_bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqd, new_bfqq, ++ "wr start after merge with %d, rais_max_time %u", ++ bfqq->pid, ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ } ++ ++ if (bfqq->wr_coeff > 1) { /* bfqq has given its wr to new_bfqq */ ++ bfqq->wr_coeff = 1; ++ bfqq->entity.prio_changed = 1; ++ if (bfq_bfqq_busy(bfqq)) { ++ bfqd->wr_busy_queues--; ++ BUG_ON(bfqd->wr_busy_queues < 0); ++ } ++ ++ } ++ ++ bfq_log_bfqq(bfqd, new_bfqq, "merge_bfqqs: wr_busy %d", ++ bfqd->wr_busy_queues); ++ ++ /* ++ * Grab a reference to the bic, to prevent it from being destroyed ++ * before being possibly touched by a bfq_split_bfqq(). ++ */ ++ bfq_get_bic_reference(bfqq); ++ bfq_get_bic_reference(new_bfqq); ++ /* ++ * Merge queues (that is, let bic redirect its requests to new_bfqq) ++ */ ++ bic_set_bfqq(bic, new_bfqq, 1); ++ bfq_mark_bfqq_coop(new_bfqq); ++ /* ++ * new_bfqq now belongs to at least two bics (it is a shared queue): ++ * set new_bfqq->bic to NULL. bfqq either: ++ * - does not belong to any bic any more, and hence bfqq->bic must ++ * be set to NULL, or ++ * - is a queue whose owning bics have already been redirected to a ++ * different queue, hence the queue is destined to not belong to ++ * any bic soon and bfqq->bic is already NULL (therefore the next ++ * assignment causes no harm). ++ */ ++ new_bfqq->bic = NULL; ++ bfqq->bic = NULL; ++ /* release process reference to bfqq */ ++ bfq_put_queue(bfqq); ++} ++ ++static int bfq_allow_bio_merge(struct request_queue *q, struct request *rq, ++ struct bio *bio) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ bool is_sync = op_is_sync(bio->bi_opf); ++ struct bfq_io_cq *bic; ++ struct bfq_queue *bfqq, *new_bfqq; ++ ++ /* ++ * Disallow merge of a sync bio into an async request. ++ */ ++ if (is_sync && !rq_is_sync(rq)) ++ return false; ++ ++ /* ++ * Lookup the bfqq that this bio will be queued with. Allow ++ * merge only if rq is queued there. ++ * Queue lock is held here. ++ */ ++ bic = bfq_bic_lookup(bfqd, current->io_context); ++ if (!bic) ++ return false; ++ ++ bfqq = bic_to_bfqq(bic, is_sync); ++ /* ++ * We take advantage of this function to perform an early merge ++ * of the queues of possible cooperating processes. ++ */ ++ if (bfqq) { ++ new_bfqq = bfq_setup_cooperator(bfqd, bfqq, bio, false); ++ if (new_bfqq) { ++ bfq_merge_bfqqs(bfqd, bic, bfqq, new_bfqq); ++ /* ++ * If we get here, the bio will be queued in the ++ * shared queue, i.e., new_bfqq, so use new_bfqq ++ * to decide whether bio and rq can be merged. ++ */ ++ bfqq = new_bfqq; ++ } ++ } ++ ++ return bfqq == RQ_BFQQ(rq); ++} ++ ++static int bfq_allow_rq_merge(struct request_queue *q, struct request *rq, ++ struct request *next) ++{ ++ return RQ_BFQQ(rq) == RQ_BFQQ(next); ++} ++ ++/* ++ * Set the maximum time for the in-service queue to consume its ++ * budget. This prevents seeky processes from lowering the throughput. ++ * In practice, a time-slice service scheme is used with seeky ++ * processes. ++ */ ++static void bfq_set_budget_timeout(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ unsigned int timeout_coeff; ++ ++ if (bfqq->wr_cur_max_time == bfqd->bfq_wr_rt_max_time) ++ timeout_coeff = 1; ++ else ++ timeout_coeff = bfqq->entity.weight / bfqq->entity.orig_weight; ++ ++ bfqd->last_budget_start = ktime_get(); ++ ++ bfqq->budget_timeout = jiffies + ++ bfqd->bfq_timeout * timeout_coeff; ++ ++ bfq_log_bfqq(bfqd, bfqq, "set budget_timeout %u", ++ jiffies_to_msecs(bfqd->bfq_timeout * timeout_coeff)); ++} ++ ++static void __bfq_set_in_service_queue(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ if (bfqq) { ++ bfqg_stats_update_avg_queue_size(bfqq_group(bfqq)); ++ bfq_mark_bfqq_must_alloc(bfqq); ++ bfq_clear_bfqq_fifo_expire(bfqq); ++ ++ bfqd->budgets_assigned = (bfqd->budgets_assigned*7 + 256) / 8; ++ ++ BUG_ON(bfqq == bfqd->in_service_queue); ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list)); ++ ++ if (time_is_before_jiffies(bfqq->last_wr_start_finish) && ++ bfqq->wr_coeff > 1 && ++ bfqq->wr_cur_max_time == bfqd->bfq_wr_rt_max_time && ++ time_is_before_jiffies(bfqq->budget_timeout)) { ++ /* ++ * For soft real-time queues, move the start ++ * of the weight-raising period forward by the ++ * time the queue has not received any ++ * service. Otherwise, a relatively long ++ * service delay is likely to cause the ++ * weight-raising period of the queue to end, ++ * because of the short duration of the ++ * weight-raising period of a soft real-time ++ * queue. It is worth noting that this move ++ * is not so dangerous for the other queues, ++ * because soft real-time queues are not ++ * greedy. ++ * ++ * To not add a further variable, we use the ++ * overloaded field budget_timeout to ++ * determine for how long the queue has not ++ * received service, i.e., how much time has ++ * elapsed since the queue expired. However, ++ * this is a little imprecise, because ++ * budget_timeout is set to jiffies if bfqq ++ * not only expires, but also remains with no ++ * request. ++ */ ++ if (time_after(bfqq->budget_timeout, ++ bfqq->last_wr_start_finish)) ++ bfqq->last_wr_start_finish += ++ jiffies - bfqq->budget_timeout; ++ else ++ bfqq->last_wr_start_finish = jiffies; ++ ++ if (time_is_after_jiffies(bfqq->last_wr_start_finish)) { ++ pr_crit( ++ "BFQ WARNING:last %lu budget %lu jiffies %lu", ++ bfqq->last_wr_start_finish, ++ bfqq->budget_timeout, ++ jiffies); ++ pr_crit("diff %lu", jiffies - ++ max_t(unsigned long, ++ bfqq->last_wr_start_finish, ++ bfqq->budget_timeout)); ++ bfqq->last_wr_start_finish = jiffies; ++ } ++ } ++ ++ bfq_set_budget_timeout(bfqd, bfqq); ++ bfq_log_bfqq(bfqd, bfqq, ++ "set_in_service_queue, cur-budget = %d", ++ bfqq->entity.budget); ++ } else ++ bfq_log(bfqd, "set_in_service_queue: NULL"); ++ ++ bfqd->in_service_queue = bfqq; ++} ++ ++/* ++ * Get and set a new queue for service. ++ */ ++static struct bfq_queue *bfq_set_in_service_queue(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq = bfq_get_next_queue(bfqd); ++ ++ __bfq_set_in_service_queue(bfqd, bfqq); ++ return bfqq; ++} ++ ++static void bfq_arm_slice_timer(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq = bfqd->in_service_queue; ++ struct bfq_io_cq *bic; ++ u32 sl; ++ ++ BUG_ON(!RB_EMPTY_ROOT(&bfqq->sort_list)); ++ ++ /* Processes have exited, don't wait. */ ++ bic = bfqd->in_service_bic; ++ if (!bic || atomic_read(&bic->icq.ioc->active_ref) == 0) ++ return; ++ ++ bfq_mark_bfqq_wait_request(bfqq); ++ ++ /* ++ * We don't want to idle for seeks, but we do want to allow ++ * fair distribution of slice time for a process doing back-to-back ++ * seeks. So allow a little bit of time for him to submit a new rq. ++ * ++ * To prevent processes with (partly) seeky workloads from ++ * being too ill-treated, grant them a small fraction of the ++ * assigned budget before reducing the waiting time to ++ * BFQ_MIN_TT. This happened to help reduce latency. ++ */ ++ sl = bfqd->bfq_slice_idle; ++ /* ++ * Unless the queue is being weight-raised or the scenario is ++ * asymmetric, grant only minimum idle time if the queue ++ * is seeky. A long idling is preserved for a weight-raised ++ * queue, or, more in general, in an asymemtric scenario, ++ * because a long idling is needed for guaranteeing to a queue ++ * its reserved share of the throughput (in particular, it is ++ * needed if the queue has a higher weight than some other ++ * queue). ++ */ ++ if (BFQQ_SEEKY(bfqq) && bfqq->wr_coeff == 1 && ++ bfq_symmetric_scenario(bfqd)) ++ sl = min_t(u32, sl, BFQ_MIN_TT); ++ ++ bfqd->last_idling_start = ktime_get(); ++ hrtimer_start(&bfqd->idle_slice_timer, ns_to_ktime(sl), ++ HRTIMER_MODE_REL); ++ bfqg_stats_set_start_idle_time(bfqq_group(bfqq)); ++ bfq_log(bfqd, "arm idle: %ld/%ld ms", ++ sl / NSEC_PER_MSEC, bfqd->bfq_slice_idle / NSEC_PER_MSEC); ++} ++ ++/* ++ * In autotuning mode, max_budget is dynamically recomputed as the ++ * amount of sectors transferred in timeout at the estimated peak ++ * rate. This enables BFQ to utilize a full timeslice with a full ++ * budget, even if the in-service queue is served at peak rate. And ++ * this maximises throughput with sequential workloads. ++ */ ++static unsigned long bfq_calc_max_budget(struct bfq_data *bfqd) ++{ ++ return (u64)bfqd->peak_rate * USEC_PER_MSEC * ++ jiffies_to_msecs(bfqd->bfq_timeout)>>BFQ_RATE_SHIFT; ++} ++ ++/* ++ * Update parameters related to throughput and responsiveness, as a ++ * function of the estimated peak rate. See comments on ++ * bfq_calc_max_budget(), and on T_slow and T_fast arrays. ++ */ ++static void update_thr_responsiveness_params(struct bfq_data *bfqd) ++{ ++ int dev_type = blk_queue_nonrot(bfqd->queue); ++ ++ if (bfqd->bfq_user_max_budget == 0) { ++ bfqd->bfq_max_budget = ++ bfq_calc_max_budget(bfqd); ++ BUG_ON(bfqd->bfq_max_budget < 0); ++ bfq_log(bfqd, "new max_budget = %d", ++ bfqd->bfq_max_budget); ++ } ++ ++ if (bfqd->device_speed == BFQ_BFQD_FAST && ++ bfqd->peak_rate < device_speed_thresh[dev_type]) { ++ bfqd->device_speed = BFQ_BFQD_SLOW; ++ bfqd->RT_prod = R_slow[dev_type] * ++ T_slow[dev_type]; ++ } else if (bfqd->device_speed == BFQ_BFQD_SLOW && ++ bfqd->peak_rate > device_speed_thresh[dev_type]) { ++ bfqd->device_speed = BFQ_BFQD_FAST; ++ bfqd->RT_prod = R_fast[dev_type] * ++ T_fast[dev_type]; ++ } ++ ++ bfq_log(bfqd, ++"dev_type %s dev_speed_class = %s (%llu sects/sec), thresh %llu setcs/sec", ++ dev_type == 0 ? "ROT" : "NONROT", ++ bfqd->device_speed == BFQ_BFQD_FAST ? "FAST" : "SLOW", ++ bfqd->device_speed == BFQ_BFQD_FAST ? ++ (USEC_PER_SEC*(u64)R_fast[dev_type])>>BFQ_RATE_SHIFT : ++ (USEC_PER_SEC*(u64)R_slow[dev_type])>>BFQ_RATE_SHIFT, ++ (USEC_PER_SEC*(u64)device_speed_thresh[dev_type])>> ++ BFQ_RATE_SHIFT); ++} ++ ++static void bfq_reset_rate_computation(struct bfq_data *bfqd, struct request *rq) ++{ ++ if (rq != NULL) { /* new rq dispatch now, reset accordingly */ ++ bfqd->last_dispatch = bfqd->first_dispatch = ktime_get_ns() ; ++ bfqd->peak_rate_samples = 1; ++ bfqd->sequential_samples = 0; ++ bfqd->tot_sectors_dispatched = bfqd->last_rq_max_size = ++ blk_rq_sectors(rq); ++ } else /* no new rq dispatched, just reset the number of samples */ ++ bfqd->peak_rate_samples = 0; /* full re-init on next disp. */ ++ ++ bfq_log(bfqd, ++ "reset_rate_computation at end, sample %u/%u tot_sects %llu", ++ bfqd->peak_rate_samples, bfqd->sequential_samples, ++ bfqd->tot_sectors_dispatched); ++} ++ ++static void bfq_update_rate_reset(struct bfq_data *bfqd, struct request *rq) ++{ ++ u32 rate, weight, divisor; ++ ++ /* ++ * For the convergence property to hold (see comments on ++ * bfq_update_peak_rate()) and for the assessment to be ++ * reliable, a minimum number of samples must be present, and ++ * a minimum amount of time must have elapsed. If not so, do ++ * not compute new rate. Just reset parameters, to get ready ++ * for a new evaluation attempt. ++ */ ++ if (bfqd->peak_rate_samples < BFQ_RATE_MIN_SAMPLES || ++ bfqd->delta_from_first < BFQ_RATE_MIN_INTERVAL) { ++ bfq_log(bfqd, ++ "update_rate_reset: only resetting, delta_first %lluus samples %d", ++ bfqd->delta_from_first>>10, bfqd->peak_rate_samples); ++ goto reset_computation; ++ } ++ ++ /* ++ * If a new request completion has occurred after last ++ * dispatch, then, to approximate the rate at which requests ++ * have been served by the device, it is more precise to ++ * extend the observation interval to the last completion. ++ */ ++ bfqd->delta_from_first = ++ max_t(u64, bfqd->delta_from_first, ++ bfqd->last_completion - bfqd->first_dispatch); ++ ++ BUG_ON(bfqd->delta_from_first == 0); ++ /* ++ * Rate computed in sects/usec, and not sects/nsec, for ++ * precision issues. ++ */ ++ rate = div64_ul(bfqd->tot_sectors_dispatched<delta_from_first, NSEC_PER_USEC)); ++ ++ bfq_log(bfqd, ++"update_rate_reset: tot_sects %llu delta_first %lluus rate %llu sects/s (%d)", ++ bfqd->tot_sectors_dispatched, bfqd->delta_from_first>>10, ++ ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), ++ rate > 20< 20M sectors/sec) ++ */ ++ if ((bfqd->sequential_samples < (3 * bfqd->peak_rate_samples)>>2 && ++ rate <= bfqd->peak_rate) || ++ rate > 20<peak_rate_samples, bfqd->sequential_samples, ++ ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), ++ ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); ++ goto reset_computation; ++ } else { ++ bfq_log(bfqd, ++ "update_rate_reset: do update, samples %u/%u rate/peak %llu/%llu", ++ bfqd->peak_rate_samples, bfqd->sequential_samples, ++ ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), ++ ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); ++ } ++ ++ /* ++ * We have to update the peak rate, at last! To this purpose, ++ * we use a low-pass filter. We compute the smoothing constant ++ * of the filter as a function of the 'weight' of the new ++ * measured rate. ++ * ++ * As can be seen in next formulas, we define this weight as a ++ * quantity proportional to how sequential the workload is, ++ * and to how long the observation time interval is. ++ * ++ * The weight runs from 0 to 8. The maximum value of the ++ * weight, 8, yields the minimum value for the smoothing ++ * constant. At this minimum value for the smoothing constant, ++ * the measured rate contributes for half of the next value of ++ * the estimated peak rate. ++ * ++ * So, the first step is to compute the weight as a function ++ * of how sequential the workload is. Note that the weight ++ * cannot reach 9, because bfqd->sequential_samples cannot ++ * become equal to bfqd->peak_rate_samples, which, in its ++ * turn, holds true because bfqd->sequential_samples is not ++ * incremented for the first sample. ++ */ ++ weight = (9 * bfqd->sequential_samples) / bfqd->peak_rate_samples; ++ ++ /* ++ * Second step: further refine the weight as a function of the ++ * duration of the observation interval. ++ */ ++ weight = min_t(u32, 8, ++ div_u64(weight * bfqd->delta_from_first, ++ BFQ_RATE_REF_INTERVAL)); ++ ++ /* ++ * Divisor ranging from 10, for minimum weight, to 2, for ++ * maximum weight. ++ */ ++ divisor = 10 - weight; ++ BUG_ON(divisor == 0); ++ ++ /* ++ * Finally, update peak rate: ++ * ++ * peak_rate = peak_rate * (divisor-1) / divisor + rate / divisor ++ */ ++ bfqd->peak_rate *= divisor-1; ++ bfqd->peak_rate /= divisor; ++ rate /= divisor; /* smoothing constant alpha = 1/divisor */ ++ ++ bfq_log(bfqd, ++ "update_rate_reset: divisor %d tmp_peak_rate %llu tmp_rate %u", ++ divisor, ++ ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT), ++ (u32)((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT)); ++ ++ BUG_ON(bfqd->peak_rate == 0); ++ BUG_ON(bfqd->peak_rate > 20<peak_rate += rate; ++ update_thr_responsiveness_params(bfqd); ++ BUG_ON(bfqd->peak_rate > 20<peak_rate_samples == 0) { /* first dispatch */ ++ bfq_log(bfqd, ++ "update_peak_rate: goto reset, samples %d", ++ bfqd->peak_rate_samples) ; ++ bfq_reset_rate_computation(bfqd, rq); ++ goto update_last_values; /* will add one sample */ ++ } ++ ++ /* ++ * Device idle for very long: the observation interval lasting ++ * up to this dispatch cannot be a valid observation interval ++ * for computing a new peak rate (similarly to the late- ++ * completion event in bfq_completed_request()). Go to ++ * update_rate_and_reset to have the following three steps ++ * taken: ++ * - close the observation interval at the last (previous) ++ * request dispatch or completion ++ * - compute rate, if possible, for that observation interval ++ * - start a new observation interval with this dispatch ++ */ ++ if (now_ns - bfqd->last_dispatch > 100*NSEC_PER_MSEC && ++ bfqd->rq_in_driver == 0) { ++ bfq_log(bfqd, ++"update_peak_rate: jumping to updating&resetting delta_last %lluus samples %d", ++ (now_ns - bfqd->last_dispatch)>>10, ++ bfqd->peak_rate_samples) ; ++ goto update_rate_and_reset; ++ } ++ ++ /* Update sampling information */ ++ bfqd->peak_rate_samples++; ++ ++ if ((bfqd->rq_in_driver > 0 || ++ now_ns - bfqd->last_completion < BFQ_MIN_TT) ++ && get_sdist(bfqd->last_position, rq) < BFQQ_SEEK_THR) ++ bfqd->sequential_samples++; ++ ++ bfqd->tot_sectors_dispatched += blk_rq_sectors(rq); ++ ++ /* Reset max observed rq size every 32 dispatches */ ++ if (likely(bfqd->peak_rate_samples % 32)) ++ bfqd->last_rq_max_size = ++ max_t(u32, blk_rq_sectors(rq), bfqd->last_rq_max_size); ++ else ++ bfqd->last_rq_max_size = blk_rq_sectors(rq); ++ ++ bfqd->delta_from_first = now_ns - bfqd->first_dispatch; ++ ++ bfq_log(bfqd, ++ "update_peak_rate: added samples %u/%u tot_sects %llu delta_first %lluus", ++ bfqd->peak_rate_samples, bfqd->sequential_samples, ++ bfqd->tot_sectors_dispatched, ++ bfqd->delta_from_first>>10); ++ ++ /* Target observation interval not yet reached, go on sampling */ ++ if (bfqd->delta_from_first < BFQ_RATE_REF_INTERVAL) ++ goto update_last_values; ++ ++update_rate_and_reset: ++ bfq_update_rate_reset(bfqd, rq); ++update_last_values: ++ bfqd->last_position = blk_rq_pos(rq) + blk_rq_sectors(rq); ++ bfqd->last_dispatch = now_ns; ++ ++ bfq_log(bfqd, ++ "update_peak_rate: delta_first %lluus last_pos %llu peak_rate %llu", ++ (now_ns - bfqd->first_dispatch)>>10, ++ (unsigned long long) bfqd->last_position, ++ ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); ++ bfq_log(bfqd, ++ "update_peak_rate: samples at end %d", bfqd->peak_rate_samples); ++} ++ ++/* ++ * Move request from internal lists to the dispatch list of the request queue ++ */ ++static void bfq_dispatch_insert(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ ++ /* ++ * For consistency, the next instruction should have been executed ++ * after removing the request from the queue and dispatching it. ++ * We execute instead this instruction before bfq_remove_request() ++ * (and hence introduce a temporary inconsistency), for efficiency. ++ * In fact, in a forced_dispatch, this prevents two counters related ++ * to bfqq->dispatched to risk to be uselessly decremented if bfqq ++ * is not in service, and then to be incremented again after ++ * incrementing bfqq->dispatched. ++ */ ++ bfqq->dispatched++; ++ bfq_update_peak_rate(q->elevator->elevator_data, rq); ++ ++ bfq_remove_request(rq); ++ elv_dispatch_sort(q, rq); ++} ++ ++static void __bfq_bfqq_expire(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ BUG_ON(bfqq != bfqd->in_service_queue); ++ ++ /* ++ * If this bfqq is shared between multiple processes, check ++ * to make sure that those processes are still issuing I/Os ++ * within the mean seek distance. If not, it may be time to ++ * break the queues apart again. ++ */ ++ if (bfq_bfqq_coop(bfqq) && BFQQ_SEEKY(bfqq)) ++ bfq_mark_bfqq_split_coop(bfqq); ++ ++ if (RB_EMPTY_ROOT(&bfqq->sort_list)) { ++ if (bfqq->dispatched == 0) ++ /* ++ * Overloading budget_timeout field to store ++ * the time at which the queue remains with no ++ * backlog and no outstanding request; used by ++ * the weight-raising mechanism. ++ */ ++ bfqq->budget_timeout = jiffies; ++ ++ bfq_del_bfqq_busy(bfqd, bfqq, true); ++ } else { ++ bfq_requeue_bfqq(bfqd, bfqq); ++ /* ++ * Resort priority tree of potential close cooperators. ++ */ ++ bfq_pos_tree_add_move(bfqd, bfqq); ++ } ++ ++ /* ++ * All in-service entities must have been properly deactivated ++ * or requeued before executing the next function, which ++ * resets all in-service entites as no more in service. ++ */ ++ __bfq_bfqd_reset_in_service(bfqd); ++} ++ ++/** ++ * __bfq_bfqq_recalc_budget - try to adapt the budget to the @bfqq behavior. ++ * @bfqd: device data. ++ * @bfqq: queue to update. ++ * @reason: reason for expiration. ++ * ++ * Handle the feedback on @bfqq budget at queue expiration. ++ * See the body for detailed comments. ++ */ ++static void __bfq_bfqq_recalc_budget(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ enum bfqq_expiration reason) ++{ ++ struct request *next_rq; ++ int budget, min_budget; ++ ++ BUG_ON(bfqq != bfqd->in_service_queue); ++ ++ min_budget = bfq_min_budget(bfqd); ++ ++ if (bfqq->wr_coeff == 1) ++ budget = bfqq->max_budget; ++ else /* ++ * Use a constant, low budget for weight-raised queues, ++ * to help achieve a low latency. Keep it slightly higher ++ * than the minimum possible budget, to cause a little ++ * bit fewer expirations. ++ */ ++ budget = 2 * min_budget; ++ ++ bfq_log_bfqq(bfqd, bfqq, "recalc_budg: last budg %d, budg left %d", ++ bfqq->entity.budget, bfq_bfqq_budget_left(bfqq)); ++ bfq_log_bfqq(bfqd, bfqq, "recalc_budg: last max_budg %d, min budg %d", ++ budget, bfq_min_budget(bfqd)); ++ bfq_log_bfqq(bfqd, bfqq, "recalc_budg: sync %d, seeky %d", ++ bfq_bfqq_sync(bfqq), BFQQ_SEEKY(bfqd->in_service_queue)); ++ ++ if (bfq_bfqq_sync(bfqq) && bfqq->wr_coeff == 1) { ++ switch (reason) { ++ /* ++ * Caveat: in all the following cases we trade latency ++ * for throughput. ++ */ ++ case BFQ_BFQQ_TOO_IDLE: ++ /* ++ * This is the only case where we may reduce ++ * the budget: if there is no request of the ++ * process still waiting for completion, then ++ * we assume (tentatively) that the timer has ++ * expired because the batch of requests of ++ * the process could have been served with a ++ * smaller budget. Hence, betting that ++ * process will behave in the same way when it ++ * becomes backlogged again, we reduce its ++ * next budget. As long as we guess right, ++ * this budget cut reduces the latency ++ * experienced by the process. ++ * ++ * However, if there are still outstanding ++ * requests, then the process may have not yet ++ * issued its next request just because it is ++ * still waiting for the completion of some of ++ * the still outstanding ones. So in this ++ * subcase we do not reduce its budget, on the ++ * contrary we increase it to possibly boost ++ * the throughput, as discussed in the ++ * comments to the BUDGET_TIMEOUT case. ++ */ ++ if (bfqq->dispatched > 0) /* still outstanding reqs */ ++ budget = min(budget * 2, bfqd->bfq_max_budget); ++ else { ++ if (budget > 5 * min_budget) ++ budget -= 4 * min_budget; ++ else ++ budget = min_budget; ++ } ++ break; ++ case BFQ_BFQQ_BUDGET_TIMEOUT: ++ /* ++ * We double the budget here because it gives ++ * the chance to boost the throughput if this ++ * is not a seeky process (and has bumped into ++ * this timeout because of, e.g., ZBR). ++ */ ++ budget = min(budget * 2, bfqd->bfq_max_budget); ++ break; ++ case BFQ_BFQQ_BUDGET_EXHAUSTED: ++ /* ++ * The process still has backlog, and did not ++ * let either the budget timeout or the disk ++ * idling timeout expire. Hence it is not ++ * seeky, has a short thinktime and may be ++ * happy with a higher budget too. So ++ * definitely increase the budget of this good ++ * candidate to boost the disk throughput. ++ */ ++ budget = min(budget * 4, bfqd->bfq_max_budget); ++ break; ++ case BFQ_BFQQ_NO_MORE_REQUESTS: ++ /* ++ * For queues that expire for this reason, it ++ * is particularly important to keep the ++ * budget close to the actual service they ++ * need. Doing so reduces the timestamp ++ * misalignment problem described in the ++ * comments in the body of ++ * __bfq_activate_entity. In fact, suppose ++ * that a queue systematically expires for ++ * BFQ_BFQQ_NO_MORE_REQUESTS and presents a ++ * new request in time to enjoy timestamp ++ * back-shifting. The larger the budget of the ++ * queue is with respect to the service the ++ * queue actually requests in each service ++ * slot, the more times the queue can be ++ * reactivated with the same virtual finish ++ * time. It follows that, even if this finish ++ * time is pushed to the system virtual time ++ * to reduce the consequent timestamp ++ * misalignment, the queue unjustly enjoys for ++ * many re-activations a lower finish time ++ * than all newly activated queues. ++ * ++ * The service needed by bfqq is measured ++ * quite precisely by bfqq->entity.service. ++ * Since bfqq does not enjoy device idling, ++ * bfqq->entity.service is equal to the number ++ * of sectors that the process associated with ++ * bfqq requested to read/write before waiting ++ * for request completions, or blocking for ++ * other reasons. ++ */ ++ budget = max_t(int, bfqq->entity.service, min_budget); ++ break; ++ default: ++ return; ++ } ++ } else if (!bfq_bfqq_sync(bfqq)) ++ /* ++ * Async queues get always the maximum possible ++ * budget, as for them we do not care about latency ++ * (in addition, their ability to dispatch is limited ++ * by the charging factor). ++ */ ++ budget = bfqd->bfq_max_budget; ++ ++ bfqq->max_budget = budget; ++ ++ if (bfqd->budgets_assigned >= bfq_stats_min_budgets && ++ !bfqd->bfq_user_max_budget) ++ bfqq->max_budget = min(bfqq->max_budget, bfqd->bfq_max_budget); ++ ++ /* ++ * If there is still backlog, then assign a new budget, making ++ * sure that it is large enough for the next request. Since ++ * the finish time of bfqq must be kept in sync with the ++ * budget, be sure to call __bfq_bfqq_expire() *after* this ++ * update. ++ * ++ * If there is no backlog, then no need to update the budget; ++ * it will be updated on the arrival of a new request. ++ */ ++ next_rq = bfqq->next_rq; ++ if (next_rq) { ++ BUG_ON(reason == BFQ_BFQQ_TOO_IDLE || ++ reason == BFQ_BFQQ_NO_MORE_REQUESTS); ++ bfqq->entity.budget = max_t(unsigned long, bfqq->max_budget, ++ bfq_serv_to_charge(next_rq, bfqq)); ++ BUG_ON(!bfq_bfqq_busy(bfqq)); ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list)); ++ } ++ ++ bfq_log_bfqq(bfqd, bfqq, "head sect: %u, new budget %d", ++ next_rq ? blk_rq_sectors(next_rq) : 0, ++ bfqq->entity.budget); ++} ++ ++/* ++ * Return true if the process associated with bfqq is "slow". The slow ++ * flag is used, in addition to the budget timeout, to reduce the ++ * amount of service provided to seeky processes, and thus reduce ++ * their chances to lower the throughput. More details in the comments ++ * on the function bfq_bfqq_expire(). ++ * ++ * An important observation is in order: as discussed in the comments ++ * on the function bfq_update_peak_rate(), with devices with internal ++ * queues, it is hard if ever possible to know when and for how long ++ * an I/O request is processed by the device (apart from the trivial ++ * I/O pattern where a new request is dispatched only after the ++ * previous one has been completed). This makes it hard to evaluate ++ * the real rate at which the I/O requests of each bfq_queue are ++ * served. In fact, for an I/O scheduler like BFQ, serving a ++ * bfq_queue means just dispatching its requests during its service ++ * slot (i.e., until the budget of the queue is exhausted, or the ++ * queue remains idle, or, finally, a timeout fires). But, during the ++ * service slot of a bfq_queue, around 100 ms at most, the device may ++ * be even still processing requests of bfq_queues served in previous ++ * service slots. On the opposite end, the requests of the in-service ++ * bfq_queue may be completed after the service slot of the queue ++ * finishes. ++ * ++ * Anyway, unless more sophisticated solutions are used ++ * (where possible), the sum of the sizes of the requests dispatched ++ * during the service slot of a bfq_queue is probably the only ++ * approximation available for the service received by the bfq_queue ++ * during its service slot. And this sum is the quantity used in this ++ * function to evaluate the I/O speed of a process. ++ */ ++static bool bfq_bfqq_is_slow(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ bool compensate, enum bfqq_expiration reason, ++ unsigned long *delta_ms) ++{ ++ ktime_t delta_ktime; ++ u32 delta_usecs; ++ bool slow = BFQQ_SEEKY(bfqq); /* if delta too short, use seekyness */ ++ ++ if (!bfq_bfqq_sync(bfqq)) ++ return false; ++ ++ if (compensate) ++ delta_ktime = bfqd->last_idling_start; ++ else ++ delta_ktime = ktime_get(); ++ delta_ktime = ktime_sub(delta_ktime, bfqd->last_budget_start); ++ delta_usecs = ktime_to_us(delta_ktime); ++ ++ /* don't use too short time intervals */ ++ if (delta_usecs < 1000) { ++ if (blk_queue_nonrot(bfqd->queue)) ++ /* ++ * give same worst-case guarantees as idling ++ * for seeky ++ */ ++ *delta_ms = BFQ_MIN_TT / NSEC_PER_MSEC; ++ else /* charge at least one seek */ ++ *delta_ms = bfq_slice_idle / NSEC_PER_MSEC; ++ ++ bfq_log(bfqd, "bfq_bfqq_is_slow: too short %u", delta_usecs); ++ ++ return slow; ++ } ++ ++ *delta_ms = delta_usecs / USEC_PER_MSEC; ++ ++ /* ++ * Use only long (> 20ms) intervals to filter out excessive ++ * spikes in service rate estimation. ++ */ ++ if (delta_usecs > 20000) { ++ /* ++ * Caveat for rotational devices: processes doing I/O ++ * in the slower disk zones tend to be slow(er) even ++ * if not seeky. In this respect, the estimated peak ++ * rate is likely to be an average over the disk ++ * surface. Accordingly, to not be too harsh with ++ * unlucky processes, a process is deemed slow only if ++ * its rate has been lower than half of the estimated ++ * peak rate. ++ */ ++ slow = bfqq->entity.service < bfqd->bfq_max_budget / 2; ++ bfq_log(bfqd, "bfq_bfqq_is_slow: relative rate %d/%d", ++ bfqq->entity.service, bfqd->bfq_max_budget); ++ } ++ ++ bfq_log_bfqq(bfqd, bfqq, "bfq_bfqq_is_slow: slow %d", slow); ++ ++ return slow; ++} ++ ++/* ++ * To be deemed as soft real-time, an application must meet two ++ * requirements. First, the application must not require an average ++ * bandwidth higher than the approximate bandwidth required to playback or ++ * record a compressed high-definition video. ++ * The next function is invoked on the completion of the last request of a ++ * batch, to compute the next-start time instant, soft_rt_next_start, such ++ * that, if the next request of the application does not arrive before ++ * soft_rt_next_start, then the above requirement on the bandwidth is met. ++ * ++ * The second requirement is that the request pattern of the application is ++ * isochronous, i.e., that, after issuing a request or a batch of requests, ++ * the application stops issuing new requests until all its pending requests ++ * have been completed. After that, the application may issue a new batch, ++ * and so on. ++ * For this reason the next function is invoked to compute ++ * soft_rt_next_start only for applications that meet this requirement, ++ * whereas soft_rt_next_start is set to infinity for applications that do ++ * not. ++ * ++ * Unfortunately, even a greedy application may happen to behave in an ++ * isochronous way if the CPU load is high. In fact, the application may ++ * stop issuing requests while the CPUs are busy serving other processes, ++ * then restart, then stop again for a while, and so on. In addition, if ++ * the disk achieves a low enough throughput with the request pattern ++ * issued by the application (e.g., because the request pattern is random ++ * and/or the device is slow), then the application may meet the above ++ * bandwidth requirement too. To prevent such a greedy application to be ++ * deemed as soft real-time, a further rule is used in the computation of ++ * soft_rt_next_start: soft_rt_next_start must be higher than the current ++ * time plus the maximum time for which the arrival of a request is waited ++ * for when a sync queue becomes idle, namely bfqd->bfq_slice_idle. ++ * This filters out greedy applications, as the latter issue instead their ++ * next request as soon as possible after the last one has been completed ++ * (in contrast, when a batch of requests is completed, a soft real-time ++ * application spends some time processing data). ++ * ++ * Unfortunately, the last filter may easily generate false positives if ++ * only bfqd->bfq_slice_idle is used as a reference time interval and one ++ * or both the following cases occur: ++ * 1) HZ is so low that the duration of a jiffy is comparable to or higher ++ * than bfqd->bfq_slice_idle. This happens, e.g., on slow devices with ++ * HZ=100. ++ * 2) jiffies, instead of increasing at a constant rate, may stop increasing ++ * for a while, then suddenly 'jump' by several units to recover the lost ++ * increments. This seems to happen, e.g., inside virtual machines. ++ * To address this issue, we do not use as a reference time interval just ++ * bfqd->bfq_slice_idle, but bfqd->bfq_slice_idle plus a few jiffies. In ++ * particular we add the minimum number of jiffies for which the filter ++ * seems to be quite precise also in embedded systems and KVM/QEMU virtual ++ * machines. ++ */ ++static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ bfq_log_bfqq(bfqd, bfqq, ++"softrt_next_start: service_blkg %lu soft_rate %u sects/sec interval %u", ++ bfqq->service_from_backlogged, ++ bfqd->bfq_wr_max_softrt_rate, ++ jiffies_to_msecs(HZ * bfqq->service_from_backlogged / ++ bfqd->bfq_wr_max_softrt_rate)); ++ ++ return max(bfqq->last_idle_bklogged + ++ HZ * bfqq->service_from_backlogged / ++ bfqd->bfq_wr_max_softrt_rate, ++ jiffies + nsecs_to_jiffies(bfqq->bfqd->bfq_slice_idle) + 4); ++} ++ ++/* ++ * Return the farthest future time instant according to jiffies ++ * macros. ++ */ ++static unsigned long bfq_greatest_from_now(void) ++{ ++ return jiffies + MAX_JIFFY_OFFSET; ++} ++ ++/* ++ * Return the farthest past time instant according to jiffies ++ * macros. ++ */ ++static unsigned long bfq_smallest_from_now(void) ++{ ++ return jiffies - MAX_JIFFY_OFFSET; ++} ++ ++/** ++ * bfq_bfqq_expire - expire a queue. ++ * @bfqd: device owning the queue. ++ * @bfqq: the queue to expire. ++ * @compensate: if true, compensate for the time spent idling. ++ * @reason: the reason causing the expiration. ++ * ++ * If the process associated with bfqq does slow I/O (e.g., because it ++ * issues random requests), we charge bfqq with the time it has been ++ * in service instead of the service it has received (see ++ * bfq_bfqq_charge_time for details on how this goal is achieved). As ++ * a consequence, bfqq will typically get higher timestamps upon ++ * reactivation, and hence it will be rescheduled as if it had ++ * received more service than what it has actually received. In the ++ * end, bfqq receives less service in proportion to how slowly its ++ * associated process consumes its budgets (and hence how seriously it ++ * tends to lower the throughput). In addition, this time-charging ++ * strategy guarantees time fairness among slow processes. In ++ * contrast, if the process associated with bfqq is not slow, we ++ * charge bfqq exactly with the service it has received. ++ * ++ * Charging time to the first type of queues and the exact service to ++ * the other has the effect of using the WF2Q+ policy to schedule the ++ * former on a timeslice basis, without violating service domain ++ * guarantees among the latter. ++ */ ++static void bfq_bfqq_expire(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ bool compensate, ++ enum bfqq_expiration reason) ++{ ++ bool slow; ++ unsigned long delta = 0; ++ struct bfq_entity *entity = &bfqq->entity; ++ int ref; ++ ++ BUG_ON(bfqq != bfqd->in_service_queue); ++ ++ /* ++ * Check whether the process is slow (see bfq_bfqq_is_slow). ++ */ ++ slow = bfq_bfqq_is_slow(bfqd, bfqq, compensate, reason, &delta); ++ ++ /* ++ * Increase service_from_backlogged before next statement, ++ * because the possible next invocation of ++ * bfq_bfqq_charge_time would likely inflate ++ * entity->service. In contrast, service_from_backlogged must ++ * contain real service, to enable the soft real-time ++ * heuristic to correctly compute the bandwidth consumed by ++ * bfqq. ++ */ ++ bfqq->service_from_backlogged += entity->service; ++ ++ /* ++ * As above explained, charge slow (typically seeky) and ++ * timed-out queues with the time and not the service ++ * received, to favor sequential workloads. ++ * ++ * Processes doing I/O in the slower disk zones will tend to ++ * be slow(er) even if not seeky. Therefore, since the ++ * estimated peak rate is actually an average over the disk ++ * surface, these processes may timeout just for bad luck. To ++ * avoid punishing them, do not charge time to processes that ++ * succeeded in consuming at least 2/3 of their budget. This ++ * allows BFQ to preserve enough elasticity to still perform ++ * bandwidth, and not time, distribution with little unlucky ++ * or quasi-sequential processes. ++ */ ++ if (bfqq->wr_coeff == 1 && ++ (slow || ++ (reason == BFQ_BFQQ_BUDGET_TIMEOUT && ++ bfq_bfqq_budget_left(bfqq) >= entity->budget / 3))) ++ bfq_bfqq_charge_time(bfqd, bfqq, delta); ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ if (reason == BFQ_BFQQ_TOO_IDLE && ++ entity->service <= 2 * entity->budget / 10) ++ bfq_clear_bfqq_IO_bound(bfqq); ++ ++ if (bfqd->low_latency && bfqq->wr_coeff == 1) ++ bfqq->last_wr_start_finish = jiffies; ++ ++ if (bfqd->low_latency && bfqd->bfq_wr_max_softrt_rate > 0 && ++ RB_EMPTY_ROOT(&bfqq->sort_list)) { ++ /* ++ * If we get here, and there are no outstanding ++ * requests, then the request pattern is isochronous ++ * (see the comments on the function ++ * bfq_bfqq_softrt_next_start()). Thus we can compute ++ * soft_rt_next_start. If, instead, the queue still ++ * has outstanding requests, then we have to wait for ++ * the completion of all the outstanding requests to ++ * discover whether the request pattern is actually ++ * isochronous. ++ */ ++ BUG_ON(bfqd->busy_queues < 1); ++ if (bfqq->dispatched == 0) { ++ bfqq->soft_rt_next_start = ++ bfq_bfqq_softrt_next_start(bfqd, bfqq); ++ bfq_log_bfqq(bfqd, bfqq, "new soft_rt_next %lu", ++ bfqq->soft_rt_next_start); ++ } else { ++ /* ++ * The application is still waiting for the ++ * completion of one or more requests: ++ * prevent it from possibly being incorrectly ++ * deemed as soft real-time by setting its ++ * soft_rt_next_start to infinity. In fact, ++ * without this assignment, the application ++ * would be incorrectly deemed as soft ++ * real-time if: ++ * 1) it issued a new request before the ++ * completion of all its in-flight ++ * requests, and ++ * 2) at that time, its soft_rt_next_start ++ * happened to be in the past. ++ */ ++ bfqq->soft_rt_next_start = ++ bfq_greatest_from_now(); ++ /* ++ * Schedule an update of soft_rt_next_start to when ++ * the task may be discovered to be isochronous. ++ */ ++ bfq_mark_bfqq_softrt_update(bfqq); ++ } ++ } ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "expire (%d, slow %d, num_disp %d, idle_win %d, weight %d)", ++ reason, slow, bfqq->dispatched, ++ bfq_bfqq_idle_window(bfqq), entity->weight); ++ ++ /* ++ * Increase, decrease or leave budget unchanged according to ++ * reason. ++ */ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ __bfq_bfqq_recalc_budget(bfqd, bfqq, reason); ++ BUG_ON(bfqq->next_rq == NULL && ++ bfqq->entity.budget < bfqq->entity.service); ++ ref = bfqq->ref; ++ __bfq_bfqq_expire(bfqd, bfqq); ++ ++ BUG_ON(ref > 1 && ++ !bfq_bfqq_busy(bfqq) && reason == BFQ_BFQQ_BUDGET_EXHAUSTED && ++ !bfq_class_idle(bfqq)); ++ ++ /* mark bfqq as waiting a request only if a bic still points to it */ ++ if (ref > 1 && !bfq_bfqq_busy(bfqq) && ++ reason != BFQ_BFQQ_BUDGET_TIMEOUT && ++ reason != BFQ_BFQQ_BUDGET_EXHAUSTED) ++ bfq_mark_bfqq_non_blocking_wait_rq(bfqq); ++} ++ ++/* ++ * Budget timeout is not implemented through a dedicated timer, but ++ * just checked on request arrivals and completions, as well as on ++ * idle timer expirations. ++ */ ++static bool bfq_bfqq_budget_timeout(struct bfq_queue *bfqq) ++{ ++ return time_is_before_eq_jiffies(bfqq->budget_timeout); ++} ++ ++/* ++ * If we expire a queue that is actively waiting (i.e., with the ++ * device idled) for the arrival of a new request, then we may incur ++ * the timestamp misalignment problem described in the body of the ++ * function __bfq_activate_entity. Hence we return true only if this ++ * condition does not hold, or if the queue is slow enough to deserve ++ * only to be kicked off for preserving a high throughput. ++ */ ++static bool bfq_may_expire_for_budg_timeout(struct bfq_queue *bfqq) ++{ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "may_budget_timeout: wait_request %d left %d timeout %d", ++ bfq_bfqq_wait_request(bfqq), ++ bfq_bfqq_budget_left(bfqq) >= bfqq->entity.budget / 3, ++ bfq_bfqq_budget_timeout(bfqq)); ++ ++ return (!bfq_bfqq_wait_request(bfqq) || ++ bfq_bfqq_budget_left(bfqq) >= bfqq->entity.budget / 3) ++ && ++ bfq_bfqq_budget_timeout(bfqq); ++} ++ ++/* ++ * For a queue that becomes empty, device idling is allowed only if ++ * this function returns true for that queue. As a consequence, since ++ * device idling plays a critical role for both throughput boosting ++ * and service guarantees, the return value of this function plays a ++ * critical role as well. ++ * ++ * In a nutshell, this function returns true only if idling is ++ * beneficial for throughput or, even if detrimental for throughput, ++ * idling is however necessary to preserve service guarantees (low ++ * latency, desired throughput distribution, ...). In particular, on ++ * NCQ-capable devices, this function tries to return false, so as to ++ * help keep the drives' internal queues full, whenever this helps the ++ * device boost the throughput without causing any service-guarantee ++ * issue. ++ * ++ * In more detail, the return value of this function is obtained by, ++ * first, computing a number of boolean variables that take into ++ * account throughput and service-guarantee issues, and, then, ++ * combining these variables in a logical expression. Most of the ++ * issues taken into account are not trivial. We discuss these issues ++ * while introducing the variables. ++ */ ++static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq) ++{ ++ struct bfq_data *bfqd = bfqq->bfqd; ++ bool idling_boosts_thr, idling_boosts_thr_without_issues, ++ idling_needed_for_service_guarantees, ++ asymmetric_scenario; ++ ++ if (bfqd->strict_guarantees) ++ return true; ++ ++ /* ++ * The next variable takes into account the cases where idling ++ * boosts the throughput. ++ * ++ * The value of the variable is computed considering, first, that ++ * idling is virtually always beneficial for the throughput if: ++ * (a) the device is not NCQ-capable, or ++ * (b) regardless of the presence of NCQ, the device is rotational ++ * and the request pattern for bfqq is I/O-bound and sequential. ++ * ++ * Secondly, and in contrast to the above item (b), idling an ++ * NCQ-capable flash-based device would not boost the ++ * throughput even with sequential I/O; rather it would lower ++ * the throughput in proportion to how fast the device ++ * is. Accordingly, the next variable is true if any of the ++ * above conditions (a) and (b) is true, and, in particular, ++ * happens to be false if bfqd is an NCQ-capable flash-based ++ * device. ++ */ ++ idling_boosts_thr = !bfqd->hw_tag || ++ (!blk_queue_nonrot(bfqd->queue) && bfq_bfqq_IO_bound(bfqq) && ++ bfq_bfqq_idle_window(bfqq)); ++ ++ /* ++ * The value of the next variable, ++ * idling_boosts_thr_without_issues, is equal to that of ++ * idling_boosts_thr, unless a special case holds. In this ++ * special case, described below, idling may cause problems to ++ * weight-raised queues. ++ * ++ * When the request pool is saturated (e.g., in the presence ++ * of write hogs), if the processes associated with ++ * non-weight-raised queues ask for requests at a lower rate, ++ * then processes associated with weight-raised queues have a ++ * higher probability to get a request from the pool ++ * immediately (or at least soon) when they need one. Thus ++ * they have a higher probability to actually get a fraction ++ * of the device throughput proportional to their high ++ * weight. This is especially true with NCQ-capable drives, ++ * which enqueue several requests in advance, and further ++ * reorder internally-queued requests. ++ * ++ * For this reason, we force to false the value of ++ * idling_boosts_thr_without_issues if there are weight-raised ++ * busy queues. In this case, and if bfqq is not weight-raised, ++ * this guarantees that the device is not idled for bfqq (if, ++ * instead, bfqq is weight-raised, then idling will be ++ * guaranteed by another variable, see below). Combined with ++ * the timestamping rules of BFQ (see [1] for details), this ++ * behavior causes bfqq, and hence any sync non-weight-raised ++ * queue, to get a lower number of requests served, and thus ++ * to ask for a lower number of requests from the request ++ * pool, before the busy weight-raised queues get served ++ * again. This often mitigates starvation problems in the ++ * presence of heavy write workloads and NCQ, thereby ++ * guaranteeing a higher application and system responsiveness ++ * in these hostile scenarios. ++ */ ++ idling_boosts_thr_without_issues = idling_boosts_thr && ++ bfqd->wr_busy_queues == 0; ++ ++ /* ++ * There is then a case where idling must be performed not ++ * for throughput concerns, but to preserve service ++ * guarantees. ++ * ++ * To introduce this case, we can note that allowing the drive ++ * to enqueue more than one request at a time, and hence ++ * delegating de facto final scheduling decisions to the ++ * drive's internal scheduler, entails loss of control on the ++ * actual request service order. In particular, the critical ++ * situation is when requests from different processes happen ++ * to be present, at the same time, in the internal queue(s) ++ * of the drive. In such a situation, the drive, by deciding ++ * the service order of the internally-queued requests, does ++ * determine also the actual throughput distribution among ++ * these processes. But the drive typically has no notion or ++ * concern about per-process throughput distribution, and ++ * makes its decisions only on a per-request basis. Therefore, ++ * the service distribution enforced by the drive's internal ++ * scheduler is likely to coincide with the desired ++ * device-throughput distribution only in a completely ++ * symmetric scenario where: ++ * (i) each of these processes must get the same throughput as ++ * the others; ++ * (ii) all these processes have the same I/O pattern ++ * (either sequential or random). ++ * In fact, in such a scenario, the drive will tend to treat ++ * the requests of each of these processes in about the same ++ * way as the requests of the others, and thus to provide ++ * each of these processes with about the same throughput ++ * (which is exactly the desired throughput distribution). In ++ * contrast, in any asymmetric scenario, device idling is ++ * certainly needed to guarantee that bfqq receives its ++ * assigned fraction of the device throughput (see [1] for ++ * details). ++ * ++ * We address this issue by controlling, actually, only the ++ * symmetry sub-condition (i), i.e., provided that ++ * sub-condition (i) holds, idling is not performed, ++ * regardless of whether sub-condition (ii) holds. In other ++ * words, only if sub-condition (i) holds, then idling is ++ * allowed, and the device tends to be prevented from queueing ++ * many requests, possibly of several processes. The reason ++ * for not controlling also sub-condition (ii) is that we ++ * exploit preemption to preserve guarantees in case of ++ * symmetric scenarios, even if (ii) does not hold, as ++ * explained in the next two paragraphs. ++ * ++ * Even if a queue, say Q, is expired when it remains idle, Q ++ * can still preempt the new in-service queue if the next ++ * request of Q arrives soon (see the comments on ++ * bfq_bfqq_update_budg_for_activation). If all queues and ++ * groups have the same weight, this form of preemption, ++ * combined with the hole-recovery heuristic described in the ++ * comments on function bfq_bfqq_update_budg_for_activation, ++ * are enough to preserve a correct bandwidth distribution in ++ * the mid term, even without idling. In fact, even if not ++ * idling allows the internal queues of the device to contain ++ * many requests, and thus to reorder requests, we can rather ++ * safely assume that the internal scheduler still preserves a ++ * minimum of mid-term fairness. The motivation for using ++ * preemption instead of idling is that, by not idling, ++ * service guarantees are preserved without minimally ++ * sacrificing throughput. In other words, both a high ++ * throughput and its desired distribution are obtained. ++ * ++ * More precisely, this preemption-based, idleless approach ++ * provides fairness in terms of IOPS, and not sectors per ++ * second. This can be seen with a simple example. Suppose ++ * that there are two queues with the same weight, but that ++ * the first queue receives requests of 8 sectors, while the ++ * second queue receives requests of 1024 sectors. In ++ * addition, suppose that each of the two queues contains at ++ * most one request at a time, which implies that each queue ++ * always remains idle after it is served. Finally, after ++ * remaining idle, each queue receives very quickly a new ++ * request. It follows that the two queues are served ++ * alternatively, preempting each other if needed. This ++ * implies that, although both queues have the same weight, ++ * the queue with large requests receives a service that is ++ * 1024/8 times as high as the service received by the other ++ * queue. ++ * ++ * On the other hand, device idling is performed, and thus ++ * pure sector-domain guarantees are provided, for the ++ * following queues, which are likely to need stronger ++ * throughput guarantees: weight-raised queues, and queues ++ * with a higher weight than other queues. When such queues ++ * are active, sub-condition (i) is false, which triggers ++ * device idling. ++ * ++ * According to the above considerations, the next variable is ++ * true (only) if sub-condition (i) holds. To compute the ++ * value of this variable, we not only use the return value of ++ * the function bfq_symmetric_scenario(), but also check ++ * whether bfqq is being weight-raised, because ++ * bfq_symmetric_scenario() does not take into account also ++ * weight-raised queues (see comments on ++ * bfq_weights_tree_add()). ++ * ++ * As a side note, it is worth considering that the above ++ * device-idling countermeasures may however fail in the ++ * following unlucky scenario: if idling is (correctly) ++ * disabled in a time period during which all symmetry ++ * sub-conditions hold, and hence the device is allowed to ++ * enqueue many requests, but at some later point in time some ++ * sub-condition stops to hold, then it may become impossible ++ * to let requests be served in the desired order until all ++ * the requests already queued in the device have been served. ++ */ ++ asymmetric_scenario = bfqq->wr_coeff > 1 || ++ !bfq_symmetric_scenario(bfqd); ++ ++ /* ++ * Finally, there is a case where maximizing throughput is the ++ * best choice even if it may cause unfairness toward ++ * bfqq. Such a case is when bfqq became active in a burst of ++ * queue activations. Queues that became active during a large ++ * burst benefit only from throughput, as discussed in the ++ * comments on bfq_handle_burst. Thus, if bfqq became active ++ * in a burst and not idling the device maximizes throughput, ++ * then the device must no be idled, because not idling the ++ * device provides bfqq and all other queues in the burst with ++ * maximum benefit. Combining this and the above case, we can ++ * now establish when idling is actually needed to preserve ++ * service guarantees. ++ */ ++ idling_needed_for_service_guarantees = ++ asymmetric_scenario && !bfq_bfqq_in_large_burst(bfqq); ++ ++ /* ++ * We have now all the components we need to compute the return ++ * value of the function, which is true only if both the following ++ * conditions hold: ++ * 1) bfqq is sync, because idling make sense only for sync queues; ++ * 2) idling either boosts the throughput (without issues), or ++ * is necessary to preserve service guarantees. ++ */ ++ bfq_log_bfqq(bfqd, bfqq, "may_idle: sync %d idling_boosts_thr %d", ++ bfq_bfqq_sync(bfqq), idling_boosts_thr); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "may_idle: wr_busy %d boosts %d IO-bound %d guar %d", ++ bfqd->wr_busy_queues, ++ idling_boosts_thr_without_issues, ++ bfq_bfqq_IO_bound(bfqq), ++ idling_needed_for_service_guarantees); ++ ++ return bfq_bfqq_sync(bfqq) && ++ (idling_boosts_thr_without_issues || ++ idling_needed_for_service_guarantees); ++} ++ ++/* ++ * If the in-service queue is empty but the function bfq_bfqq_may_idle ++ * returns true, then: ++ * 1) the queue must remain in service and cannot be expired, and ++ * 2) the device must be idled to wait for the possible arrival of a new ++ * request for the queue. ++ * See the comments on the function bfq_bfqq_may_idle for the reasons ++ * why performing device idling is the best choice to boost the throughput ++ * and preserve service guarantees when bfq_bfqq_may_idle itself ++ * returns true. ++ */ ++static bool bfq_bfqq_must_idle(struct bfq_queue *bfqq) ++{ ++ struct bfq_data *bfqd = bfqq->bfqd; ++ ++ return RB_EMPTY_ROOT(&bfqq->sort_list) && bfqd->bfq_slice_idle != 0 && ++ bfq_bfqq_may_idle(bfqq); ++} ++ ++/* ++ * Select a queue for service. If we have a current queue in service, ++ * check whether to continue servicing it, or retrieve and set a new one. ++ */ ++static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq; ++ struct request *next_rq; ++ enum bfqq_expiration reason = BFQ_BFQQ_BUDGET_TIMEOUT; ++ ++ bfqq = bfqd->in_service_queue; ++ if (!bfqq) ++ goto new_queue; ++ ++ bfq_log_bfqq(bfqd, bfqq, "select_queue: already in-service queue"); ++ ++ if (bfq_may_expire_for_budg_timeout(bfqq) && ++ !hrtimer_active(&bfqd->idle_slice_timer) && ++ !bfq_bfqq_must_idle(bfqq)) ++ goto expire; ++ ++check_queue: ++ /* ++ * This loop is rarely executed more than once. Even when it ++ * happens, it is much more convenient to re-execute this loop ++ * than to return NULL and trigger a new dispatch to get a ++ * request served. ++ */ ++ next_rq = bfqq->next_rq; ++ /* ++ * If bfqq has requests queued and it has enough budget left to ++ * serve them, keep the queue, otherwise expire it. ++ */ ++ if (next_rq) { ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list)); ++ ++ if (bfq_serv_to_charge(next_rq, bfqq) > ++ bfq_bfqq_budget_left(bfqq)) { ++ /* ++ * Expire the queue for budget exhaustion, ++ * which makes sure that the next budget is ++ * enough to serve the next request, even if ++ * it comes from the fifo expired path. ++ */ ++ reason = BFQ_BFQQ_BUDGET_EXHAUSTED; ++ goto expire; ++ } else { ++ /* ++ * The idle timer may be pending because we may ++ * not disable disk idling even when a new request ++ * arrives. ++ */ ++ if (bfq_bfqq_wait_request(bfqq)) { ++ BUG_ON(!hrtimer_active(&bfqd->idle_slice_timer)); ++ /* ++ * If we get here: 1) at least a new request ++ * has arrived but we have not disabled the ++ * timer because the request was too small, ++ * 2) then the block layer has unplugged ++ * the device, causing the dispatch to be ++ * invoked. ++ * ++ * Since the device is unplugged, now the ++ * requests are probably large enough to ++ * provide a reasonable throughput. ++ * So we disable idling. ++ */ ++ bfq_clear_bfqq_wait_request(bfqq); ++ hrtimer_try_to_cancel(&bfqd->idle_slice_timer); ++ bfqg_stats_update_idle_time(bfqq_group(bfqq)); ++ } ++ goto keep_queue; ++ } ++ } ++ ++ /* ++ * No requests pending. However, if the in-service queue is idling ++ * for a new request, or has requests waiting for a completion and ++ * may idle after their completion, then keep it anyway. ++ */ ++ if (hrtimer_active(&bfqd->idle_slice_timer) || ++ (bfqq->dispatched != 0 && bfq_bfqq_may_idle(bfqq))) { ++ bfqq = NULL; ++ goto keep_queue; ++ } ++ ++ reason = BFQ_BFQQ_NO_MORE_REQUESTS; ++expire: ++ bfq_bfqq_expire(bfqd, bfqq, false, reason); ++new_queue: ++ bfqq = bfq_set_in_service_queue(bfqd); ++ if (bfqq) { ++ bfq_log_bfqq(bfqd, bfqq, "select_queue: checking new queue"); ++ goto check_queue; ++ } ++keep_queue: ++ if (bfqq) ++ bfq_log_bfqq(bfqd, bfqq, "select_queue: returned this queue"); ++ else ++ bfq_log(bfqd, "select_queue: no queue returned"); ++ ++ return bfqq; ++} ++ ++static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ ++ if (bfqq->wr_coeff > 1) { /* queue is being weight-raised */ ++ BUG_ON(bfqq->wr_cur_max_time == bfqd->bfq_wr_rt_max_time && ++ time_is_after_jiffies(bfqq->last_wr_start_finish)); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "raising period dur %u/%u msec, old coeff %u, w %d(%d)", ++ jiffies_to_msecs(jiffies - bfqq->last_wr_start_finish), ++ jiffies_to_msecs(bfqq->wr_cur_max_time), ++ bfqq->wr_coeff, ++ bfqq->entity.weight, bfqq->entity.orig_weight); ++ ++ BUG_ON(bfqq != bfqd->in_service_queue && entity->weight != ++ entity->orig_weight * bfqq->wr_coeff); ++ if (entity->prio_changed) ++ bfq_log_bfqq(bfqd, bfqq, "WARN: pending prio change"); ++ ++ /* ++ * If the queue was activated in a burst, or too much ++ * time has elapsed from the beginning of this ++ * weight-raising period, then end weight raising. ++ */ ++ if (bfq_bfqq_in_large_burst(bfqq)) ++ bfq_bfqq_end_wr(bfqq); ++ else if (time_is_before_jiffies(bfqq->last_wr_start_finish + ++ bfqq->wr_cur_max_time)) { ++ if (bfqq->wr_cur_max_time != bfqd->bfq_wr_rt_max_time || ++ time_is_before_jiffies(bfqq->wr_start_at_switch_to_srt + ++ bfq_wr_duration(bfqd))) ++ bfq_bfqq_end_wr(bfqq); ++ else { ++ /* switch back to interactive wr */ ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ bfqq->last_wr_start_finish = ++ bfqq->wr_start_at_switch_to_srt; ++ BUG_ON(time_is_after_jiffies( ++ bfqq->last_wr_start_finish)); ++ bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqd, bfqq, ++ "back to interactive wr"); ++ } ++ } ++ } ++ /* ++ * To improve latency (for this or other queues), immediately ++ * update weight both if it must be raised and if it must be ++ * lowered. Since, entity may be on some active tree here, and ++ * might have a pending change of its ioprio class, invoke ++ * next function with the last parameter unset (see the ++ * comments on the function). ++ */ ++ if ((entity->weight > entity->orig_weight) != (bfqq->wr_coeff > 1)) ++ __bfq_entity_update_weight_prio(bfq_entity_service_tree(entity), ++ entity, false); ++} ++ ++/* ++ * Dispatch one request from bfqq, moving it to the request queue ++ * dispatch list. ++ */ ++static int bfq_dispatch_request(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ int dispatched = 0; ++ struct request *rq = bfqq->next_rq; ++ unsigned long service_to_charge; ++ ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list)); ++ BUG_ON(!rq); ++ service_to_charge = bfq_serv_to_charge(rq, bfqq); ++ ++ BUG_ON(service_to_charge > bfq_bfqq_budget_left(bfqq)); ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ bfq_bfqq_served(bfqq, service_to_charge); ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ bfq_dispatch_insert(bfqd->queue, rq); ++ ++ /* ++ * If weight raising has to terminate for bfqq, then next ++ * function causes an immediate update of bfqq's weight, ++ * without waiting for next activation. As a consequence, on ++ * expiration, bfqq will be timestamped as if has never been ++ * weight-raised during this service slot, even if it has ++ * received part or even most of the service as a ++ * weight-raised queue. This inflates bfqq's timestamps, which ++ * is beneficial, as bfqq is then more willing to leave the ++ * device immediately to possible other weight-raised queues. ++ */ ++ bfq_update_wr_data(bfqd, bfqq); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "dispatched %u sec req (%llu), budg left %d", ++ blk_rq_sectors(rq), ++ (unsigned long long) blk_rq_pos(rq), ++ bfq_bfqq_budget_left(bfqq)); ++ ++ dispatched++; ++ ++ if (!bfqd->in_service_bic) { ++ atomic_long_inc(&RQ_BIC(rq)->icq.ioc->refcount); ++ bfqd->in_service_bic = RQ_BIC(rq); ++ } ++ ++ if (bfqd->busy_queues > 1 && bfq_class_idle(bfqq)) ++ goto expire; ++ ++ return dispatched; ++ ++expire: ++ bfq_bfqq_expire(bfqd, bfqq, false, BFQ_BFQQ_BUDGET_EXHAUSTED); ++ return dispatched; ++} ++ ++static int __bfq_forced_dispatch_bfqq(struct bfq_queue *bfqq) ++{ ++ int dispatched = 0; ++ ++ while (bfqq->next_rq) { ++ bfq_dispatch_insert(bfqq->bfqd->queue, bfqq->next_rq); ++ dispatched++; ++ } ++ ++ BUG_ON(!list_empty(&bfqq->fifo)); ++ return dispatched; ++} ++ ++/* ++ * Drain our current requests. ++ * Used for barriers and when switching io schedulers on-the-fly. ++ */ ++static int bfq_forced_dispatch(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq, *n; ++ struct bfq_service_tree *st; ++ int dispatched = 0; ++ ++ bfqq = bfqd->in_service_queue; ++ if (bfqq) ++ __bfq_bfqq_expire(bfqd, bfqq); ++ ++ /* ++ * Loop through classes, and be careful to leave the scheduler ++ * in a consistent state, as feedback mechanisms and vtime ++ * updates cannot be disabled during the process. ++ */ ++ list_for_each_entry_safe(bfqq, n, &bfqd->active_list, bfqq_list) { ++ st = bfq_entity_service_tree(&bfqq->entity); ++ ++ dispatched += __bfq_forced_dispatch_bfqq(bfqq); ++ ++ bfqq->max_budget = bfq_max_budget(bfqd); ++ bfq_forget_idle(st); ++ } ++ ++ BUG_ON(bfqd->busy_queues != 0); ++ ++ return dispatched; ++} ++ ++static int bfq_dispatch_requests(struct request_queue *q, int force) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct bfq_queue *bfqq; ++ ++ bfq_log(bfqd, "dispatch requests: %d busy queues", bfqd->busy_queues); ++ ++ if (bfqd->busy_queues == 0) ++ return 0; ++ ++ if (unlikely(force)) ++ return bfq_forced_dispatch(bfqd); ++ ++ /* ++ * Force device to serve one request at a time if ++ * strict_guarantees is true. Forcing this service scheme is ++ * currently the ONLY way to guarantee that the request ++ * service order enforced by the scheduler is respected by a ++ * queueing device. Otherwise the device is free even to make ++ * some unlucky request wait for as long as the device ++ * wishes. ++ * ++ * Of course, serving one request at at time may cause loss of ++ * throughput. ++ */ ++ if (bfqd->strict_guarantees && bfqd->rq_in_driver > 0) ++ return 0; ++ ++ bfqq = bfq_select_queue(bfqd); ++ if (!bfqq) ++ return 0; ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ BUG_ON(bfq_bfqq_wait_request(bfqq)); ++ ++ if (!bfq_dispatch_request(bfqd, bfqq)) ++ return 0; ++ ++ bfq_log_bfqq(bfqd, bfqq, "dispatched %s request", ++ bfq_bfqq_sync(bfqq) ? "sync" : "async"); ++ ++ BUG_ON(bfqq->next_rq == NULL && ++ bfqq->entity.budget < bfqq->entity.service); ++ return 1; ++} ++ ++/* ++ * Task holds one reference to the queue, dropped when task exits. Each rq ++ * in-flight on this queue also holds a reference, dropped when rq is freed. ++ * ++ * Queue lock must be held here. Recall not to use bfqq after calling ++ * this function on it. ++ */ ++static void bfq_put_queue(struct bfq_queue *bfqq) ++{ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ struct bfq_group *bfqg = bfqq_group(bfqq); ++#endif ++ ++ BUG_ON(bfqq->ref <= 0); ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p %d", bfqq, bfqq->ref); ++ bfqq->ref--; ++ if (bfqq->ref) ++ return; ++ ++ BUG_ON(rb_first(&bfqq->sort_list)); ++ BUG_ON(bfqq->allocated[READ] + bfqq->allocated[WRITE] != 0); ++ BUG_ON(bfqq->entity.tree); ++ BUG_ON(bfq_bfqq_busy(bfqq)); ++ ++ if (bfq_bfqq_sync(bfqq)) ++ /* ++ * The fact that this queue is being destroyed does not ++ * invalidate the fact that this queue may have been ++ * activated during the current burst. As a consequence, ++ * although the queue does not exist anymore, and hence ++ * needs to be removed from the burst list if there, ++ * the burst size has not to be decremented. ++ */ ++ hlist_del_init(&bfqq->burst_list_node); ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); ++ ++ kmem_cache_free(bfq_pool, bfqq); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ bfqg_put(bfqg); ++#endif ++} ++ ++static void bfq_put_cooperator(struct bfq_queue *bfqq) ++{ ++ struct bfq_queue *__bfqq, *next; ++ ++ /* ++ * If this queue was scheduled to merge with another queue, be ++ * sure to drop the reference taken on that queue (and others in ++ * the merge chain). See bfq_setup_merge and bfq_merge_bfqqs. ++ */ ++ __bfqq = bfqq->new_bfqq; ++ while (__bfqq) { ++ if (__bfqq == bfqq) ++ break; ++ next = __bfqq->new_bfqq; ++ bfq_put_queue(__bfqq); ++ __bfqq = next; ++ } ++} ++ ++static void bfq_exit_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ if (bfqq == bfqd->in_service_queue) { ++ __bfq_bfqq_expire(bfqd, bfqq); ++ bfq_schedule_dispatch(bfqd); ++ } ++ ++ bfq_log_bfqq(bfqd, bfqq, "exit_bfqq: %p, %d", bfqq, bfqq->ref); ++ ++ bfq_put_cooperator(bfqq); ++ ++ bfq_put_queue(bfqq); /* release process reference */ ++} ++ ++static void bfq_init_icq(struct io_cq *icq) ++{ ++ icq_to_bic(icq)->ttime.last_end_request = ktime_get_ns() - (1ULL<<32); ++} ++ ++static void bfq_exit_icq(struct io_cq *icq) ++{ ++ struct bfq_io_cq *bic = icq_to_bic(icq); ++ struct bfq_data *bfqd = bic_to_bfqd(bic); ++ ++ if (bic_to_bfqq(bic, false)) { ++ bfq_exit_bfqq(bfqd, bic_to_bfqq(bic, false)); ++ bic_set_bfqq(bic, NULL, false); ++ } ++ ++ if (bic_to_bfqq(bic, true)) { ++ /* ++ * If the bic is using a shared queue, put the reference ++ * taken on the io_context when the bic started using a ++ * shared bfq_queue. ++ */ ++ if (bfq_bfqq_coop(bic_to_bfqq(bic, true))) ++ put_io_context(icq->ioc); ++ bfq_exit_bfqq(bfqd, bic_to_bfqq(bic, true)); ++ bic_set_bfqq(bic, NULL, true); ++ } ++} ++ ++/* ++ * Update the entity prio values; note that the new values will not ++ * be used until the next (re)activation. ++ */ ++static void bfq_set_next_ioprio_data(struct bfq_queue *bfqq, ++ struct bfq_io_cq *bic) ++{ ++ struct task_struct *tsk = current; ++ int ioprio_class; ++ ++ ioprio_class = IOPRIO_PRIO_CLASS(bic->ioprio); ++ switch (ioprio_class) { ++ default: ++ dev_err(bfqq->bfqd->queue->backing_dev_info->dev, ++ "bfq: bad prio class %d\n", ioprio_class); ++ case IOPRIO_CLASS_NONE: ++ /* ++ * No prio set, inherit CPU scheduling settings. ++ */ ++ bfqq->new_ioprio = task_nice_ioprio(tsk); ++ bfqq->new_ioprio_class = task_nice_ioclass(tsk); ++ break; ++ case IOPRIO_CLASS_RT: ++ bfqq->new_ioprio = IOPRIO_PRIO_DATA(bic->ioprio); ++ bfqq->new_ioprio_class = IOPRIO_CLASS_RT; ++ break; ++ case IOPRIO_CLASS_BE: ++ bfqq->new_ioprio = IOPRIO_PRIO_DATA(bic->ioprio); ++ bfqq->new_ioprio_class = IOPRIO_CLASS_BE; ++ break; ++ case IOPRIO_CLASS_IDLE: ++ bfqq->new_ioprio_class = IOPRIO_CLASS_IDLE; ++ bfqq->new_ioprio = 7; ++ bfq_clear_bfqq_idle_window(bfqq); ++ break; ++ } ++ ++ if (bfqq->new_ioprio >= IOPRIO_BE_NR) { ++ pr_crit("bfq_set_next_ioprio_data: new_ioprio %d\n", ++ bfqq->new_ioprio); ++ BUG(); ++ } ++ ++ bfqq->entity.new_weight = bfq_ioprio_to_weight(bfqq->new_ioprio); ++ bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "set_next_ioprio_data: bic_class %d prio %d class %d", ++ ioprio_class, bfqq->new_ioprio, bfqq->new_ioprio_class); ++} ++ ++static void bfq_check_ioprio_change(struct bfq_io_cq *bic, struct bio *bio) ++{ ++ struct bfq_data *bfqd = bic_to_bfqd(bic); ++ struct bfq_queue *bfqq; ++ unsigned long uninitialized_var(flags); ++ int ioprio = bic->icq.ioc->ioprio; ++ ++ /* ++ * This condition may trigger on a newly created bic, be sure to ++ * drop the lock before returning. ++ */ ++ if (unlikely(!bfqd) || likely(bic->ioprio == ioprio)) ++ return; ++ ++ bic->ioprio = ioprio; ++ ++ bfqq = bic_to_bfqq(bic, false); ++ if (bfqq) { ++ /* release process reference on this queue */ ++ bfq_put_queue(bfqq); ++ bfqq = bfq_get_queue(bfqd, bio, BLK_RW_ASYNC, bic); ++ bic_set_bfqq(bic, bfqq, false); ++ bfq_log_bfqq(bfqd, bfqq, ++ "check_ioprio_change: bfqq %p %d", ++ bfqq, bfqq->ref); ++ } ++ ++ bfqq = bic_to_bfqq(bic, true); ++ if (bfqq) ++ bfq_set_next_ioprio_data(bfqq, bic); ++} ++ ++static void bfq_init_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ struct bfq_io_cq *bic, pid_t pid, int is_sync) ++{ ++ RB_CLEAR_NODE(&bfqq->entity.rb_node); ++ INIT_LIST_HEAD(&bfqq->fifo); ++ INIT_HLIST_NODE(&bfqq->burst_list_node); ++ BUG_ON(!hlist_unhashed(&bfqq->burst_list_node)); ++ ++ bfqq->ref = 0; ++ bfqq->bfqd = bfqd; ++ ++ if (bic) ++ bfq_set_next_ioprio_data(bfqq, bic); ++ ++ if (is_sync) { ++ if (!bfq_class_idle(bfqq)) ++ bfq_mark_bfqq_idle_window(bfqq); ++ bfq_mark_bfqq_sync(bfqq); ++ bfq_mark_bfqq_just_created(bfqq); ++ } else ++ bfq_clear_bfqq_sync(bfqq); ++ bfq_mark_bfqq_IO_bound(bfqq); ++ ++ /* Tentative initial value to trade off between thr and lat */ ++ bfqq->max_budget = (2 * bfq_max_budget(bfqd)) / 3; ++ bfqq->pid = pid; ++ ++ bfqq->wr_coeff = 1; ++ bfqq->last_wr_start_finish = jiffies; ++ bfqq->wr_start_at_switch_to_srt = bfq_smallest_from_now(); ++ bfqq->budget_timeout = bfq_smallest_from_now(); ++ bfqq->split_time = bfq_smallest_from_now(); ++ ++ /* ++ * Set to the value for which bfqq will not be deemed as ++ * soft rt when it becomes backlogged. ++ */ ++ bfqq->soft_rt_next_start = bfq_greatest_from_now(); ++ ++ /* first request is almost certainly seeky */ ++ bfqq->seek_history = 1; ++} ++ ++static struct bfq_queue **bfq_async_queue_prio(struct bfq_data *bfqd, ++ struct bfq_group *bfqg, ++ int ioprio_class, int ioprio) ++{ ++ switch (ioprio_class) { ++ case IOPRIO_CLASS_RT: ++ return &bfqg->async_bfqq[0][ioprio]; ++ case IOPRIO_CLASS_NONE: ++ ioprio = IOPRIO_NORM; ++ /* fall through */ ++ case IOPRIO_CLASS_BE: ++ return &bfqg->async_bfqq[1][ioprio]; ++ case IOPRIO_CLASS_IDLE: ++ return &bfqg->async_idle_bfqq; ++ default: ++ BUG(); ++ } ++} ++ ++static struct bfq_queue *bfq_get_queue(struct bfq_data *bfqd, ++ struct bio *bio, bool is_sync, ++ struct bfq_io_cq *bic) ++{ ++ const int ioprio = IOPRIO_PRIO_DATA(bic->ioprio); ++ const int ioprio_class = IOPRIO_PRIO_CLASS(bic->ioprio); ++ struct bfq_queue **async_bfqq = NULL; ++ struct bfq_queue *bfqq; ++ struct bfq_group *bfqg; ++ ++ rcu_read_lock(); ++ ++ bfqg = bfq_find_set_group(bfqd, bio_blkcg(bio)); ++ if (!bfqg) { ++ bfqq = &bfqd->oom_bfqq; ++ goto out; ++ } ++ ++ if (!is_sync) { ++ async_bfqq = bfq_async_queue_prio(bfqd, bfqg, ioprio_class, ++ ioprio); ++ bfqq = *async_bfqq; ++ if (bfqq) ++ goto out; ++ } ++ ++ bfqq = kmem_cache_alloc_node(bfq_pool, ++ GFP_NOWAIT | __GFP_ZERO | __GFP_NOWARN, ++ bfqd->queue->node); ++ ++ if (bfqq) { ++ bfq_init_bfqq(bfqd, bfqq, bic, current->pid, ++ is_sync); ++ bfq_init_entity(&bfqq->entity, bfqg); ++ bfq_log_bfqq(bfqd, bfqq, "allocated"); ++ } else { ++ bfqq = &bfqd->oom_bfqq; ++ bfq_log_bfqq(bfqd, bfqq, "using oom bfqq"); ++ goto out; ++ } ++ ++ /* ++ * Pin the queue now that it's allocated, scheduler exit will ++ * prune it. ++ */ ++ if (async_bfqq) { ++ bfqq->ref++; /* ++ * Extra group reference, w.r.t. sync ++ * queue. This extra reference is removed ++ * only if bfqq->bfqg disappears, to ++ * guarantee that this queue is not freed ++ * until its group goes away. ++ */ ++ bfq_log_bfqq(bfqd, bfqq, "get_queue, bfqq not in async: %p, %d", ++ bfqq, bfqq->ref); ++ *async_bfqq = bfqq; ++ } ++ ++out: ++ bfqq->ref++; /* get a process reference to this queue */ ++ bfq_log_bfqq(bfqd, bfqq, "get_queue, at end: %p, %d", bfqq, bfqq->ref); ++ rcu_read_unlock(); ++ return bfqq; ++} ++ ++static void bfq_update_io_thinktime(struct bfq_data *bfqd, ++ struct bfq_io_cq *bic) ++{ ++ struct bfq_ttime *ttime = &bic->ttime; ++ u64 elapsed = ktime_get_ns() - bic->ttime.last_end_request; ++ ++ elapsed = min_t(u64, elapsed, 2 * bfqd->bfq_slice_idle); ++ ++ ttime->ttime_samples = (7*bic->ttime.ttime_samples + 256) / 8; ++ ttime->ttime_total = div_u64(7*ttime->ttime_total + 256*elapsed, 8); ++ ttime->ttime_mean = div64_ul(ttime->ttime_total + 128, ++ ttime->ttime_samples); ++} ++ ++static void ++bfq_update_io_seektime(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ struct request *rq) ++{ ++ bfqq->seek_history <<= 1; ++ bfqq->seek_history |= ++ get_sdist(bfqq->last_request_pos, rq) > BFQQ_SEEK_THR && ++ (!blk_queue_nonrot(bfqd->queue) || ++ blk_rq_sectors(rq) < BFQQ_SECT_THR_NONROT); ++} ++ ++/* ++ * Disable idle window if the process thinks too long or seeks so much that ++ * it doesn't matter. ++ */ ++static void bfq_update_idle_window(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ struct bfq_io_cq *bic) ++{ ++ int enable_idle; ++ ++ /* Don't idle for async or idle io prio class. */ ++ if (!bfq_bfqq_sync(bfqq) || bfq_class_idle(bfqq)) ++ return; ++ ++ /* Idle window just restored, statistics are meaningless. */ ++ if (time_is_after_eq_jiffies(bfqq->split_time + ++ bfqd->bfq_wr_min_idle_time)) ++ return; ++ ++ enable_idle = bfq_bfqq_idle_window(bfqq); ++ ++ if (atomic_read(&bic->icq.ioc->active_ref) == 0 || ++ bfqd->bfq_slice_idle == 0 || ++ (bfqd->hw_tag && BFQQ_SEEKY(bfqq) && ++ bfqq->wr_coeff == 1)) ++ enable_idle = 0; ++ else if (bfq_sample_valid(bic->ttime.ttime_samples)) { ++ if (bic->ttime.ttime_mean > bfqd->bfq_slice_idle && ++ bfqq->wr_coeff == 1) ++ enable_idle = 0; ++ else ++ enable_idle = 1; ++ } ++ bfq_log_bfqq(bfqd, bfqq, "update_idle_window: enable_idle %d", ++ enable_idle); ++ ++ if (enable_idle) ++ bfq_mark_bfqq_idle_window(bfqq); ++ else ++ bfq_clear_bfqq_idle_window(bfqq); ++} ++ ++/* ++ * Called when a new fs request (rq) is added to bfqq. Check if there's ++ * something we should do about it. ++ */ ++static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ struct request *rq) ++{ ++ struct bfq_io_cq *bic = RQ_BIC(rq); ++ ++ if (rq->cmd_flags & REQ_META) ++ bfqq->meta_pending++; ++ ++ bfq_update_io_thinktime(bfqd, bic); ++ bfq_update_io_seektime(bfqd, bfqq, rq); ++ if (bfqq->entity.service > bfq_max_budget(bfqd) / 8 || ++ !BFQQ_SEEKY(bfqq)) ++ bfq_update_idle_window(bfqd, bfqq, bic); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "rq_enqueued: idle_window=%d (seeky %d)", ++ bfq_bfqq_idle_window(bfqq), BFQQ_SEEKY(bfqq)); ++ ++ bfqq->last_request_pos = blk_rq_pos(rq) + blk_rq_sectors(rq); ++ ++ if (bfqq == bfqd->in_service_queue && bfq_bfqq_wait_request(bfqq)) { ++ bool small_req = bfqq->queued[rq_is_sync(rq)] == 1 && ++ blk_rq_sectors(rq) < 32; ++ bool budget_timeout = bfq_bfqq_budget_timeout(bfqq); ++ ++ /* ++ * There is just this request queued: if the request ++ * is small and the queue is not to be expired, then ++ * just exit. ++ * ++ * In this way, if the device is being idled to wait ++ * for a new request from the in-service queue, we ++ * avoid unplugging the device and committing the ++ * device to serve just a small request. On the ++ * contrary, we wait for the block layer to decide ++ * when to unplug the device: hopefully, new requests ++ * will be merged to this one quickly, then the device ++ * will be unplugged and larger requests will be ++ * dispatched. ++ */ ++ if (small_req && !budget_timeout) ++ return; ++ ++ /* ++ * A large enough request arrived, or the queue is to ++ * be expired: in both cases disk idling is to be ++ * stopped, so clear wait_request flag and reset ++ * timer. ++ */ ++ bfq_clear_bfqq_wait_request(bfqq); ++ hrtimer_try_to_cancel(&bfqd->idle_slice_timer); ++ bfqg_stats_update_idle_time(bfqq_group(bfqq)); ++ ++ /* ++ * The queue is not empty, because a new request just ++ * arrived. Hence we can safely expire the queue, in ++ * case of budget timeout, without risking that the ++ * timestamps of the queue are not updated correctly. ++ * See [1] for more details. ++ */ ++ if (budget_timeout) ++ bfq_bfqq_expire(bfqd, bfqq, false, ++ BFQ_BFQQ_BUDGET_TIMEOUT); ++ ++ /* ++ * Let the request rip immediately, or let a new queue be ++ * selected if bfqq has just been expired. ++ */ ++ __blk_run_queue(bfqd->queue); ++ } ++} ++ ++static void bfq_insert_request(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct bfq_queue *bfqq = RQ_BFQQ(rq), *new_bfqq; ++ ++ assert_spin_locked(bfqd->queue->queue_lock); ++ ++ /* ++ * An unplug may trigger a requeue of a request from the device ++ * driver: make sure we are in process context while trying to ++ * merge two bfq_queues. ++ */ ++ if (!in_interrupt()) { ++ new_bfqq = bfq_setup_cooperator(bfqd, bfqq, rq, true); ++ if (new_bfqq) { ++ if (bic_to_bfqq(RQ_BIC(rq), 1) != bfqq) ++ new_bfqq = bic_to_bfqq(RQ_BIC(rq), 1); ++ /* ++ * Release the request's reference to the old bfqq ++ * and make sure one is taken to the shared queue. ++ */ ++ new_bfqq->allocated[rq_data_dir(rq)]++; ++ bfqq->allocated[rq_data_dir(rq)]--; ++ new_bfqq->ref++; ++ bfq_clear_bfqq_just_created(bfqq); ++ if (bic_to_bfqq(RQ_BIC(rq), 1) == bfqq) ++ bfq_merge_bfqqs(bfqd, RQ_BIC(rq), ++ bfqq, new_bfqq); ++ /* ++ * rq is about to be enqueued into new_bfqq, ++ * release rq reference on bfqq ++ */ ++ bfq_put_queue(bfqq); ++ rq->elv.priv[1] = new_bfqq; ++ bfqq = new_bfqq; ++ } ++ } ++ ++ bfq_add_request(rq); ++ ++ rq->fifo_time = ktime_get_ns() + bfqd->bfq_fifo_expire[rq_is_sync(rq)]; ++ list_add_tail(&rq->queuelist, &bfqq->fifo); ++ ++ bfq_rq_enqueued(bfqd, bfqq, rq); ++} ++ ++static void bfq_update_hw_tag(struct bfq_data *bfqd) ++{ ++ bfqd->max_rq_in_driver = max_t(int, bfqd->max_rq_in_driver, ++ bfqd->rq_in_driver); ++ ++ if (bfqd->hw_tag == 1) ++ return; ++ ++ /* ++ * This sample is valid if the number of outstanding requests ++ * is large enough to allow a queueing behavior. Note that the ++ * sum is not exact, as it's not taking into account deactivated ++ * requests. ++ */ ++ if (bfqd->rq_in_driver + bfqd->queued < BFQ_HW_QUEUE_THRESHOLD) ++ return; ++ ++ if (bfqd->hw_tag_samples++ < BFQ_HW_QUEUE_SAMPLES) ++ return; ++ ++ bfqd->hw_tag = bfqd->max_rq_in_driver > BFQ_HW_QUEUE_THRESHOLD; ++ bfqd->max_rq_in_driver = 0; ++ bfqd->hw_tag_samples = 0; ++} ++ ++static void bfq_completed_request(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ struct bfq_data *bfqd = bfqq->bfqd; ++ u64 now_ns; ++ u32 delta_us; ++ ++ bfq_log_bfqq(bfqd, bfqq, "completed one req with %u sects left", ++ blk_rq_sectors(rq)); ++ ++ assert_spin_locked(bfqd->queue->queue_lock); ++ bfq_update_hw_tag(bfqd); ++ ++ BUG_ON(!bfqd->rq_in_driver); ++ BUG_ON(!bfqq->dispatched); ++ bfqd->rq_in_driver--; ++ bfqq->dispatched--; ++ bfqg_stats_update_completion(bfqq_group(bfqq), ++ rq_start_time_ns(rq), ++ rq_io_start_time_ns(rq), ++ rq->cmd_flags); ++ ++ if (!bfqq->dispatched && !bfq_bfqq_busy(bfqq)) { ++ BUG_ON(!RB_EMPTY_ROOT(&bfqq->sort_list)); ++ /* ++ * Set budget_timeout (which we overload to store the ++ * time at which the queue remains with no backlog and ++ * no outstanding request; used by the weight-raising ++ * mechanism). ++ */ ++ bfqq->budget_timeout = jiffies; ++ ++ bfq_weights_tree_remove(bfqd, &bfqq->entity, ++ &bfqd->queue_weights_tree); ++ } ++ ++ now_ns = ktime_get_ns(); ++ ++ RQ_BIC(rq)->ttime.last_end_request = now_ns; ++ ++ /* ++ * Using us instead of ns, to get a reasonable precision in ++ * computing rate in next check. ++ */ ++ delta_us = div_u64(now_ns - bfqd->last_completion, NSEC_PER_USEC); ++ ++ bfq_log(bfqd, "rq_completed: delta %uus/%luus max_size %u rate %llu/%llu", ++ delta_us, BFQ_MIN_TT/NSEC_PER_USEC, bfqd->last_rq_max_size, ++ (USEC_PER_SEC* ++ (u64)((bfqd->last_rq_max_size<>BFQ_RATE_SHIFT, ++ (USEC_PER_SEC*(u64)(1UL<<(BFQ_RATE_SHIFT-10)))>>BFQ_RATE_SHIFT); ++ ++ /* ++ * If the request took rather long to complete, and, according ++ * to the maximum request size recorded, this completion latency ++ * implies that the request was certainly served at a very low ++ * rate (less than 1M sectors/sec), then the whole observation ++ * interval that lasts up to this time instant cannot be a ++ * valid time interval for computing a new peak rate. Invoke ++ * bfq_update_rate_reset to have the following three steps ++ * taken: ++ * - close the observation interval at the last (previous) ++ * request dispatch or completion ++ * - compute rate, if possible, for that observation interval ++ * - reset to zero samples, which will trigger a proper ++ * re-initialization of the observation interval on next ++ * dispatch ++ */ ++ if (delta_us > BFQ_MIN_TT/NSEC_PER_USEC && ++ (bfqd->last_rq_max_size<last_completion = now_ns; ++ ++ /* ++ * If we are waiting to discover whether the request pattern ++ * of the task associated with the queue is actually ++ * isochronous, and both requisites for this condition to hold ++ * are now satisfied, then compute soft_rt_next_start (see the ++ * comments on the function bfq_bfqq_softrt_next_start()). We ++ * schedule this delayed check when bfqq expires, if it still ++ * has in-flight requests. ++ */ ++ if (bfq_bfqq_softrt_update(bfqq) && bfqq->dispatched == 0 && ++ RB_EMPTY_ROOT(&bfqq->sort_list)) ++ bfqq->soft_rt_next_start = ++ bfq_bfqq_softrt_next_start(bfqd, bfqq); ++ ++ /* ++ * If this is the in-service queue, check if it needs to be expired, ++ * or if we want to idle in case it has no pending requests. ++ */ ++ if (bfqd->in_service_queue == bfqq) { ++ if (bfqq->dispatched == 0 && bfq_bfqq_must_idle(bfqq)) { ++ bfq_arm_slice_timer(bfqd); ++ goto out; ++ } else if (bfq_may_expire_for_budg_timeout(bfqq)) ++ bfq_bfqq_expire(bfqd, bfqq, false, ++ BFQ_BFQQ_BUDGET_TIMEOUT); ++ else if (RB_EMPTY_ROOT(&bfqq->sort_list) && ++ (bfqq->dispatched == 0 || ++ !bfq_bfqq_may_idle(bfqq))) ++ bfq_bfqq_expire(bfqd, bfqq, false, ++ BFQ_BFQQ_NO_MORE_REQUESTS); ++ } ++ ++ if (!bfqd->rq_in_driver) ++ bfq_schedule_dispatch(bfqd); ++ ++out: ++ return; ++} ++ ++static int __bfq_may_queue(struct bfq_queue *bfqq) ++{ ++ if (bfq_bfqq_wait_request(bfqq) && bfq_bfqq_must_alloc(bfqq)) { ++ bfq_clear_bfqq_must_alloc(bfqq); ++ return ELV_MQUEUE_MUST; ++ } ++ ++ return ELV_MQUEUE_MAY; ++} ++ ++static int bfq_may_queue(struct request_queue *q, unsigned int op) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct task_struct *tsk = current; ++ struct bfq_io_cq *bic; ++ struct bfq_queue *bfqq; ++ ++ /* ++ * Don't force setup of a queue from here, as a call to may_queue ++ * does not necessarily imply that a request actually will be ++ * queued. So just lookup a possibly existing queue, or return ++ * 'may queue' if that fails. ++ */ ++ bic = bfq_bic_lookup(bfqd, tsk->io_context); ++ if (!bic) ++ return ELV_MQUEUE_MAY; ++ ++ bfqq = bic_to_bfqq(bic, op_is_sync(op)); ++ if (bfqq) ++ return __bfq_may_queue(bfqq); ++ ++ return ELV_MQUEUE_MAY; ++} ++ ++/* ++ * Queue lock held here. ++ */ ++static void bfq_put_request(struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ ++ if (bfqq) { ++ const int rw = rq_data_dir(rq); ++ ++ BUG_ON(!bfqq->allocated[rw]); ++ bfqq->allocated[rw]--; ++ ++ rq->elv.priv[0] = NULL; ++ rq->elv.priv[1] = NULL; ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "put_request %p, %d", ++ bfqq, bfqq->ref); ++ bfq_put_queue(bfqq); ++ } ++} ++ ++/* ++ * Returns NULL if a new bfqq should be allocated, or the old bfqq if this ++ * was the last process referring to that bfqq. ++ */ ++static struct bfq_queue * ++bfq_split_bfqq(struct bfq_io_cq *bic, struct bfq_queue *bfqq) ++{ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "splitting queue"); ++ ++ put_io_context(bic->icq.ioc); ++ ++ if (bfqq_process_refs(bfqq) == 1) { ++ bfqq->pid = current->pid; ++ bfq_clear_bfqq_coop(bfqq); ++ bfq_clear_bfqq_split_coop(bfqq); ++ return bfqq; ++ } ++ ++ bic_set_bfqq(bic, NULL, 1); ++ ++ bfq_put_cooperator(bfqq); ++ ++ bfq_put_queue(bfqq); ++ return NULL; ++} ++ ++/* ++ * Allocate bfq data structures associated with this request. ++ */ ++static int bfq_set_request(struct request_queue *q, struct request *rq, ++ struct bio *bio, gfp_t gfp_mask) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct bfq_io_cq *bic = icq_to_bic(rq->elv.icq); ++ const int rw = rq_data_dir(rq); ++ const int is_sync = rq_is_sync(rq); ++ struct bfq_queue *bfqq; ++ unsigned long flags; ++ bool bfqq_already_existing = false, split = false; ++ ++ spin_lock_irqsave(q->queue_lock, flags); ++ ++ if (!bic) ++ goto queue_fail; ++ ++ bfq_check_ioprio_change(bic, bio); ++ ++ bfq_bic_update_cgroup(bic, bio); ++ ++new_queue: ++ bfqq = bic_to_bfqq(bic, is_sync); ++ if (!bfqq || bfqq == &bfqd->oom_bfqq) { ++ if (bfqq) ++ bfq_put_queue(bfqq); ++ bfqq = bfq_get_queue(bfqd, bio, is_sync, bic); ++ BUG_ON(!hlist_unhashed(&bfqq->burst_list_node)); ++ ++ bic_set_bfqq(bic, bfqq, is_sync); ++ if (split && is_sync) { ++ bfq_log_bfqq(bfqd, bfqq, ++ "set_request: was_in_list %d " ++ "was_in_large_burst %d " ++ "large burst in progress %d", ++ bic->was_in_burst_list, ++ bic->saved_in_large_burst, ++ bfqd->large_burst); ++ ++ if ((bic->was_in_burst_list && bfqd->large_burst) || ++ bic->saved_in_large_burst) { ++ bfq_log_bfqq(bfqd, bfqq, ++ "set_request: marking in " ++ "large burst"); ++ bfq_mark_bfqq_in_large_burst(bfqq); ++ } else { ++ bfq_log_bfqq(bfqd, bfqq, ++ "set_request: clearing in " ++ "large burst"); ++ bfq_clear_bfqq_in_large_burst(bfqq); ++ if (bic->was_in_burst_list) ++ hlist_add_head(&bfqq->burst_list_node, ++ &bfqd->burst_list); ++ } ++ bfqq->split_time = jiffies; ++ } ++ } else { ++ /* If the queue was seeky for too long, break it apart. */ ++ if (bfq_bfqq_coop(bfqq) && bfq_bfqq_split_coop(bfqq)) { ++ bfq_log_bfqq(bfqd, bfqq, "breaking apart bfqq"); ++ ++ /* Update bic before losing reference to bfqq */ ++ if (bfq_bfqq_in_large_burst(bfqq)) ++ bic->saved_in_large_burst = true; ++ ++ bfqq = bfq_split_bfqq(bic, bfqq); ++ split = true; ++ if (!bfqq) ++ goto new_queue; ++ else ++ bfqq_already_existing = true; ++ } ++ } ++ ++ bfqq->allocated[rw]++; ++ bfqq->ref++; ++ bfq_log_bfqq(bfqd, bfqq, "set_request: bfqq %p, %d", bfqq, bfqq->ref); ++ ++ rq->elv.priv[0] = bic; ++ rq->elv.priv[1] = bfqq; ++ ++ /* ++ * If a bfq_queue has only one process reference, it is owned ++ * by only one bfq_io_cq: we can set the bic field of the ++ * bfq_queue to the address of that structure. Also, if the ++ * queue has just been split, mark a flag so that the ++ * information is available to the other scheduler hooks. ++ */ ++ if (likely(bfqq != &bfqd->oom_bfqq) && bfqq_process_refs(bfqq) == 1) { ++ bfqq->bic = bic; ++ if (split) { ++ /* ++ * If the queue has just been split from a shared ++ * queue, restore the idle window and the possible ++ * weight raising period. ++ */ ++ bfq_bfqq_resume_state(bfqq, bfqd, bic, ++ bfqq_already_existing); ++ } ++ } ++ ++ if (unlikely(bfq_bfqq_just_created(bfqq))) ++ bfq_handle_burst(bfqd, bfqq); ++ ++ spin_unlock_irqrestore(q->queue_lock, flags); ++ ++ return 0; ++ ++queue_fail: ++ bfq_schedule_dispatch(bfqd); ++ spin_unlock_irqrestore(q->queue_lock, flags); ++ ++ return 1; ++} ++ ++static void bfq_kick_queue(struct work_struct *work) ++{ ++ struct bfq_data *bfqd = ++ container_of(work, struct bfq_data, unplug_work); ++ struct request_queue *q = bfqd->queue; ++ ++ spin_lock_irq(q->queue_lock); ++ __blk_run_queue(q); ++ spin_unlock_irq(q->queue_lock); ++} ++ ++/* ++ * Handler of the expiration of the timer running if the in-service queue ++ * is idling inside its time slice. ++ */ ++static enum hrtimer_restart bfq_idle_slice_timer(struct hrtimer *timer) ++{ ++ struct bfq_data *bfqd = container_of(timer, struct bfq_data, ++ idle_slice_timer); ++ struct bfq_queue *bfqq; ++ unsigned long flags; ++ enum bfqq_expiration reason; ++ ++ spin_lock_irqsave(bfqd->queue->queue_lock, flags); ++ ++ bfqq = bfqd->in_service_queue; ++ /* ++ * Theoretical race here: the in-service queue can be NULL or ++ * different from the queue that was idling if the timer handler ++ * spins on the queue_lock and a new request arrives for the ++ * current queue and there is a full dispatch cycle that changes ++ * the in-service queue. This can hardly happen, but in the worst ++ * case we just expire a queue too early. ++ */ ++ if (bfqq) { ++ bfq_log_bfqq(bfqd, bfqq, "slice_timer expired"); ++ bfq_clear_bfqq_wait_request(bfqq); ++ ++ if (bfq_bfqq_budget_timeout(bfqq)) ++ /* ++ * Also here the queue can be safely expired ++ * for budget timeout without wasting ++ * guarantees ++ */ ++ reason = BFQ_BFQQ_BUDGET_TIMEOUT; ++ else if (bfqq->queued[0] == 0 && bfqq->queued[1] == 0) ++ /* ++ * The queue may not be empty upon timer expiration, ++ * because we may not disable the timer when the ++ * first request of the in-service queue arrives ++ * during disk idling. ++ */ ++ reason = BFQ_BFQQ_TOO_IDLE; ++ else ++ goto schedule_dispatch; ++ ++ bfq_bfqq_expire(bfqd, bfqq, true, reason); ++ } ++ ++schedule_dispatch: ++ bfq_schedule_dispatch(bfqd); ++ ++ spin_unlock_irqrestore(bfqd->queue->queue_lock, flags); ++ return HRTIMER_NORESTART; ++} ++ ++static void bfq_shutdown_timer_wq(struct bfq_data *bfqd) ++{ ++ hrtimer_cancel(&bfqd->idle_slice_timer); ++ cancel_work_sync(&bfqd->unplug_work); ++} ++ ++static void __bfq_put_async_bfqq(struct bfq_data *bfqd, ++ struct bfq_queue **bfqq_ptr) ++{ ++ struct bfq_group *root_group = bfqd->root_group; ++ struct bfq_queue *bfqq = *bfqq_ptr; ++ ++ bfq_log(bfqd, "put_async_bfqq: %p", bfqq); ++ if (bfqq) { ++ bfq_bfqq_move(bfqd, bfqq, root_group); ++ bfq_log_bfqq(bfqd, bfqq, "put_async_bfqq: putting %p, %d", ++ bfqq, bfqq->ref); ++ bfq_put_queue(bfqq); ++ *bfqq_ptr = NULL; ++ } ++} ++ ++/* ++ * Release all the bfqg references to its async queues. If we are ++ * deallocating the group these queues may still contain requests, so ++ * we reparent them to the root cgroup (i.e., the only one that will ++ * exist for sure until all the requests on a device are gone). ++ */ ++static void bfq_put_async_queues(struct bfq_data *bfqd, struct bfq_group *bfqg) ++{ ++ int i, j; ++ ++ for (i = 0; i < 2; i++) ++ for (j = 0; j < IOPRIO_BE_NR; j++) ++ __bfq_put_async_bfqq(bfqd, &bfqg->async_bfqq[i][j]); ++ ++ __bfq_put_async_bfqq(bfqd, &bfqg->async_idle_bfqq); ++} ++ ++static void bfq_exit_queue(struct elevator_queue *e) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ struct request_queue *q = bfqd->queue; ++ struct bfq_queue *bfqq, *n; ++ ++ bfq_shutdown_timer_wq(bfqd); ++ ++ spin_lock_irq(q->queue_lock); ++ ++ BUG_ON(bfqd->in_service_queue); ++ list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) ++ bfq_deactivate_bfqq(bfqd, bfqq, false, false); ++ ++ spin_unlock_irq(q->queue_lock); ++ ++ bfq_shutdown_timer_wq(bfqd); ++ ++ BUG_ON(hrtimer_active(&bfqd->idle_slice_timer)); ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ blkcg_deactivate_policy(q, &blkcg_policy_bfq); ++#else ++ bfq_put_async_queues(bfqd, bfqd->root_group); ++ kfree(bfqd->root_group); ++#endif ++ ++ kfree(bfqd); ++} ++ ++static void bfq_init_root_group(struct bfq_group *root_group, ++ struct bfq_data *bfqd) ++{ ++ int i; ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ root_group->entity.parent = NULL; ++ root_group->my_entity = NULL; ++ root_group->bfqd = bfqd; ++#endif ++ root_group->rq_pos_tree = RB_ROOT; ++ for (i = 0; i < BFQ_IOPRIO_CLASSES; i++) ++ root_group->sched_data.service_tree[i] = BFQ_SERVICE_TREE_INIT; ++ root_group->sched_data.bfq_class_idle_last_service = jiffies; ++} ++ ++static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) ++{ ++ struct bfq_data *bfqd; ++ struct elevator_queue *eq; ++ ++ eq = elevator_alloc(q, e); ++ if (!eq) ++ return -ENOMEM; ++ ++ bfqd = kzalloc_node(sizeof(*bfqd), GFP_KERNEL, q->node); ++ if (!bfqd) { ++ kobject_put(&eq->kobj); ++ return -ENOMEM; ++ } ++ eq->elevator_data = bfqd; ++ ++ /* ++ * Our fallback bfqq if bfq_find_alloc_queue() runs into OOM issues. ++ * Grab a permanent reference to it, so that the normal code flow ++ * will not attempt to free it. ++ */ ++ bfq_init_bfqq(bfqd, &bfqd->oom_bfqq, NULL, 1, 0); ++ bfqd->oom_bfqq.ref++; ++ bfqd->oom_bfqq.new_ioprio = BFQ_DEFAULT_QUEUE_IOPRIO; ++ bfqd->oom_bfqq.new_ioprio_class = IOPRIO_CLASS_BE; ++ bfqd->oom_bfqq.entity.new_weight = ++ bfq_ioprio_to_weight(bfqd->oom_bfqq.new_ioprio); ++ ++ /* oom_bfqq does not participate to bursts */ ++ bfq_clear_bfqq_just_created(&bfqd->oom_bfqq); ++ /* ++ * Trigger weight initialization, according to ioprio, at the ++ * oom_bfqq's first activation. The oom_bfqq's ioprio and ioprio ++ * class won't be changed any more. ++ */ ++ bfqd->oom_bfqq.entity.prio_changed = 1; ++ ++ bfqd->queue = q; ++ ++ spin_lock_irq(q->queue_lock); ++ q->elevator = eq; ++ spin_unlock_irq(q->queue_lock); ++ ++ bfqd->root_group = bfq_create_group_hierarchy(bfqd, q->node); ++ if (!bfqd->root_group) ++ goto out_free; ++ bfq_init_root_group(bfqd->root_group, bfqd); ++ bfq_init_entity(&bfqd->oom_bfqq.entity, bfqd->root_group); ++ ++ hrtimer_init(&bfqd->idle_slice_timer, CLOCK_MONOTONIC, ++ HRTIMER_MODE_REL); ++ bfqd->idle_slice_timer.function = bfq_idle_slice_timer; ++ ++ bfqd->queue_weights_tree = RB_ROOT; ++ bfqd->group_weights_tree = RB_ROOT; ++ ++ INIT_WORK(&bfqd->unplug_work, bfq_kick_queue); ++ ++ INIT_LIST_HEAD(&bfqd->active_list); ++ INIT_LIST_HEAD(&bfqd->idle_list); ++ INIT_HLIST_HEAD(&bfqd->burst_list); ++ ++ bfqd->hw_tag = -1; ++ ++ bfqd->bfq_max_budget = bfq_default_max_budget; ++ ++ bfqd->bfq_fifo_expire[0] = bfq_fifo_expire[0]; ++ bfqd->bfq_fifo_expire[1] = bfq_fifo_expire[1]; ++ bfqd->bfq_back_max = bfq_back_max; ++ bfqd->bfq_back_penalty = bfq_back_penalty; ++ bfqd->bfq_slice_idle = bfq_slice_idle; ++ bfqd->bfq_timeout = bfq_timeout; ++ ++ bfqd->bfq_requests_within_timer = 120; ++ ++ bfqd->bfq_large_burst_thresh = 8; ++ bfqd->bfq_burst_interval = msecs_to_jiffies(180); ++ ++ bfqd->low_latency = true; ++ ++ /* ++ * Trade-off between responsiveness and fairness. ++ */ ++ bfqd->bfq_wr_coeff = 30; ++ bfqd->bfq_wr_rt_max_time = msecs_to_jiffies(300); ++ bfqd->bfq_wr_max_time = 0; ++ bfqd->bfq_wr_min_idle_time = msecs_to_jiffies(2000); ++ bfqd->bfq_wr_min_inter_arr_async = msecs_to_jiffies(500); ++ bfqd->bfq_wr_max_softrt_rate = 7000; /* ++ * Approximate rate required ++ * to playback or record a ++ * high-definition compressed ++ * video. ++ */ ++ bfqd->wr_busy_queues = 0; ++ ++ /* ++ * Begin by assuming, optimistically, that the device is a ++ * high-speed one, and that its peak rate is equal to 2/3 of ++ * the highest reference rate. ++ */ ++ bfqd->RT_prod = R_fast[blk_queue_nonrot(bfqd->queue)] * ++ T_fast[blk_queue_nonrot(bfqd->queue)]; ++ bfqd->peak_rate = R_fast[blk_queue_nonrot(bfqd->queue)] * 2 / 3; ++ bfqd->device_speed = BFQ_BFQD_FAST; ++ ++ return 0; ++ ++out_free: ++ kfree(bfqd); ++ kobject_put(&eq->kobj); ++ return -ENOMEM; ++} ++ ++static void bfq_slab_kill(void) ++{ ++ kmem_cache_destroy(bfq_pool); ++} ++ ++static int __init bfq_slab_setup(void) ++{ ++ bfq_pool = KMEM_CACHE(bfq_queue, 0); ++ if (!bfq_pool) ++ return -ENOMEM; ++ return 0; ++} ++ ++static ssize_t bfq_var_show(unsigned int var, char *page) ++{ ++ return sprintf(page, "%u\n", var); ++} ++ ++static ssize_t bfq_var_store(unsigned long *var, const char *page, ++ size_t count) ++{ ++ unsigned long new_val; ++ int ret = kstrtoul(page, 10, &new_val); ++ ++ if (ret == 0) ++ *var = new_val; ++ ++ return count; ++} ++ ++static ssize_t bfq_wr_max_time_show(struct elevator_queue *e, char *page) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ ++ return sprintf(page, "%d\n", bfqd->bfq_wr_max_time > 0 ? ++ jiffies_to_msecs(bfqd->bfq_wr_max_time) : ++ jiffies_to_msecs(bfq_wr_duration(bfqd))); ++} ++ ++static ssize_t bfq_weights_show(struct elevator_queue *e, char *page) ++{ ++ struct bfq_queue *bfqq; ++ struct bfq_data *bfqd = e->elevator_data; ++ ssize_t num_char = 0; ++ ++ num_char += sprintf(page + num_char, "Tot reqs queued %d\n\n", ++ bfqd->queued); ++ ++ spin_lock_irq(bfqd->queue->queue_lock); ++ ++ num_char += sprintf(page + num_char, "Active:\n"); ++ list_for_each_entry(bfqq, &bfqd->active_list, bfqq_list) { ++ num_char += sprintf(page + num_char, ++ "pid%d: weight %hu, nr_queued %d %d, ", ++ bfqq->pid, ++ bfqq->entity.weight, ++ bfqq->queued[0], ++ bfqq->queued[1]); ++ num_char += sprintf(page + num_char, ++ "dur %d/%u\n", ++ jiffies_to_msecs( ++ jiffies - ++ bfqq->last_wr_start_finish), ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ } ++ ++ num_char += sprintf(page + num_char, "Idle:\n"); ++ list_for_each_entry(bfqq, &bfqd->idle_list, bfqq_list) { ++ num_char += sprintf(page + num_char, ++ "pid%d: weight %hu, dur %d/%u\n", ++ bfqq->pid, ++ bfqq->entity.weight, ++ jiffies_to_msecs(jiffies - ++ bfqq->last_wr_start_finish), ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ } ++ ++ spin_unlock_irq(bfqd->queue->queue_lock); ++ ++ return num_char; ++} ++ ++#define SHOW_FUNCTION(__FUNC, __VAR, __CONV) \ ++static ssize_t __FUNC(struct elevator_queue *e, char *page) \ ++{ \ ++ struct bfq_data *bfqd = e->elevator_data; \ ++ u64 __data = __VAR; \ ++ if (__CONV == 1) \ ++ __data = jiffies_to_msecs(__data); \ ++ else if (__CONV == 2) \ ++ __data = div_u64(__data, NSEC_PER_MSEC); \ ++ return bfq_var_show(__data, (page)); \ ++} ++SHOW_FUNCTION(bfq_fifo_expire_sync_show, bfqd->bfq_fifo_expire[1], 2); ++SHOW_FUNCTION(bfq_fifo_expire_async_show, bfqd->bfq_fifo_expire[0], 2); ++SHOW_FUNCTION(bfq_back_seek_max_show, bfqd->bfq_back_max, 0); ++SHOW_FUNCTION(bfq_back_seek_penalty_show, bfqd->bfq_back_penalty, 0); ++SHOW_FUNCTION(bfq_slice_idle_show, bfqd->bfq_slice_idle, 2); ++SHOW_FUNCTION(bfq_max_budget_show, bfqd->bfq_user_max_budget, 0); ++SHOW_FUNCTION(bfq_timeout_sync_show, bfqd->bfq_timeout, 1); ++SHOW_FUNCTION(bfq_strict_guarantees_show, bfqd->strict_guarantees, 0); ++SHOW_FUNCTION(bfq_low_latency_show, bfqd->low_latency, 0); ++SHOW_FUNCTION(bfq_wr_coeff_show, bfqd->bfq_wr_coeff, 0); ++SHOW_FUNCTION(bfq_wr_rt_max_time_show, bfqd->bfq_wr_rt_max_time, 1); ++SHOW_FUNCTION(bfq_wr_min_idle_time_show, bfqd->bfq_wr_min_idle_time, 1); ++SHOW_FUNCTION(bfq_wr_min_inter_arr_async_show, bfqd->bfq_wr_min_inter_arr_async, ++ 1); ++SHOW_FUNCTION(bfq_wr_max_softrt_rate_show, bfqd->bfq_wr_max_softrt_rate, 0); ++#undef SHOW_FUNCTION ++ ++#define USEC_SHOW_FUNCTION(__FUNC, __VAR) \ ++static ssize_t __FUNC(struct elevator_queue *e, char *page) \ ++{ \ ++ struct bfq_data *bfqd = e->elevator_data; \ ++ u64 __data = __VAR; \ ++ __data = div_u64(__data, NSEC_PER_USEC); \ ++ return bfq_var_show(__data, (page)); \ ++} ++USEC_SHOW_FUNCTION(bfq_slice_idle_us_show, bfqd->bfq_slice_idle); ++#undef USEC_SHOW_FUNCTION ++ ++#define STORE_FUNCTION(__FUNC, __PTR, MIN, MAX, __CONV) \ ++static ssize_t \ ++__FUNC(struct elevator_queue *e, const char *page, size_t count) \ ++{ \ ++ struct bfq_data *bfqd = e->elevator_data; \ ++ unsigned long uninitialized_var(__data); \ ++ int ret = bfq_var_store(&__data, (page), count); \ ++ if (__data < (MIN)) \ ++ __data = (MIN); \ ++ else if (__data > (MAX)) \ ++ __data = (MAX); \ ++ if (__CONV == 1) \ ++ *(__PTR) = msecs_to_jiffies(__data); \ ++ else if (__CONV == 2) \ ++ *(__PTR) = (u64)__data * NSEC_PER_MSEC; \ ++ else \ ++ *(__PTR) = __data; \ ++ return ret; \ ++} ++STORE_FUNCTION(bfq_fifo_expire_sync_store, &bfqd->bfq_fifo_expire[1], 1, ++ INT_MAX, 2); ++STORE_FUNCTION(bfq_fifo_expire_async_store, &bfqd->bfq_fifo_expire[0], 1, ++ INT_MAX, 2); ++STORE_FUNCTION(bfq_back_seek_max_store, &bfqd->bfq_back_max, 0, INT_MAX, 0); ++STORE_FUNCTION(bfq_back_seek_penalty_store, &bfqd->bfq_back_penalty, 1, ++ INT_MAX, 0); ++STORE_FUNCTION(bfq_slice_idle_store, &bfqd->bfq_slice_idle, 0, INT_MAX, 2); ++STORE_FUNCTION(bfq_wr_coeff_store, &bfqd->bfq_wr_coeff, 1, INT_MAX, 0); ++STORE_FUNCTION(bfq_wr_max_time_store, &bfqd->bfq_wr_max_time, 0, INT_MAX, 1); ++STORE_FUNCTION(bfq_wr_rt_max_time_store, &bfqd->bfq_wr_rt_max_time, 0, INT_MAX, ++ 1); ++STORE_FUNCTION(bfq_wr_min_idle_time_store, &bfqd->bfq_wr_min_idle_time, 0, ++ INT_MAX, 1); ++STORE_FUNCTION(bfq_wr_min_inter_arr_async_store, ++ &bfqd->bfq_wr_min_inter_arr_async, 0, INT_MAX, 1); ++STORE_FUNCTION(bfq_wr_max_softrt_rate_store, &bfqd->bfq_wr_max_softrt_rate, 0, ++ INT_MAX, 0); ++#undef STORE_FUNCTION ++ ++#define USEC_STORE_FUNCTION(__FUNC, __PTR, MIN, MAX) \ ++static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)\ ++{ \ ++ struct bfq_data *bfqd = e->elevator_data; \ ++ unsigned long uninitialized_var(__data); \ ++ int ret = bfq_var_store(&__data, (page), count); \ ++ if (__data < (MIN)) \ ++ __data = (MIN); \ ++ else if (__data > (MAX)) \ ++ __data = (MAX); \ ++ *(__PTR) = (u64)__data * NSEC_PER_USEC; \ ++ return ret; \ ++} ++USEC_STORE_FUNCTION(bfq_slice_idle_us_store, &bfqd->bfq_slice_idle, 0, ++ UINT_MAX); ++#undef USEC_STORE_FUNCTION ++ ++/* do nothing for the moment */ ++static ssize_t bfq_weights_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ return count; ++} ++ ++static ssize_t bfq_max_budget_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ unsigned long uninitialized_var(__data); ++ int ret = bfq_var_store(&__data, (page), count); ++ ++ if (__data == 0) ++ bfqd->bfq_max_budget = bfq_calc_max_budget(bfqd); ++ else { ++ if (__data > INT_MAX) ++ __data = INT_MAX; ++ bfqd->bfq_max_budget = __data; ++ } ++ ++ bfqd->bfq_user_max_budget = __data; ++ ++ return ret; ++} ++ ++/* ++ * Leaving this name to preserve name compatibility with cfq ++ * parameters, but this timeout is used for both sync and async. ++ */ ++static ssize_t bfq_timeout_sync_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ unsigned long uninitialized_var(__data); ++ int ret = bfq_var_store(&__data, (page), count); ++ ++ if (__data < 1) ++ __data = 1; ++ else if (__data > INT_MAX) ++ __data = INT_MAX; ++ ++ bfqd->bfq_timeout = msecs_to_jiffies(__data); ++ if (bfqd->bfq_user_max_budget == 0) ++ bfqd->bfq_max_budget = bfq_calc_max_budget(bfqd); ++ ++ return ret; ++} ++ ++static ssize_t bfq_strict_guarantees_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ unsigned long uninitialized_var(__data); ++ int ret = bfq_var_store(&__data, (page), count); ++ ++ if (__data > 1) ++ __data = 1; ++ if (!bfqd->strict_guarantees && __data == 1 ++ && bfqd->bfq_slice_idle < 8 * NSEC_PER_MSEC) ++ bfqd->bfq_slice_idle = 8 * NSEC_PER_MSEC; ++ ++ bfqd->strict_guarantees = __data; ++ ++ return ret; ++} ++ ++static ssize_t bfq_low_latency_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ unsigned long uninitialized_var(__data); ++ int ret = bfq_var_store(&__data, (page), count); ++ ++ if (__data > 1) ++ __data = 1; ++ if (__data == 0 && bfqd->low_latency != 0) ++ bfq_end_wr(bfqd); ++ bfqd->low_latency = __data; ++ ++ return ret; ++} ++ ++#define BFQ_ATTR(name) \ ++ __ATTR(name, S_IRUGO|S_IWUSR, bfq_##name##_show, bfq_##name##_store) ++ ++static struct elv_fs_entry bfq_attrs[] = { ++ BFQ_ATTR(fifo_expire_sync), ++ BFQ_ATTR(fifo_expire_async), ++ BFQ_ATTR(back_seek_max), ++ BFQ_ATTR(back_seek_penalty), ++ BFQ_ATTR(slice_idle), ++ BFQ_ATTR(slice_idle_us), ++ BFQ_ATTR(max_budget), ++ BFQ_ATTR(timeout_sync), ++ BFQ_ATTR(strict_guarantees), ++ BFQ_ATTR(low_latency), ++ BFQ_ATTR(wr_coeff), ++ BFQ_ATTR(wr_max_time), ++ BFQ_ATTR(wr_rt_max_time), ++ BFQ_ATTR(wr_min_idle_time), ++ BFQ_ATTR(wr_min_inter_arr_async), ++ BFQ_ATTR(wr_max_softrt_rate), ++ BFQ_ATTR(weights), ++ __ATTR_NULL ++}; ++ ++static struct elevator_type iosched_bfq = { ++ .ops.sq = { ++ .elevator_merge_fn = bfq_merge, ++ .elevator_merged_fn = bfq_merged_request, ++ .elevator_merge_req_fn = bfq_merged_requests, ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ .elevator_bio_merged_fn = bfq_bio_merged, ++#endif ++ .elevator_allow_bio_merge_fn = bfq_allow_bio_merge, ++ .elevator_allow_rq_merge_fn = bfq_allow_rq_merge, ++ .elevator_dispatch_fn = bfq_dispatch_requests, ++ .elevator_add_req_fn = bfq_insert_request, ++ .elevator_activate_req_fn = bfq_activate_request, ++ .elevator_deactivate_req_fn = bfq_deactivate_request, ++ .elevator_completed_req_fn = bfq_completed_request, ++ .elevator_former_req_fn = elv_rb_former_request, ++ .elevator_latter_req_fn = elv_rb_latter_request, ++ .elevator_init_icq_fn = bfq_init_icq, ++ .elevator_exit_icq_fn = bfq_exit_icq, ++ .elevator_set_req_fn = bfq_set_request, ++ .elevator_put_req_fn = bfq_put_request, ++ .elevator_may_queue_fn = bfq_may_queue, ++ .elevator_init_fn = bfq_init_queue, ++ .elevator_exit_fn = bfq_exit_queue, ++ }, ++ .icq_size = sizeof(struct bfq_io_cq), ++ .icq_align = __alignof__(struct bfq_io_cq), ++ .elevator_attrs = bfq_attrs, ++ .elevator_name = "bfq-sq", ++ .elevator_owner = THIS_MODULE, ++}; ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++static struct blkcg_policy blkcg_policy_bfq = { ++ .dfl_cftypes = bfq_blkg_files, ++ .legacy_cftypes = bfq_blkcg_legacy_files, ++ ++ .cpd_alloc_fn = bfq_cpd_alloc, ++ .cpd_init_fn = bfq_cpd_init, ++ .cpd_bind_fn = bfq_cpd_init, ++ .cpd_free_fn = bfq_cpd_free, ++ ++ .pd_alloc_fn = bfq_pd_alloc, ++ .pd_init_fn = bfq_pd_init, ++ .pd_offline_fn = bfq_pd_offline, ++ .pd_free_fn = bfq_pd_free, ++ .pd_reset_stats_fn = bfq_pd_reset_stats, ++}; ++#endif ++ ++static int __init bfq_init(void) ++{ ++ int ret; ++ char msg[60] = "BFQ I/O-scheduler: v8r12"; ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ ret = blkcg_policy_register(&blkcg_policy_bfq); ++ if (ret) ++ return ret; ++#endif ++ ++ ret = -ENOMEM; ++ if (bfq_slab_setup()) ++ goto err_pol_unreg; ++ ++ /* ++ * Times to load large popular applications for the typical ++ * systems installed on the reference devices (see the ++ * comments before the definitions of the next two ++ * arrays). Actually, we use slightly slower values, as the ++ * estimated peak rate tends to be smaller than the actual ++ * peak rate. The reason for this last fact is that estimates ++ * are computed over much shorter time intervals than the long ++ * intervals typically used for benchmarking. Why? First, to ++ * adapt more quickly to variations. Second, because an I/O ++ * scheduler cannot rely on a peak-rate-evaluation workload to ++ * be run for a long time. ++ */ ++ T_slow[0] = msecs_to_jiffies(3500); /* actually 4 sec */ ++ T_slow[1] = msecs_to_jiffies(6000); /* actually 6.5 sec */ ++ T_fast[0] = msecs_to_jiffies(7000); /* actually 8 sec */ ++ T_fast[1] = msecs_to_jiffies(2500); /* actually 3 sec */ ++ ++ /* ++ * Thresholds that determine the switch between speed classes ++ * (see the comments before the definition of the array ++ * device_speed_thresh). These thresholds are biased towards ++ * transitions to the fast class. This is safer than the ++ * opposite bias. In fact, a wrong transition to the slow ++ * class results in short weight-raising periods, because the ++ * speed of the device then tends to be higher that the ++ * reference peak rate. On the opposite end, a wrong ++ * transition to the fast class tends to increase ++ * weight-raising periods, because of the opposite reason. ++ */ ++ device_speed_thresh[0] = (4 * R_slow[0]) / 3; ++ device_speed_thresh[1] = (4 * R_slow[1]) / 3; ++ ++ ret = elv_register(&iosched_bfq); ++ if (ret) ++ goto err_pol_unreg; ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ strcat(msg, " (with cgroups support)"); ++#endif ++ pr_info("%s", msg); ++ ++ return 0; ++ ++err_pol_unreg: ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ blkcg_policy_unregister(&blkcg_policy_bfq); ++#endif ++ return ret; ++} ++ ++static void __exit bfq_exit(void) ++{ ++ elv_unregister(&iosched_bfq); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ blkcg_policy_unregister(&blkcg_policy_bfq); ++#endif ++ bfq_slab_kill(); ++} ++ ++module_init(bfq_init); ++module_exit(bfq_exit); ++ ++MODULE_AUTHOR("Arianna Avanzini, Fabio Checconi, Paolo Valente"); ++MODULE_LICENSE("GPL"); +diff --git a/block/bfq.h b/block/bfq.h +new file mode 100644 +index 000000000000..f5751ea59d98 +--- /dev/null ++++ b/block/bfq.h +@@ -0,0 +1,948 @@ ++/* ++ * BFQ v8r12 for 4.11.0: data structures and common functions prototypes. ++ * ++ * Based on ideas and code from CFQ: ++ * Copyright (C) 2003 Jens Axboe ++ * ++ * Copyright (C) 2008 Fabio Checconi ++ * Paolo Valente ++ * ++ * Copyright (C) 2015 Paolo Valente ++ * ++ * Copyright (C) 2017 Paolo Valente ++ */ ++ ++#ifndef _BFQ_H ++#define _BFQ_H ++ ++#include ++#include ++#include ++ ++#define BFQ_IOPRIO_CLASSES 3 ++#define BFQ_CL_IDLE_TIMEOUT (HZ/5) ++ ++#define BFQ_MIN_WEIGHT 1 ++#define BFQ_MAX_WEIGHT 1000 ++#define BFQ_WEIGHT_CONVERSION_COEFF 10 ++ ++#define BFQ_DEFAULT_QUEUE_IOPRIO 4 ++ ++#define BFQ_WEIGHT_LEGACY_DFL 100 ++#define BFQ_DEFAULT_GRP_IOPRIO 0 ++#define BFQ_DEFAULT_GRP_CLASS IOPRIO_CLASS_BE ++ ++/* ++ * Soft real-time applications are extremely more latency sensitive ++ * than interactive ones. Over-raise the weight of the former to ++ * privilege them against the latter. ++ */ ++#define BFQ_SOFTRT_WEIGHT_FACTOR 100 ++ ++struct bfq_entity; ++ ++/** ++ * struct bfq_service_tree - per ioprio_class service tree. ++ * ++ * Each service tree represents a B-WF2Q+ scheduler on its own. Each ++ * ioprio_class has its own independent scheduler, and so its own ++ * bfq_service_tree. All the fields are protected by the queue lock ++ * of the containing bfqd. ++ */ ++struct bfq_service_tree { ++ /* tree for active entities (i.e., those backlogged) */ ++ struct rb_root active; ++ /* tree for idle entities (i.e., not backlogged, with V <= F_i)*/ ++ struct rb_root idle; ++ ++ struct bfq_entity *first_idle; /* idle entity with minimum F_i */ ++ struct bfq_entity *last_idle; /* idle entity with maximum F_i */ ++ ++ u64 vtime; /* scheduler virtual time */ ++ /* scheduler weight sum; active and idle entities contribute to it */ ++ unsigned long wsum; ++}; ++ ++/** ++ * struct bfq_sched_data - multi-class scheduler. ++ * ++ * bfq_sched_data is the basic scheduler queue. It supports three ++ * ioprio_classes, and can be used either as a toplevel queue or as an ++ * intermediate queue on a hierarchical setup. @next_in_service ++ * points to the active entity of the sched_data service trees that ++ * will be scheduled next. It is used to reduce the number of steps ++ * needed for each hierarchical-schedule update. ++ * ++ * The supported ioprio_classes are the same as in CFQ, in descending ++ * priority order, IOPRIO_CLASS_RT, IOPRIO_CLASS_BE, IOPRIO_CLASS_IDLE. ++ * Requests from higher priority queues are served before all the ++ * requests from lower priority queues; among requests of the same ++ * queue requests are served according to B-WF2Q+. ++ * All the fields are protected by the queue lock of the containing bfqd. ++ */ ++struct bfq_sched_data { ++ struct bfq_entity *in_service_entity; /* entity in service */ ++ /* head-of-the-line entity in the scheduler (see comments above) */ ++ struct bfq_entity *next_in_service; ++ /* array of service trees, one per ioprio_class */ ++ struct bfq_service_tree service_tree[BFQ_IOPRIO_CLASSES]; ++ /* last time CLASS_IDLE was served */ ++ unsigned long bfq_class_idle_last_service; ++ ++}; ++ ++/** ++ * struct bfq_weight_counter - counter of the number of all active entities ++ * with a given weight. ++ */ ++struct bfq_weight_counter { ++ unsigned int weight; /* weight of the entities this counter refers to */ ++ unsigned int num_active; /* nr of active entities with this weight */ ++ /* ++ * Weights tree member (see bfq_data's @queue_weights_tree and ++ * @group_weights_tree) ++ */ ++ struct rb_node weights_node; ++}; ++ ++/** ++ * struct bfq_entity - schedulable entity. ++ * ++ * A bfq_entity is used to represent either a bfq_queue (leaf node in the ++ * cgroup hierarchy) or a bfq_group into the upper level scheduler. Each ++ * entity belongs to the sched_data of the parent group in the cgroup ++ * hierarchy. Non-leaf entities have also their own sched_data, stored ++ * in @my_sched_data. ++ * ++ * Each entity stores independently its priority values; this would ++ * allow different weights on different devices, but this ++ * functionality is not exported to userspace by now. Priorities and ++ * weights are updated lazily, first storing the new values into the ++ * new_* fields, then setting the @prio_changed flag. As soon as ++ * there is a transition in the entity state that allows the priority ++ * update to take place the effective and the requested priority ++ * values are synchronized. ++ * ++ * Unless cgroups are used, the weight value is calculated from the ++ * ioprio to export the same interface as CFQ. When dealing with ++ * ``well-behaved'' queues (i.e., queues that do not spend too much ++ * time to consume their budget and have true sequential behavior, and ++ * when there are no external factors breaking anticipation) the ++ * relative weights at each level of the cgroups hierarchy should be ++ * guaranteed. All the fields are protected by the queue lock of the ++ * containing bfqd. ++ */ ++struct bfq_entity { ++ struct rb_node rb_node; /* service_tree member */ ++ /* pointer to the weight counter associated with this entity */ ++ struct bfq_weight_counter *weight_counter; ++ ++ /* ++ * Flag, true if the entity is on a tree (either the active or ++ * the idle one of its service_tree) or is in service. ++ */ ++ bool on_st; ++ ++ u64 finish; /* B-WF2Q+ finish timestamp (aka F_i) */ ++ u64 start; /* B-WF2Q+ start timestamp (aka S_i) */ ++ ++ /* tree the entity is enqueued into; %NULL if not on a tree */ ++ struct rb_root *tree; ++ ++ /* ++ * minimum start time of the (active) subtree rooted at this ++ * entity; used for O(log N) lookups into active trees ++ */ ++ u64 min_start; ++ ++ /* amount of service received during the last service slot */ ++ int service; ++ ++ /* budget, used also to calculate F_i: F_i = S_i + @budget / @weight */ ++ int budget; ++ ++ unsigned int weight; /* weight of the queue */ ++ unsigned int new_weight; /* next weight if a change is in progress */ ++ ++ /* original weight, used to implement weight boosting */ ++ unsigned int orig_weight; ++ ++ /* parent entity, for hierarchical scheduling */ ++ struct bfq_entity *parent; ++ ++ /* ++ * For non-leaf nodes in the hierarchy, the associated ++ * scheduler queue, %NULL on leaf nodes. ++ */ ++ struct bfq_sched_data *my_sched_data; ++ /* the scheduler queue this entity belongs to */ ++ struct bfq_sched_data *sched_data; ++ ++ /* flag, set to request a weight, ioprio or ioprio_class change */ ++ int prio_changed; ++}; ++ ++struct bfq_group; ++ ++/** ++ * struct bfq_queue - leaf schedulable entity. ++ * ++ * A bfq_queue is a leaf request queue; it can be associated with an ++ * io_context or more, if it is async or shared between cooperating ++ * processes. @cgroup holds a reference to the cgroup, to be sure that it ++ * does not disappear while a bfqq still references it (mostly to avoid ++ * races between request issuing and task migration followed by cgroup ++ * destruction). ++ * All the fields are protected by the queue lock of the containing bfqd. ++ */ ++struct bfq_queue { ++ /* reference counter */ ++ int ref; ++ /* parent bfq_data */ ++ struct bfq_data *bfqd; ++ ++ /* current ioprio and ioprio class */ ++ unsigned short ioprio, ioprio_class; ++ /* next ioprio and ioprio class if a change is in progress */ ++ unsigned short new_ioprio, new_ioprio_class; ++ ++ /* ++ * Shared bfq_queue if queue is cooperating with one or more ++ * other queues. ++ */ ++ struct bfq_queue *new_bfqq; ++ /* request-position tree member (see bfq_group's @rq_pos_tree) */ ++ struct rb_node pos_node; ++ /* request-position tree root (see bfq_group's @rq_pos_tree) */ ++ struct rb_root *pos_root; ++ ++ /* sorted list of pending requests */ ++ struct rb_root sort_list; ++ /* if fifo isn't expired, next request to serve */ ++ struct request *next_rq; ++ /* number of sync and async requests queued */ ++ int queued[2]; ++ /* number of sync and async requests currently allocated */ ++ int allocated[2]; ++ /* number of pending metadata requests */ ++ int meta_pending; ++ /* fifo list of requests in sort_list */ ++ struct list_head fifo; ++ ++ /* entity representing this queue in the scheduler */ ++ struct bfq_entity entity; ++ ++ /* maximum budget allowed from the feedback mechanism */ ++ int max_budget; ++ /* budget expiration (in jiffies) */ ++ unsigned long budget_timeout; ++ ++ /* number of requests on the dispatch list or inside driver */ ++ int dispatched; ++ ++ unsigned int flags; /* status flags.*/ ++ ++ /* node for active/idle bfqq list inside parent bfqd */ ++ struct list_head bfqq_list; ++ ++ /* bit vector: a 1 for each seeky requests in history */ ++ u32 seek_history; ++ ++ /* node for the device's burst list */ ++ struct hlist_node burst_list_node; ++ ++ /* position of the last request enqueued */ ++ sector_t last_request_pos; ++ ++ /* Number of consecutive pairs of request completion and ++ * arrival, such that the queue becomes idle after the ++ * completion, but the next request arrives within an idle ++ * time slice; used only if the queue's IO_bound flag has been ++ * cleared. ++ */ ++ unsigned int requests_within_timer; ++ ++ /* pid of the process owning the queue, used for logging purposes */ ++ pid_t pid; ++ ++ /* ++ * Pointer to the bfq_io_cq owning the bfq_queue, set to %NULL ++ * if the queue is shared. ++ */ ++ struct bfq_io_cq *bic; ++ ++ /* current maximum weight-raising time for this queue */ ++ unsigned long wr_cur_max_time; ++ /* ++ * Minimum time instant such that, only if a new request is ++ * enqueued after this time instant in an idle @bfq_queue with ++ * no outstanding requests, then the task associated with the ++ * queue it is deemed as soft real-time (see the comments on ++ * the function bfq_bfqq_softrt_next_start()) ++ */ ++ unsigned long soft_rt_next_start; ++ /* ++ * Start time of the current weight-raising period if ++ * the @bfq-queue is being weight-raised, otherwise ++ * finish time of the last weight-raising period. ++ */ ++ unsigned long last_wr_start_finish; ++ /* factor by which the weight of this queue is multiplied */ ++ unsigned int wr_coeff; ++ /* ++ * Time of the last transition of the @bfq_queue from idle to ++ * backlogged. ++ */ ++ unsigned long last_idle_bklogged; ++ /* ++ * Cumulative service received from the @bfq_queue since the ++ * last transition from idle to backlogged. ++ */ ++ unsigned long service_from_backlogged; ++ /* ++ * Value of wr start time when switching to soft rt ++ */ ++ unsigned long wr_start_at_switch_to_srt; ++ ++ unsigned long split_time; /* time of last split */ ++}; ++ ++/** ++ * struct bfq_ttime - per process thinktime stats. ++ */ ++struct bfq_ttime { ++ u64 last_end_request; /* completion time of last request */ ++ ++ u64 ttime_total; /* total process thinktime */ ++ unsigned long ttime_samples; /* number of thinktime samples */ ++ u64 ttime_mean; /* average process thinktime */ ++ ++}; ++ ++/** ++ * struct bfq_io_cq - per (request_queue, io_context) structure. ++ */ ++struct bfq_io_cq { ++ /* associated io_cq structure */ ++ struct io_cq icq; /* must be the first member */ ++ /* array of two process queues, the sync and the async */ ++ struct bfq_queue *bfqq[2]; ++ /* associated @bfq_ttime struct */ ++ struct bfq_ttime ttime; ++ /* per (request_queue, blkcg) ioprio */ ++ int ioprio; ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ uint64_t blkcg_serial_nr; /* the current blkcg serial */ ++#endif ++ ++ /* ++ * Snapshot of the idle window before merging; taken to ++ * remember this value while the queue is merged, so as to be ++ * able to restore it in case of split. ++ */ ++ bool saved_idle_window; ++ /* ++ * Same purpose as the previous two fields for the I/O bound ++ * classification of a queue. ++ */ ++ bool saved_IO_bound; ++ ++ /* ++ * Same purpose as the previous fields for the value of the ++ * field keeping the queue's belonging to a large burst ++ */ ++ bool saved_in_large_burst; ++ /* ++ * True if the queue belonged to a burst list before its merge ++ * with another cooperating queue. ++ */ ++ bool was_in_burst_list; ++ ++ /* ++ * Similar to previous fields: save wr information. ++ */ ++ unsigned long saved_wr_coeff; ++ unsigned long saved_last_wr_start_finish; ++ unsigned long saved_wr_start_at_switch_to_srt; ++ unsigned int saved_wr_cur_max_time; ++}; ++ ++enum bfq_device_speed { ++ BFQ_BFQD_FAST, ++ BFQ_BFQD_SLOW, ++}; ++ ++/** ++ * struct bfq_data - per-device data structure. ++ * ++ * All the fields are protected by the @queue lock. ++ */ ++struct bfq_data { ++ /* request queue for the device */ ++ struct request_queue *queue; ++ ++ /* root bfq_group for the device */ ++ struct bfq_group *root_group; ++ ++ /* ++ * rbtree of weight counters of @bfq_queues, sorted by ++ * weight. Used to keep track of whether all @bfq_queues have ++ * the same weight. The tree contains one counter for each ++ * distinct weight associated to some active and not ++ * weight-raised @bfq_queue (see the comments to the functions ++ * bfq_weights_tree_[add|remove] for further details). ++ */ ++ struct rb_root queue_weights_tree; ++ /* ++ * rbtree of non-queue @bfq_entity weight counters, sorted by ++ * weight. Used to keep track of whether all @bfq_groups have ++ * the same weight. The tree contains one counter for each ++ * distinct weight associated to some active @bfq_group (see ++ * the comments to the functions bfq_weights_tree_[add|remove] ++ * for further details). ++ */ ++ struct rb_root group_weights_tree; ++ ++ /* ++ * Number of bfq_queues containing requests (including the ++ * queue in service, even if it is idling). ++ */ ++ int busy_queues; ++ /* number of weight-raised busy @bfq_queues */ ++ int wr_busy_queues; ++ /* number of queued requests */ ++ int queued; ++ /* number of requests dispatched and waiting for completion */ ++ int rq_in_driver; ++ ++ /* ++ * Maximum number of requests in driver in the last ++ * @hw_tag_samples completed requests. ++ */ ++ int max_rq_in_driver; ++ /* number of samples used to calculate hw_tag */ ++ int hw_tag_samples; ++ /* flag set to one if the driver is showing a queueing behavior */ ++ int hw_tag; ++ ++ /* number of budgets assigned */ ++ int budgets_assigned; ++ ++ /* ++ * Timer set when idling (waiting) for the next request from ++ * the queue in service. ++ */ ++ struct hrtimer idle_slice_timer; ++ /* delayed work to restart dispatching on the request queue */ ++ struct work_struct unplug_work; ++ ++ /* bfq_queue in service */ ++ struct bfq_queue *in_service_queue; ++ /* bfq_io_cq (bic) associated with the @in_service_queue */ ++ struct bfq_io_cq *in_service_bic; ++ ++ /* on-disk position of the last served request */ ++ sector_t last_position; ++ ++ /* time of last request completion (ns) */ ++ u64 last_completion; ++ ++ /* time of first rq dispatch in current observation interval (ns) */ ++ u64 first_dispatch; ++ /* time of last rq dispatch in current observation interval (ns) */ ++ u64 last_dispatch; ++ ++ /* beginning of the last budget */ ++ ktime_t last_budget_start; ++ /* beginning of the last idle slice */ ++ ktime_t last_idling_start; ++ ++ /* number of samples in current observation interval */ ++ int peak_rate_samples; ++ /* num of samples of seq dispatches in current observation interval */ ++ u32 sequential_samples; ++ /* total num of sectors transferred in current observation interval */ ++ u64 tot_sectors_dispatched; ++ /* max rq size seen during current observation interval (sectors) */ ++ u32 last_rq_max_size; ++ /* time elapsed from first dispatch in current observ. interval (us) */ ++ u64 delta_from_first; ++ /* current estimate of device peak rate */ ++ u32 peak_rate; ++ ++ /* maximum budget allotted to a bfq_queue before rescheduling */ ++ int bfq_max_budget; ++ ++ /* list of all the bfq_queues active on the device */ ++ struct list_head active_list; ++ /* list of all the bfq_queues idle on the device */ ++ struct list_head idle_list; ++ ++ /* ++ * Timeout for async/sync requests; when it fires, requests ++ * are served in fifo order. ++ */ ++ u64 bfq_fifo_expire[2]; ++ /* weight of backward seeks wrt forward ones */ ++ unsigned int bfq_back_penalty; ++ /* maximum allowed backward seek */ ++ unsigned int bfq_back_max; ++ /* maximum idling time */ ++ u32 bfq_slice_idle; ++ ++ /* user-configured max budget value (0 for auto-tuning) */ ++ int bfq_user_max_budget; ++ /* ++ * Timeout for bfq_queues to consume their budget; used to ++ * prevent seeky queues from imposing long latencies to ++ * sequential or quasi-sequential ones (this also implies that ++ * seeky queues cannot receive guarantees in the service ++ * domain; after a timeout they are charged for the time they ++ * have been in service, to preserve fairness among them, but ++ * without service-domain guarantees). ++ */ ++ unsigned int bfq_timeout; ++ ++ /* ++ * Number of consecutive requests that must be issued within ++ * the idle time slice to set again idling to a queue which ++ * was marked as non-I/O-bound (see the definition of the ++ * IO_bound flag for further details). ++ */ ++ unsigned int bfq_requests_within_timer; ++ ++ /* ++ * Force device idling whenever needed to provide accurate ++ * service guarantees, without caring about throughput ++ * issues. CAVEAT: this may even increase latencies, in case ++ * of useless idling for processes that did stop doing I/O. ++ */ ++ bool strict_guarantees; ++ ++ /* ++ * Last time at which a queue entered the current burst of ++ * queues being activated shortly after each other; for more ++ * details about this and the following parameters related to ++ * a burst of activations, see the comments on the function ++ * bfq_handle_burst. ++ */ ++ unsigned long last_ins_in_burst; ++ /* ++ * Reference time interval used to decide whether a queue has ++ * been activated shortly after @last_ins_in_burst. ++ */ ++ unsigned long bfq_burst_interval; ++ /* number of queues in the current burst of queue activations */ ++ int burst_size; ++ ++ /* common parent entity for the queues in the burst */ ++ struct bfq_entity *burst_parent_entity; ++ /* Maximum burst size above which the current queue-activation ++ * burst is deemed as 'large'. ++ */ ++ unsigned long bfq_large_burst_thresh; ++ /* true if a large queue-activation burst is in progress */ ++ bool large_burst; ++ /* ++ * Head of the burst list (as for the above fields, more ++ * details in the comments on the function bfq_handle_burst). ++ */ ++ struct hlist_head burst_list; ++ ++ /* if set to true, low-latency heuristics are enabled */ ++ bool low_latency; ++ /* ++ * Maximum factor by which the weight of a weight-raised queue ++ * is multiplied. ++ */ ++ unsigned int bfq_wr_coeff; ++ /* maximum duration of a weight-raising period (jiffies) */ ++ unsigned int bfq_wr_max_time; ++ ++ /* Maximum weight-raising duration for soft real-time processes */ ++ unsigned int bfq_wr_rt_max_time; ++ /* ++ * Minimum idle period after which weight-raising may be ++ * reactivated for a queue (in jiffies). ++ */ ++ unsigned int bfq_wr_min_idle_time; ++ /* ++ * Minimum period between request arrivals after which ++ * weight-raising may be reactivated for an already busy async ++ * queue (in jiffies). ++ */ ++ unsigned long bfq_wr_min_inter_arr_async; ++ ++ /* Max service-rate for a soft real-time queue, in sectors/sec */ ++ unsigned int bfq_wr_max_softrt_rate; ++ /* ++ * Cached value of the product R*T, used for computing the ++ * maximum duration of weight raising automatically. ++ */ ++ u64 RT_prod; ++ /* device-speed class for the low-latency heuristic */ ++ enum bfq_device_speed device_speed; ++ ++ /* fallback dummy bfqq for extreme OOM conditions */ ++ struct bfq_queue oom_bfqq; ++}; ++ ++enum bfqq_state_flags { ++ BFQ_BFQQ_FLAG_just_created = 0, /* queue just allocated */ ++ BFQ_BFQQ_FLAG_busy, /* has requests or is in service */ ++ BFQ_BFQQ_FLAG_wait_request, /* waiting for a request */ ++ BFQ_BFQQ_FLAG_non_blocking_wait_rq, /* ++ * waiting for a request ++ * without idling the device ++ */ ++ BFQ_BFQQ_FLAG_must_alloc, /* must be allowed rq alloc */ ++ BFQ_BFQQ_FLAG_fifo_expire, /* FIFO checked in this slice */ ++ BFQ_BFQQ_FLAG_idle_window, /* slice idling enabled */ ++ BFQ_BFQQ_FLAG_sync, /* synchronous queue */ ++ BFQ_BFQQ_FLAG_IO_bound, /* ++ * bfqq has timed-out at least once ++ * having consumed at most 2/10 of ++ * its budget ++ */ ++ BFQ_BFQQ_FLAG_in_large_burst, /* ++ * bfqq activated in a large burst, ++ * see comments to bfq_handle_burst. ++ */ ++ BFQ_BFQQ_FLAG_softrt_update, /* ++ * may need softrt-next-start ++ * update ++ */ ++ BFQ_BFQQ_FLAG_coop, /* bfqq is shared */ ++ BFQ_BFQQ_FLAG_split_coop /* shared bfqq will be split */ ++}; ++ ++#define BFQ_BFQQ_FNS(name) \ ++static void bfq_mark_bfqq_##name(struct bfq_queue *bfqq) \ ++{ \ ++ (bfqq)->flags |= (1 << BFQ_BFQQ_FLAG_##name); \ ++} \ ++static void bfq_clear_bfqq_##name(struct bfq_queue *bfqq) \ ++{ \ ++ (bfqq)->flags &= ~(1 << BFQ_BFQQ_FLAG_##name); \ ++} \ ++static int bfq_bfqq_##name(const struct bfq_queue *bfqq) \ ++{ \ ++ return ((bfqq)->flags & (1 << BFQ_BFQQ_FLAG_##name)) != 0; \ ++} ++ ++BFQ_BFQQ_FNS(just_created); ++BFQ_BFQQ_FNS(busy); ++BFQ_BFQQ_FNS(wait_request); ++BFQ_BFQQ_FNS(non_blocking_wait_rq); ++BFQ_BFQQ_FNS(must_alloc); ++BFQ_BFQQ_FNS(fifo_expire); ++BFQ_BFQQ_FNS(idle_window); ++BFQ_BFQQ_FNS(sync); ++BFQ_BFQQ_FNS(IO_bound); ++BFQ_BFQQ_FNS(in_large_burst); ++BFQ_BFQQ_FNS(coop); ++BFQ_BFQQ_FNS(split_coop); ++BFQ_BFQQ_FNS(softrt_update); ++#undef BFQ_BFQQ_FNS ++ ++/* Logging facilities. */ ++#ifdef CONFIG_BFQ_REDIRECT_TO_CONSOLE ++ ++static const char *checked_dev_name(const struct device *dev) ++{ ++ static const char nodev[] = "nodev"; ++ ++ if (dev) ++ return dev_name(dev); ++ ++ return nodev; ++} ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); ++static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ ++ char __pbuf[128]; \ ++ \ ++ assert_spin_locked((bfqd)->queue->queue_lock); \ ++ blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ ++ pr_crit("%s bfq%d%c %s " fmt "\n", \ ++ checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ ++ (bfqq)->pid, \ ++ bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ ++ __pbuf, ##args); \ ++} while (0) ++ ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ ++ char __pbuf[128]; \ ++ \ ++ blkg_path(bfqg_to_blkg(bfqg), __pbuf, sizeof(__pbuf)); \ ++ pr_crit("%s %s " fmt "\n", \ ++ checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ ++ __pbuf, ##args); \ ++} while (0) ++ ++#else /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ ++ pr_crit("%s bfq%d%c " fmt "\n", \ ++ checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ ++ (bfqq)->pid, bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ ++ ##args) ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) ++ ++#endif /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++ ++#define bfq_log(bfqd, fmt, args...) \ ++ pr_crit("%s bfq " fmt "\n", \ ++ checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ ++ ##args) ++ ++#else /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); ++static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ ++ char __pbuf[128]; \ ++ \ ++ assert_spin_locked((bfqd)->queue->queue_lock); \ ++ blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ ++ blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s " fmt, \ ++ (bfqq)->pid, \ ++ bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ ++ __pbuf, ##args); \ ++} while (0) ++ ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ ++ char __pbuf[128]; \ ++ \ ++ blkg_path(bfqg_to_blkg(bfqg), __pbuf, sizeof(__pbuf)); \ ++ blk_add_trace_msg((bfqd)->queue, "%s " fmt, __pbuf, ##args); \ ++} while (0) ++ ++#else /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ ++ blk_add_trace_msg((bfqd)->queue, "bfq%d%c " fmt, (bfqq)->pid, \ ++ bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ ++ ##args) ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) ++ ++#endif /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++ ++#define bfq_log(bfqd, fmt, args...) \ ++ blk_add_trace_msg((bfqd)->queue, "bfq " fmt, ##args) ++#endif /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ ++ ++/* Expiration reasons. */ ++enum bfqq_expiration { ++ BFQ_BFQQ_TOO_IDLE = 0, /* ++ * queue has been idling for ++ * too long ++ */ ++ BFQ_BFQQ_BUDGET_TIMEOUT, /* budget took too long to be used */ ++ BFQ_BFQQ_BUDGET_EXHAUSTED, /* budget consumed */ ++ BFQ_BFQQ_NO_MORE_REQUESTS, /* the queue has no more requests */ ++ BFQ_BFQQ_PREEMPTED /* preemption in progress */ ++}; ++ ++ ++struct bfqg_stats { ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ /* number of ios merged */ ++ struct blkg_rwstat merged; ++ /* total time spent on device in ns, may not be accurate w/ queueing */ ++ struct blkg_rwstat service_time; ++ /* total time spent waiting in scheduler queue in ns */ ++ struct blkg_rwstat wait_time; ++ /* number of IOs queued up */ ++ struct blkg_rwstat queued; ++ /* total disk time and nr sectors dispatched by this group */ ++ struct blkg_stat time; ++ /* sum of number of ios queued across all samples */ ++ struct blkg_stat avg_queue_size_sum; ++ /* count of samples taken for average */ ++ struct blkg_stat avg_queue_size_samples; ++ /* how many times this group has been removed from service tree */ ++ struct blkg_stat dequeue; ++ /* total time spent waiting for it to be assigned a timeslice. */ ++ struct blkg_stat group_wait_time; ++ /* time spent idling for this blkcg_gq */ ++ struct blkg_stat idle_time; ++ /* total time with empty current active q with other requests queued */ ++ struct blkg_stat empty_time; ++ /* fields after this shouldn't be cleared on stat reset */ ++ uint64_t start_group_wait_time; ++ uint64_t start_idle_time; ++ uint64_t start_empty_time; ++ uint16_t flags; ++#endif ++}; ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++/* ++ * struct bfq_group_data - per-blkcg storage for the blkio subsystem. ++ * ++ * @ps: @blkcg_policy_storage that this structure inherits ++ * @weight: weight of the bfq_group ++ */ ++struct bfq_group_data { ++ /* must be the first member */ ++ struct blkcg_policy_data pd; ++ ++ unsigned int weight; ++}; ++ ++/** ++ * struct bfq_group - per (device, cgroup) data structure. ++ * @entity: schedulable entity to insert into the parent group sched_data. ++ * @sched_data: own sched_data, to contain child entities (they may be ++ * both bfq_queues and bfq_groups). ++ * @bfqd: the bfq_data for the device this group acts upon. ++ * @async_bfqq: array of async queues for all the tasks belonging to ++ * the group, one queue per ioprio value per ioprio_class, ++ * except for the idle class that has only one queue. ++ * @async_idle_bfqq: async queue for the idle class (ioprio is ignored). ++ * @my_entity: pointer to @entity, %NULL for the toplevel group; used ++ * to avoid too many special cases during group creation/ ++ * migration. ++ * @active_entities: number of active entities belonging to the group; ++ * unused for the root group. Used to know whether there ++ * are groups with more than one active @bfq_entity ++ * (see the comments to the function ++ * bfq_bfqq_may_idle()). ++ * @rq_pos_tree: rbtree sorted by next_request position, used when ++ * determining if two or more queues have interleaving ++ * requests (see bfq_find_close_cooperator()). ++ * ++ * Each (device, cgroup) pair has its own bfq_group, i.e., for each cgroup ++ * there is a set of bfq_groups, each one collecting the lower-level ++ * entities belonging to the group that are acting on the same device. ++ * ++ * Locking works as follows: ++ * o @bfqd is protected by the queue lock, RCU is used to access it ++ * from the readers. ++ * o All the other fields are protected by the @bfqd queue lock. ++ */ ++struct bfq_group { ++ /* must be the first member */ ++ struct blkg_policy_data pd; ++ ++ struct bfq_entity entity; ++ struct bfq_sched_data sched_data; ++ ++ void *bfqd; ++ ++ struct bfq_queue *async_bfqq[2][IOPRIO_BE_NR]; ++ struct bfq_queue *async_idle_bfqq; ++ ++ struct bfq_entity *my_entity; ++ ++ int active_entities; ++ ++ struct rb_root rq_pos_tree; ++ ++ struct bfqg_stats stats; ++}; ++ ++#else ++struct bfq_group { ++ struct bfq_sched_data sched_data; ++ ++ struct bfq_queue *async_bfqq[2][IOPRIO_BE_NR]; ++ struct bfq_queue *async_idle_bfqq; ++ ++ struct rb_root rq_pos_tree; ++}; ++#endif ++ ++static struct bfq_queue *bfq_entity_to_bfqq(struct bfq_entity *entity); ++ ++static unsigned int bfq_class_idx(struct bfq_entity *entity) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ ++ return bfqq ? bfqq->ioprio_class - 1 : ++ BFQ_DEFAULT_GRP_CLASS - 1; ++} ++ ++static struct bfq_service_tree * ++bfq_entity_service_tree(struct bfq_entity *entity) ++{ ++ struct bfq_sched_data *sched_data = entity->sched_data; ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ unsigned int idx = bfq_class_idx(entity); ++ ++ BUG_ON(idx >= BFQ_IOPRIO_CLASSES); ++ BUG_ON(sched_data == NULL); ++ ++ if (bfqq) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "entity_service_tree %p %d", ++ sched_data->service_tree + idx, idx); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ else { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "entity_service_tree %p %d", ++ sched_data->service_tree + idx, idx); ++ } ++#endif ++ return sched_data->service_tree + idx; ++} ++ ++static struct bfq_queue *bic_to_bfqq(struct bfq_io_cq *bic, bool is_sync) ++{ ++ return bic->bfqq[is_sync]; ++} ++ ++static void bic_set_bfqq(struct bfq_io_cq *bic, struct bfq_queue *bfqq, ++ bool is_sync) ++{ ++ bic->bfqq[is_sync] = bfqq; ++} ++ ++static struct bfq_data *bic_to_bfqd(struct bfq_io_cq *bic) ++{ ++ return bic->icq.q->elevator->elevator_data; ++} ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ ++static struct bfq_group *bfq_bfqq_to_bfqg(struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *group_entity = bfqq->entity.parent; ++ ++ if (!group_entity) ++ group_entity = &bfqq->bfqd->root_group->entity; ++ ++ return container_of(group_entity, struct bfq_group, entity); ++} ++ ++#else ++ ++static struct bfq_group *bfq_bfqq_to_bfqg(struct bfq_queue *bfqq) ++{ ++ return bfqq->bfqd->root_group; ++} ++ ++#endif ++ ++static void bfq_check_ioprio_change(struct bfq_io_cq *bic, struct bio *bio); ++static void bfq_put_queue(struct bfq_queue *bfqq); ++static void bfq_dispatch_insert(struct request_queue *q, struct request *rq); ++static struct bfq_queue *bfq_get_queue(struct bfq_data *bfqd, ++ struct bio *bio, bool is_sync, ++ struct bfq_io_cq *bic); ++static void bfq_end_wr_async_queues(struct bfq_data *bfqd, ++ struct bfq_group *bfqg); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++static void bfq_put_async_queues(struct bfq_data *bfqd, struct bfq_group *bfqg); ++#endif ++static void bfq_exit_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq); ++ ++#endif /* _BFQ_H */ +diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h +index 8da66379f7ea..bf000c58644b 100644 +--- a/include/linux/blkdev.h ++++ b/include/linux/blkdev.h +@@ -54,7 +54,7 @@ struct blk_stat_callback; + * Maximum number of blkcg policies allowed to be registered concurrently. + * Defined here to simplify include dependency. + */ +-#define BLKCG_MAX_POLS 3 ++#define BLKCG_MAX_POLS 4 + + typedef void (rq_end_io_fn)(struct request *, blk_status_t); + + +From 9916fed6c89c61a2b26053be04501784570bbec8 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 20 Jul 2017 10:46:39 +0200 +Subject: [PATCH 02/51] Add extra checks related to entity scheduling + +- extra checks related to ioprioi-class changes +- specific check on st->idle in __bfq_requeue_entity + +Signed-off-by: Paolo Valente +--- + block/bfq-sched.c | 9 ++++++++- + 1 file changed, 8 insertions(+), 1 deletion(-) + +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index ac8991bca9fa..5ddf9af4261e 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -812,6 +812,7 @@ __bfq_entity_update_weight_prio(struct bfq_service_tree *old_st, + } + #endif + ++ BUG_ON(entity->tree && update_class_too); + BUG_ON(old_st->wsum < entity->weight); + old_st->wsum -= entity->weight; + +@@ -883,8 +884,10 @@ __bfq_entity_update_weight_prio(struct bfq_service_tree *old_st, + + new_st->wsum += entity->weight; + +- if (new_st != old_st) ++ if (new_st != old_st) { ++ BUG_ON(!update_class_too); + entity->start = new_st->vtime; ++ } + } + + return new_st; +@@ -993,6 +996,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + * tree, then it is safe to invoke next function with the last + * parameter set (see the comments on the function). + */ ++ BUG_ON(entity->tree); + st = __bfq_entity_update_weight_prio(st, entity, true); + bfq_calc_finish(entity, entity->budget); + +@@ -1113,9 +1117,11 @@ static void __bfq_activate_entity(struct bfq_entity *entity, + * check for that. + */ + bfq_idle_extract(st, entity); ++ BUG_ON(entity->tree); + entity->start = bfq_gt(min_vstart, entity->finish) ? + min_vstart : entity->finish; + } else { ++ BUG_ON(entity->tree); + /* + * The finish time of the entity may be invalid, and + * it is in the past for sure, otherwise the queue +@@ -1203,6 +1209,7 @@ static void __bfq_requeue_entity(struct bfq_entity *entity) + */ + bfq_calc_finish(entity, entity->service); + entity->start = entity->finish; ++ BUG_ON(entity->tree && entity->tree == &st->idle); + BUG_ON(entity->tree && entity->tree != &st->active); + /* + * In addition, if the entity had more than one child + +From 8f5b2c25dcbe31dda524e85b921b3aa1fe11d111 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 21 Jul 2017 12:08:57 +0200 +Subject: [PATCH 03/51] block, bfq: reset in_service_entity if it becomes idle + +BFQ implements hierarchical scheduling by representing each group of +queues with a generic parent entity. For each parent entity, BFQ +maintains an in_service_entity pointer: if one of the child entities +happens to be in service, in_service_entity points to it. The +resetting of these pointers happens only on queue expirations: when +the in-service queue is expired, i.e., stops to be the queue in +service, BFQ resets all in_service_entity pointers along the +parent-entity path from this queue to the root entity. + +Functions handling the scheduling of entities assume, naturally, that +in-service entities are active, i.e., have pending I/O requests (or, +as a special case, even if they have no pending requests, they are +expected to receive a new request very soon, with the scheduler idling +the storage device while waiting for such an event). Unfortunately, +the above resetting scheme of the in_service_entity pointers may cause +this assumption to be violated. For example, the in-service queue may +happen to remain without requests because of a request merge. In this +case the queue does become idle, and all related data structures are +updated accordingly. But in_service_entity still points to the queue +in the parent entity. This inconsistency may even propagate to +higher-level parent entities, if they happen to become idle as well, +as a consequence of the leaf queue becoming idle. For this queue and +parent entities, scheduling functions have an undefined behaviour, +and, as reported, may easily lead to kernel crashes or hangs. + +This commit addresses this issue by simply resetting the +in_service_entity field also when it is detected to point to an entity +becoming idle (regardless of why the entity becomes idle). + +Reported-by: Laurentiu Nicola +Signed-off-by: Paolo Valente +Tested-by: Laurentiu Nicola +--- + block/bfq-sched.c | 4 +++- + 1 file changed, 3 insertions(+), 1 deletion(-) + +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index 5ddf9af4261e..a07a06eb5c72 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -1336,8 +1336,10 @@ static bool __bfq_deactivate_entity(struct bfq_entity *entity, + + BUG_ON(is_in_service && entity->tree && entity->tree != &st->active); + +- if (is_in_service) ++ if (is_in_service) { + bfq_calc_finish(entity, entity->service); ++ sd->in_service_entity = NULL; ++ } + + if (entity->tree == &st->active) + bfq_active_extract(st, entity); + +From 600ea668e2d340c95724bcf981d88812d6900342 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 28 Jul 2017 21:09:51 +0200 +Subject: [PATCH 04/51] block, bfq: consider also in_service_entity to state + whether an entity is active + +Groups of BFQ queues are represented by generic entities in BFQ. When +a queue belonging to a parent entity is deactivated, the parent entity +may need to be deactivated too, in case the deactivated queue was the +only active queue for the parent entity. This deactivation may need to +be propagated upwards if the entity belongs, in its turn, to a further +higher-level entity, and so on. In particular, the upward propagation +of deactivation stops at the first parent entity that remains active +even if one of its child entities has been deactivated. + +To decide whether the last non-deactivation condition holds for a +parent entity, BFQ checks whether the field next_in_service is still +not NULL for the parent entity, after the deactivation of one of its +child entity. If it is not NULL, then there are certainly other active +entities in the parent entity, and deactivations can stop. + +Unfortunately, this check misses a corner case: if in_service_entity +is not NULL, then next_in_service may happen to be NULL, although the +parent entity is evidently active. This happens if: 1) the entity +pointed by in_service_entity is the only active entity in the parent +entity, and 2) according to the definition of next_in_service, the +in_service_entity cannot be considered as next_in_service. See the +comments on the definition of next_in_service for details on this +second point. + +Hitting the above corner case causes crashes. + +To address this issue, this commit: +1) Extends the above check on only next_in_service to controlling both +next_in_service and in_service_entity (if any of them is not NULL, +then no further deactivation is performed) +2) Improves the (important) comments on how next_in_service is defined +and updated; in particular it fixes a few rather obscure paragraphs + +Reported-by: Eric Wheeler +Reported-by: Rick Yiu +Reported-by: Tom X Nguyen +Signed-off-by: Paolo Valente +Tested-by: Eric Wheeler +Tested-by: Rick Yiu +Tested-by: Laurentiu Nicola +Tested-by: Tom X Nguyen +--- + block/bfq-sched.c | 140 ++++++++++++++++++++++++++++++------------------------ + block/bfq.h | 23 +++++++-- + 2 files changed, 95 insertions(+), 68 deletions(-) + +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index a07a06eb5c72..5c0f9290a79c 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -196,21 +196,23 @@ static bool bfq_update_parent_budget(struct bfq_entity *next_in_service) + + /* + * This function tells whether entity stops being a candidate for next +- * service, according to the following logic. ++ * service, according to the restrictive definition of the field ++ * next_in_service. In particular, this function is invoked for an ++ * entity that is about to be set in service. + * +- * This function is invoked for an entity that is about to be set in +- * service. If such an entity is a queue, then the entity is no longer +- * a candidate for next service (i.e, a candidate entity to serve +- * after the in-service entity is expired). The function then returns +- * true. ++ * If entity is a queue, then the entity is no longer a candidate for ++ * next service according to the that definition, because entity is ++ * about to become the in-service queue. This function then returns ++ * true if entity is a queue. + * +- * In contrast, the entity could stil be a candidate for next service +- * if it is not a queue, and has more than one child. In fact, even if +- * one of its children is about to be set in service, other children +- * may still be the next to serve. As a consequence, a non-queue +- * entity is not a candidate for next-service only if it has only one +- * child. And only if this condition holds, then the function returns +- * true for a non-queue entity. ++ * In contrast, entity could still be a candidate for next service if ++ * it is not a queue, and has more than one active child. In fact, ++ * even if one of its children is about to be set in service, other ++ * active children may still be the next to serve, for the parent ++ * entity, even according to the above definition. As a consequence, a ++ * non-queue entity is not a candidate for next-service only if it has ++ * only one active child. And only if this condition holds, then this ++ * function returns true for a non-queue entity. + */ + static bool bfq_no_longer_next_in_service(struct bfq_entity *entity) + { +@@ -223,6 +225,18 @@ static bool bfq_no_longer_next_in_service(struct bfq_entity *entity) + + BUG_ON(bfqg == ((struct bfq_data *)(bfqg->bfqd))->root_group); + BUG_ON(bfqg->active_entities == 0); ++ /* ++ * The field active_entities does not always contain the ++ * actual number of active children entities: it happens to ++ * not account for the in-service entity in case the latter is ++ * removed from its active tree (which may get done after ++ * invoking the function bfq_no_longer_next_in_service in ++ * bfq_get_next_queue). Fortunately, here, i.e., while ++ * bfq_no_longer_next_in_service is not yet completed in ++ * bfq_get_next_queue, bfq_active_extract has not yet been ++ * invoked, and thus active_entities still coincides with the ++ * actual number of active entities. ++ */ + if (bfqg->active_entities == 1) + return true; + +@@ -1089,7 +1103,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + * one of its children receives a new request. + * + * Basically, this function updates the timestamps of entity and +- * inserts entity into its active tree, ater possible extracting it ++ * inserts entity into its active tree, ater possibly extracting it + * from its idle tree. + */ + static void __bfq_activate_entity(struct bfq_entity *entity, +@@ -1213,7 +1227,7 @@ static void __bfq_requeue_entity(struct bfq_entity *entity) + BUG_ON(entity->tree && entity->tree != &st->active); + /* + * In addition, if the entity had more than one child +- * when set in service, then was not extracted from ++ * when set in service, then it was not extracted from + * the active tree. This implies that the position of + * the entity in the active tree may need to be + * changed now, because we have just updated the start +@@ -1221,9 +1235,8 @@ static void __bfq_requeue_entity(struct bfq_entity *entity) + * time in a moment (the requeueing is then, more + * precisely, a repositioning in this case). To + * implement this repositioning, we: 1) dequeue the +- * entity here, 2) update the finish time and +- * requeue the entity according to the new +- * timestamps below. ++ * entity here, 2) update the finish time and requeue ++ * the entity according to the new timestamps below. + */ + if (entity->tree) + bfq_active_extract(st, entity); +@@ -1270,9 +1283,9 @@ static void __bfq_activate_requeue_entity(struct bfq_entity *entity, + + + /** +- * bfq_activate_entity - activate or requeue an entity representing a bfq_queue, +- * and activate, requeue or reposition all ancestors +- * for which such an update becomes necessary. ++ * bfq_activate_requeue_entity - activate or requeue an entity representing a bfq_queue, ++ * and activate, requeue or reposition all ancestors ++ * for which such an update becomes necessary. + * @entity: the entity to activate. + * @non_blocking_wait_rq: true if this entity was waiting for a request + * @requeue: true if this is a requeue, which implies that bfqq is +@@ -1308,9 +1321,9 @@ static void bfq_activate_requeue_entity(struct bfq_entity *entity, + * @ins_into_idle_tree: if false, the entity will not be put into the + * idle tree. + * +- * Deactivates an entity, independently from its previous state. Must ++ * Deactivates an entity, independently of its previous state. Must + * be invoked only if entity is on a service tree. Extracts the entity +- * from that tree, and if necessary and allowed, puts it on the idle ++ * from that tree, and if necessary and allowed, puts it into the idle + * tree. + */ + static bool __bfq_deactivate_entity(struct bfq_entity *entity, +@@ -1359,7 +1372,7 @@ static bool __bfq_deactivate_entity(struct bfq_entity *entity, + /** + * bfq_deactivate_entity - deactivate an entity representing a bfq_queue. + * @entity: the entity to deactivate. +- * @ins_into_idle_tree: true if the entity can be put on the idle tree ++ * @ins_into_idle_tree: true if the entity can be put into the idle tree + */ + static void bfq_deactivate_entity(struct bfq_entity *entity, + bool ins_into_idle_tree, +@@ -1406,16 +1419,29 @@ static void bfq_deactivate_entity(struct bfq_entity *entity, + */ + bfq_update_next_in_service(sd, NULL); + +- if (sd->next_in_service) { ++ if (sd->next_in_service || sd->in_service_entity) { + /* +- * The parent entity is still backlogged, +- * because next_in_service is not NULL. So, no +- * further upwards deactivation must be +- * performed. Yet, next_in_service has +- * changed. Then the schedule does need to be +- * updated upwards. ++ * The parent entity is still active, because ++ * either next_in_service or in_service_entity ++ * is not NULL. So, no further upwards ++ * deactivation must be performed. Yet, ++ * next_in_service has changed. Then the ++ * schedule does need to be updated upwards. ++ * ++ * NOTE If in_service_entity is not NULL, then ++ * next_in_service may happen to be NULL, ++ * although the parent entity is evidently ++ * active. This happens if 1) the entity ++ * pointed by in_service_entity is the only ++ * active entity in the parent entity, and 2) ++ * according to the definition of ++ * next_in_service, the in_service_entity ++ * cannot be considered as ++ * next_in_service. See the comments on the ++ * definition of next_in_service for details. + */ + BUG_ON(sd->next_in_service == entity); ++ BUG_ON(sd->in_service_entity == entity); + break; + } + +@@ -1806,45 +1832,33 @@ static struct bfq_queue *bfq_get_next_queue(struct bfq_data *bfqd) + + /* + * If entity is no longer a candidate for next +- * service, then we extract it from its active tree, +- * for the following reason. To further boost the +- * throughput in some special case, BFQ needs to know +- * which is the next candidate entity to serve, while +- * there is already an entity in service. In this +- * respect, to make it easy to compute/update the next +- * candidate entity to serve after the current +- * candidate has been set in service, there is a case +- * where it is necessary to extract the current +- * candidate from its service tree. Such a case is +- * when the entity just set in service cannot be also +- * a candidate for next service. Details about when +- * this conditions holds are reported in the comments +- * on the function bfq_no_longer_next_in_service() +- * invoked below. ++ * service, then it must be extracted from its active ++ * tree, so as to make sure that it won't be ++ * considered when computing next_in_service. See the ++ * comments on the function ++ * bfq_no_longer_next_in_service() for details. + */ + if (bfq_no_longer_next_in_service(entity)) + bfq_active_extract(bfq_entity_service_tree(entity), + entity); + + /* +- * For the same reason why we may have just extracted +- * entity from its active tree, we may need to update +- * next_in_service for the sched_data of entity too, +- * regardless of whether entity has been extracted. +- * In fact, even if entity has not been extracted, a +- * descendant entity may get extracted. Such an event +- * would cause a change in next_in_service for the +- * level of the descendant entity, and thus possibly +- * back to upper levels. ++ * Even if entity is not to be extracted according to ++ * the above check, a descendant entity may get ++ * extracted in one of the next iterations of this ++ * loop. Such an event could cause a change in ++ * next_in_service for the level of the descendant ++ * entity, and thus possibly back to this level. + * +- * We cannot perform the resulting needed update +- * before the end of this loop, because, to know which +- * is the correct next-to-serve candidate entity for +- * each level, we need first to find the leaf entity +- * to set in service. In fact, only after we know +- * which is the next-to-serve leaf entity, we can +- * discover whether the parent entity of the leaf +- * entity becomes the next-to-serve, and so on. ++ * However, we cannot perform the resulting needed ++ * update of next_in_service for this level before the ++ * end of the whole loop, because, to know which is ++ * the correct next-to-serve candidate entity for each ++ * level, we need first to find the leaf entity to set ++ * in service. In fact, only after we know which is ++ * the next-to-serve leaf entity, we can discover ++ * whether the parent entity of the leaf entity ++ * becomes the next-to-serve, and so on. + */ + + /* Log some information */ +diff --git a/block/bfq.h b/block/bfq.h +index f5751ea59d98..ebd9688b9f61 100644 +--- a/block/bfq.h ++++ b/block/bfq.h +@@ -68,17 +68,30 @@ struct bfq_service_tree { + * + * bfq_sched_data is the basic scheduler queue. It supports three + * ioprio_classes, and can be used either as a toplevel queue or as an +- * intermediate queue on a hierarchical setup. @next_in_service +- * points to the active entity of the sched_data service trees that +- * will be scheduled next. It is used to reduce the number of steps +- * needed for each hierarchical-schedule update. ++ * intermediate queue in a hierarchical setup. + * + * The supported ioprio_classes are the same as in CFQ, in descending + * priority order, IOPRIO_CLASS_RT, IOPRIO_CLASS_BE, IOPRIO_CLASS_IDLE. + * Requests from higher priority queues are served before all the + * requests from lower priority queues; among requests of the same + * queue requests are served according to B-WF2Q+. +- * All the fields are protected by the queue lock of the containing bfqd. ++ * ++ * The schedule is implemented by the service trees, plus the field ++ * @next_in_service, which points to the entity on the active trees ++ * that will be served next, if 1) no changes in the schedule occurs ++ * before the current in-service entity is expired, 2) the in-service ++ * queue becomes idle when it expires, and 3) if the entity pointed by ++ * in_service_entity is not a queue, then the in-service child entity ++ * of the entity pointed by in_service_entity becomes idle on ++ * expiration. This peculiar definition allows for the following ++ * optimization, not yet exploited: while a given entity is still in ++ * service, we already know which is the best candidate for next ++ * service among the other active entitities in the same parent ++ * entity. We can then quickly compare the timestamps of the ++ * in-service entity with those of such best candidate. ++ * ++ * All the fields are protected by the queue lock of the containing ++ * bfqd. + */ + struct bfq_sched_data { + struct bfq_entity *in_service_entity; /* entity in service */ + +From 6b5effd10bc6711a862e7cbd7cd2dd0146defa01 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 4 May 2017 10:53:43 +0200 +Subject: [PATCH 05/51] block, bfq: improve and refactor throughput-boosting + logic + +When a queue associated with a process remains empty, there are cases +where throughput gets boosted if the device is idled to await the +arrival of a new I/O request for that queue. Currently, BFQ assumes +that one of these cases is when the device has no internal queueing +(regardless of the properties of the I/O being served). Unfortunately, +this condition has proved to be too general. So, this commit refines it +as "the device has no internal queueing and is rotational". + +This refinement provides a significant throughput boost with random +I/O, on flash-based storage without internal queueing. For example, on +a HiKey board, throughput increases by up to 125%, growing, e.g., from +6.9MB/s to 15.6MB/s with two or three random readers in parallel. + +This commit also refactors the code related to device idling, for the +following reason. Finding the change that provides the above large +improvement has been slightly more difficult than it had to be, +because the logic that decides whether to idle the device is still +scattered across three functions. Almost all of the logic is in the +function bfq_bfqq_may_idle, but (1) part of the decision is made in +bfq_update_idle_window, and (2) the function bfq_bfqq_must_idle may +switch off idling regardless of the output of bfq_bfqq_may_idle. In +addition, both bfq_update_idle_window and bfq_bfqq_must_idle make +their decisions as a function of parameters that are used, for similar +purposes, also in bfq_bfqq_may_idle. This commit addresses this issue +by moving all the logic into bfq_bfqq_may_idle. + +Signed-off-by: Paolo Valente +Signed-off-by: Luca Miccio +--- + block/bfq-sq-iosched.c | 141 +++++++++++++++++++++++++++---------------------- + block/bfq.h | 12 ++--- + 2 files changed, 83 insertions(+), 70 deletions(-) + +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 65e7c7e77f3c..30d019fc67e0 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -684,10 +684,10 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + unsigned int old_wr_coeff; + bool busy = bfq_already_existing && bfq_bfqq_busy(bfqq); + +- if (bic->saved_idle_window) +- bfq_mark_bfqq_idle_window(bfqq); ++ if (bic->saved_has_short_ttime) ++ bfq_mark_bfqq_has_short_ttime(bfqq); + else +- bfq_clear_bfqq_idle_window(bfqq); ++ bfq_clear_bfqq_has_short_ttime(bfqq); + + if (bic->saved_IO_bound) + bfq_mark_bfqq_IO_bound(bfqq); +@@ -2047,7 +2047,7 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq) + if (!bic) + return; + +- bic->saved_idle_window = bfq_bfqq_idle_window(bfqq); ++ bic->saved_has_short_ttime = bfq_bfqq_has_short_ttime(bfqq); + bic->saved_IO_bound = bfq_bfqq_IO_bound(bfqq); + bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq); + bic->was_in_burst_list = !hlist_unhashed(&bfqq->burst_list_node); +@@ -3214,9 +3214,9 @@ static void bfq_bfqq_expire(struct bfq_data *bfqd, + } + + bfq_log_bfqq(bfqd, bfqq, +- "expire (%d, slow %d, num_disp %d, idle_win %d, weight %d)", ++ "expire (%d, slow %d, num_disp %d, short_ttime %d, weight %d)", + reason, slow, bfqq->dispatched, +- bfq_bfqq_idle_window(bfqq), entity->weight); ++ bfq_bfqq_has_short_ttime(bfqq), entity->weight); + + /* + * Increase, decrease or leave budget unchanged according to +@@ -3298,7 +3298,10 @@ static bool bfq_may_expire_for_budg_timeout(struct bfq_queue *bfqq) + static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq) + { + struct bfq_data *bfqd = bfqq->bfqd; +- bool idling_boosts_thr, idling_boosts_thr_without_issues, ++ bool rot_without_queueing = ++ !blk_queue_nonrot(bfqd->queue) && !bfqd->hw_tag, ++ bfqq_sequential_and_IO_bound, ++ idling_boosts_thr, idling_boosts_thr_without_issues, + idling_needed_for_service_guarantees, + asymmetric_scenario; + +@@ -3306,27 +3309,44 @@ static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq) + return true; + + /* ++ * Idling is performed only if slice_idle > 0. In addition, we ++ * do not idle if ++ * (a) bfqq is async ++ * (b) bfqq is in the idle io prio class: in this case we do ++ * not idle because we want to minimize the bandwidth that ++ * queues in this class can steal to higher-priority queues ++ */ ++ if (bfqd->bfq_slice_idle == 0 || !bfq_bfqq_sync(bfqq) || ++ bfq_class_idle(bfqq)) ++ return false; ++ ++ bfqq_sequential_and_IO_bound = !BFQQ_SEEKY(bfqq) && ++ bfq_bfqq_IO_bound(bfqq) && bfq_bfqq_has_short_ttime(bfqq); ++ /* + * The next variable takes into account the cases where idling + * boosts the throughput. + * + * The value of the variable is computed considering, first, that + * idling is virtually always beneficial for the throughput if: +- * (a) the device is not NCQ-capable, or +- * (b) regardless of the presence of NCQ, the device is rotational +- * and the request pattern for bfqq is I/O-bound and sequential. ++ * (a) the device is not NCQ-capable and rotational, or ++ * (b) regardless of the presence of NCQ, the device is rotational and ++ * the request pattern for bfqq is I/O-bound and sequential, or ++ * (c) regardless of whether it is rotational, the device is ++ * not NCQ-capable and the request pattern for bfqq is ++ * I/O-bound and sequential. + * + * Secondly, and in contrast to the above item (b), idling an + * NCQ-capable flash-based device would not boost the + * throughput even with sequential I/O; rather it would lower + * the throughput in proportion to how fast the device + * is. Accordingly, the next variable is true if any of the +- * above conditions (a) and (b) is true, and, in particular, +- * happens to be false if bfqd is an NCQ-capable flash-based +- * device. ++ * above conditions (a), (b) or (c) is true, and, in ++ * particular, happens to be false if bfqd is an NCQ-capable ++ * flash-based device. + */ +- idling_boosts_thr = !bfqd->hw_tag || +- (!blk_queue_nonrot(bfqd->queue) && bfq_bfqq_IO_bound(bfqq) && +- bfq_bfqq_idle_window(bfqq)); ++ idling_boosts_thr = rot_without_queueing || ++ ((!blk_queue_nonrot(bfqd->queue) || !bfqd->hw_tag) && ++ bfqq_sequential_and_IO_bound); + + /* + * The value of the next variable, +@@ -3497,12 +3517,10 @@ static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq) + asymmetric_scenario && !bfq_bfqq_in_large_burst(bfqq); + + /* +- * We have now all the components we need to compute the return +- * value of the function, which is true only if both the following +- * conditions hold: +- * 1) bfqq is sync, because idling make sense only for sync queues; +- * 2) idling either boosts the throughput (without issues), or +- * is necessary to preserve service guarantees. ++ * We have now all the components we need to compute the ++ * return value of the function, which is true only if idling ++ * either boosts the throughput (without issues), or is ++ * necessary to preserve service guarantees. + */ + bfq_log_bfqq(bfqd, bfqq, "may_idle: sync %d idling_boosts_thr %d", + bfq_bfqq_sync(bfqq), idling_boosts_thr); +@@ -3514,9 +3532,8 @@ static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq) + bfq_bfqq_IO_bound(bfqq), + idling_needed_for_service_guarantees); + +- return bfq_bfqq_sync(bfqq) && +- (idling_boosts_thr_without_issues || +- idling_needed_for_service_guarantees); ++ return idling_boosts_thr_without_issues || ++ idling_needed_for_service_guarantees; + } + + /* +@@ -3532,10 +3549,7 @@ static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq) + */ + static bool bfq_bfqq_must_idle(struct bfq_queue *bfqq) + { +- struct bfq_data *bfqd = bfqq->bfqd; +- +- return RB_EMPTY_ROOT(&bfqq->sort_list) && bfqd->bfq_slice_idle != 0 && +- bfq_bfqq_may_idle(bfqq); ++ return RB_EMPTY_ROOT(&bfqq->sort_list) && bfq_bfqq_may_idle(bfqq); + } + + /* +@@ -3994,7 +4008,6 @@ static void bfq_set_next_ioprio_data(struct bfq_queue *bfqq, + case IOPRIO_CLASS_IDLE: + bfqq->new_ioprio_class = IOPRIO_CLASS_IDLE; + bfqq->new_ioprio = 7; +- bfq_clear_bfqq_idle_window(bfqq); + break; + } + +@@ -4058,8 +4071,14 @@ static void bfq_init_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, + bfq_set_next_ioprio_data(bfqq, bic); + + if (is_sync) { ++ /* ++ * No need to mark as has_short_ttime if in ++ * idle_class, because no device idling is performed ++ * for queues in idle class ++ */ + if (!bfq_class_idle(bfqq)) +- bfq_mark_bfqq_idle_window(bfqq); ++ /* tentatively mark as has_short_ttime */ ++ bfq_mark_bfqq_has_short_ttime(bfqq); + bfq_mark_bfqq_sync(bfqq); + bfq_mark_bfqq_just_created(bfqq); + } else +@@ -4195,18 +4214,19 @@ bfq_update_io_seektime(struct bfq_data *bfqd, struct bfq_queue *bfqq, + blk_rq_sectors(rq) < BFQQ_SECT_THR_NONROT); + } + +-/* +- * Disable idle window if the process thinks too long or seeks so much that +- * it doesn't matter. +- */ +-static void bfq_update_idle_window(struct bfq_data *bfqd, +- struct bfq_queue *bfqq, +- struct bfq_io_cq *bic) ++static void bfq_update_has_short_ttime(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ struct bfq_io_cq *bic) + { +- int enable_idle; ++ bool has_short_ttime = true; + +- /* Don't idle for async or idle io prio class. */ +- if (!bfq_bfqq_sync(bfqq) || bfq_class_idle(bfqq)) ++ /* ++ * No need to update has_short_ttime if bfqq is async or in ++ * idle io prio class, or if bfq_slice_idle is zero, because ++ * no device idling is performed for bfqq in this case. ++ */ ++ if (!bfq_bfqq_sync(bfqq) || bfq_class_idle(bfqq) || ++ bfqd->bfq_slice_idle == 0) + return; + + /* Idle window just restored, statistics are meaningless. */ +@@ -4214,27 +4234,22 @@ static void bfq_update_idle_window(struct bfq_data *bfqd, + bfqd->bfq_wr_min_idle_time)) + return; + +- enable_idle = bfq_bfqq_idle_window(bfqq); +- ++ /* Think time is infinite if no process is linked to ++ * bfqq. Otherwise check average think time to ++ * decide whether to mark as has_short_ttime ++ */ + if (atomic_read(&bic->icq.ioc->active_ref) == 0 || +- bfqd->bfq_slice_idle == 0 || +- (bfqd->hw_tag && BFQQ_SEEKY(bfqq) && +- bfqq->wr_coeff == 1)) +- enable_idle = 0; +- else if (bfq_sample_valid(bic->ttime.ttime_samples)) { +- if (bic->ttime.ttime_mean > bfqd->bfq_slice_idle && +- bfqq->wr_coeff == 1) +- enable_idle = 0; +- else +- enable_idle = 1; +- } +- bfq_log_bfqq(bfqd, bfqq, "update_idle_window: enable_idle %d", +- enable_idle); ++ (bfq_sample_valid(bic->ttime.ttime_samples) && ++ bic->ttime.ttime_mean > bfqd->bfq_slice_idle)) ++ has_short_ttime = false; ++ ++ bfq_log_bfqq(bfqd, bfqq, "update_has_short_ttime: has_short_ttime %d", ++ has_short_ttime); + +- if (enable_idle) +- bfq_mark_bfqq_idle_window(bfqq); ++ if (has_short_ttime) ++ bfq_mark_bfqq_has_short_ttime(bfqq); + else +- bfq_clear_bfqq_idle_window(bfqq); ++ bfq_clear_bfqq_has_short_ttime(bfqq); + } + + /* +@@ -4250,14 +4265,12 @@ static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, + bfqq->meta_pending++; + + bfq_update_io_thinktime(bfqd, bic); ++ bfq_update_has_short_ttime(bfqd, bfqq, bic); + bfq_update_io_seektime(bfqd, bfqq, rq); +- if (bfqq->entity.service > bfq_max_budget(bfqd) / 8 || +- !BFQQ_SEEKY(bfqq)) +- bfq_update_idle_window(bfqd, bfqq, bic); + + bfq_log_bfqq(bfqd, bfqq, +- "rq_enqueued: idle_window=%d (seeky %d)", +- bfq_bfqq_idle_window(bfqq), BFQQ_SEEKY(bfqq)); ++ "rq_enqueued: has_short_ttime=%d (seeky %d)", ++ bfq_bfqq_has_short_ttime(bfqq), BFQQ_SEEKY(bfqq)); + + bfqq->last_request_pos = blk_rq_pos(rq) + blk_rq_sectors(rq); + +diff --git a/block/bfq.h b/block/bfq.h +index ebd9688b9f61..34fc4697fd89 100644 +--- a/block/bfq.h ++++ b/block/bfq.h +@@ -349,11 +349,11 @@ struct bfq_io_cq { + #endif + + /* +- * Snapshot of the idle window before merging; taken to +- * remember this value while the queue is merged, so as to be +- * able to restore it in case of split. ++ * Snapshot of the has_short_time flag before merging; taken ++ * to remember its value while the queue is merged, so as to ++ * be able to restore it in case of split. + */ +- bool saved_idle_window; ++ bool saved_has_short_ttime; + /* + * Same purpose as the previous two fields for the I/O bound + * classification of a queue. +@@ -610,7 +610,7 @@ enum bfqq_state_flags { + */ + BFQ_BFQQ_FLAG_must_alloc, /* must be allowed rq alloc */ + BFQ_BFQQ_FLAG_fifo_expire, /* FIFO checked in this slice */ +- BFQ_BFQQ_FLAG_idle_window, /* slice idling enabled */ ++ BFQ_BFQQ_FLAG_has_short_ttime, /* queue has a short think time */ + BFQ_BFQQ_FLAG_sync, /* synchronous queue */ + BFQ_BFQQ_FLAG_IO_bound, /* + * bfqq has timed-out at least once +@@ -649,7 +649,7 @@ BFQ_BFQQ_FNS(wait_request); + BFQ_BFQQ_FNS(non_blocking_wait_rq); + BFQ_BFQQ_FNS(must_alloc); + BFQ_BFQQ_FNS(fifo_expire); +-BFQ_BFQQ_FNS(idle_window); ++BFQ_BFQQ_FNS(has_short_ttime); + BFQ_BFQQ_FNS(sync); + BFQ_BFQQ_FNS(IO_bound); + BFQ_BFQQ_FNS(in_large_burst); + +From b5e746fa99d961a5642cffb27c19a77e8b638007 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Mon, 19 Dec 2016 16:59:33 +0100 +Subject: [PATCH 06/51] FIRST BFQ-MQ COMMIT: Copy bfq-sq-iosched.c as + bfq-mq-iosched.c + +This commit introduces bfq-mq-iosched.c, the main source file that +will contain the code of bfq for blk-mq. I name tentatively +bfq-mq this version of bfq. + +For the moment, the file bfq-mq-iosched.c is just a copy of +bfq-sq-iosched.c, i.e, of the main source file of bfq for blk. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 5392 ++++++++++++++++++++++++++++++++++++++++++++++++ + 1 file changed, 5392 insertions(+) + create mode 100644 block/bfq-mq-iosched.c + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +new file mode 100644 +index 000000000000..30d019fc67e0 +--- /dev/null ++++ b/block/bfq-mq-iosched.c +@@ -0,0 +1,5392 @@ ++/* ++ * Budget Fair Queueing (BFQ) I/O scheduler. ++ * ++ * Based on ideas and code from CFQ: ++ * Copyright (C) 2003 Jens Axboe ++ * ++ * Copyright (C) 2008 Fabio Checconi ++ * Paolo Valente ++ * ++ * Copyright (C) 2015 Paolo Valente ++ * ++ * Copyright (C) 2017 Paolo Valente ++ * ++ * Licensed under the GPL-2 as detailed in the accompanying COPYING.BFQ ++ * file. ++ * ++ * BFQ is a proportional-share I/O scheduler, with some extra ++ * low-latency capabilities. BFQ also supports full hierarchical ++ * scheduling through cgroups. Next paragraphs provide an introduction ++ * on BFQ inner workings. Details on BFQ benefits and usage can be ++ * found in Documentation/block/bfq-iosched.txt. ++ * ++ * BFQ is a proportional-share storage-I/O scheduling algorithm based ++ * on the slice-by-slice service scheme of CFQ. But BFQ assigns ++ * budgets, measured in number of sectors, to processes instead of ++ * time slices. The device is not granted to the in-service process ++ * for a given time slice, but until it has exhausted its assigned ++ * budget. This change from the time to the service domain enables BFQ ++ * to distribute the device throughput among processes as desired, ++ * without any distortion due to throughput fluctuations, or to device ++ * internal queueing. BFQ uses an ad hoc internal scheduler, called ++ * B-WF2Q+, to schedule processes according to their budgets. More ++ * precisely, BFQ schedules queues associated with processes. Thanks to ++ * the accurate policy of B-WF2Q+, BFQ can afford to assign high ++ * budgets to I/O-bound processes issuing sequential requests (to ++ * boost the throughput), and yet guarantee a low latency to ++ * interactive and soft real-time applications. ++ * ++ * NOTE: if the main or only goal, with a given device, is to achieve ++ * the maximum-possible throughput at all times, then do switch off ++ * all low-latency heuristics for that device, by setting low_latency ++ * to 0. ++ * ++ * BFQ is described in [1], where also a reference to the initial, more ++ * theoretical paper on BFQ can be found. The interested reader can find ++ * in the latter paper full details on the main algorithm, as well as ++ * formulas of the guarantees and formal proofs of all the properties. ++ * With respect to the version of BFQ presented in these papers, this ++ * implementation adds a few more heuristics, such as the one that ++ * guarantees a low latency to soft real-time applications, and a ++ * hierarchical extension based on H-WF2Q+. ++ * ++ * B-WF2Q+ is based on WF2Q+, that is described in [2], together with ++ * H-WF2Q+, while the augmented tree used to implement B-WF2Q+ with O(log N) ++ * complexity derives from the one introduced with EEVDF in [3]. ++ * ++ * [1] P. Valente, A. Avanzini, "Evolution of the BFQ Storage I/O ++ * Scheduler", Proceedings of the First Workshop on Mobile System ++ * Technologies (MST-2015), May 2015. ++ * http://algogroup.unimore.it/people/paolo/disk_sched/mst-2015.pdf ++ * ++ * http://algogroup.unimo.it/people/paolo/disk_sched/bf1-v1-suite-results.pdf ++ * ++ * [2] Jon C.R. Bennett and H. Zhang, ``Hierarchical Packet Fair Queueing ++ * Algorithms,'' IEEE/ACM Transactions on Networking, 5(5):675-689, ++ * Oct 1997. ++ * ++ * http://www.cs.cmu.edu/~hzhang/papers/TON-97-Oct.ps.gz ++ * ++ * [3] I. Stoica and H. Abdel-Wahab, ``Earliest Eligible Virtual Deadline ++ * First: A Flexible and Accurate Mechanism for Proportional Share ++ * Resource Allocation,'' technical report. ++ * ++ * http://www.cs.berkeley.edu/~istoica/papers/eevdf-tr-95.pdf ++ */ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include "blk.h" ++#include "bfq.h" ++ ++/* Expiration time of sync (0) and async (1) requests, in ns. */ ++static const u64 bfq_fifo_expire[2] = { NSEC_PER_SEC / 4, NSEC_PER_SEC / 8 }; ++ ++/* Maximum backwards seek, in KiB. */ ++static const int bfq_back_max = (16 * 1024); ++ ++/* Penalty of a backwards seek, in number of sectors. */ ++static const int bfq_back_penalty = 2; ++ ++/* Idling period duration, in ns. */ ++static u32 bfq_slice_idle = (NSEC_PER_SEC / 125); ++ ++/* Minimum number of assigned budgets for which stats are safe to compute. */ ++static const int bfq_stats_min_budgets = 194; ++ ++/* Default maximum budget values, in sectors and number of requests. */ ++static const int bfq_default_max_budget = (16 * 1024); ++ ++/* ++ * Async to sync throughput distribution is controlled as follows: ++ * when an async request is served, the entity is charged the number ++ * of sectors of the request, multiplied by the factor below ++ */ ++static const int bfq_async_charge_factor = 10; ++ ++/* Default timeout values, in jiffies, approximating CFQ defaults. */ ++static const int bfq_timeout = (HZ / 8); ++ ++static struct kmem_cache *bfq_pool; ++ ++/* Below this threshold (in ns), we consider thinktime immediate. */ ++#define BFQ_MIN_TT (2 * NSEC_PER_MSEC) ++ ++/* hw_tag detection: parallel requests threshold and min samples needed. */ ++#define BFQ_HW_QUEUE_THRESHOLD 4 ++#define BFQ_HW_QUEUE_SAMPLES 32 ++ ++#define BFQQ_SEEK_THR (sector_t)(8 * 100) ++#define BFQQ_SECT_THR_NONROT (sector_t)(2 * 32) ++#define BFQQ_CLOSE_THR (sector_t)(8 * 1024) ++#define BFQQ_SEEKY(bfqq) (hweight32(bfqq->seek_history) > 32/8) ++ ++/* Min number of samples required to perform peak-rate update */ ++#define BFQ_RATE_MIN_SAMPLES 32 ++/* Min observation time interval required to perform a peak-rate update (ns) */ ++#define BFQ_RATE_MIN_INTERVAL (300*NSEC_PER_MSEC) ++/* Target observation time interval for a peak-rate update (ns) */ ++#define BFQ_RATE_REF_INTERVAL NSEC_PER_SEC ++ ++/* Shift used for peak rate fixed precision calculations. */ ++#define BFQ_RATE_SHIFT 16 ++ ++/* ++ * By default, BFQ computes the duration of the weight raising for ++ * interactive applications automatically, using the following formula: ++ * duration = (R / r) * T, where r is the peak rate of the device, and ++ * R and T are two reference parameters. ++ * In particular, R is the peak rate of the reference device (see below), ++ * and T is a reference time: given the systems that are likely to be ++ * installed on the reference device according to its speed class, T is ++ * about the maximum time needed, under BFQ and while reading two files in ++ * parallel, to load typical large applications on these systems. ++ * In practice, the slower/faster the device at hand is, the more/less it ++ * takes to load applications with respect to the reference device. ++ * Accordingly, the longer/shorter BFQ grants weight raising to interactive ++ * applications. ++ * ++ * BFQ uses four different reference pairs (R, T), depending on: ++ * . whether the device is rotational or non-rotational; ++ * . whether the device is slow, such as old or portable HDDs, as well as ++ * SD cards, or fast, such as newer HDDs and SSDs. ++ * ++ * The device's speed class is dynamically (re)detected in ++ * bfq_update_peak_rate() every time the estimated peak rate is updated. ++ * ++ * In the following definitions, R_slow[0]/R_fast[0] and ++ * T_slow[0]/T_fast[0] are the reference values for a slow/fast ++ * rotational device, whereas R_slow[1]/R_fast[1] and ++ * T_slow[1]/T_fast[1] are the reference values for a slow/fast ++ * non-rotational device. Finally, device_speed_thresh are the ++ * thresholds used to switch between speed classes. The reference ++ * rates are not the actual peak rates of the devices used as a ++ * reference, but slightly lower values. The reason for using these ++ * slightly lower values is that the peak-rate estimator tends to ++ * yield slightly lower values than the actual peak rate (it can yield ++ * the actual peak rate only if there is only one process doing I/O, ++ * and the process does sequential I/O). ++ * ++ * Both the reference peak rates and the thresholds are measured in ++ * sectors/usec, left-shifted by BFQ_RATE_SHIFT. ++ */ ++static int R_slow[2] = {1000, 10700}; ++static int R_fast[2] = {14000, 33000}; ++/* ++ * To improve readability, a conversion function is used to initialize the ++ * following arrays, which entails that they can be initialized only in a ++ * function. ++ */ ++static int T_slow[2]; ++static int T_fast[2]; ++static int device_speed_thresh[2]; ++ ++#define BFQ_SERVICE_TREE_INIT ((struct bfq_service_tree) \ ++ { RB_ROOT, RB_ROOT, NULL, NULL, 0, 0 }) ++ ++#define RQ_BIC(rq) ((struct bfq_io_cq *) (rq)->elv.priv[0]) ++#define RQ_BFQQ(rq) ((rq)->elv.priv[1]) ++ ++static void bfq_schedule_dispatch(struct bfq_data *bfqd); ++ ++#include "bfq-ioc.c" ++#include "bfq-sched.c" ++#include "bfq-cgroup-included.c" ++ ++#define bfq_class_idle(bfqq) ((bfqq)->ioprio_class == IOPRIO_CLASS_IDLE) ++#define bfq_class_rt(bfqq) ((bfqq)->ioprio_class == IOPRIO_CLASS_RT) ++ ++#define bfq_sample_valid(samples) ((samples) > 80) ++ ++/* ++ * Scheduler run of queue, if there are requests pending and no one in the ++ * driver that will restart queueing. ++ */ ++static void bfq_schedule_dispatch(struct bfq_data *bfqd) ++{ ++ if (bfqd->queued != 0) { ++ bfq_log(bfqd, "schedule dispatch"); ++ kblockd_schedule_work(&bfqd->unplug_work); ++ } ++} ++ ++/* ++ * Lifted from AS - choose which of rq1 and rq2 that is best served now. ++ * We choose the request that is closesr to the head right now. Distance ++ * behind the head is penalized and only allowed to a certain extent. ++ */ ++static struct request *bfq_choose_req(struct bfq_data *bfqd, ++ struct request *rq1, ++ struct request *rq2, ++ sector_t last) ++{ ++ sector_t s1, s2, d1 = 0, d2 = 0; ++ unsigned long back_max; ++#define BFQ_RQ1_WRAP 0x01 /* request 1 wraps */ ++#define BFQ_RQ2_WRAP 0x02 /* request 2 wraps */ ++ unsigned int wrap = 0; /* bit mask: requests behind the disk head? */ ++ ++ if (!rq1 || rq1 == rq2) ++ return rq2; ++ if (!rq2) ++ return rq1; ++ ++ if (rq_is_sync(rq1) && !rq_is_sync(rq2)) ++ return rq1; ++ else if (rq_is_sync(rq2) && !rq_is_sync(rq1)) ++ return rq2; ++ if ((rq1->cmd_flags & REQ_META) && !(rq2->cmd_flags & REQ_META)) ++ return rq1; ++ else if ((rq2->cmd_flags & REQ_META) && !(rq1->cmd_flags & REQ_META)) ++ return rq2; ++ ++ s1 = blk_rq_pos(rq1); ++ s2 = blk_rq_pos(rq2); ++ ++ /* ++ * By definition, 1KiB is 2 sectors. ++ */ ++ back_max = bfqd->bfq_back_max * 2; ++ ++ /* ++ * Strict one way elevator _except_ in the case where we allow ++ * short backward seeks which are biased as twice the cost of a ++ * similar forward seek. ++ */ ++ if (s1 >= last) ++ d1 = s1 - last; ++ else if (s1 + back_max >= last) ++ d1 = (last - s1) * bfqd->bfq_back_penalty; ++ else ++ wrap |= BFQ_RQ1_WRAP; ++ ++ if (s2 >= last) ++ d2 = s2 - last; ++ else if (s2 + back_max >= last) ++ d2 = (last - s2) * bfqd->bfq_back_penalty; ++ else ++ wrap |= BFQ_RQ2_WRAP; ++ ++ /* Found required data */ ++ ++ /* ++ * By doing switch() on the bit mask "wrap" we avoid having to ++ * check two variables for all permutations: --> faster! ++ */ ++ switch (wrap) { ++ case 0: /* common case for CFQ: rq1 and rq2 not wrapped */ ++ if (d1 < d2) ++ return rq1; ++ else if (d2 < d1) ++ return rq2; ++ ++ if (s1 >= s2) ++ return rq1; ++ else ++ return rq2; ++ ++ case BFQ_RQ2_WRAP: ++ return rq1; ++ case BFQ_RQ1_WRAP: ++ return rq2; ++ case (BFQ_RQ1_WRAP|BFQ_RQ2_WRAP): /* both rqs wrapped */ ++ default: ++ /* ++ * Since both rqs are wrapped, ++ * start with the one that's further behind head ++ * (--> only *one* back seek required), ++ * since back seek takes more time than forward. ++ */ ++ if (s1 <= s2) ++ return rq1; ++ else ++ return rq2; ++ } ++} ++ ++static struct bfq_queue * ++bfq_rq_pos_tree_lookup(struct bfq_data *bfqd, struct rb_root *root, ++ sector_t sector, struct rb_node **ret_parent, ++ struct rb_node ***rb_link) ++{ ++ struct rb_node **p, *parent; ++ struct bfq_queue *bfqq = NULL; ++ ++ parent = NULL; ++ p = &root->rb_node; ++ while (*p) { ++ struct rb_node **n; ++ ++ parent = *p; ++ bfqq = rb_entry(parent, struct bfq_queue, pos_node); ++ ++ /* ++ * Sort strictly based on sector. Smallest to the left, ++ * largest to the right. ++ */ ++ if (sector > blk_rq_pos(bfqq->next_rq)) ++ n = &(*p)->rb_right; ++ else if (sector < blk_rq_pos(bfqq->next_rq)) ++ n = &(*p)->rb_left; ++ else ++ break; ++ p = n; ++ bfqq = NULL; ++ } ++ ++ *ret_parent = parent; ++ if (rb_link) ++ *rb_link = p; ++ ++ bfq_log(bfqd, "rq_pos_tree_lookup %llu: returning %d", ++ (unsigned long long) sector, ++ bfqq ? bfqq->pid : 0); ++ ++ return bfqq; ++} ++ ++static void bfq_pos_tree_add_move(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ struct rb_node **p, *parent; ++ struct bfq_queue *__bfqq; ++ ++ if (bfqq->pos_root) { ++ rb_erase(&bfqq->pos_node, bfqq->pos_root); ++ bfqq->pos_root = NULL; ++ } ++ ++ if (bfq_class_idle(bfqq)) ++ return; ++ if (!bfqq->next_rq) ++ return; ++ ++ bfqq->pos_root = &bfq_bfqq_to_bfqg(bfqq)->rq_pos_tree; ++ __bfqq = bfq_rq_pos_tree_lookup(bfqd, bfqq->pos_root, ++ blk_rq_pos(bfqq->next_rq), &parent, &p); ++ if (!__bfqq) { ++ rb_link_node(&bfqq->pos_node, parent, p); ++ rb_insert_color(&bfqq->pos_node, bfqq->pos_root); ++ } else ++ bfqq->pos_root = NULL; ++} ++ ++/* ++ * Tell whether there are active queues or groups with differentiated weights. ++ */ ++static bool bfq_differentiated_weights(struct bfq_data *bfqd) ++{ ++ /* ++ * For weights to differ, at least one of the trees must contain ++ * at least two nodes. ++ */ ++ return (!RB_EMPTY_ROOT(&bfqd->queue_weights_tree) && ++ (bfqd->queue_weights_tree.rb_node->rb_left || ++ bfqd->queue_weights_tree.rb_node->rb_right) ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ ) || ++ (!RB_EMPTY_ROOT(&bfqd->group_weights_tree) && ++ (bfqd->group_weights_tree.rb_node->rb_left || ++ bfqd->group_weights_tree.rb_node->rb_right) ++#endif ++ ); ++} ++ ++/* ++ * The following function returns true if every queue must receive the ++ * same share of the throughput (this condition is used when deciding ++ * whether idling may be disabled, see the comments in the function ++ * bfq_bfqq_may_idle()). ++ * ++ * Such a scenario occurs when: ++ * 1) all active queues have the same weight, ++ * 2) all active groups at the same level in the groups tree have the same ++ * weight, ++ * 3) all active groups at the same level in the groups tree have the same ++ * number of children. ++ * ++ * Unfortunately, keeping the necessary state for evaluating exactly the ++ * above symmetry conditions would be quite complex and time-consuming. ++ * Therefore this function evaluates, instead, the following stronger ++ * sub-conditions, for which it is much easier to maintain the needed ++ * state: ++ * 1) all active queues have the same weight, ++ * 2) all active groups have the same weight, ++ * 3) all active groups have at most one active child each. ++ * In particular, the last two conditions are always true if hierarchical ++ * support and the cgroups interface are not enabled, thus no state needs ++ * to be maintained in this case. ++ */ ++static bool bfq_symmetric_scenario(struct bfq_data *bfqd) ++{ ++ return !bfq_differentiated_weights(bfqd); ++} ++ ++/* ++ * If the weight-counter tree passed as input contains no counter for ++ * the weight of the input entity, then add that counter; otherwise just ++ * increment the existing counter. ++ * ++ * Note that weight-counter trees contain few nodes in mostly symmetric ++ * scenarios. For example, if all queues have the same weight, then the ++ * weight-counter tree for the queues may contain at most one node. ++ * This holds even if low_latency is on, because weight-raised queues ++ * are not inserted in the tree. ++ * In most scenarios, the rate at which nodes are created/destroyed ++ * should be low too. ++ */ ++static void bfq_weights_tree_add(struct bfq_data *bfqd, ++ struct bfq_entity *entity, ++ struct rb_root *root) ++{ ++ struct rb_node **new = &(root->rb_node), *parent = NULL; ++ ++ /* ++ * Do not insert if the entity is already associated with a ++ * counter, which happens if: ++ * 1) the entity is associated with a queue, ++ * 2) a request arrival has caused the queue to become both ++ * non-weight-raised, and hence change its weight, and ++ * backlogged; in this respect, each of the two events ++ * causes an invocation of this function, ++ * 3) this is the invocation of this function caused by the ++ * second event. This second invocation is actually useless, ++ * and we handle this fact by exiting immediately. More ++ * efficient or clearer solutions might possibly be adopted. ++ */ ++ if (entity->weight_counter) ++ return; ++ ++ while (*new) { ++ struct bfq_weight_counter *__counter = container_of(*new, ++ struct bfq_weight_counter, ++ weights_node); ++ parent = *new; ++ ++ if (entity->weight == __counter->weight) { ++ entity->weight_counter = __counter; ++ goto inc_counter; ++ } ++ if (entity->weight < __counter->weight) ++ new = &((*new)->rb_left); ++ else ++ new = &((*new)->rb_right); ++ } ++ ++ entity->weight_counter = kzalloc(sizeof(struct bfq_weight_counter), ++ GFP_ATOMIC); ++ ++ /* ++ * In the unlucky event of an allocation failure, we just ++ * exit. This will cause the weight of entity to not be ++ * considered in bfq_differentiated_weights, which, in its ++ * turn, causes the scenario to be deemed wrongly symmetric in ++ * case entity's weight would have been the only weight making ++ * the scenario asymmetric. On the bright side, no unbalance ++ * will however occur when entity becomes inactive again (the ++ * invocation of this function is triggered by an activation ++ * of entity). In fact, bfq_weights_tree_remove does nothing ++ * if !entity->weight_counter. ++ */ ++ if (unlikely(!entity->weight_counter)) ++ return; ++ ++ entity->weight_counter->weight = entity->weight; ++ rb_link_node(&entity->weight_counter->weights_node, parent, new); ++ rb_insert_color(&entity->weight_counter->weights_node, root); ++ ++inc_counter: ++ entity->weight_counter->num_active++; ++} ++ ++/* ++ * Decrement the weight counter associated with the entity, and, if the ++ * counter reaches 0, remove the counter from the tree. ++ * See the comments to the function bfq_weights_tree_add() for considerations ++ * about overhead. ++ */ ++static void bfq_weights_tree_remove(struct bfq_data *bfqd, ++ struct bfq_entity *entity, ++ struct rb_root *root) ++{ ++ if (!entity->weight_counter) ++ return; ++ ++ BUG_ON(RB_EMPTY_ROOT(root)); ++ BUG_ON(entity->weight_counter->weight != entity->weight); ++ ++ BUG_ON(!entity->weight_counter->num_active); ++ entity->weight_counter->num_active--; ++ if (entity->weight_counter->num_active > 0) ++ goto reset_entity_pointer; ++ ++ rb_erase(&entity->weight_counter->weights_node, root); ++ kfree(entity->weight_counter); ++ ++reset_entity_pointer: ++ entity->weight_counter = NULL; ++} ++ ++/* ++ * Return expired entry, or NULL to just start from scratch in rbtree. ++ */ ++static struct request *bfq_check_fifo(struct bfq_queue *bfqq, ++ struct request *last) ++{ ++ struct request *rq; ++ ++ if (bfq_bfqq_fifo_expire(bfqq)) ++ return NULL; ++ ++ bfq_mark_bfqq_fifo_expire(bfqq); ++ ++ rq = rq_entry_fifo(bfqq->fifo.next); ++ ++ if (rq == last || ktime_get_ns() < rq->fifo_time) ++ return NULL; ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "check_fifo: returned %p", rq); ++ BUG_ON(RB_EMPTY_NODE(&rq->rb_node)); ++ return rq; ++} ++ ++static struct request *bfq_find_next_rq(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ struct request *last) ++{ ++ struct rb_node *rbnext = rb_next(&last->rb_node); ++ struct rb_node *rbprev = rb_prev(&last->rb_node); ++ struct request *next, *prev = NULL; ++ ++ BUG_ON(list_empty(&bfqq->fifo)); ++ ++ /* Follow expired path, else get first next available. */ ++ next = bfq_check_fifo(bfqq, last); ++ if (next) { ++ BUG_ON(next == last); ++ return next; ++ } ++ ++ BUG_ON(RB_EMPTY_NODE(&last->rb_node)); ++ ++ if (rbprev) ++ prev = rb_entry_rq(rbprev); ++ ++ if (rbnext) ++ next = rb_entry_rq(rbnext); ++ else { ++ rbnext = rb_first(&bfqq->sort_list); ++ if (rbnext && rbnext != &last->rb_node) ++ next = rb_entry_rq(rbnext); ++ } ++ ++ return bfq_choose_req(bfqd, next, prev, blk_rq_pos(last)); ++} ++ ++/* see the definition of bfq_async_charge_factor for details */ ++static unsigned long bfq_serv_to_charge(struct request *rq, ++ struct bfq_queue *bfqq) ++{ ++ if (bfq_bfqq_sync(bfqq) || bfqq->wr_coeff > 1) ++ return blk_rq_sectors(rq); ++ ++ /* ++ * If there are no weight-raised queues, then amplify service ++ * by just the async charge factor; otherwise amplify service ++ * by twice the async charge factor, to further reduce latency ++ * for weight-raised queues. ++ */ ++ if (bfqq->bfqd->wr_busy_queues == 0) ++ return blk_rq_sectors(rq) * bfq_async_charge_factor; ++ ++ return blk_rq_sectors(rq) * 2 * bfq_async_charge_factor; ++} ++ ++/** ++ * bfq_updated_next_req - update the queue after a new next_rq selection. ++ * @bfqd: the device data the queue belongs to. ++ * @bfqq: the queue to update. ++ * ++ * If the first request of a queue changes we make sure that the queue ++ * has enough budget to serve at least its first request (if the ++ * request has grown). We do this because if the queue has not enough ++ * budget for its first request, it has to go through two dispatch ++ * rounds to actually get it dispatched. ++ */ ++static void bfq_updated_next_req(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ struct bfq_service_tree *st = bfq_entity_service_tree(entity); ++ struct request *next_rq = bfqq->next_rq; ++ unsigned long new_budget; ++ ++ if (!next_rq) ++ return; ++ ++ if (bfqq == bfqd->in_service_queue) ++ /* ++ * In order not to break guarantees, budgets cannot be ++ * changed after an entity has been selected. ++ */ ++ return; ++ ++ BUG_ON(entity->tree != &st->active); ++ BUG_ON(entity == entity->sched_data->in_service_entity); ++ ++ new_budget = max_t(unsigned long, bfqq->max_budget, ++ bfq_serv_to_charge(next_rq, bfqq)); ++ if (entity->budget != new_budget) { ++ entity->budget = new_budget; ++ bfq_log_bfqq(bfqd, bfqq, "updated next rq: new budget %lu", ++ new_budget); ++ bfq_requeue_bfqq(bfqd, bfqq); ++ } ++} ++ ++static unsigned int bfq_wr_duration(struct bfq_data *bfqd) ++{ ++ u64 dur; ++ ++ if (bfqd->bfq_wr_max_time > 0) ++ return bfqd->bfq_wr_max_time; ++ ++ dur = bfqd->RT_prod; ++ do_div(dur, bfqd->peak_rate); ++ ++ /* ++ * Limit duration between 3 and 13 seconds. Tests show that ++ * higher values than 13 seconds often yield the opposite of ++ * the desired result, i.e., worsen responsiveness by letting ++ * non-interactive and non-soft-real-time applications ++ * preserve weight raising for a too long time interval. ++ * ++ * On the other end, lower values than 3 seconds make it ++ * difficult for most interactive tasks to complete their jobs ++ * before weight-raising finishes. ++ */ ++ if (dur > msecs_to_jiffies(13000)) ++ dur = msecs_to_jiffies(13000); ++ else if (dur < msecs_to_jiffies(3000)) ++ dur = msecs_to_jiffies(3000); ++ ++ return dur; ++} ++ ++static void ++bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, ++ struct bfq_io_cq *bic, bool bfq_already_existing) ++{ ++ unsigned int old_wr_coeff; ++ bool busy = bfq_already_existing && bfq_bfqq_busy(bfqq); ++ ++ if (bic->saved_has_short_ttime) ++ bfq_mark_bfqq_has_short_ttime(bfqq); ++ else ++ bfq_clear_bfqq_has_short_ttime(bfqq); ++ ++ if (bic->saved_IO_bound) ++ bfq_mark_bfqq_IO_bound(bfqq); ++ else ++ bfq_clear_bfqq_IO_bound(bfqq); ++ ++ if (unlikely(busy)) ++ old_wr_coeff = bfqq->wr_coeff; ++ ++ bfqq->wr_coeff = bic->saved_wr_coeff; ++ bfqq->wr_start_at_switch_to_srt = bic->saved_wr_start_at_switch_to_srt; ++ BUG_ON(time_is_after_jiffies(bfqq->wr_start_at_switch_to_srt)); ++ bfqq->last_wr_start_finish = bic->saved_last_wr_start_finish; ++ bfqq->wr_cur_max_time = bic->saved_wr_cur_max_time; ++ BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); ++ ++ if (bfqq->wr_coeff > 1 && (bfq_bfqq_in_large_burst(bfqq) || ++ time_is_before_jiffies(bfqq->last_wr_start_finish + ++ bfqq->wr_cur_max_time))) { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "resume state: switching off wr (%lu + %lu < %lu)", ++ bfqq->last_wr_start_finish, bfqq->wr_cur_max_time, ++ jiffies); ++ ++ bfqq->wr_coeff = 1; ++ } ++ ++ /* make sure weight will be updated, however we got here */ ++ bfqq->entity.prio_changed = 1; ++ ++ if (likely(!busy)) ++ return; ++ ++ if (old_wr_coeff == 1 && bfqq->wr_coeff > 1) { ++ bfqd->wr_busy_queues++; ++ BUG_ON(bfqd->wr_busy_queues > bfqd->busy_queues); ++ } else if (old_wr_coeff > 1 && bfqq->wr_coeff == 1) { ++ bfqd->wr_busy_queues--; ++ BUG_ON(bfqd->wr_busy_queues < 0); ++ } ++} ++ ++static int bfqq_process_refs(struct bfq_queue *bfqq) ++{ ++ int process_refs, io_refs; ++ ++ lockdep_assert_held(bfqq->bfqd->queue->queue_lock); ++ ++ io_refs = bfqq->allocated[READ] + bfqq->allocated[WRITE]; ++ process_refs = bfqq->ref - io_refs - bfqq->entity.on_st; ++ BUG_ON(process_refs < 0); ++ return process_refs; ++} ++ ++/* Empty burst list and add just bfqq (see comments to bfq_handle_burst) */ ++static void bfq_reset_burst_list(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ struct bfq_queue *item; ++ struct hlist_node *n; ++ ++ hlist_for_each_entry_safe(item, n, &bfqd->burst_list, burst_list_node) ++ hlist_del_init(&item->burst_list_node); ++ hlist_add_head(&bfqq->burst_list_node, &bfqd->burst_list); ++ bfqd->burst_size = 1; ++ bfqd->burst_parent_entity = bfqq->entity.parent; ++} ++ ++/* Add bfqq to the list of queues in current burst (see bfq_handle_burst) */ ++static void bfq_add_to_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ /* Increment burst size to take into account also bfqq */ ++ bfqd->burst_size++; ++ ++ bfq_log_bfqq(bfqd, bfqq, "add_to_burst %d", bfqd->burst_size); ++ ++ BUG_ON(bfqd->burst_size > bfqd->bfq_large_burst_thresh); ++ ++ if (bfqd->burst_size == bfqd->bfq_large_burst_thresh) { ++ struct bfq_queue *pos, *bfqq_item; ++ struct hlist_node *n; ++ ++ /* ++ * Enough queues have been activated shortly after each ++ * other to consider this burst as large. ++ */ ++ bfqd->large_burst = true; ++ bfq_log_bfqq(bfqd, bfqq, "add_to_burst: large burst started"); ++ ++ /* ++ * We can now mark all queues in the burst list as ++ * belonging to a large burst. ++ */ ++ hlist_for_each_entry(bfqq_item, &bfqd->burst_list, ++ burst_list_node) { ++ bfq_mark_bfqq_in_large_burst(bfqq_item); ++ bfq_log_bfqq(bfqd, bfqq_item, "marked in large burst"); ++ } ++ bfq_mark_bfqq_in_large_burst(bfqq); ++ bfq_log_bfqq(bfqd, bfqq, "marked in large burst"); ++ ++ /* ++ * From now on, and until the current burst finishes, any ++ * new queue being activated shortly after the last queue ++ * was inserted in the burst can be immediately marked as ++ * belonging to a large burst. So the burst list is not ++ * needed any more. Remove it. ++ */ ++ hlist_for_each_entry_safe(pos, n, &bfqd->burst_list, ++ burst_list_node) ++ hlist_del_init(&pos->burst_list_node); ++ } else /* ++ * Burst not yet large: add bfqq to the burst list. Do ++ * not increment the ref counter for bfqq, because bfqq ++ * is removed from the burst list before freeing bfqq ++ * in put_queue. ++ */ ++ hlist_add_head(&bfqq->burst_list_node, &bfqd->burst_list); ++} ++ ++/* ++ * If many queues belonging to the same group happen to be created ++ * shortly after each other, then the processes associated with these ++ * queues have typically a common goal. In particular, bursts of queue ++ * creations are usually caused by services or applications that spawn ++ * many parallel threads/processes. Examples are systemd during boot, ++ * or git grep. To help these processes get their job done as soon as ++ * possible, it is usually better to not grant either weight-raising ++ * or device idling to their queues. ++ * ++ * In this comment we describe, firstly, the reasons why this fact ++ * holds, and, secondly, the next function, which implements the main ++ * steps needed to properly mark these queues so that they can then be ++ * treated in a different way. ++ * ++ * The above services or applications benefit mostly from a high ++ * throughput: the quicker the requests of the activated queues are ++ * cumulatively served, the sooner the target job of these queues gets ++ * completed. As a consequence, weight-raising any of these queues, ++ * which also implies idling the device for it, is almost always ++ * counterproductive. In most cases it just lowers throughput. ++ * ++ * On the other hand, a burst of queue creations may be caused also by ++ * the start of an application that does not consist of a lot of ++ * parallel I/O-bound threads. In fact, with a complex application, ++ * several short processes may need to be executed to start-up the ++ * application. In this respect, to start an application as quickly as ++ * possible, the best thing to do is in any case to privilege the I/O ++ * related to the application with respect to all other ++ * I/O. Therefore, the best strategy to start as quickly as possible ++ * an application that causes a burst of queue creations is to ++ * weight-raise all the queues created during the burst. This is the ++ * exact opposite of the best strategy for the other type of bursts. ++ * ++ * In the end, to take the best action for each of the two cases, the ++ * two types of bursts need to be distinguished. Fortunately, this ++ * seems relatively easy, by looking at the sizes of the bursts. In ++ * particular, we found a threshold such that only bursts with a ++ * larger size than that threshold are apparently caused by ++ * services or commands such as systemd or git grep. For brevity, ++ * hereafter we call just 'large' these bursts. BFQ *does not* ++ * weight-raise queues whose creation occurs in a large burst. In ++ * addition, for each of these queues BFQ performs or does not perform ++ * idling depending on which choice boosts the throughput more. The ++ * exact choice depends on the device and request pattern at ++ * hand. ++ * ++ * Unfortunately, false positives may occur while an interactive task ++ * is starting (e.g., an application is being started). The ++ * consequence is that the queues associated with the task do not ++ * enjoy weight raising as expected. Fortunately these false positives ++ * are very rare. They typically occur if some service happens to ++ * start doing I/O exactly when the interactive task starts. ++ * ++ * Turning back to the next function, it implements all the steps ++ * needed to detect the occurrence of a large burst and to properly ++ * mark all the queues belonging to it (so that they can then be ++ * treated in a different way). This goal is achieved by maintaining a ++ * "burst list" that holds, temporarily, the queues that belong to the ++ * burst in progress. The list is then used to mark these queues as ++ * belonging to a large burst if the burst does become large. The main ++ * steps are the following. ++ * ++ * . when the very first queue is created, the queue is inserted into the ++ * list (as it could be the first queue in a possible burst) ++ * ++ * . if the current burst has not yet become large, and a queue Q that does ++ * not yet belong to the burst is activated shortly after the last time ++ * at which a new queue entered the burst list, then the function appends ++ * Q to the burst list ++ * ++ * . if, as a consequence of the previous step, the burst size reaches ++ * the large-burst threshold, then ++ * ++ * . all the queues in the burst list are marked as belonging to a ++ * large burst ++ * ++ * . the burst list is deleted; in fact, the burst list already served ++ * its purpose (keeping temporarily track of the queues in a burst, ++ * so as to be able to mark them as belonging to a large burst in the ++ * previous sub-step), and now is not needed any more ++ * ++ * . the device enters a large-burst mode ++ * ++ * . if a queue Q that does not belong to the burst is created while ++ * the device is in large-burst mode and shortly after the last time ++ * at which a queue either entered the burst list or was marked as ++ * belonging to the current large burst, then Q is immediately marked ++ * as belonging to a large burst. ++ * ++ * . if a queue Q that does not belong to the burst is created a while ++ * later, i.e., not shortly after, than the last time at which a queue ++ * either entered the burst list or was marked as belonging to the ++ * current large burst, then the current burst is deemed as finished and: ++ * ++ * . the large-burst mode is reset if set ++ * ++ * . the burst list is emptied ++ * ++ * . Q is inserted in the burst list, as Q may be the first queue ++ * in a possible new burst (then the burst list contains just Q ++ * after this step). ++ */ ++static void bfq_handle_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ /* ++ * If bfqq is already in the burst list or is part of a large ++ * burst, or finally has just been split, then there is ++ * nothing else to do. ++ */ ++ if (!hlist_unhashed(&bfqq->burst_list_node) || ++ bfq_bfqq_in_large_burst(bfqq) || ++ time_is_after_eq_jiffies(bfqq->split_time + ++ msecs_to_jiffies(10))) ++ return; ++ ++ /* ++ * If bfqq's creation happens late enough, or bfqq belongs to ++ * a different group than the burst group, then the current ++ * burst is finished, and related data structures must be ++ * reset. ++ * ++ * In this respect, consider the special case where bfqq is ++ * the very first queue created after BFQ is selected for this ++ * device. In this case, last_ins_in_burst and ++ * burst_parent_entity are not yet significant when we get ++ * here. But it is easy to verify that, whether or not the ++ * following condition is true, bfqq will end up being ++ * inserted into the burst list. In particular the list will ++ * happen to contain only bfqq. And this is exactly what has ++ * to happen, as bfqq may be the first queue of the first ++ * burst. ++ */ ++ if (time_is_before_jiffies(bfqd->last_ins_in_burst + ++ bfqd->bfq_burst_interval) || ++ bfqq->entity.parent != bfqd->burst_parent_entity) { ++ bfqd->large_burst = false; ++ bfq_reset_burst_list(bfqd, bfqq); ++ bfq_log_bfqq(bfqd, bfqq, ++ "handle_burst: late activation or different group"); ++ goto end; ++ } ++ ++ /* ++ * If we get here, then bfqq is being activated shortly after the ++ * last queue. So, if the current burst is also large, we can mark ++ * bfqq as belonging to this large burst immediately. ++ */ ++ if (bfqd->large_burst) { ++ bfq_log_bfqq(bfqd, bfqq, "handle_burst: marked in burst"); ++ bfq_mark_bfqq_in_large_burst(bfqq); ++ goto end; ++ } ++ ++ /* ++ * If we get here, then a large-burst state has not yet been ++ * reached, but bfqq is being activated shortly after the last ++ * queue. Then we add bfqq to the burst. ++ */ ++ bfq_add_to_burst(bfqd, bfqq); ++end: ++ /* ++ * At this point, bfqq either has been added to the current ++ * burst or has caused the current burst to terminate and a ++ * possible new burst to start. In particular, in the second ++ * case, bfqq has become the first queue in the possible new ++ * burst. In both cases last_ins_in_burst needs to be moved ++ * forward. ++ */ ++ bfqd->last_ins_in_burst = jiffies; ++ ++} ++ ++static int bfq_bfqq_budget_left(struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ ++ return entity->budget - entity->service; ++} ++ ++/* ++ * If enough samples have been computed, return the current max budget ++ * stored in bfqd, which is dynamically updated according to the ++ * estimated disk peak rate; otherwise return the default max budget ++ */ ++static int bfq_max_budget(struct bfq_data *bfqd) ++{ ++ if (bfqd->budgets_assigned < bfq_stats_min_budgets) ++ return bfq_default_max_budget; ++ else ++ return bfqd->bfq_max_budget; ++} ++ ++/* ++ * Return min budget, which is a fraction of the current or default ++ * max budget (trying with 1/32) ++ */ ++static int bfq_min_budget(struct bfq_data *bfqd) ++{ ++ if (bfqd->budgets_assigned < bfq_stats_min_budgets) ++ return bfq_default_max_budget / 32; ++ else ++ return bfqd->bfq_max_budget / 32; ++} ++ ++static void bfq_bfqq_expire(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ bool compensate, ++ enum bfqq_expiration reason); ++ ++/* ++ * The next function, invoked after the input queue bfqq switches from ++ * idle to busy, updates the budget of bfqq. The function also tells ++ * whether the in-service queue should be expired, by returning ++ * true. The purpose of expiring the in-service queue is to give bfqq ++ * the chance to possibly preempt the in-service queue, and the reason ++ * for preempting the in-service queue is to achieve one of the two ++ * goals below. ++ * ++ * 1. Guarantee to bfqq its reserved bandwidth even if bfqq has ++ * expired because it has remained idle. In particular, bfqq may have ++ * expired for one of the following two reasons: ++ * ++ * - BFQ_BFQQ_NO_MORE_REQUEST bfqq did not enjoy any device idling and ++ * did not make it to issue a new request before its last request ++ * was served; ++ * ++ * - BFQ_BFQQ_TOO_IDLE bfqq did enjoy device idling, but did not issue ++ * a new request before the expiration of the idling-time. ++ * ++ * Even if bfqq has expired for one of the above reasons, the process ++ * associated with the queue may be however issuing requests greedily, ++ * and thus be sensitive to the bandwidth it receives (bfqq may have ++ * remained idle for other reasons: CPU high load, bfqq not enjoying ++ * idling, I/O throttling somewhere in the path from the process to ++ * the I/O scheduler, ...). But if, after every expiration for one of ++ * the above two reasons, bfqq has to wait for the service of at least ++ * one full budget of another queue before being served again, then ++ * bfqq is likely to get a much lower bandwidth or resource time than ++ * its reserved ones. To address this issue, two countermeasures need ++ * to be taken. ++ * ++ * First, the budget and the timestamps of bfqq need to be updated in ++ * a special way on bfqq reactivation: they need to be updated as if ++ * bfqq did not remain idle and did not expire. In fact, if they are ++ * computed as if bfqq expired and remained idle until reactivation, ++ * then the process associated with bfqq is treated as if, instead of ++ * being greedy, it stopped issuing requests when bfqq remained idle, ++ * and restarts issuing requests only on this reactivation. In other ++ * words, the scheduler does not help the process recover the "service ++ * hole" between bfqq expiration and reactivation. As a consequence, ++ * the process receives a lower bandwidth than its reserved one. In ++ * contrast, to recover this hole, the budget must be updated as if ++ * bfqq was not expired at all before this reactivation, i.e., it must ++ * be set to the value of the remaining budget when bfqq was ++ * expired. Along the same line, timestamps need to be assigned the ++ * value they had the last time bfqq was selected for service, i.e., ++ * before last expiration. Thus timestamps need to be back-shifted ++ * with respect to their normal computation (see [1] for more details ++ * on this tricky aspect). ++ * ++ * Secondly, to allow the process to recover the hole, the in-service ++ * queue must be expired too, to give bfqq the chance to preempt it ++ * immediately. In fact, if bfqq has to wait for a full budget of the ++ * in-service queue to be completed, then it may become impossible to ++ * let the process recover the hole, even if the back-shifted ++ * timestamps of bfqq are lower than those of the in-service queue. If ++ * this happens for most or all of the holes, then the process may not ++ * receive its reserved bandwidth. In this respect, it is worth noting ++ * that, being the service of outstanding requests unpreemptible, a ++ * little fraction of the holes may however be unrecoverable, thereby ++ * causing a little loss of bandwidth. ++ * ++ * The last important point is detecting whether bfqq does need this ++ * bandwidth recovery. In this respect, the next function deems the ++ * process associated with bfqq greedy, and thus allows it to recover ++ * the hole, if: 1) the process is waiting for the arrival of a new ++ * request (which implies that bfqq expired for one of the above two ++ * reasons), and 2) such a request has arrived soon. The first ++ * condition is controlled through the flag non_blocking_wait_rq, ++ * while the second through the flag arrived_in_time. If both ++ * conditions hold, then the function computes the budget in the ++ * above-described special way, and signals that the in-service queue ++ * should be expired. Timestamp back-shifting is done later in ++ * __bfq_activate_entity. ++ * ++ * 2. Reduce latency. Even if timestamps are not backshifted to let ++ * the process associated with bfqq recover a service hole, bfqq may ++ * however happen to have, after being (re)activated, a lower finish ++ * timestamp than the in-service queue. That is, the next budget of ++ * bfqq may have to be completed before the one of the in-service ++ * queue. If this is the case, then preempting the in-service queue ++ * allows this goal to be achieved, apart from the unpreemptible, ++ * outstanding requests mentioned above. ++ * ++ * Unfortunately, regardless of which of the above two goals one wants ++ * to achieve, service trees need first to be updated to know whether ++ * the in-service queue must be preempted. To have service trees ++ * correctly updated, the in-service queue must be expired and ++ * rescheduled, and bfqq must be scheduled too. This is one of the ++ * most costly operations (in future versions, the scheduling ++ * mechanism may be re-designed in such a way to make it possible to ++ * know whether preemption is needed without needing to update service ++ * trees). In addition, queue preemptions almost always cause random ++ * I/O, and thus loss of throughput. Because of these facts, the next ++ * function adopts the following simple scheme to avoid both costly ++ * operations and too frequent preemptions: it requests the expiration ++ * of the in-service queue (unconditionally) only for queues that need ++ * to recover a hole, or that either are weight-raised or deserve to ++ * be weight-raised. ++ */ ++static bool bfq_bfqq_update_budg_for_activation(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ bool arrived_in_time, ++ bool wr_or_deserves_wr) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ ++ if (bfq_bfqq_non_blocking_wait_rq(bfqq) && arrived_in_time) { ++ /* ++ * We do not clear the flag non_blocking_wait_rq here, as ++ * the latter is used in bfq_activate_bfqq to signal ++ * that timestamps need to be back-shifted (and is ++ * cleared right after). ++ */ ++ ++ /* ++ * In next assignment we rely on that either ++ * entity->service or entity->budget are not updated ++ * on expiration if bfqq is empty (see ++ * __bfq_bfqq_recalc_budget). Thus both quantities ++ * remain unchanged after such an expiration, and the ++ * following statement therefore assigns to ++ * entity->budget the remaining budget on such an ++ * expiration. For clarity, entity->service is not ++ * updated on expiration in any case, and, in normal ++ * operation, is reset only when bfqq is selected for ++ * service (see bfq_get_next_queue). ++ */ ++ BUG_ON(bfqq->max_budget < 0); ++ entity->budget = min_t(unsigned long, ++ bfq_bfqq_budget_left(bfqq), ++ bfqq->max_budget); ++ ++ BUG_ON(entity->budget < 0); ++ return true; ++ } ++ ++ BUG_ON(bfqq->max_budget < 0); ++ entity->budget = max_t(unsigned long, bfqq->max_budget, ++ bfq_serv_to_charge(bfqq->next_rq, bfqq)); ++ BUG_ON(entity->budget < 0); ++ ++ bfq_clear_bfqq_non_blocking_wait_rq(bfqq); ++ return wr_or_deserves_wr; ++} ++ ++static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ unsigned int old_wr_coeff, ++ bool wr_or_deserves_wr, ++ bool interactive, ++ bool in_burst, ++ bool soft_rt) ++{ ++ if (old_wr_coeff == 1 && wr_or_deserves_wr) { ++ /* start a weight-raising period */ ++ if (interactive) { ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ } else { ++ bfqq->wr_start_at_switch_to_srt = jiffies; ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff * ++ BFQ_SOFTRT_WEIGHT_FACTOR; ++ bfqq->wr_cur_max_time = ++ bfqd->bfq_wr_rt_max_time; ++ } ++ /* ++ * If needed, further reduce budget to make sure it is ++ * close to bfqq's backlog, so as to reduce the ++ * scheduling-error component due to a too large ++ * budget. Do not care about throughput consequences, ++ * but only about latency. Finally, do not assign a ++ * too small budget either, to avoid increasing ++ * latency by causing too frequent expirations. ++ */ ++ bfqq->entity.budget = min_t(unsigned long, ++ bfqq->entity.budget, ++ 2 * bfq_min_budget(bfqd)); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "wrais starting at %lu, rais_max_time %u", ++ jiffies, ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ } else if (old_wr_coeff > 1) { ++ if (interactive) { /* update wr coeff and duration */ ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ } else if (in_burst) { ++ bfqq->wr_coeff = 1; ++ bfq_log_bfqq(bfqd, bfqq, ++ "wrais ending at %lu, rais_max_time %u", ++ jiffies, ++ jiffies_to_msecs(bfqq-> ++ wr_cur_max_time)); ++ } else if (soft_rt) { ++ /* ++ * The application is now or still meeting the ++ * requirements for being deemed soft rt. We ++ * can then correctly and safely (re)charge ++ * the weight-raising duration for the ++ * application with the weight-raising ++ * duration for soft rt applications. ++ * ++ * In particular, doing this recharge now, i.e., ++ * before the weight-raising period for the ++ * application finishes, reduces the probability ++ * of the following negative scenario: ++ * 1) the weight of a soft rt application is ++ * raised at startup (as for any newly ++ * created application), ++ * 2) since the application is not interactive, ++ * at a certain time weight-raising is ++ * stopped for the application, ++ * 3) at that time the application happens to ++ * still have pending requests, and hence ++ * is destined to not have a chance to be ++ * deemed soft rt before these requests are ++ * completed (see the comments to the ++ * function bfq_bfqq_softrt_next_start() ++ * for details on soft rt detection), ++ * 4) these pending requests experience a high ++ * latency because the application is not ++ * weight-raised while they are pending. ++ */ ++ if (bfqq->wr_cur_max_time != ++ bfqd->bfq_wr_rt_max_time) { ++ bfqq->wr_start_at_switch_to_srt = ++ bfqq->last_wr_start_finish; ++ BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); ++ ++ bfqq->wr_cur_max_time = ++ bfqd->bfq_wr_rt_max_time; ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff * ++ BFQ_SOFTRT_WEIGHT_FACTOR; ++ bfq_log_bfqq(bfqd, bfqq, ++ "switching to soft_rt wr"); ++ } else ++ bfq_log_bfqq(bfqd, bfqq, ++ "moving forward soft_rt wr duration"); ++ bfqq->last_wr_start_finish = jiffies; ++ } ++ } ++} ++ ++static bool bfq_bfqq_idle_for_long_time(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ return bfqq->dispatched == 0 && ++ time_is_before_jiffies( ++ bfqq->budget_timeout + ++ bfqd->bfq_wr_min_idle_time); ++} ++ ++static void bfq_bfqq_handle_idle_busy_switch(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ int old_wr_coeff, ++ struct request *rq, ++ bool *interactive) ++{ ++ bool soft_rt, in_burst, wr_or_deserves_wr, ++ bfqq_wants_to_preempt, ++ idle_for_long_time = bfq_bfqq_idle_for_long_time(bfqd, bfqq), ++ /* ++ * See the comments on ++ * bfq_bfqq_update_budg_for_activation for ++ * details on the usage of the next variable. ++ */ ++ arrived_in_time = ktime_get_ns() <= ++ RQ_BIC(rq)->ttime.last_end_request + ++ bfqd->bfq_slice_idle * 3; ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "bfq_add_request non-busy: " ++ "jiffies %lu, in_time %d, idle_long %d busyw %d " ++ "wr_coeff %u", ++ jiffies, arrived_in_time, ++ idle_for_long_time, ++ bfq_bfqq_non_blocking_wait_rq(bfqq), ++ old_wr_coeff); ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ BUG_ON(bfqq == bfqd->in_service_queue); ++ bfqg_stats_update_io_add(bfqq_group(RQ_BFQQ(rq)), bfqq, rq->cmd_flags); ++ ++ /* ++ * bfqq deserves to be weight-raised if: ++ * - it is sync, ++ * - it does not belong to a large burst, ++ * - it has been idle for enough time or is soft real-time, ++ * - is linked to a bfq_io_cq (it is not shared in any sense) ++ */ ++ in_burst = bfq_bfqq_in_large_burst(bfqq); ++ soft_rt = bfqd->bfq_wr_max_softrt_rate > 0 && ++ !in_burst && ++ time_is_before_jiffies(bfqq->soft_rt_next_start); ++ *interactive = ++ !in_burst && ++ idle_for_long_time; ++ wr_or_deserves_wr = bfqd->low_latency && ++ (bfqq->wr_coeff > 1 || ++ (bfq_bfqq_sync(bfqq) && ++ bfqq->bic && (*interactive || soft_rt))); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "bfq_add_request: " ++ "in_burst %d, " ++ "soft_rt %d (next %lu), inter %d, bic %p", ++ bfq_bfqq_in_large_burst(bfqq), soft_rt, ++ bfqq->soft_rt_next_start, ++ *interactive, ++ bfqq->bic); ++ ++ /* ++ * Using the last flag, update budget and check whether bfqq ++ * may want to preempt the in-service queue. ++ */ ++ bfqq_wants_to_preempt = ++ bfq_bfqq_update_budg_for_activation(bfqd, bfqq, ++ arrived_in_time, ++ wr_or_deserves_wr); ++ ++ /* ++ * If bfqq happened to be activated in a burst, but has been ++ * idle for much more than an interactive queue, then we ++ * assume that, in the overall I/O initiated in the burst, the ++ * I/O associated with bfqq is finished. So bfqq does not need ++ * to be treated as a queue belonging to a burst ++ * anymore. Accordingly, we reset bfqq's in_large_burst flag ++ * if set, and remove bfqq from the burst list if it's ++ * there. We do not decrement burst_size, because the fact ++ * that bfqq does not need to belong to the burst list any ++ * more does not invalidate the fact that bfqq was created in ++ * a burst. ++ */ ++ if (likely(!bfq_bfqq_just_created(bfqq)) && ++ idle_for_long_time && ++ time_is_before_jiffies( ++ bfqq->budget_timeout + ++ msecs_to_jiffies(10000))) { ++ hlist_del_init(&bfqq->burst_list_node); ++ bfq_clear_bfqq_in_large_burst(bfqq); ++ } ++ ++ bfq_clear_bfqq_just_created(bfqq); ++ ++ if (!bfq_bfqq_IO_bound(bfqq)) { ++ if (arrived_in_time) { ++ bfqq->requests_within_timer++; ++ if (bfqq->requests_within_timer >= ++ bfqd->bfq_requests_within_timer) ++ bfq_mark_bfqq_IO_bound(bfqq); ++ } else ++ bfqq->requests_within_timer = 0; ++ bfq_log_bfqq(bfqd, bfqq, "requests in time %d", ++ bfqq->requests_within_timer); ++ } ++ ++ if (bfqd->low_latency) { ++ if (unlikely(time_is_after_jiffies(bfqq->split_time))) ++ /* wraparound */ ++ bfqq->split_time = ++ jiffies - bfqd->bfq_wr_min_idle_time - 1; ++ ++ if (time_is_before_jiffies(bfqq->split_time + ++ bfqd->bfq_wr_min_idle_time)) { ++ bfq_update_bfqq_wr_on_rq_arrival(bfqd, bfqq, ++ old_wr_coeff, ++ wr_or_deserves_wr, ++ *interactive, ++ in_burst, ++ soft_rt); ++ ++ if (old_wr_coeff != bfqq->wr_coeff) ++ bfqq->entity.prio_changed = 1; ++ } ++ } ++ ++ bfqq->last_idle_bklogged = jiffies; ++ bfqq->service_from_backlogged = 0; ++ bfq_clear_bfqq_softrt_update(bfqq); ++ ++ bfq_add_bfqq_busy(bfqd, bfqq); ++ ++ /* ++ * Expire in-service queue only if preemption may be needed ++ * for guarantees. In this respect, the function ++ * next_queue_may_preempt just checks a simple, necessary ++ * condition, and not a sufficient condition based on ++ * timestamps. In fact, for the latter condition to be ++ * evaluated, timestamps would need first to be updated, and ++ * this operation is quite costly (see the comments on the ++ * function bfq_bfqq_update_budg_for_activation). ++ */ ++ if (bfqd->in_service_queue && bfqq_wants_to_preempt && ++ bfqd->in_service_queue->wr_coeff < bfqq->wr_coeff && ++ next_queue_may_preempt(bfqd)) { ++ struct bfq_queue *in_serv = ++ bfqd->in_service_queue; ++ BUG_ON(in_serv == bfqq); ++ ++ bfq_bfqq_expire(bfqd, bfqd->in_service_queue, ++ false, BFQ_BFQQ_PREEMPTED); ++ } ++} ++ ++static void bfq_add_request(struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ struct bfq_data *bfqd = bfqq->bfqd; ++ struct request *next_rq, *prev; ++ unsigned int old_wr_coeff = bfqq->wr_coeff; ++ bool interactive = false; ++ ++ bfq_log_bfqq(bfqd, bfqq, "add_request: size %u %s", ++ blk_rq_sectors(rq), rq_is_sync(rq) ? "S" : "A"); ++ ++ if (bfqq->wr_coeff > 1) /* queue is being weight-raised */ ++ bfq_log_bfqq(bfqd, bfqq, ++ "raising period dur %u/%u msec, old coeff %u, w %d(%d)", ++ jiffies_to_msecs(jiffies - bfqq->last_wr_start_finish), ++ jiffies_to_msecs(bfqq->wr_cur_max_time), ++ bfqq->wr_coeff, ++ bfqq->entity.weight, bfqq->entity.orig_weight); ++ ++ bfqq->queued[rq_is_sync(rq)]++; ++ bfqd->queued++; ++ ++ elv_rb_add(&bfqq->sort_list, rq); ++ ++ /* ++ * Check if this request is a better next-to-serve candidate. ++ */ ++ prev = bfqq->next_rq; ++ next_rq = bfq_choose_req(bfqd, bfqq->next_rq, rq, bfqd->last_position); ++ BUG_ON(!next_rq); ++ bfqq->next_rq = next_rq; ++ ++ /* ++ * Adjust priority tree position, if next_rq changes. ++ */ ++ if (prev != bfqq->next_rq) ++ bfq_pos_tree_add_move(bfqd, bfqq); ++ ++ if (!bfq_bfqq_busy(bfqq)) /* switching to busy ... */ ++ bfq_bfqq_handle_idle_busy_switch(bfqd, bfqq, old_wr_coeff, ++ rq, &interactive); ++ else { ++ if (bfqd->low_latency && old_wr_coeff == 1 && !rq_is_sync(rq) && ++ time_is_before_jiffies( ++ bfqq->last_wr_start_finish + ++ bfqd->bfq_wr_min_inter_arr_async)) { ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ ++ bfqd->wr_busy_queues++; ++ BUG_ON(bfqd->wr_busy_queues > bfqd->busy_queues); ++ bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqd, bfqq, ++ "non-idle wrais starting, " ++ "wr_max_time %u wr_busy %d", ++ jiffies_to_msecs(bfqq->wr_cur_max_time), ++ bfqd->wr_busy_queues); ++ } ++ if (prev != bfqq->next_rq) ++ bfq_updated_next_req(bfqd, bfqq); ++ } ++ ++ /* ++ * Assign jiffies to last_wr_start_finish in the following ++ * cases: ++ * ++ * . if bfqq is not going to be weight-raised, because, for ++ * non weight-raised queues, last_wr_start_finish stores the ++ * arrival time of the last request; as of now, this piece ++ * of information is used only for deciding whether to ++ * weight-raise async queues ++ * ++ * . if bfqq is not weight-raised, because, if bfqq is now ++ * switching to weight-raised, then last_wr_start_finish ++ * stores the time when weight-raising starts ++ * ++ * . if bfqq is interactive, because, regardless of whether ++ * bfqq is currently weight-raised, the weight-raising ++ * period must start or restart (this case is considered ++ * separately because it is not detected by the above ++ * conditions, if bfqq is already weight-raised) ++ * ++ * last_wr_start_finish has to be updated also if bfqq is soft ++ * real-time, because the weight-raising period is constantly ++ * restarted on idle-to-busy transitions for these queues, but ++ * this is already done in bfq_bfqq_handle_idle_busy_switch if ++ * needed. ++ */ ++ if (bfqd->low_latency && ++ (old_wr_coeff == 1 || bfqq->wr_coeff == 1 || interactive)) ++ bfqq->last_wr_start_finish = jiffies; ++} ++ ++static struct request *bfq_find_rq_fmerge(struct bfq_data *bfqd, ++ struct bio *bio) ++{ ++ struct task_struct *tsk = current; ++ struct bfq_io_cq *bic; ++ struct bfq_queue *bfqq; ++ ++ bic = bfq_bic_lookup(bfqd, tsk->io_context); ++ if (!bic) ++ return NULL; ++ ++ bfqq = bic_to_bfqq(bic, op_is_sync(bio->bi_opf)); ++ if (bfqq) ++ return elv_rb_find(&bfqq->sort_list, bio_end_sector(bio)); ++ ++ return NULL; ++} ++ ++static sector_t get_sdist(sector_t last_pos, struct request *rq) ++{ ++ sector_t sdist = 0; ++ ++ if (last_pos) { ++ if (last_pos < blk_rq_pos(rq)) ++ sdist = blk_rq_pos(rq) - last_pos; ++ else ++ sdist = last_pos - blk_rq_pos(rq); ++ } ++ ++ return sdist; ++} ++ ++static void bfq_activate_request(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ bfqd->rq_in_driver++; ++} ++ ++static void bfq_deactivate_request(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ ++ BUG_ON(bfqd->rq_in_driver == 0); ++ bfqd->rq_in_driver--; ++} ++ ++static void bfq_remove_request(struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ struct bfq_data *bfqd = bfqq->bfqd; ++ const int sync = rq_is_sync(rq); ++ ++ BUG_ON(bfqq->entity.service > bfqq->entity.budget && ++ bfqq == bfqd->in_service_queue); ++ ++ if (bfqq->next_rq == rq) { ++ bfqq->next_rq = bfq_find_next_rq(bfqd, bfqq, rq); ++ bfq_updated_next_req(bfqd, bfqq); ++ } ++ ++ if (rq->queuelist.prev != &rq->queuelist) ++ list_del_init(&rq->queuelist); ++ BUG_ON(bfqq->queued[sync] == 0); ++ bfqq->queued[sync]--; ++ bfqd->queued--; ++ elv_rb_del(&bfqq->sort_list, rq); ++ ++ if (RB_EMPTY_ROOT(&bfqq->sort_list)) { ++ bfqq->next_rq = NULL; ++ ++ BUG_ON(bfqq->entity.budget < 0); ++ ++ if (bfq_bfqq_busy(bfqq) && bfqq != bfqd->in_service_queue) { ++ BUG_ON(bfqq->ref < 2); /* referred by rq and on tree */ ++ bfq_del_bfqq_busy(bfqd, bfqq, false); ++ /* ++ * bfqq emptied. In normal operation, when ++ * bfqq is empty, bfqq->entity.service and ++ * bfqq->entity.budget must contain, ++ * respectively, the service received and the ++ * budget used last time bfqq emptied. These ++ * facts do not hold in this case, as at least ++ * this last removal occurred while bfqq is ++ * not in service. To avoid inconsistencies, ++ * reset both bfqq->entity.service and ++ * bfqq->entity.budget, if bfqq has still a ++ * process that may issue I/O requests to it. ++ */ ++ bfqq->entity.budget = bfqq->entity.service = 0; ++ } ++ ++ /* ++ * Remove queue from request-position tree as it is empty. ++ */ ++ if (bfqq->pos_root) { ++ rb_erase(&bfqq->pos_node, bfqq->pos_root); ++ bfqq->pos_root = NULL; ++ } ++ } ++ ++ if (rq->cmd_flags & REQ_META) { ++ BUG_ON(bfqq->meta_pending == 0); ++ bfqq->meta_pending--; ++ } ++ bfqg_stats_update_io_remove(bfqq_group(bfqq), rq->cmd_flags); ++} ++ ++static enum elv_merge bfq_merge(struct request_queue *q, struct request **req, ++ struct bio *bio) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct request *__rq; ++ ++ __rq = bfq_find_rq_fmerge(bfqd, bio); ++ if (__rq && elv_bio_merge_ok(__rq, bio)) { ++ *req = __rq; ++ return ELEVATOR_FRONT_MERGE; ++ } ++ ++ return ELEVATOR_NO_MERGE; ++} ++ ++static void bfq_merged_request(struct request_queue *q, struct request *req, ++ enum elv_merge type) ++{ ++ if (type == ELEVATOR_FRONT_MERGE && ++ rb_prev(&req->rb_node) && ++ blk_rq_pos(req) < ++ blk_rq_pos(container_of(rb_prev(&req->rb_node), ++ struct request, rb_node))) { ++ struct bfq_queue *bfqq = RQ_BFQQ(req); ++ struct bfq_data *bfqd = bfqq->bfqd; ++ struct request *prev, *next_rq; ++ ++ /* Reposition request in its sort_list */ ++ elv_rb_del(&bfqq->sort_list, req); ++ elv_rb_add(&bfqq->sort_list, req); ++ /* Choose next request to be served for bfqq */ ++ prev = bfqq->next_rq; ++ next_rq = bfq_choose_req(bfqd, bfqq->next_rq, req, ++ bfqd->last_position); ++ BUG_ON(!next_rq); ++ bfqq->next_rq = next_rq; ++ /* ++ * If next_rq changes, update both the queue's budget to ++ * fit the new request and the queue's position in its ++ * rq_pos_tree. ++ */ ++ if (prev != bfqq->next_rq) { ++ bfq_updated_next_req(bfqd, bfqq); ++ bfq_pos_tree_add_move(bfqd, bfqq); ++ } ++ } ++} ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++static void bfq_bio_merged(struct request_queue *q, struct request *req, ++ struct bio *bio) ++{ ++ bfqg_stats_update_io_merged(bfqq_group(RQ_BFQQ(req)), bio->bi_opf); ++} ++#endif ++ ++static void bfq_merged_requests(struct request_queue *q, struct request *rq, ++ struct request *next) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq), *next_bfqq = RQ_BFQQ(next); ++ ++ /* ++ * If next and rq belong to the same bfq_queue and next is older ++ * than rq, then reposition rq in the fifo (by substituting next ++ * with rq). Otherwise, if next and rq belong to different ++ * bfq_queues, never reposition rq: in fact, we would have to ++ * reposition it with respect to next's position in its own fifo, ++ * which would most certainly be too expensive with respect to ++ * the benefits. ++ */ ++ if (bfqq == next_bfqq && ++ !list_empty(&rq->queuelist) && !list_empty(&next->queuelist) && ++ next->fifo_time < rq->fifo_time) { ++ list_del_init(&rq->queuelist); ++ list_replace_init(&next->queuelist, &rq->queuelist); ++ rq->fifo_time = next->fifo_time; ++ } ++ ++ if (bfqq->next_rq == next) ++ bfqq->next_rq = rq; ++ ++ bfq_remove_request(next); ++ bfqg_stats_update_io_merged(bfqq_group(bfqq), next->cmd_flags); ++} ++ ++/* Must be called with bfqq != NULL */ ++static void bfq_bfqq_end_wr(struct bfq_queue *bfqq) ++{ ++ BUG_ON(!bfqq); ++ ++ if (bfq_bfqq_busy(bfqq)) { ++ bfqq->bfqd->wr_busy_queues--; ++ BUG_ON(bfqq->bfqd->wr_busy_queues < 0); ++ } ++ bfqq->wr_coeff = 1; ++ bfqq->wr_cur_max_time = 0; ++ bfqq->last_wr_start_finish = jiffies; ++ /* ++ * Trigger a weight change on the next invocation of ++ * __bfq_entity_update_weight_prio. ++ */ ++ bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "end_wr: wrais ending at %lu, rais_max_time %u", ++ bfqq->last_wr_start_finish, ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "end_wr: wr_busy %d", ++ bfqq->bfqd->wr_busy_queues); ++} ++ ++static void bfq_end_wr_async_queues(struct bfq_data *bfqd, ++ struct bfq_group *bfqg) ++{ ++ int i, j; ++ ++ for (i = 0; i < 2; i++) ++ for (j = 0; j < IOPRIO_BE_NR; j++) ++ if (bfqg->async_bfqq[i][j]) ++ bfq_bfqq_end_wr(bfqg->async_bfqq[i][j]); ++ if (bfqg->async_idle_bfqq) ++ bfq_bfqq_end_wr(bfqg->async_idle_bfqq); ++} ++ ++static void bfq_end_wr(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq; ++ ++ spin_lock_irq(bfqd->queue->queue_lock); ++ ++ list_for_each_entry(bfqq, &bfqd->active_list, bfqq_list) ++ bfq_bfqq_end_wr(bfqq); ++ list_for_each_entry(bfqq, &bfqd->idle_list, bfqq_list) ++ bfq_bfqq_end_wr(bfqq); ++ bfq_end_wr_async(bfqd); ++ ++ spin_unlock_irq(bfqd->queue->queue_lock); ++} ++ ++static sector_t bfq_io_struct_pos(void *io_struct, bool request) ++{ ++ if (request) ++ return blk_rq_pos(io_struct); ++ else ++ return ((struct bio *)io_struct)->bi_iter.bi_sector; ++} ++ ++static int bfq_rq_close_to_sector(void *io_struct, bool request, ++ sector_t sector) ++{ ++ return abs(bfq_io_struct_pos(io_struct, request) - sector) <= ++ BFQQ_CLOSE_THR; ++} ++ ++static struct bfq_queue *bfqq_find_close(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ sector_t sector) ++{ ++ struct rb_root *root = &bfq_bfqq_to_bfqg(bfqq)->rq_pos_tree; ++ struct rb_node *parent, *node; ++ struct bfq_queue *__bfqq; ++ ++ if (RB_EMPTY_ROOT(root)) ++ return NULL; ++ ++ /* ++ * First, if we find a request starting at the end of the last ++ * request, choose it. ++ */ ++ __bfqq = bfq_rq_pos_tree_lookup(bfqd, root, sector, &parent, NULL); ++ if (__bfqq) ++ return __bfqq; ++ ++ /* ++ * If the exact sector wasn't found, the parent of the NULL leaf ++ * will contain the closest sector (rq_pos_tree sorted by ++ * next_request position). ++ */ ++ __bfqq = rb_entry(parent, struct bfq_queue, pos_node); ++ if (bfq_rq_close_to_sector(__bfqq->next_rq, true, sector)) ++ return __bfqq; ++ ++ if (blk_rq_pos(__bfqq->next_rq) < sector) ++ node = rb_next(&__bfqq->pos_node); ++ else ++ node = rb_prev(&__bfqq->pos_node); ++ if (!node) ++ return NULL; ++ ++ __bfqq = rb_entry(node, struct bfq_queue, pos_node); ++ if (bfq_rq_close_to_sector(__bfqq->next_rq, true, sector)) ++ return __bfqq; ++ ++ return NULL; ++} ++ ++static struct bfq_queue *bfq_find_close_cooperator(struct bfq_data *bfqd, ++ struct bfq_queue *cur_bfqq, ++ sector_t sector) ++{ ++ struct bfq_queue *bfqq; ++ ++ /* ++ * We shall notice if some of the queues are cooperating, ++ * e.g., working closely on the same area of the device. In ++ * that case, we can group them together and: 1) don't waste ++ * time idling, and 2) serve the union of their requests in ++ * the best possible order for throughput. ++ */ ++ bfqq = bfqq_find_close(bfqd, cur_bfqq, sector); ++ if (!bfqq || bfqq == cur_bfqq) ++ return NULL; ++ ++ return bfqq; ++} ++ ++static struct bfq_queue * ++bfq_setup_merge(struct bfq_queue *bfqq, struct bfq_queue *new_bfqq) ++{ ++ int process_refs, new_process_refs; ++ struct bfq_queue *__bfqq; ++ ++ /* ++ * If there are no process references on the new_bfqq, then it is ++ * unsafe to follow the ->new_bfqq chain as other bfqq's in the chain ++ * may have dropped their last reference (not just their last process ++ * reference). ++ */ ++ if (!bfqq_process_refs(new_bfqq)) ++ return NULL; ++ ++ /* Avoid a circular list and skip interim queue merges. */ ++ while ((__bfqq = new_bfqq->new_bfqq)) { ++ if (__bfqq == bfqq) ++ return NULL; ++ new_bfqq = __bfqq; ++ } ++ ++ process_refs = bfqq_process_refs(bfqq); ++ new_process_refs = bfqq_process_refs(new_bfqq); ++ /* ++ * If the process for the bfqq has gone away, there is no ++ * sense in merging the queues. ++ */ ++ if (process_refs == 0 || new_process_refs == 0) ++ return NULL; ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "scheduling merge with queue %d", ++ new_bfqq->pid); ++ ++ /* ++ * Merging is just a redirection: the requests of the process ++ * owning one of the two queues are redirected to the other queue. ++ * The latter queue, in its turn, is set as shared if this is the ++ * first time that the requests of some process are redirected to ++ * it. ++ * ++ * We redirect bfqq to new_bfqq and not the opposite, because we ++ * are in the context of the process owning bfqq, hence we have ++ * the io_cq of this process. So we can immediately configure this ++ * io_cq to redirect the requests of the process to new_bfqq. ++ * ++ * NOTE, even if new_bfqq coincides with the in-service queue, the ++ * io_cq of new_bfqq is not available, because, if the in-service ++ * queue is shared, bfqd->in_service_bic may not point to the ++ * io_cq of the in-service queue. ++ * Redirecting the requests of the process owning bfqq to the ++ * currently in-service queue is in any case the best option, as ++ * we feed the in-service queue with new requests close to the ++ * last request served and, by doing so, hopefully increase the ++ * throughput. ++ */ ++ bfqq->new_bfqq = new_bfqq; ++ new_bfqq->ref += process_refs; ++ return new_bfqq; ++} ++ ++static bool bfq_may_be_close_cooperator(struct bfq_queue *bfqq, ++ struct bfq_queue *new_bfqq) ++{ ++ if (bfq_class_idle(bfqq) || bfq_class_idle(new_bfqq) || ++ (bfqq->ioprio_class != new_bfqq->ioprio_class)) ++ return false; ++ ++ /* ++ * If either of the queues has already been detected as seeky, ++ * then merging it with the other queue is unlikely to lead to ++ * sequential I/O. ++ */ ++ if (BFQQ_SEEKY(bfqq) || BFQQ_SEEKY(new_bfqq)) ++ return false; ++ ++ /* ++ * Interleaved I/O is known to be done by (some) applications ++ * only for reads, so it does not make sense to merge async ++ * queues. ++ */ ++ if (!bfq_bfqq_sync(bfqq) || !bfq_bfqq_sync(new_bfqq)) ++ return false; ++ ++ return true; ++} ++ ++/* ++ * If this function returns true, then bfqq cannot be merged. The idea ++ * is that true cooperation happens very early after processes start ++ * to do I/O. Usually, late cooperations are just accidental false ++ * positives. In case bfqq is weight-raised, such false positives ++ * would evidently degrade latency guarantees for bfqq. ++ */ ++static bool wr_from_too_long(struct bfq_queue *bfqq) ++{ ++ return bfqq->wr_coeff > 1 && ++ time_is_before_jiffies(bfqq->last_wr_start_finish + ++ msecs_to_jiffies(100)); ++} ++ ++/* ++ * Attempt to schedule a merge of bfqq with the currently in-service ++ * queue or with a close queue among the scheduled queues. Return ++ * NULL if no merge was scheduled, a pointer to the shared bfq_queue ++ * structure otherwise. ++ * ++ * The OOM queue is not allowed to participate to cooperation: in fact, since ++ * the requests temporarily redirected to the OOM queue could be redirected ++ * again to dedicated queues at any time, the state needed to correctly ++ * handle merging with the OOM queue would be quite complex and expensive ++ * to maintain. Besides, in such a critical condition as an out of memory, ++ * the benefits of queue merging may be little relevant, or even negligible. ++ * ++ * Weight-raised queues can be merged only if their weight-raising ++ * period has just started. In fact cooperating processes are usually ++ * started together. Thus, with this filter we avoid false positives ++ * that would jeopardize low-latency guarantees. ++ * ++ * WARNING: queue merging may impair fairness among non-weight raised ++ * queues, for at least two reasons: 1) the original weight of a ++ * merged queue may change during the merged state, 2) even being the ++ * weight the same, a merged queue may be bloated with many more ++ * requests than the ones produced by its originally-associated ++ * process. ++ */ ++static struct bfq_queue * ++bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ void *io_struct, bool request) ++{ ++ struct bfq_queue *in_service_bfqq, *new_bfqq; ++ ++ if (bfqq->new_bfqq) ++ return bfqq->new_bfqq; ++ ++ if (io_struct && wr_from_too_long(bfqq) && ++ likely(bfqq != &bfqd->oom_bfqq)) ++ bfq_log_bfqq(bfqd, bfqq, ++ "would have looked for coop, but bfq%d wr", ++ bfqq->pid); ++ ++ if (!io_struct || ++ wr_from_too_long(bfqq) || ++ unlikely(bfqq == &bfqd->oom_bfqq)) ++ return NULL; ++ ++ /* If there is only one backlogged queue, don't search. */ ++ if (bfqd->busy_queues == 1) ++ return NULL; ++ ++ in_service_bfqq = bfqd->in_service_queue; ++ ++ if (in_service_bfqq && in_service_bfqq != bfqq && ++ bfqd->in_service_bic && wr_from_too_long(in_service_bfqq) ++ && likely(in_service_bfqq == &bfqd->oom_bfqq)) ++ bfq_log_bfqq(bfqd, bfqq, ++ "would have tried merge with in-service-queue, but wr"); ++ ++ if (!in_service_bfqq || in_service_bfqq == bfqq || ++ !bfqd->in_service_bic || wr_from_too_long(in_service_bfqq) || ++ unlikely(in_service_bfqq == &bfqd->oom_bfqq)) ++ goto check_scheduled; ++ ++ if (bfq_rq_close_to_sector(io_struct, request, bfqd->last_position) && ++ bfqq->entity.parent == in_service_bfqq->entity.parent && ++ bfq_may_be_close_cooperator(bfqq, in_service_bfqq)) { ++ new_bfqq = bfq_setup_merge(bfqq, in_service_bfqq); ++ if (new_bfqq) ++ return new_bfqq; ++ } ++ /* ++ * Check whether there is a cooperator among currently scheduled ++ * queues. The only thing we need is that the bio/request is not ++ * NULL, as we need it to establish whether a cooperator exists. ++ */ ++check_scheduled: ++ new_bfqq = bfq_find_close_cooperator(bfqd, bfqq, ++ bfq_io_struct_pos(io_struct, request)); ++ ++ BUG_ON(new_bfqq && bfqq->entity.parent != new_bfqq->entity.parent); ++ ++ if (new_bfqq && wr_from_too_long(new_bfqq) && ++ likely(new_bfqq != &bfqd->oom_bfqq) && ++ bfq_may_be_close_cooperator(bfqq, new_bfqq)) ++ bfq_log_bfqq(bfqd, bfqq, ++ "would have merged with bfq%d, but wr", ++ new_bfqq->pid); ++ ++ if (new_bfqq && !wr_from_too_long(new_bfqq) && ++ likely(new_bfqq != &bfqd->oom_bfqq) && ++ bfq_may_be_close_cooperator(bfqq, new_bfqq)) ++ return bfq_setup_merge(bfqq, new_bfqq); ++ ++ return NULL; ++} ++ ++static void bfq_bfqq_save_state(struct bfq_queue *bfqq) ++{ ++ struct bfq_io_cq *bic = bfqq->bic; ++ ++ /* ++ * If !bfqq->bic, the queue is already shared or its requests ++ * have already been redirected to a shared queue; both idle window ++ * and weight raising state have already been saved. Do nothing. ++ */ ++ if (!bic) ++ return; ++ ++ bic->saved_has_short_ttime = bfq_bfqq_has_short_ttime(bfqq); ++ bic->saved_IO_bound = bfq_bfqq_IO_bound(bfqq); ++ bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq); ++ bic->was_in_burst_list = !hlist_unhashed(&bfqq->burst_list_node); ++ bic->saved_wr_coeff = bfqq->wr_coeff; ++ bic->saved_wr_start_at_switch_to_srt = bfqq->wr_start_at_switch_to_srt; ++ bic->saved_last_wr_start_finish = bfqq->last_wr_start_finish; ++ bic->saved_wr_cur_max_time = bfqq->wr_cur_max_time; ++ BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); ++} ++ ++static void bfq_get_bic_reference(struct bfq_queue *bfqq) ++{ ++ /* ++ * If bfqq->bic has a non-NULL value, the bic to which it belongs ++ * is about to begin using a shared bfq_queue. ++ */ ++ if (bfqq->bic) ++ atomic_long_inc(&bfqq->bic->icq.ioc->refcount); ++} ++ ++static void ++bfq_merge_bfqqs(struct bfq_data *bfqd, struct bfq_io_cq *bic, ++ struct bfq_queue *bfqq, struct bfq_queue *new_bfqq) ++{ ++ bfq_log_bfqq(bfqd, bfqq, "merging with queue %lu", ++ (unsigned long) new_bfqq->pid); ++ /* Save weight raising and idle window of the merged queues */ ++ bfq_bfqq_save_state(bfqq); ++ bfq_bfqq_save_state(new_bfqq); ++ if (bfq_bfqq_IO_bound(bfqq)) ++ bfq_mark_bfqq_IO_bound(new_bfqq); ++ bfq_clear_bfqq_IO_bound(bfqq); ++ ++ /* ++ * If bfqq is weight-raised, then let new_bfqq inherit ++ * weight-raising. To reduce false positives, neglect the case ++ * where bfqq has just been created, but has not yet made it ++ * to be weight-raised (which may happen because EQM may merge ++ * bfqq even before bfq_add_request is executed for the first ++ * time for bfqq). Handling this case would however be very ++ * easy, thanks to the flag just_created. ++ */ ++ if (new_bfqq->wr_coeff == 1 && bfqq->wr_coeff > 1) { ++ new_bfqq->wr_coeff = bfqq->wr_coeff; ++ new_bfqq->wr_cur_max_time = bfqq->wr_cur_max_time; ++ new_bfqq->last_wr_start_finish = bfqq->last_wr_start_finish; ++ new_bfqq->wr_start_at_switch_to_srt = ++ bfqq->wr_start_at_switch_to_srt; ++ if (bfq_bfqq_busy(new_bfqq)) { ++ bfqd->wr_busy_queues++; ++ BUG_ON(bfqd->wr_busy_queues > bfqd->busy_queues); ++ } ++ ++ new_bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqd, new_bfqq, ++ "wr start after merge with %d, rais_max_time %u", ++ bfqq->pid, ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ } ++ ++ if (bfqq->wr_coeff > 1) { /* bfqq has given its wr to new_bfqq */ ++ bfqq->wr_coeff = 1; ++ bfqq->entity.prio_changed = 1; ++ if (bfq_bfqq_busy(bfqq)) { ++ bfqd->wr_busy_queues--; ++ BUG_ON(bfqd->wr_busy_queues < 0); ++ } ++ ++ } ++ ++ bfq_log_bfqq(bfqd, new_bfqq, "merge_bfqqs: wr_busy %d", ++ bfqd->wr_busy_queues); ++ ++ /* ++ * Grab a reference to the bic, to prevent it from being destroyed ++ * before being possibly touched by a bfq_split_bfqq(). ++ */ ++ bfq_get_bic_reference(bfqq); ++ bfq_get_bic_reference(new_bfqq); ++ /* ++ * Merge queues (that is, let bic redirect its requests to new_bfqq) ++ */ ++ bic_set_bfqq(bic, new_bfqq, 1); ++ bfq_mark_bfqq_coop(new_bfqq); ++ /* ++ * new_bfqq now belongs to at least two bics (it is a shared queue): ++ * set new_bfqq->bic to NULL. bfqq either: ++ * - does not belong to any bic any more, and hence bfqq->bic must ++ * be set to NULL, or ++ * - is a queue whose owning bics have already been redirected to a ++ * different queue, hence the queue is destined to not belong to ++ * any bic soon and bfqq->bic is already NULL (therefore the next ++ * assignment causes no harm). ++ */ ++ new_bfqq->bic = NULL; ++ bfqq->bic = NULL; ++ /* release process reference to bfqq */ ++ bfq_put_queue(bfqq); ++} ++ ++static int bfq_allow_bio_merge(struct request_queue *q, struct request *rq, ++ struct bio *bio) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ bool is_sync = op_is_sync(bio->bi_opf); ++ struct bfq_io_cq *bic; ++ struct bfq_queue *bfqq, *new_bfqq; ++ ++ /* ++ * Disallow merge of a sync bio into an async request. ++ */ ++ if (is_sync && !rq_is_sync(rq)) ++ return false; ++ ++ /* ++ * Lookup the bfqq that this bio will be queued with. Allow ++ * merge only if rq is queued there. ++ * Queue lock is held here. ++ */ ++ bic = bfq_bic_lookup(bfqd, current->io_context); ++ if (!bic) ++ return false; ++ ++ bfqq = bic_to_bfqq(bic, is_sync); ++ /* ++ * We take advantage of this function to perform an early merge ++ * of the queues of possible cooperating processes. ++ */ ++ if (bfqq) { ++ new_bfqq = bfq_setup_cooperator(bfqd, bfqq, bio, false); ++ if (new_bfqq) { ++ bfq_merge_bfqqs(bfqd, bic, bfqq, new_bfqq); ++ /* ++ * If we get here, the bio will be queued in the ++ * shared queue, i.e., new_bfqq, so use new_bfqq ++ * to decide whether bio and rq can be merged. ++ */ ++ bfqq = new_bfqq; ++ } ++ } ++ ++ return bfqq == RQ_BFQQ(rq); ++} ++ ++static int bfq_allow_rq_merge(struct request_queue *q, struct request *rq, ++ struct request *next) ++{ ++ return RQ_BFQQ(rq) == RQ_BFQQ(next); ++} ++ ++/* ++ * Set the maximum time for the in-service queue to consume its ++ * budget. This prevents seeky processes from lowering the throughput. ++ * In practice, a time-slice service scheme is used with seeky ++ * processes. ++ */ ++static void bfq_set_budget_timeout(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ unsigned int timeout_coeff; ++ ++ if (bfqq->wr_cur_max_time == bfqd->bfq_wr_rt_max_time) ++ timeout_coeff = 1; ++ else ++ timeout_coeff = bfqq->entity.weight / bfqq->entity.orig_weight; ++ ++ bfqd->last_budget_start = ktime_get(); ++ ++ bfqq->budget_timeout = jiffies + ++ bfqd->bfq_timeout * timeout_coeff; ++ ++ bfq_log_bfqq(bfqd, bfqq, "set budget_timeout %u", ++ jiffies_to_msecs(bfqd->bfq_timeout * timeout_coeff)); ++} ++ ++static void __bfq_set_in_service_queue(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ if (bfqq) { ++ bfqg_stats_update_avg_queue_size(bfqq_group(bfqq)); ++ bfq_mark_bfqq_must_alloc(bfqq); ++ bfq_clear_bfqq_fifo_expire(bfqq); ++ ++ bfqd->budgets_assigned = (bfqd->budgets_assigned*7 + 256) / 8; ++ ++ BUG_ON(bfqq == bfqd->in_service_queue); ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list)); ++ ++ if (time_is_before_jiffies(bfqq->last_wr_start_finish) && ++ bfqq->wr_coeff > 1 && ++ bfqq->wr_cur_max_time == bfqd->bfq_wr_rt_max_time && ++ time_is_before_jiffies(bfqq->budget_timeout)) { ++ /* ++ * For soft real-time queues, move the start ++ * of the weight-raising period forward by the ++ * time the queue has not received any ++ * service. Otherwise, a relatively long ++ * service delay is likely to cause the ++ * weight-raising period of the queue to end, ++ * because of the short duration of the ++ * weight-raising period of a soft real-time ++ * queue. It is worth noting that this move ++ * is not so dangerous for the other queues, ++ * because soft real-time queues are not ++ * greedy. ++ * ++ * To not add a further variable, we use the ++ * overloaded field budget_timeout to ++ * determine for how long the queue has not ++ * received service, i.e., how much time has ++ * elapsed since the queue expired. However, ++ * this is a little imprecise, because ++ * budget_timeout is set to jiffies if bfqq ++ * not only expires, but also remains with no ++ * request. ++ */ ++ if (time_after(bfqq->budget_timeout, ++ bfqq->last_wr_start_finish)) ++ bfqq->last_wr_start_finish += ++ jiffies - bfqq->budget_timeout; ++ else ++ bfqq->last_wr_start_finish = jiffies; ++ ++ if (time_is_after_jiffies(bfqq->last_wr_start_finish)) { ++ pr_crit( ++ "BFQ WARNING:last %lu budget %lu jiffies %lu", ++ bfqq->last_wr_start_finish, ++ bfqq->budget_timeout, ++ jiffies); ++ pr_crit("diff %lu", jiffies - ++ max_t(unsigned long, ++ bfqq->last_wr_start_finish, ++ bfqq->budget_timeout)); ++ bfqq->last_wr_start_finish = jiffies; ++ } ++ } ++ ++ bfq_set_budget_timeout(bfqd, bfqq); ++ bfq_log_bfqq(bfqd, bfqq, ++ "set_in_service_queue, cur-budget = %d", ++ bfqq->entity.budget); ++ } else ++ bfq_log(bfqd, "set_in_service_queue: NULL"); ++ ++ bfqd->in_service_queue = bfqq; ++} ++ ++/* ++ * Get and set a new queue for service. ++ */ ++static struct bfq_queue *bfq_set_in_service_queue(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq = bfq_get_next_queue(bfqd); ++ ++ __bfq_set_in_service_queue(bfqd, bfqq); ++ return bfqq; ++} ++ ++static void bfq_arm_slice_timer(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq = bfqd->in_service_queue; ++ struct bfq_io_cq *bic; ++ u32 sl; ++ ++ BUG_ON(!RB_EMPTY_ROOT(&bfqq->sort_list)); ++ ++ /* Processes have exited, don't wait. */ ++ bic = bfqd->in_service_bic; ++ if (!bic || atomic_read(&bic->icq.ioc->active_ref) == 0) ++ return; ++ ++ bfq_mark_bfqq_wait_request(bfqq); ++ ++ /* ++ * We don't want to idle for seeks, but we do want to allow ++ * fair distribution of slice time for a process doing back-to-back ++ * seeks. So allow a little bit of time for him to submit a new rq. ++ * ++ * To prevent processes with (partly) seeky workloads from ++ * being too ill-treated, grant them a small fraction of the ++ * assigned budget before reducing the waiting time to ++ * BFQ_MIN_TT. This happened to help reduce latency. ++ */ ++ sl = bfqd->bfq_slice_idle; ++ /* ++ * Unless the queue is being weight-raised or the scenario is ++ * asymmetric, grant only minimum idle time if the queue ++ * is seeky. A long idling is preserved for a weight-raised ++ * queue, or, more in general, in an asymemtric scenario, ++ * because a long idling is needed for guaranteeing to a queue ++ * its reserved share of the throughput (in particular, it is ++ * needed if the queue has a higher weight than some other ++ * queue). ++ */ ++ if (BFQQ_SEEKY(bfqq) && bfqq->wr_coeff == 1 && ++ bfq_symmetric_scenario(bfqd)) ++ sl = min_t(u32, sl, BFQ_MIN_TT); ++ ++ bfqd->last_idling_start = ktime_get(); ++ hrtimer_start(&bfqd->idle_slice_timer, ns_to_ktime(sl), ++ HRTIMER_MODE_REL); ++ bfqg_stats_set_start_idle_time(bfqq_group(bfqq)); ++ bfq_log(bfqd, "arm idle: %ld/%ld ms", ++ sl / NSEC_PER_MSEC, bfqd->bfq_slice_idle / NSEC_PER_MSEC); ++} ++ ++/* ++ * In autotuning mode, max_budget is dynamically recomputed as the ++ * amount of sectors transferred in timeout at the estimated peak ++ * rate. This enables BFQ to utilize a full timeslice with a full ++ * budget, even if the in-service queue is served at peak rate. And ++ * this maximises throughput with sequential workloads. ++ */ ++static unsigned long bfq_calc_max_budget(struct bfq_data *bfqd) ++{ ++ return (u64)bfqd->peak_rate * USEC_PER_MSEC * ++ jiffies_to_msecs(bfqd->bfq_timeout)>>BFQ_RATE_SHIFT; ++} ++ ++/* ++ * Update parameters related to throughput and responsiveness, as a ++ * function of the estimated peak rate. See comments on ++ * bfq_calc_max_budget(), and on T_slow and T_fast arrays. ++ */ ++static void update_thr_responsiveness_params(struct bfq_data *bfqd) ++{ ++ int dev_type = blk_queue_nonrot(bfqd->queue); ++ ++ if (bfqd->bfq_user_max_budget == 0) { ++ bfqd->bfq_max_budget = ++ bfq_calc_max_budget(bfqd); ++ BUG_ON(bfqd->bfq_max_budget < 0); ++ bfq_log(bfqd, "new max_budget = %d", ++ bfqd->bfq_max_budget); ++ } ++ ++ if (bfqd->device_speed == BFQ_BFQD_FAST && ++ bfqd->peak_rate < device_speed_thresh[dev_type]) { ++ bfqd->device_speed = BFQ_BFQD_SLOW; ++ bfqd->RT_prod = R_slow[dev_type] * ++ T_slow[dev_type]; ++ } else if (bfqd->device_speed == BFQ_BFQD_SLOW && ++ bfqd->peak_rate > device_speed_thresh[dev_type]) { ++ bfqd->device_speed = BFQ_BFQD_FAST; ++ bfqd->RT_prod = R_fast[dev_type] * ++ T_fast[dev_type]; ++ } ++ ++ bfq_log(bfqd, ++"dev_type %s dev_speed_class = %s (%llu sects/sec), thresh %llu setcs/sec", ++ dev_type == 0 ? "ROT" : "NONROT", ++ bfqd->device_speed == BFQ_BFQD_FAST ? "FAST" : "SLOW", ++ bfqd->device_speed == BFQ_BFQD_FAST ? ++ (USEC_PER_SEC*(u64)R_fast[dev_type])>>BFQ_RATE_SHIFT : ++ (USEC_PER_SEC*(u64)R_slow[dev_type])>>BFQ_RATE_SHIFT, ++ (USEC_PER_SEC*(u64)device_speed_thresh[dev_type])>> ++ BFQ_RATE_SHIFT); ++} ++ ++static void bfq_reset_rate_computation(struct bfq_data *bfqd, struct request *rq) ++{ ++ if (rq != NULL) { /* new rq dispatch now, reset accordingly */ ++ bfqd->last_dispatch = bfqd->first_dispatch = ktime_get_ns() ; ++ bfqd->peak_rate_samples = 1; ++ bfqd->sequential_samples = 0; ++ bfqd->tot_sectors_dispatched = bfqd->last_rq_max_size = ++ blk_rq_sectors(rq); ++ } else /* no new rq dispatched, just reset the number of samples */ ++ bfqd->peak_rate_samples = 0; /* full re-init on next disp. */ ++ ++ bfq_log(bfqd, ++ "reset_rate_computation at end, sample %u/%u tot_sects %llu", ++ bfqd->peak_rate_samples, bfqd->sequential_samples, ++ bfqd->tot_sectors_dispatched); ++} ++ ++static void bfq_update_rate_reset(struct bfq_data *bfqd, struct request *rq) ++{ ++ u32 rate, weight, divisor; ++ ++ /* ++ * For the convergence property to hold (see comments on ++ * bfq_update_peak_rate()) and for the assessment to be ++ * reliable, a minimum number of samples must be present, and ++ * a minimum amount of time must have elapsed. If not so, do ++ * not compute new rate. Just reset parameters, to get ready ++ * for a new evaluation attempt. ++ */ ++ if (bfqd->peak_rate_samples < BFQ_RATE_MIN_SAMPLES || ++ bfqd->delta_from_first < BFQ_RATE_MIN_INTERVAL) { ++ bfq_log(bfqd, ++ "update_rate_reset: only resetting, delta_first %lluus samples %d", ++ bfqd->delta_from_first>>10, bfqd->peak_rate_samples); ++ goto reset_computation; ++ } ++ ++ /* ++ * If a new request completion has occurred after last ++ * dispatch, then, to approximate the rate at which requests ++ * have been served by the device, it is more precise to ++ * extend the observation interval to the last completion. ++ */ ++ bfqd->delta_from_first = ++ max_t(u64, bfqd->delta_from_first, ++ bfqd->last_completion - bfqd->first_dispatch); ++ ++ BUG_ON(bfqd->delta_from_first == 0); ++ /* ++ * Rate computed in sects/usec, and not sects/nsec, for ++ * precision issues. ++ */ ++ rate = div64_ul(bfqd->tot_sectors_dispatched<delta_from_first, NSEC_PER_USEC)); ++ ++ bfq_log(bfqd, ++"update_rate_reset: tot_sects %llu delta_first %lluus rate %llu sects/s (%d)", ++ bfqd->tot_sectors_dispatched, bfqd->delta_from_first>>10, ++ ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), ++ rate > 20< 20M sectors/sec) ++ */ ++ if ((bfqd->sequential_samples < (3 * bfqd->peak_rate_samples)>>2 && ++ rate <= bfqd->peak_rate) || ++ rate > 20<peak_rate_samples, bfqd->sequential_samples, ++ ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), ++ ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); ++ goto reset_computation; ++ } else { ++ bfq_log(bfqd, ++ "update_rate_reset: do update, samples %u/%u rate/peak %llu/%llu", ++ bfqd->peak_rate_samples, bfqd->sequential_samples, ++ ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), ++ ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); ++ } ++ ++ /* ++ * We have to update the peak rate, at last! To this purpose, ++ * we use a low-pass filter. We compute the smoothing constant ++ * of the filter as a function of the 'weight' of the new ++ * measured rate. ++ * ++ * As can be seen in next formulas, we define this weight as a ++ * quantity proportional to how sequential the workload is, ++ * and to how long the observation time interval is. ++ * ++ * The weight runs from 0 to 8. The maximum value of the ++ * weight, 8, yields the minimum value for the smoothing ++ * constant. At this minimum value for the smoothing constant, ++ * the measured rate contributes for half of the next value of ++ * the estimated peak rate. ++ * ++ * So, the first step is to compute the weight as a function ++ * of how sequential the workload is. Note that the weight ++ * cannot reach 9, because bfqd->sequential_samples cannot ++ * become equal to bfqd->peak_rate_samples, which, in its ++ * turn, holds true because bfqd->sequential_samples is not ++ * incremented for the first sample. ++ */ ++ weight = (9 * bfqd->sequential_samples) / bfqd->peak_rate_samples; ++ ++ /* ++ * Second step: further refine the weight as a function of the ++ * duration of the observation interval. ++ */ ++ weight = min_t(u32, 8, ++ div_u64(weight * bfqd->delta_from_first, ++ BFQ_RATE_REF_INTERVAL)); ++ ++ /* ++ * Divisor ranging from 10, for minimum weight, to 2, for ++ * maximum weight. ++ */ ++ divisor = 10 - weight; ++ BUG_ON(divisor == 0); ++ ++ /* ++ * Finally, update peak rate: ++ * ++ * peak_rate = peak_rate * (divisor-1) / divisor + rate / divisor ++ */ ++ bfqd->peak_rate *= divisor-1; ++ bfqd->peak_rate /= divisor; ++ rate /= divisor; /* smoothing constant alpha = 1/divisor */ ++ ++ bfq_log(bfqd, ++ "update_rate_reset: divisor %d tmp_peak_rate %llu tmp_rate %u", ++ divisor, ++ ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT), ++ (u32)((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT)); ++ ++ BUG_ON(bfqd->peak_rate == 0); ++ BUG_ON(bfqd->peak_rate > 20<peak_rate += rate; ++ update_thr_responsiveness_params(bfqd); ++ BUG_ON(bfqd->peak_rate > 20<peak_rate_samples == 0) { /* first dispatch */ ++ bfq_log(bfqd, ++ "update_peak_rate: goto reset, samples %d", ++ bfqd->peak_rate_samples) ; ++ bfq_reset_rate_computation(bfqd, rq); ++ goto update_last_values; /* will add one sample */ ++ } ++ ++ /* ++ * Device idle for very long: the observation interval lasting ++ * up to this dispatch cannot be a valid observation interval ++ * for computing a new peak rate (similarly to the late- ++ * completion event in bfq_completed_request()). Go to ++ * update_rate_and_reset to have the following three steps ++ * taken: ++ * - close the observation interval at the last (previous) ++ * request dispatch or completion ++ * - compute rate, if possible, for that observation interval ++ * - start a new observation interval with this dispatch ++ */ ++ if (now_ns - bfqd->last_dispatch > 100*NSEC_PER_MSEC && ++ bfqd->rq_in_driver == 0) { ++ bfq_log(bfqd, ++"update_peak_rate: jumping to updating&resetting delta_last %lluus samples %d", ++ (now_ns - bfqd->last_dispatch)>>10, ++ bfqd->peak_rate_samples) ; ++ goto update_rate_and_reset; ++ } ++ ++ /* Update sampling information */ ++ bfqd->peak_rate_samples++; ++ ++ if ((bfqd->rq_in_driver > 0 || ++ now_ns - bfqd->last_completion < BFQ_MIN_TT) ++ && get_sdist(bfqd->last_position, rq) < BFQQ_SEEK_THR) ++ bfqd->sequential_samples++; ++ ++ bfqd->tot_sectors_dispatched += blk_rq_sectors(rq); ++ ++ /* Reset max observed rq size every 32 dispatches */ ++ if (likely(bfqd->peak_rate_samples % 32)) ++ bfqd->last_rq_max_size = ++ max_t(u32, blk_rq_sectors(rq), bfqd->last_rq_max_size); ++ else ++ bfqd->last_rq_max_size = blk_rq_sectors(rq); ++ ++ bfqd->delta_from_first = now_ns - bfqd->first_dispatch; ++ ++ bfq_log(bfqd, ++ "update_peak_rate: added samples %u/%u tot_sects %llu delta_first %lluus", ++ bfqd->peak_rate_samples, bfqd->sequential_samples, ++ bfqd->tot_sectors_dispatched, ++ bfqd->delta_from_first>>10); ++ ++ /* Target observation interval not yet reached, go on sampling */ ++ if (bfqd->delta_from_first < BFQ_RATE_REF_INTERVAL) ++ goto update_last_values; ++ ++update_rate_and_reset: ++ bfq_update_rate_reset(bfqd, rq); ++update_last_values: ++ bfqd->last_position = blk_rq_pos(rq) + blk_rq_sectors(rq); ++ bfqd->last_dispatch = now_ns; ++ ++ bfq_log(bfqd, ++ "update_peak_rate: delta_first %lluus last_pos %llu peak_rate %llu", ++ (now_ns - bfqd->first_dispatch)>>10, ++ (unsigned long long) bfqd->last_position, ++ ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); ++ bfq_log(bfqd, ++ "update_peak_rate: samples at end %d", bfqd->peak_rate_samples); ++} ++ ++/* ++ * Move request from internal lists to the dispatch list of the request queue ++ */ ++static void bfq_dispatch_insert(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ ++ /* ++ * For consistency, the next instruction should have been executed ++ * after removing the request from the queue and dispatching it. ++ * We execute instead this instruction before bfq_remove_request() ++ * (and hence introduce a temporary inconsistency), for efficiency. ++ * In fact, in a forced_dispatch, this prevents two counters related ++ * to bfqq->dispatched to risk to be uselessly decremented if bfqq ++ * is not in service, and then to be incremented again after ++ * incrementing bfqq->dispatched. ++ */ ++ bfqq->dispatched++; ++ bfq_update_peak_rate(q->elevator->elevator_data, rq); ++ ++ bfq_remove_request(rq); ++ elv_dispatch_sort(q, rq); ++} ++ ++static void __bfq_bfqq_expire(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ BUG_ON(bfqq != bfqd->in_service_queue); ++ ++ /* ++ * If this bfqq is shared between multiple processes, check ++ * to make sure that those processes are still issuing I/Os ++ * within the mean seek distance. If not, it may be time to ++ * break the queues apart again. ++ */ ++ if (bfq_bfqq_coop(bfqq) && BFQQ_SEEKY(bfqq)) ++ bfq_mark_bfqq_split_coop(bfqq); ++ ++ if (RB_EMPTY_ROOT(&bfqq->sort_list)) { ++ if (bfqq->dispatched == 0) ++ /* ++ * Overloading budget_timeout field to store ++ * the time at which the queue remains with no ++ * backlog and no outstanding request; used by ++ * the weight-raising mechanism. ++ */ ++ bfqq->budget_timeout = jiffies; ++ ++ bfq_del_bfqq_busy(bfqd, bfqq, true); ++ } else { ++ bfq_requeue_bfqq(bfqd, bfqq); ++ /* ++ * Resort priority tree of potential close cooperators. ++ */ ++ bfq_pos_tree_add_move(bfqd, bfqq); ++ } ++ ++ /* ++ * All in-service entities must have been properly deactivated ++ * or requeued before executing the next function, which ++ * resets all in-service entites as no more in service. ++ */ ++ __bfq_bfqd_reset_in_service(bfqd); ++} ++ ++/** ++ * __bfq_bfqq_recalc_budget - try to adapt the budget to the @bfqq behavior. ++ * @bfqd: device data. ++ * @bfqq: queue to update. ++ * @reason: reason for expiration. ++ * ++ * Handle the feedback on @bfqq budget at queue expiration. ++ * See the body for detailed comments. ++ */ ++static void __bfq_bfqq_recalc_budget(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ enum bfqq_expiration reason) ++{ ++ struct request *next_rq; ++ int budget, min_budget; ++ ++ BUG_ON(bfqq != bfqd->in_service_queue); ++ ++ min_budget = bfq_min_budget(bfqd); ++ ++ if (bfqq->wr_coeff == 1) ++ budget = bfqq->max_budget; ++ else /* ++ * Use a constant, low budget for weight-raised queues, ++ * to help achieve a low latency. Keep it slightly higher ++ * than the minimum possible budget, to cause a little ++ * bit fewer expirations. ++ */ ++ budget = 2 * min_budget; ++ ++ bfq_log_bfqq(bfqd, bfqq, "recalc_budg: last budg %d, budg left %d", ++ bfqq->entity.budget, bfq_bfqq_budget_left(bfqq)); ++ bfq_log_bfqq(bfqd, bfqq, "recalc_budg: last max_budg %d, min budg %d", ++ budget, bfq_min_budget(bfqd)); ++ bfq_log_bfqq(bfqd, bfqq, "recalc_budg: sync %d, seeky %d", ++ bfq_bfqq_sync(bfqq), BFQQ_SEEKY(bfqd->in_service_queue)); ++ ++ if (bfq_bfqq_sync(bfqq) && bfqq->wr_coeff == 1) { ++ switch (reason) { ++ /* ++ * Caveat: in all the following cases we trade latency ++ * for throughput. ++ */ ++ case BFQ_BFQQ_TOO_IDLE: ++ /* ++ * This is the only case where we may reduce ++ * the budget: if there is no request of the ++ * process still waiting for completion, then ++ * we assume (tentatively) that the timer has ++ * expired because the batch of requests of ++ * the process could have been served with a ++ * smaller budget. Hence, betting that ++ * process will behave in the same way when it ++ * becomes backlogged again, we reduce its ++ * next budget. As long as we guess right, ++ * this budget cut reduces the latency ++ * experienced by the process. ++ * ++ * However, if there are still outstanding ++ * requests, then the process may have not yet ++ * issued its next request just because it is ++ * still waiting for the completion of some of ++ * the still outstanding ones. So in this ++ * subcase we do not reduce its budget, on the ++ * contrary we increase it to possibly boost ++ * the throughput, as discussed in the ++ * comments to the BUDGET_TIMEOUT case. ++ */ ++ if (bfqq->dispatched > 0) /* still outstanding reqs */ ++ budget = min(budget * 2, bfqd->bfq_max_budget); ++ else { ++ if (budget > 5 * min_budget) ++ budget -= 4 * min_budget; ++ else ++ budget = min_budget; ++ } ++ break; ++ case BFQ_BFQQ_BUDGET_TIMEOUT: ++ /* ++ * We double the budget here because it gives ++ * the chance to boost the throughput if this ++ * is not a seeky process (and has bumped into ++ * this timeout because of, e.g., ZBR). ++ */ ++ budget = min(budget * 2, bfqd->bfq_max_budget); ++ break; ++ case BFQ_BFQQ_BUDGET_EXHAUSTED: ++ /* ++ * The process still has backlog, and did not ++ * let either the budget timeout or the disk ++ * idling timeout expire. Hence it is not ++ * seeky, has a short thinktime and may be ++ * happy with a higher budget too. So ++ * definitely increase the budget of this good ++ * candidate to boost the disk throughput. ++ */ ++ budget = min(budget * 4, bfqd->bfq_max_budget); ++ break; ++ case BFQ_BFQQ_NO_MORE_REQUESTS: ++ /* ++ * For queues that expire for this reason, it ++ * is particularly important to keep the ++ * budget close to the actual service they ++ * need. Doing so reduces the timestamp ++ * misalignment problem described in the ++ * comments in the body of ++ * __bfq_activate_entity. In fact, suppose ++ * that a queue systematically expires for ++ * BFQ_BFQQ_NO_MORE_REQUESTS and presents a ++ * new request in time to enjoy timestamp ++ * back-shifting. The larger the budget of the ++ * queue is with respect to the service the ++ * queue actually requests in each service ++ * slot, the more times the queue can be ++ * reactivated with the same virtual finish ++ * time. It follows that, even if this finish ++ * time is pushed to the system virtual time ++ * to reduce the consequent timestamp ++ * misalignment, the queue unjustly enjoys for ++ * many re-activations a lower finish time ++ * than all newly activated queues. ++ * ++ * The service needed by bfqq is measured ++ * quite precisely by bfqq->entity.service. ++ * Since bfqq does not enjoy device idling, ++ * bfqq->entity.service is equal to the number ++ * of sectors that the process associated with ++ * bfqq requested to read/write before waiting ++ * for request completions, or blocking for ++ * other reasons. ++ */ ++ budget = max_t(int, bfqq->entity.service, min_budget); ++ break; ++ default: ++ return; ++ } ++ } else if (!bfq_bfqq_sync(bfqq)) ++ /* ++ * Async queues get always the maximum possible ++ * budget, as for them we do not care about latency ++ * (in addition, their ability to dispatch is limited ++ * by the charging factor). ++ */ ++ budget = bfqd->bfq_max_budget; ++ ++ bfqq->max_budget = budget; ++ ++ if (bfqd->budgets_assigned >= bfq_stats_min_budgets && ++ !bfqd->bfq_user_max_budget) ++ bfqq->max_budget = min(bfqq->max_budget, bfqd->bfq_max_budget); ++ ++ /* ++ * If there is still backlog, then assign a new budget, making ++ * sure that it is large enough for the next request. Since ++ * the finish time of bfqq must be kept in sync with the ++ * budget, be sure to call __bfq_bfqq_expire() *after* this ++ * update. ++ * ++ * If there is no backlog, then no need to update the budget; ++ * it will be updated on the arrival of a new request. ++ */ ++ next_rq = bfqq->next_rq; ++ if (next_rq) { ++ BUG_ON(reason == BFQ_BFQQ_TOO_IDLE || ++ reason == BFQ_BFQQ_NO_MORE_REQUESTS); ++ bfqq->entity.budget = max_t(unsigned long, bfqq->max_budget, ++ bfq_serv_to_charge(next_rq, bfqq)); ++ BUG_ON(!bfq_bfqq_busy(bfqq)); ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list)); ++ } ++ ++ bfq_log_bfqq(bfqd, bfqq, "head sect: %u, new budget %d", ++ next_rq ? blk_rq_sectors(next_rq) : 0, ++ bfqq->entity.budget); ++} ++ ++/* ++ * Return true if the process associated with bfqq is "slow". The slow ++ * flag is used, in addition to the budget timeout, to reduce the ++ * amount of service provided to seeky processes, and thus reduce ++ * their chances to lower the throughput. More details in the comments ++ * on the function bfq_bfqq_expire(). ++ * ++ * An important observation is in order: as discussed in the comments ++ * on the function bfq_update_peak_rate(), with devices with internal ++ * queues, it is hard if ever possible to know when and for how long ++ * an I/O request is processed by the device (apart from the trivial ++ * I/O pattern where a new request is dispatched only after the ++ * previous one has been completed). This makes it hard to evaluate ++ * the real rate at which the I/O requests of each bfq_queue are ++ * served. In fact, for an I/O scheduler like BFQ, serving a ++ * bfq_queue means just dispatching its requests during its service ++ * slot (i.e., until the budget of the queue is exhausted, or the ++ * queue remains idle, or, finally, a timeout fires). But, during the ++ * service slot of a bfq_queue, around 100 ms at most, the device may ++ * be even still processing requests of bfq_queues served in previous ++ * service slots. On the opposite end, the requests of the in-service ++ * bfq_queue may be completed after the service slot of the queue ++ * finishes. ++ * ++ * Anyway, unless more sophisticated solutions are used ++ * (where possible), the sum of the sizes of the requests dispatched ++ * during the service slot of a bfq_queue is probably the only ++ * approximation available for the service received by the bfq_queue ++ * during its service slot. And this sum is the quantity used in this ++ * function to evaluate the I/O speed of a process. ++ */ ++static bool bfq_bfqq_is_slow(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ bool compensate, enum bfqq_expiration reason, ++ unsigned long *delta_ms) ++{ ++ ktime_t delta_ktime; ++ u32 delta_usecs; ++ bool slow = BFQQ_SEEKY(bfqq); /* if delta too short, use seekyness */ ++ ++ if (!bfq_bfqq_sync(bfqq)) ++ return false; ++ ++ if (compensate) ++ delta_ktime = bfqd->last_idling_start; ++ else ++ delta_ktime = ktime_get(); ++ delta_ktime = ktime_sub(delta_ktime, bfqd->last_budget_start); ++ delta_usecs = ktime_to_us(delta_ktime); ++ ++ /* don't use too short time intervals */ ++ if (delta_usecs < 1000) { ++ if (blk_queue_nonrot(bfqd->queue)) ++ /* ++ * give same worst-case guarantees as idling ++ * for seeky ++ */ ++ *delta_ms = BFQ_MIN_TT / NSEC_PER_MSEC; ++ else /* charge at least one seek */ ++ *delta_ms = bfq_slice_idle / NSEC_PER_MSEC; ++ ++ bfq_log(bfqd, "bfq_bfqq_is_slow: too short %u", delta_usecs); ++ ++ return slow; ++ } ++ ++ *delta_ms = delta_usecs / USEC_PER_MSEC; ++ ++ /* ++ * Use only long (> 20ms) intervals to filter out excessive ++ * spikes in service rate estimation. ++ */ ++ if (delta_usecs > 20000) { ++ /* ++ * Caveat for rotational devices: processes doing I/O ++ * in the slower disk zones tend to be slow(er) even ++ * if not seeky. In this respect, the estimated peak ++ * rate is likely to be an average over the disk ++ * surface. Accordingly, to not be too harsh with ++ * unlucky processes, a process is deemed slow only if ++ * its rate has been lower than half of the estimated ++ * peak rate. ++ */ ++ slow = bfqq->entity.service < bfqd->bfq_max_budget / 2; ++ bfq_log(bfqd, "bfq_bfqq_is_slow: relative rate %d/%d", ++ bfqq->entity.service, bfqd->bfq_max_budget); ++ } ++ ++ bfq_log_bfqq(bfqd, bfqq, "bfq_bfqq_is_slow: slow %d", slow); ++ ++ return slow; ++} ++ ++/* ++ * To be deemed as soft real-time, an application must meet two ++ * requirements. First, the application must not require an average ++ * bandwidth higher than the approximate bandwidth required to playback or ++ * record a compressed high-definition video. ++ * The next function is invoked on the completion of the last request of a ++ * batch, to compute the next-start time instant, soft_rt_next_start, such ++ * that, if the next request of the application does not arrive before ++ * soft_rt_next_start, then the above requirement on the bandwidth is met. ++ * ++ * The second requirement is that the request pattern of the application is ++ * isochronous, i.e., that, after issuing a request or a batch of requests, ++ * the application stops issuing new requests until all its pending requests ++ * have been completed. After that, the application may issue a new batch, ++ * and so on. ++ * For this reason the next function is invoked to compute ++ * soft_rt_next_start only for applications that meet this requirement, ++ * whereas soft_rt_next_start is set to infinity for applications that do ++ * not. ++ * ++ * Unfortunately, even a greedy application may happen to behave in an ++ * isochronous way if the CPU load is high. In fact, the application may ++ * stop issuing requests while the CPUs are busy serving other processes, ++ * then restart, then stop again for a while, and so on. In addition, if ++ * the disk achieves a low enough throughput with the request pattern ++ * issued by the application (e.g., because the request pattern is random ++ * and/or the device is slow), then the application may meet the above ++ * bandwidth requirement too. To prevent such a greedy application to be ++ * deemed as soft real-time, a further rule is used in the computation of ++ * soft_rt_next_start: soft_rt_next_start must be higher than the current ++ * time plus the maximum time for which the arrival of a request is waited ++ * for when a sync queue becomes idle, namely bfqd->bfq_slice_idle. ++ * This filters out greedy applications, as the latter issue instead their ++ * next request as soon as possible after the last one has been completed ++ * (in contrast, when a batch of requests is completed, a soft real-time ++ * application spends some time processing data). ++ * ++ * Unfortunately, the last filter may easily generate false positives if ++ * only bfqd->bfq_slice_idle is used as a reference time interval and one ++ * or both the following cases occur: ++ * 1) HZ is so low that the duration of a jiffy is comparable to or higher ++ * than bfqd->bfq_slice_idle. This happens, e.g., on slow devices with ++ * HZ=100. ++ * 2) jiffies, instead of increasing at a constant rate, may stop increasing ++ * for a while, then suddenly 'jump' by several units to recover the lost ++ * increments. This seems to happen, e.g., inside virtual machines. ++ * To address this issue, we do not use as a reference time interval just ++ * bfqd->bfq_slice_idle, but bfqd->bfq_slice_idle plus a few jiffies. In ++ * particular we add the minimum number of jiffies for which the filter ++ * seems to be quite precise also in embedded systems and KVM/QEMU virtual ++ * machines. ++ */ ++static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ bfq_log_bfqq(bfqd, bfqq, ++"softrt_next_start: service_blkg %lu soft_rate %u sects/sec interval %u", ++ bfqq->service_from_backlogged, ++ bfqd->bfq_wr_max_softrt_rate, ++ jiffies_to_msecs(HZ * bfqq->service_from_backlogged / ++ bfqd->bfq_wr_max_softrt_rate)); ++ ++ return max(bfqq->last_idle_bklogged + ++ HZ * bfqq->service_from_backlogged / ++ bfqd->bfq_wr_max_softrt_rate, ++ jiffies + nsecs_to_jiffies(bfqq->bfqd->bfq_slice_idle) + 4); ++} ++ ++/* ++ * Return the farthest future time instant according to jiffies ++ * macros. ++ */ ++static unsigned long bfq_greatest_from_now(void) ++{ ++ return jiffies + MAX_JIFFY_OFFSET; ++} ++ ++/* ++ * Return the farthest past time instant according to jiffies ++ * macros. ++ */ ++static unsigned long bfq_smallest_from_now(void) ++{ ++ return jiffies - MAX_JIFFY_OFFSET; ++} ++ ++/** ++ * bfq_bfqq_expire - expire a queue. ++ * @bfqd: device owning the queue. ++ * @bfqq: the queue to expire. ++ * @compensate: if true, compensate for the time spent idling. ++ * @reason: the reason causing the expiration. ++ * ++ * If the process associated with bfqq does slow I/O (e.g., because it ++ * issues random requests), we charge bfqq with the time it has been ++ * in service instead of the service it has received (see ++ * bfq_bfqq_charge_time for details on how this goal is achieved). As ++ * a consequence, bfqq will typically get higher timestamps upon ++ * reactivation, and hence it will be rescheduled as if it had ++ * received more service than what it has actually received. In the ++ * end, bfqq receives less service in proportion to how slowly its ++ * associated process consumes its budgets (and hence how seriously it ++ * tends to lower the throughput). In addition, this time-charging ++ * strategy guarantees time fairness among slow processes. In ++ * contrast, if the process associated with bfqq is not slow, we ++ * charge bfqq exactly with the service it has received. ++ * ++ * Charging time to the first type of queues and the exact service to ++ * the other has the effect of using the WF2Q+ policy to schedule the ++ * former on a timeslice basis, without violating service domain ++ * guarantees among the latter. ++ */ ++static void bfq_bfqq_expire(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ bool compensate, ++ enum bfqq_expiration reason) ++{ ++ bool slow; ++ unsigned long delta = 0; ++ struct bfq_entity *entity = &bfqq->entity; ++ int ref; ++ ++ BUG_ON(bfqq != bfqd->in_service_queue); ++ ++ /* ++ * Check whether the process is slow (see bfq_bfqq_is_slow). ++ */ ++ slow = bfq_bfqq_is_slow(bfqd, bfqq, compensate, reason, &delta); ++ ++ /* ++ * Increase service_from_backlogged before next statement, ++ * because the possible next invocation of ++ * bfq_bfqq_charge_time would likely inflate ++ * entity->service. In contrast, service_from_backlogged must ++ * contain real service, to enable the soft real-time ++ * heuristic to correctly compute the bandwidth consumed by ++ * bfqq. ++ */ ++ bfqq->service_from_backlogged += entity->service; ++ ++ /* ++ * As above explained, charge slow (typically seeky) and ++ * timed-out queues with the time and not the service ++ * received, to favor sequential workloads. ++ * ++ * Processes doing I/O in the slower disk zones will tend to ++ * be slow(er) even if not seeky. Therefore, since the ++ * estimated peak rate is actually an average over the disk ++ * surface, these processes may timeout just for bad luck. To ++ * avoid punishing them, do not charge time to processes that ++ * succeeded in consuming at least 2/3 of their budget. This ++ * allows BFQ to preserve enough elasticity to still perform ++ * bandwidth, and not time, distribution with little unlucky ++ * or quasi-sequential processes. ++ */ ++ if (bfqq->wr_coeff == 1 && ++ (slow || ++ (reason == BFQ_BFQQ_BUDGET_TIMEOUT && ++ bfq_bfqq_budget_left(bfqq) >= entity->budget / 3))) ++ bfq_bfqq_charge_time(bfqd, bfqq, delta); ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ if (reason == BFQ_BFQQ_TOO_IDLE && ++ entity->service <= 2 * entity->budget / 10) ++ bfq_clear_bfqq_IO_bound(bfqq); ++ ++ if (bfqd->low_latency && bfqq->wr_coeff == 1) ++ bfqq->last_wr_start_finish = jiffies; ++ ++ if (bfqd->low_latency && bfqd->bfq_wr_max_softrt_rate > 0 && ++ RB_EMPTY_ROOT(&bfqq->sort_list)) { ++ /* ++ * If we get here, and there are no outstanding ++ * requests, then the request pattern is isochronous ++ * (see the comments on the function ++ * bfq_bfqq_softrt_next_start()). Thus we can compute ++ * soft_rt_next_start. If, instead, the queue still ++ * has outstanding requests, then we have to wait for ++ * the completion of all the outstanding requests to ++ * discover whether the request pattern is actually ++ * isochronous. ++ */ ++ BUG_ON(bfqd->busy_queues < 1); ++ if (bfqq->dispatched == 0) { ++ bfqq->soft_rt_next_start = ++ bfq_bfqq_softrt_next_start(bfqd, bfqq); ++ bfq_log_bfqq(bfqd, bfqq, "new soft_rt_next %lu", ++ bfqq->soft_rt_next_start); ++ } else { ++ /* ++ * The application is still waiting for the ++ * completion of one or more requests: ++ * prevent it from possibly being incorrectly ++ * deemed as soft real-time by setting its ++ * soft_rt_next_start to infinity. In fact, ++ * without this assignment, the application ++ * would be incorrectly deemed as soft ++ * real-time if: ++ * 1) it issued a new request before the ++ * completion of all its in-flight ++ * requests, and ++ * 2) at that time, its soft_rt_next_start ++ * happened to be in the past. ++ */ ++ bfqq->soft_rt_next_start = ++ bfq_greatest_from_now(); ++ /* ++ * Schedule an update of soft_rt_next_start to when ++ * the task may be discovered to be isochronous. ++ */ ++ bfq_mark_bfqq_softrt_update(bfqq); ++ } ++ } ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "expire (%d, slow %d, num_disp %d, short_ttime %d, weight %d)", ++ reason, slow, bfqq->dispatched, ++ bfq_bfqq_has_short_ttime(bfqq), entity->weight); ++ ++ /* ++ * Increase, decrease or leave budget unchanged according to ++ * reason. ++ */ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ __bfq_bfqq_recalc_budget(bfqd, bfqq, reason); ++ BUG_ON(bfqq->next_rq == NULL && ++ bfqq->entity.budget < bfqq->entity.service); ++ ref = bfqq->ref; ++ __bfq_bfqq_expire(bfqd, bfqq); ++ ++ BUG_ON(ref > 1 && ++ !bfq_bfqq_busy(bfqq) && reason == BFQ_BFQQ_BUDGET_EXHAUSTED && ++ !bfq_class_idle(bfqq)); ++ ++ /* mark bfqq as waiting a request only if a bic still points to it */ ++ if (ref > 1 && !bfq_bfqq_busy(bfqq) && ++ reason != BFQ_BFQQ_BUDGET_TIMEOUT && ++ reason != BFQ_BFQQ_BUDGET_EXHAUSTED) ++ bfq_mark_bfqq_non_blocking_wait_rq(bfqq); ++} ++ ++/* ++ * Budget timeout is not implemented through a dedicated timer, but ++ * just checked on request arrivals and completions, as well as on ++ * idle timer expirations. ++ */ ++static bool bfq_bfqq_budget_timeout(struct bfq_queue *bfqq) ++{ ++ return time_is_before_eq_jiffies(bfqq->budget_timeout); ++} ++ ++/* ++ * If we expire a queue that is actively waiting (i.e., with the ++ * device idled) for the arrival of a new request, then we may incur ++ * the timestamp misalignment problem described in the body of the ++ * function __bfq_activate_entity. Hence we return true only if this ++ * condition does not hold, or if the queue is slow enough to deserve ++ * only to be kicked off for preserving a high throughput. ++ */ ++static bool bfq_may_expire_for_budg_timeout(struct bfq_queue *bfqq) ++{ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "may_budget_timeout: wait_request %d left %d timeout %d", ++ bfq_bfqq_wait_request(bfqq), ++ bfq_bfqq_budget_left(bfqq) >= bfqq->entity.budget / 3, ++ bfq_bfqq_budget_timeout(bfqq)); ++ ++ return (!bfq_bfqq_wait_request(bfqq) || ++ bfq_bfqq_budget_left(bfqq) >= bfqq->entity.budget / 3) ++ && ++ bfq_bfqq_budget_timeout(bfqq); ++} ++ ++/* ++ * For a queue that becomes empty, device idling is allowed only if ++ * this function returns true for that queue. As a consequence, since ++ * device idling plays a critical role for both throughput boosting ++ * and service guarantees, the return value of this function plays a ++ * critical role as well. ++ * ++ * In a nutshell, this function returns true only if idling is ++ * beneficial for throughput or, even if detrimental for throughput, ++ * idling is however necessary to preserve service guarantees (low ++ * latency, desired throughput distribution, ...). In particular, on ++ * NCQ-capable devices, this function tries to return false, so as to ++ * help keep the drives' internal queues full, whenever this helps the ++ * device boost the throughput without causing any service-guarantee ++ * issue. ++ * ++ * In more detail, the return value of this function is obtained by, ++ * first, computing a number of boolean variables that take into ++ * account throughput and service-guarantee issues, and, then, ++ * combining these variables in a logical expression. Most of the ++ * issues taken into account are not trivial. We discuss these issues ++ * while introducing the variables. ++ */ ++static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq) ++{ ++ struct bfq_data *bfqd = bfqq->bfqd; ++ bool rot_without_queueing = ++ !blk_queue_nonrot(bfqd->queue) && !bfqd->hw_tag, ++ bfqq_sequential_and_IO_bound, ++ idling_boosts_thr, idling_boosts_thr_without_issues, ++ idling_needed_for_service_guarantees, ++ asymmetric_scenario; ++ ++ if (bfqd->strict_guarantees) ++ return true; ++ ++ /* ++ * Idling is performed only if slice_idle > 0. In addition, we ++ * do not idle if ++ * (a) bfqq is async ++ * (b) bfqq is in the idle io prio class: in this case we do ++ * not idle because we want to minimize the bandwidth that ++ * queues in this class can steal to higher-priority queues ++ */ ++ if (bfqd->bfq_slice_idle == 0 || !bfq_bfqq_sync(bfqq) || ++ bfq_class_idle(bfqq)) ++ return false; ++ ++ bfqq_sequential_and_IO_bound = !BFQQ_SEEKY(bfqq) && ++ bfq_bfqq_IO_bound(bfqq) && bfq_bfqq_has_short_ttime(bfqq); ++ /* ++ * The next variable takes into account the cases where idling ++ * boosts the throughput. ++ * ++ * The value of the variable is computed considering, first, that ++ * idling is virtually always beneficial for the throughput if: ++ * (a) the device is not NCQ-capable and rotational, or ++ * (b) regardless of the presence of NCQ, the device is rotational and ++ * the request pattern for bfqq is I/O-bound and sequential, or ++ * (c) regardless of whether it is rotational, the device is ++ * not NCQ-capable and the request pattern for bfqq is ++ * I/O-bound and sequential. ++ * ++ * Secondly, and in contrast to the above item (b), idling an ++ * NCQ-capable flash-based device would not boost the ++ * throughput even with sequential I/O; rather it would lower ++ * the throughput in proportion to how fast the device ++ * is. Accordingly, the next variable is true if any of the ++ * above conditions (a), (b) or (c) is true, and, in ++ * particular, happens to be false if bfqd is an NCQ-capable ++ * flash-based device. ++ */ ++ idling_boosts_thr = rot_without_queueing || ++ ((!blk_queue_nonrot(bfqd->queue) || !bfqd->hw_tag) && ++ bfqq_sequential_and_IO_bound); ++ ++ /* ++ * The value of the next variable, ++ * idling_boosts_thr_without_issues, is equal to that of ++ * idling_boosts_thr, unless a special case holds. In this ++ * special case, described below, idling may cause problems to ++ * weight-raised queues. ++ * ++ * When the request pool is saturated (e.g., in the presence ++ * of write hogs), if the processes associated with ++ * non-weight-raised queues ask for requests at a lower rate, ++ * then processes associated with weight-raised queues have a ++ * higher probability to get a request from the pool ++ * immediately (or at least soon) when they need one. Thus ++ * they have a higher probability to actually get a fraction ++ * of the device throughput proportional to their high ++ * weight. This is especially true with NCQ-capable drives, ++ * which enqueue several requests in advance, and further ++ * reorder internally-queued requests. ++ * ++ * For this reason, we force to false the value of ++ * idling_boosts_thr_without_issues if there are weight-raised ++ * busy queues. In this case, and if bfqq is not weight-raised, ++ * this guarantees that the device is not idled for bfqq (if, ++ * instead, bfqq is weight-raised, then idling will be ++ * guaranteed by another variable, see below). Combined with ++ * the timestamping rules of BFQ (see [1] for details), this ++ * behavior causes bfqq, and hence any sync non-weight-raised ++ * queue, to get a lower number of requests served, and thus ++ * to ask for a lower number of requests from the request ++ * pool, before the busy weight-raised queues get served ++ * again. This often mitigates starvation problems in the ++ * presence of heavy write workloads and NCQ, thereby ++ * guaranteeing a higher application and system responsiveness ++ * in these hostile scenarios. ++ */ ++ idling_boosts_thr_without_issues = idling_boosts_thr && ++ bfqd->wr_busy_queues == 0; ++ ++ /* ++ * There is then a case where idling must be performed not ++ * for throughput concerns, but to preserve service ++ * guarantees. ++ * ++ * To introduce this case, we can note that allowing the drive ++ * to enqueue more than one request at a time, and hence ++ * delegating de facto final scheduling decisions to the ++ * drive's internal scheduler, entails loss of control on the ++ * actual request service order. In particular, the critical ++ * situation is when requests from different processes happen ++ * to be present, at the same time, in the internal queue(s) ++ * of the drive. In such a situation, the drive, by deciding ++ * the service order of the internally-queued requests, does ++ * determine also the actual throughput distribution among ++ * these processes. But the drive typically has no notion or ++ * concern about per-process throughput distribution, and ++ * makes its decisions only on a per-request basis. Therefore, ++ * the service distribution enforced by the drive's internal ++ * scheduler is likely to coincide with the desired ++ * device-throughput distribution only in a completely ++ * symmetric scenario where: ++ * (i) each of these processes must get the same throughput as ++ * the others; ++ * (ii) all these processes have the same I/O pattern ++ * (either sequential or random). ++ * In fact, in such a scenario, the drive will tend to treat ++ * the requests of each of these processes in about the same ++ * way as the requests of the others, and thus to provide ++ * each of these processes with about the same throughput ++ * (which is exactly the desired throughput distribution). In ++ * contrast, in any asymmetric scenario, device idling is ++ * certainly needed to guarantee that bfqq receives its ++ * assigned fraction of the device throughput (see [1] for ++ * details). ++ * ++ * We address this issue by controlling, actually, only the ++ * symmetry sub-condition (i), i.e., provided that ++ * sub-condition (i) holds, idling is not performed, ++ * regardless of whether sub-condition (ii) holds. In other ++ * words, only if sub-condition (i) holds, then idling is ++ * allowed, and the device tends to be prevented from queueing ++ * many requests, possibly of several processes. The reason ++ * for not controlling also sub-condition (ii) is that we ++ * exploit preemption to preserve guarantees in case of ++ * symmetric scenarios, even if (ii) does not hold, as ++ * explained in the next two paragraphs. ++ * ++ * Even if a queue, say Q, is expired when it remains idle, Q ++ * can still preempt the new in-service queue if the next ++ * request of Q arrives soon (see the comments on ++ * bfq_bfqq_update_budg_for_activation). If all queues and ++ * groups have the same weight, this form of preemption, ++ * combined with the hole-recovery heuristic described in the ++ * comments on function bfq_bfqq_update_budg_for_activation, ++ * are enough to preserve a correct bandwidth distribution in ++ * the mid term, even without idling. In fact, even if not ++ * idling allows the internal queues of the device to contain ++ * many requests, and thus to reorder requests, we can rather ++ * safely assume that the internal scheduler still preserves a ++ * minimum of mid-term fairness. The motivation for using ++ * preemption instead of idling is that, by not idling, ++ * service guarantees are preserved without minimally ++ * sacrificing throughput. In other words, both a high ++ * throughput and its desired distribution are obtained. ++ * ++ * More precisely, this preemption-based, idleless approach ++ * provides fairness in terms of IOPS, and not sectors per ++ * second. This can be seen with a simple example. Suppose ++ * that there are two queues with the same weight, but that ++ * the first queue receives requests of 8 sectors, while the ++ * second queue receives requests of 1024 sectors. In ++ * addition, suppose that each of the two queues contains at ++ * most one request at a time, which implies that each queue ++ * always remains idle after it is served. Finally, after ++ * remaining idle, each queue receives very quickly a new ++ * request. It follows that the two queues are served ++ * alternatively, preempting each other if needed. This ++ * implies that, although both queues have the same weight, ++ * the queue with large requests receives a service that is ++ * 1024/8 times as high as the service received by the other ++ * queue. ++ * ++ * On the other hand, device idling is performed, and thus ++ * pure sector-domain guarantees are provided, for the ++ * following queues, which are likely to need stronger ++ * throughput guarantees: weight-raised queues, and queues ++ * with a higher weight than other queues. When such queues ++ * are active, sub-condition (i) is false, which triggers ++ * device idling. ++ * ++ * According to the above considerations, the next variable is ++ * true (only) if sub-condition (i) holds. To compute the ++ * value of this variable, we not only use the return value of ++ * the function bfq_symmetric_scenario(), but also check ++ * whether bfqq is being weight-raised, because ++ * bfq_symmetric_scenario() does not take into account also ++ * weight-raised queues (see comments on ++ * bfq_weights_tree_add()). ++ * ++ * As a side note, it is worth considering that the above ++ * device-idling countermeasures may however fail in the ++ * following unlucky scenario: if idling is (correctly) ++ * disabled in a time period during which all symmetry ++ * sub-conditions hold, and hence the device is allowed to ++ * enqueue many requests, but at some later point in time some ++ * sub-condition stops to hold, then it may become impossible ++ * to let requests be served in the desired order until all ++ * the requests already queued in the device have been served. ++ */ ++ asymmetric_scenario = bfqq->wr_coeff > 1 || ++ !bfq_symmetric_scenario(bfqd); ++ ++ /* ++ * Finally, there is a case where maximizing throughput is the ++ * best choice even if it may cause unfairness toward ++ * bfqq. Such a case is when bfqq became active in a burst of ++ * queue activations. Queues that became active during a large ++ * burst benefit only from throughput, as discussed in the ++ * comments on bfq_handle_burst. Thus, if bfqq became active ++ * in a burst and not idling the device maximizes throughput, ++ * then the device must no be idled, because not idling the ++ * device provides bfqq and all other queues in the burst with ++ * maximum benefit. Combining this and the above case, we can ++ * now establish when idling is actually needed to preserve ++ * service guarantees. ++ */ ++ idling_needed_for_service_guarantees = ++ asymmetric_scenario && !bfq_bfqq_in_large_burst(bfqq); ++ ++ /* ++ * We have now all the components we need to compute the ++ * return value of the function, which is true only if idling ++ * either boosts the throughput (without issues), or is ++ * necessary to preserve service guarantees. ++ */ ++ bfq_log_bfqq(bfqd, bfqq, "may_idle: sync %d idling_boosts_thr %d", ++ bfq_bfqq_sync(bfqq), idling_boosts_thr); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "may_idle: wr_busy %d boosts %d IO-bound %d guar %d", ++ bfqd->wr_busy_queues, ++ idling_boosts_thr_without_issues, ++ bfq_bfqq_IO_bound(bfqq), ++ idling_needed_for_service_guarantees); ++ ++ return idling_boosts_thr_without_issues || ++ idling_needed_for_service_guarantees; ++} ++ ++/* ++ * If the in-service queue is empty but the function bfq_bfqq_may_idle ++ * returns true, then: ++ * 1) the queue must remain in service and cannot be expired, and ++ * 2) the device must be idled to wait for the possible arrival of a new ++ * request for the queue. ++ * See the comments on the function bfq_bfqq_may_idle for the reasons ++ * why performing device idling is the best choice to boost the throughput ++ * and preserve service guarantees when bfq_bfqq_may_idle itself ++ * returns true. ++ */ ++static bool bfq_bfqq_must_idle(struct bfq_queue *bfqq) ++{ ++ return RB_EMPTY_ROOT(&bfqq->sort_list) && bfq_bfqq_may_idle(bfqq); ++} ++ ++/* ++ * Select a queue for service. If we have a current queue in service, ++ * check whether to continue servicing it, or retrieve and set a new one. ++ */ ++static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq; ++ struct request *next_rq; ++ enum bfqq_expiration reason = BFQ_BFQQ_BUDGET_TIMEOUT; ++ ++ bfqq = bfqd->in_service_queue; ++ if (!bfqq) ++ goto new_queue; ++ ++ bfq_log_bfqq(bfqd, bfqq, "select_queue: already in-service queue"); ++ ++ if (bfq_may_expire_for_budg_timeout(bfqq) && ++ !hrtimer_active(&bfqd->idle_slice_timer) && ++ !bfq_bfqq_must_idle(bfqq)) ++ goto expire; ++ ++check_queue: ++ /* ++ * This loop is rarely executed more than once. Even when it ++ * happens, it is much more convenient to re-execute this loop ++ * than to return NULL and trigger a new dispatch to get a ++ * request served. ++ */ ++ next_rq = bfqq->next_rq; ++ /* ++ * If bfqq has requests queued and it has enough budget left to ++ * serve them, keep the queue, otherwise expire it. ++ */ ++ if (next_rq) { ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list)); ++ ++ if (bfq_serv_to_charge(next_rq, bfqq) > ++ bfq_bfqq_budget_left(bfqq)) { ++ /* ++ * Expire the queue for budget exhaustion, ++ * which makes sure that the next budget is ++ * enough to serve the next request, even if ++ * it comes from the fifo expired path. ++ */ ++ reason = BFQ_BFQQ_BUDGET_EXHAUSTED; ++ goto expire; ++ } else { ++ /* ++ * The idle timer may be pending because we may ++ * not disable disk idling even when a new request ++ * arrives. ++ */ ++ if (bfq_bfqq_wait_request(bfqq)) { ++ BUG_ON(!hrtimer_active(&bfqd->idle_slice_timer)); ++ /* ++ * If we get here: 1) at least a new request ++ * has arrived but we have not disabled the ++ * timer because the request was too small, ++ * 2) then the block layer has unplugged ++ * the device, causing the dispatch to be ++ * invoked. ++ * ++ * Since the device is unplugged, now the ++ * requests are probably large enough to ++ * provide a reasonable throughput. ++ * So we disable idling. ++ */ ++ bfq_clear_bfqq_wait_request(bfqq); ++ hrtimer_try_to_cancel(&bfqd->idle_slice_timer); ++ bfqg_stats_update_idle_time(bfqq_group(bfqq)); ++ } ++ goto keep_queue; ++ } ++ } ++ ++ /* ++ * No requests pending. However, if the in-service queue is idling ++ * for a new request, or has requests waiting for a completion and ++ * may idle after their completion, then keep it anyway. ++ */ ++ if (hrtimer_active(&bfqd->idle_slice_timer) || ++ (bfqq->dispatched != 0 && bfq_bfqq_may_idle(bfqq))) { ++ bfqq = NULL; ++ goto keep_queue; ++ } ++ ++ reason = BFQ_BFQQ_NO_MORE_REQUESTS; ++expire: ++ bfq_bfqq_expire(bfqd, bfqq, false, reason); ++new_queue: ++ bfqq = bfq_set_in_service_queue(bfqd); ++ if (bfqq) { ++ bfq_log_bfqq(bfqd, bfqq, "select_queue: checking new queue"); ++ goto check_queue; ++ } ++keep_queue: ++ if (bfqq) ++ bfq_log_bfqq(bfqd, bfqq, "select_queue: returned this queue"); ++ else ++ bfq_log(bfqd, "select_queue: no queue returned"); ++ ++ return bfqq; ++} ++ ++static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *entity = &bfqq->entity; ++ ++ if (bfqq->wr_coeff > 1) { /* queue is being weight-raised */ ++ BUG_ON(bfqq->wr_cur_max_time == bfqd->bfq_wr_rt_max_time && ++ time_is_after_jiffies(bfqq->last_wr_start_finish)); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "raising period dur %u/%u msec, old coeff %u, w %d(%d)", ++ jiffies_to_msecs(jiffies - bfqq->last_wr_start_finish), ++ jiffies_to_msecs(bfqq->wr_cur_max_time), ++ bfqq->wr_coeff, ++ bfqq->entity.weight, bfqq->entity.orig_weight); ++ ++ BUG_ON(bfqq != bfqd->in_service_queue && entity->weight != ++ entity->orig_weight * bfqq->wr_coeff); ++ if (entity->prio_changed) ++ bfq_log_bfqq(bfqd, bfqq, "WARN: pending prio change"); ++ ++ /* ++ * If the queue was activated in a burst, or too much ++ * time has elapsed from the beginning of this ++ * weight-raising period, then end weight raising. ++ */ ++ if (bfq_bfqq_in_large_burst(bfqq)) ++ bfq_bfqq_end_wr(bfqq); ++ else if (time_is_before_jiffies(bfqq->last_wr_start_finish + ++ bfqq->wr_cur_max_time)) { ++ if (bfqq->wr_cur_max_time != bfqd->bfq_wr_rt_max_time || ++ time_is_before_jiffies(bfqq->wr_start_at_switch_to_srt + ++ bfq_wr_duration(bfqd))) ++ bfq_bfqq_end_wr(bfqq); ++ else { ++ /* switch back to interactive wr */ ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ bfqq->last_wr_start_finish = ++ bfqq->wr_start_at_switch_to_srt; ++ BUG_ON(time_is_after_jiffies( ++ bfqq->last_wr_start_finish)); ++ bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqd, bfqq, ++ "back to interactive wr"); ++ } ++ } ++ } ++ /* ++ * To improve latency (for this or other queues), immediately ++ * update weight both if it must be raised and if it must be ++ * lowered. Since, entity may be on some active tree here, and ++ * might have a pending change of its ioprio class, invoke ++ * next function with the last parameter unset (see the ++ * comments on the function). ++ */ ++ if ((entity->weight > entity->orig_weight) != (bfqq->wr_coeff > 1)) ++ __bfq_entity_update_weight_prio(bfq_entity_service_tree(entity), ++ entity, false); ++} ++ ++/* ++ * Dispatch one request from bfqq, moving it to the request queue ++ * dispatch list. ++ */ ++static int bfq_dispatch_request(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) ++{ ++ int dispatched = 0; ++ struct request *rq = bfqq->next_rq; ++ unsigned long service_to_charge; ++ ++ BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list)); ++ BUG_ON(!rq); ++ service_to_charge = bfq_serv_to_charge(rq, bfqq); ++ ++ BUG_ON(service_to_charge > bfq_bfqq_budget_left(bfqq)); ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ bfq_bfqq_served(bfqq, service_to_charge); ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ bfq_dispatch_insert(bfqd->queue, rq); ++ ++ /* ++ * If weight raising has to terminate for bfqq, then next ++ * function causes an immediate update of bfqq's weight, ++ * without waiting for next activation. As a consequence, on ++ * expiration, bfqq will be timestamped as if has never been ++ * weight-raised during this service slot, even if it has ++ * received part or even most of the service as a ++ * weight-raised queue. This inflates bfqq's timestamps, which ++ * is beneficial, as bfqq is then more willing to leave the ++ * device immediately to possible other weight-raised queues. ++ */ ++ bfq_update_wr_data(bfqd, bfqq); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "dispatched %u sec req (%llu), budg left %d", ++ blk_rq_sectors(rq), ++ (unsigned long long) blk_rq_pos(rq), ++ bfq_bfqq_budget_left(bfqq)); ++ ++ dispatched++; ++ ++ if (!bfqd->in_service_bic) { ++ atomic_long_inc(&RQ_BIC(rq)->icq.ioc->refcount); ++ bfqd->in_service_bic = RQ_BIC(rq); ++ } ++ ++ if (bfqd->busy_queues > 1 && bfq_class_idle(bfqq)) ++ goto expire; ++ ++ return dispatched; ++ ++expire: ++ bfq_bfqq_expire(bfqd, bfqq, false, BFQ_BFQQ_BUDGET_EXHAUSTED); ++ return dispatched; ++} ++ ++static int __bfq_forced_dispatch_bfqq(struct bfq_queue *bfqq) ++{ ++ int dispatched = 0; ++ ++ while (bfqq->next_rq) { ++ bfq_dispatch_insert(bfqq->bfqd->queue, bfqq->next_rq); ++ dispatched++; ++ } ++ ++ BUG_ON(!list_empty(&bfqq->fifo)); ++ return dispatched; ++} ++ ++/* ++ * Drain our current requests. ++ * Used for barriers and when switching io schedulers on-the-fly. ++ */ ++static int bfq_forced_dispatch(struct bfq_data *bfqd) ++{ ++ struct bfq_queue *bfqq, *n; ++ struct bfq_service_tree *st; ++ int dispatched = 0; ++ ++ bfqq = bfqd->in_service_queue; ++ if (bfqq) ++ __bfq_bfqq_expire(bfqd, bfqq); ++ ++ /* ++ * Loop through classes, and be careful to leave the scheduler ++ * in a consistent state, as feedback mechanisms and vtime ++ * updates cannot be disabled during the process. ++ */ ++ list_for_each_entry_safe(bfqq, n, &bfqd->active_list, bfqq_list) { ++ st = bfq_entity_service_tree(&bfqq->entity); ++ ++ dispatched += __bfq_forced_dispatch_bfqq(bfqq); ++ ++ bfqq->max_budget = bfq_max_budget(bfqd); ++ bfq_forget_idle(st); ++ } ++ ++ BUG_ON(bfqd->busy_queues != 0); ++ ++ return dispatched; ++} ++ ++static int bfq_dispatch_requests(struct request_queue *q, int force) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct bfq_queue *bfqq; ++ ++ bfq_log(bfqd, "dispatch requests: %d busy queues", bfqd->busy_queues); ++ ++ if (bfqd->busy_queues == 0) ++ return 0; ++ ++ if (unlikely(force)) ++ return bfq_forced_dispatch(bfqd); ++ ++ /* ++ * Force device to serve one request at a time if ++ * strict_guarantees is true. Forcing this service scheme is ++ * currently the ONLY way to guarantee that the request ++ * service order enforced by the scheduler is respected by a ++ * queueing device. Otherwise the device is free even to make ++ * some unlucky request wait for as long as the device ++ * wishes. ++ * ++ * Of course, serving one request at at time may cause loss of ++ * throughput. ++ */ ++ if (bfqd->strict_guarantees && bfqd->rq_in_driver > 0) ++ return 0; ++ ++ bfqq = bfq_select_queue(bfqd); ++ if (!bfqq) ++ return 0; ++ ++ BUG_ON(bfqq->entity.budget < bfqq->entity.service); ++ ++ BUG_ON(bfq_bfqq_wait_request(bfqq)); ++ ++ if (!bfq_dispatch_request(bfqd, bfqq)) ++ return 0; ++ ++ bfq_log_bfqq(bfqd, bfqq, "dispatched %s request", ++ bfq_bfqq_sync(bfqq) ? "sync" : "async"); ++ ++ BUG_ON(bfqq->next_rq == NULL && ++ bfqq->entity.budget < bfqq->entity.service); ++ return 1; ++} ++ ++/* ++ * Task holds one reference to the queue, dropped when task exits. Each rq ++ * in-flight on this queue also holds a reference, dropped when rq is freed. ++ * ++ * Queue lock must be held here. Recall not to use bfqq after calling ++ * this function on it. ++ */ ++static void bfq_put_queue(struct bfq_queue *bfqq) ++{ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ struct bfq_group *bfqg = bfqq_group(bfqq); ++#endif ++ ++ BUG_ON(bfqq->ref <= 0); ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p %d", bfqq, bfqq->ref); ++ bfqq->ref--; ++ if (bfqq->ref) ++ return; ++ ++ BUG_ON(rb_first(&bfqq->sort_list)); ++ BUG_ON(bfqq->allocated[READ] + bfqq->allocated[WRITE] != 0); ++ BUG_ON(bfqq->entity.tree); ++ BUG_ON(bfq_bfqq_busy(bfqq)); ++ ++ if (bfq_bfqq_sync(bfqq)) ++ /* ++ * The fact that this queue is being destroyed does not ++ * invalidate the fact that this queue may have been ++ * activated during the current burst. As a consequence, ++ * although the queue does not exist anymore, and hence ++ * needs to be removed from the burst list if there, ++ * the burst size has not to be decremented. ++ */ ++ hlist_del_init(&bfqq->burst_list_node); ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); ++ ++ kmem_cache_free(bfq_pool, bfqq); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ bfqg_put(bfqg); ++#endif ++} ++ ++static void bfq_put_cooperator(struct bfq_queue *bfqq) ++{ ++ struct bfq_queue *__bfqq, *next; ++ ++ /* ++ * If this queue was scheduled to merge with another queue, be ++ * sure to drop the reference taken on that queue (and others in ++ * the merge chain). See bfq_setup_merge and bfq_merge_bfqqs. ++ */ ++ __bfqq = bfqq->new_bfqq; ++ while (__bfqq) { ++ if (__bfqq == bfqq) ++ break; ++ next = __bfqq->new_bfqq; ++ bfq_put_queue(__bfqq); ++ __bfqq = next; ++ } ++} ++ ++static void bfq_exit_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++{ ++ if (bfqq == bfqd->in_service_queue) { ++ __bfq_bfqq_expire(bfqd, bfqq); ++ bfq_schedule_dispatch(bfqd); ++ } ++ ++ bfq_log_bfqq(bfqd, bfqq, "exit_bfqq: %p, %d", bfqq, bfqq->ref); ++ ++ bfq_put_cooperator(bfqq); ++ ++ bfq_put_queue(bfqq); /* release process reference */ ++} ++ ++static void bfq_init_icq(struct io_cq *icq) ++{ ++ icq_to_bic(icq)->ttime.last_end_request = ktime_get_ns() - (1ULL<<32); ++} ++ ++static void bfq_exit_icq(struct io_cq *icq) ++{ ++ struct bfq_io_cq *bic = icq_to_bic(icq); ++ struct bfq_data *bfqd = bic_to_bfqd(bic); ++ ++ if (bic_to_bfqq(bic, false)) { ++ bfq_exit_bfqq(bfqd, bic_to_bfqq(bic, false)); ++ bic_set_bfqq(bic, NULL, false); ++ } ++ ++ if (bic_to_bfqq(bic, true)) { ++ /* ++ * If the bic is using a shared queue, put the reference ++ * taken on the io_context when the bic started using a ++ * shared bfq_queue. ++ */ ++ if (bfq_bfqq_coop(bic_to_bfqq(bic, true))) ++ put_io_context(icq->ioc); ++ bfq_exit_bfqq(bfqd, bic_to_bfqq(bic, true)); ++ bic_set_bfqq(bic, NULL, true); ++ } ++} ++ ++/* ++ * Update the entity prio values; note that the new values will not ++ * be used until the next (re)activation. ++ */ ++static void bfq_set_next_ioprio_data(struct bfq_queue *bfqq, ++ struct bfq_io_cq *bic) ++{ ++ struct task_struct *tsk = current; ++ int ioprio_class; ++ ++ ioprio_class = IOPRIO_PRIO_CLASS(bic->ioprio); ++ switch (ioprio_class) { ++ default: ++ dev_err(bfqq->bfqd->queue->backing_dev_info->dev, ++ "bfq: bad prio class %d\n", ioprio_class); ++ case IOPRIO_CLASS_NONE: ++ /* ++ * No prio set, inherit CPU scheduling settings. ++ */ ++ bfqq->new_ioprio = task_nice_ioprio(tsk); ++ bfqq->new_ioprio_class = task_nice_ioclass(tsk); ++ break; ++ case IOPRIO_CLASS_RT: ++ bfqq->new_ioprio = IOPRIO_PRIO_DATA(bic->ioprio); ++ bfqq->new_ioprio_class = IOPRIO_CLASS_RT; ++ break; ++ case IOPRIO_CLASS_BE: ++ bfqq->new_ioprio = IOPRIO_PRIO_DATA(bic->ioprio); ++ bfqq->new_ioprio_class = IOPRIO_CLASS_BE; ++ break; ++ case IOPRIO_CLASS_IDLE: ++ bfqq->new_ioprio_class = IOPRIO_CLASS_IDLE; ++ bfqq->new_ioprio = 7; ++ break; ++ } ++ ++ if (bfqq->new_ioprio >= IOPRIO_BE_NR) { ++ pr_crit("bfq_set_next_ioprio_data: new_ioprio %d\n", ++ bfqq->new_ioprio); ++ BUG(); ++ } ++ ++ bfqq->entity.new_weight = bfq_ioprio_to_weight(bfqq->new_ioprio); ++ bfqq->entity.prio_changed = 1; ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "set_next_ioprio_data: bic_class %d prio %d class %d", ++ ioprio_class, bfqq->new_ioprio, bfqq->new_ioprio_class); ++} ++ ++static void bfq_check_ioprio_change(struct bfq_io_cq *bic, struct bio *bio) ++{ ++ struct bfq_data *bfqd = bic_to_bfqd(bic); ++ struct bfq_queue *bfqq; ++ unsigned long uninitialized_var(flags); ++ int ioprio = bic->icq.ioc->ioprio; ++ ++ /* ++ * This condition may trigger on a newly created bic, be sure to ++ * drop the lock before returning. ++ */ ++ if (unlikely(!bfqd) || likely(bic->ioprio == ioprio)) ++ return; ++ ++ bic->ioprio = ioprio; ++ ++ bfqq = bic_to_bfqq(bic, false); ++ if (bfqq) { ++ /* release process reference on this queue */ ++ bfq_put_queue(bfqq); ++ bfqq = bfq_get_queue(bfqd, bio, BLK_RW_ASYNC, bic); ++ bic_set_bfqq(bic, bfqq, false); ++ bfq_log_bfqq(bfqd, bfqq, ++ "check_ioprio_change: bfqq %p %d", ++ bfqq, bfqq->ref); ++ } ++ ++ bfqq = bic_to_bfqq(bic, true); ++ if (bfqq) ++ bfq_set_next_ioprio_data(bfqq, bic); ++} ++ ++static void bfq_init_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ struct bfq_io_cq *bic, pid_t pid, int is_sync) ++{ ++ RB_CLEAR_NODE(&bfqq->entity.rb_node); ++ INIT_LIST_HEAD(&bfqq->fifo); ++ INIT_HLIST_NODE(&bfqq->burst_list_node); ++ BUG_ON(!hlist_unhashed(&bfqq->burst_list_node)); ++ ++ bfqq->ref = 0; ++ bfqq->bfqd = bfqd; ++ ++ if (bic) ++ bfq_set_next_ioprio_data(bfqq, bic); ++ ++ if (is_sync) { ++ /* ++ * No need to mark as has_short_ttime if in ++ * idle_class, because no device idling is performed ++ * for queues in idle class ++ */ ++ if (!bfq_class_idle(bfqq)) ++ /* tentatively mark as has_short_ttime */ ++ bfq_mark_bfqq_has_short_ttime(bfqq); ++ bfq_mark_bfqq_sync(bfqq); ++ bfq_mark_bfqq_just_created(bfqq); ++ } else ++ bfq_clear_bfqq_sync(bfqq); ++ bfq_mark_bfqq_IO_bound(bfqq); ++ ++ /* Tentative initial value to trade off between thr and lat */ ++ bfqq->max_budget = (2 * bfq_max_budget(bfqd)) / 3; ++ bfqq->pid = pid; ++ ++ bfqq->wr_coeff = 1; ++ bfqq->last_wr_start_finish = jiffies; ++ bfqq->wr_start_at_switch_to_srt = bfq_smallest_from_now(); ++ bfqq->budget_timeout = bfq_smallest_from_now(); ++ bfqq->split_time = bfq_smallest_from_now(); ++ ++ /* ++ * Set to the value for which bfqq will not be deemed as ++ * soft rt when it becomes backlogged. ++ */ ++ bfqq->soft_rt_next_start = bfq_greatest_from_now(); ++ ++ /* first request is almost certainly seeky */ ++ bfqq->seek_history = 1; ++} ++ ++static struct bfq_queue **bfq_async_queue_prio(struct bfq_data *bfqd, ++ struct bfq_group *bfqg, ++ int ioprio_class, int ioprio) ++{ ++ switch (ioprio_class) { ++ case IOPRIO_CLASS_RT: ++ return &bfqg->async_bfqq[0][ioprio]; ++ case IOPRIO_CLASS_NONE: ++ ioprio = IOPRIO_NORM; ++ /* fall through */ ++ case IOPRIO_CLASS_BE: ++ return &bfqg->async_bfqq[1][ioprio]; ++ case IOPRIO_CLASS_IDLE: ++ return &bfqg->async_idle_bfqq; ++ default: ++ BUG(); ++ } ++} ++ ++static struct bfq_queue *bfq_get_queue(struct bfq_data *bfqd, ++ struct bio *bio, bool is_sync, ++ struct bfq_io_cq *bic) ++{ ++ const int ioprio = IOPRIO_PRIO_DATA(bic->ioprio); ++ const int ioprio_class = IOPRIO_PRIO_CLASS(bic->ioprio); ++ struct bfq_queue **async_bfqq = NULL; ++ struct bfq_queue *bfqq; ++ struct bfq_group *bfqg; ++ ++ rcu_read_lock(); ++ ++ bfqg = bfq_find_set_group(bfqd, bio_blkcg(bio)); ++ if (!bfqg) { ++ bfqq = &bfqd->oom_bfqq; ++ goto out; ++ } ++ ++ if (!is_sync) { ++ async_bfqq = bfq_async_queue_prio(bfqd, bfqg, ioprio_class, ++ ioprio); ++ bfqq = *async_bfqq; ++ if (bfqq) ++ goto out; ++ } ++ ++ bfqq = kmem_cache_alloc_node(bfq_pool, ++ GFP_NOWAIT | __GFP_ZERO | __GFP_NOWARN, ++ bfqd->queue->node); ++ ++ if (bfqq) { ++ bfq_init_bfqq(bfqd, bfqq, bic, current->pid, ++ is_sync); ++ bfq_init_entity(&bfqq->entity, bfqg); ++ bfq_log_bfqq(bfqd, bfqq, "allocated"); ++ } else { ++ bfqq = &bfqd->oom_bfqq; ++ bfq_log_bfqq(bfqd, bfqq, "using oom bfqq"); ++ goto out; ++ } ++ ++ /* ++ * Pin the queue now that it's allocated, scheduler exit will ++ * prune it. ++ */ ++ if (async_bfqq) { ++ bfqq->ref++; /* ++ * Extra group reference, w.r.t. sync ++ * queue. This extra reference is removed ++ * only if bfqq->bfqg disappears, to ++ * guarantee that this queue is not freed ++ * until its group goes away. ++ */ ++ bfq_log_bfqq(bfqd, bfqq, "get_queue, bfqq not in async: %p, %d", ++ bfqq, bfqq->ref); ++ *async_bfqq = bfqq; ++ } ++ ++out: ++ bfqq->ref++; /* get a process reference to this queue */ ++ bfq_log_bfqq(bfqd, bfqq, "get_queue, at end: %p, %d", bfqq, bfqq->ref); ++ rcu_read_unlock(); ++ return bfqq; ++} ++ ++static void bfq_update_io_thinktime(struct bfq_data *bfqd, ++ struct bfq_io_cq *bic) ++{ ++ struct bfq_ttime *ttime = &bic->ttime; ++ u64 elapsed = ktime_get_ns() - bic->ttime.last_end_request; ++ ++ elapsed = min_t(u64, elapsed, 2 * bfqd->bfq_slice_idle); ++ ++ ttime->ttime_samples = (7*bic->ttime.ttime_samples + 256) / 8; ++ ttime->ttime_total = div_u64(7*ttime->ttime_total + 256*elapsed, 8); ++ ttime->ttime_mean = div64_ul(ttime->ttime_total + 128, ++ ttime->ttime_samples); ++} ++ ++static void ++bfq_update_io_seektime(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ struct request *rq) ++{ ++ bfqq->seek_history <<= 1; ++ bfqq->seek_history |= ++ get_sdist(bfqq->last_request_pos, rq) > BFQQ_SEEK_THR && ++ (!blk_queue_nonrot(bfqd->queue) || ++ blk_rq_sectors(rq) < BFQQ_SECT_THR_NONROT); ++} ++ ++static void bfq_update_has_short_ttime(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq, ++ struct bfq_io_cq *bic) ++{ ++ bool has_short_ttime = true; ++ ++ /* ++ * No need to update has_short_ttime if bfqq is async or in ++ * idle io prio class, or if bfq_slice_idle is zero, because ++ * no device idling is performed for bfqq in this case. ++ */ ++ if (!bfq_bfqq_sync(bfqq) || bfq_class_idle(bfqq) || ++ bfqd->bfq_slice_idle == 0) ++ return; ++ ++ /* Idle window just restored, statistics are meaningless. */ ++ if (time_is_after_eq_jiffies(bfqq->split_time + ++ bfqd->bfq_wr_min_idle_time)) ++ return; ++ ++ /* Think time is infinite if no process is linked to ++ * bfqq. Otherwise check average think time to ++ * decide whether to mark as has_short_ttime ++ */ ++ if (atomic_read(&bic->icq.ioc->active_ref) == 0 || ++ (bfq_sample_valid(bic->ttime.ttime_samples) && ++ bic->ttime.ttime_mean > bfqd->bfq_slice_idle)) ++ has_short_ttime = false; ++ ++ bfq_log_bfqq(bfqd, bfqq, "update_has_short_ttime: has_short_ttime %d", ++ has_short_ttime); ++ ++ if (has_short_ttime) ++ bfq_mark_bfqq_has_short_ttime(bfqq); ++ else ++ bfq_clear_bfqq_has_short_ttime(bfqq); ++} ++ ++/* ++ * Called when a new fs request (rq) is added to bfqq. Check if there's ++ * something we should do about it. ++ */ ++static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ struct request *rq) ++{ ++ struct bfq_io_cq *bic = RQ_BIC(rq); ++ ++ if (rq->cmd_flags & REQ_META) ++ bfqq->meta_pending++; ++ ++ bfq_update_io_thinktime(bfqd, bic); ++ bfq_update_has_short_ttime(bfqd, bfqq, bic); ++ bfq_update_io_seektime(bfqd, bfqq, rq); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "rq_enqueued: has_short_ttime=%d (seeky %d)", ++ bfq_bfqq_has_short_ttime(bfqq), BFQQ_SEEKY(bfqq)); ++ ++ bfqq->last_request_pos = blk_rq_pos(rq) + blk_rq_sectors(rq); ++ ++ if (bfqq == bfqd->in_service_queue && bfq_bfqq_wait_request(bfqq)) { ++ bool small_req = bfqq->queued[rq_is_sync(rq)] == 1 && ++ blk_rq_sectors(rq) < 32; ++ bool budget_timeout = bfq_bfqq_budget_timeout(bfqq); ++ ++ /* ++ * There is just this request queued: if the request ++ * is small and the queue is not to be expired, then ++ * just exit. ++ * ++ * In this way, if the device is being idled to wait ++ * for a new request from the in-service queue, we ++ * avoid unplugging the device and committing the ++ * device to serve just a small request. On the ++ * contrary, we wait for the block layer to decide ++ * when to unplug the device: hopefully, new requests ++ * will be merged to this one quickly, then the device ++ * will be unplugged and larger requests will be ++ * dispatched. ++ */ ++ if (small_req && !budget_timeout) ++ return; ++ ++ /* ++ * A large enough request arrived, or the queue is to ++ * be expired: in both cases disk idling is to be ++ * stopped, so clear wait_request flag and reset ++ * timer. ++ */ ++ bfq_clear_bfqq_wait_request(bfqq); ++ hrtimer_try_to_cancel(&bfqd->idle_slice_timer); ++ bfqg_stats_update_idle_time(bfqq_group(bfqq)); ++ ++ /* ++ * The queue is not empty, because a new request just ++ * arrived. Hence we can safely expire the queue, in ++ * case of budget timeout, without risking that the ++ * timestamps of the queue are not updated correctly. ++ * See [1] for more details. ++ */ ++ if (budget_timeout) ++ bfq_bfqq_expire(bfqd, bfqq, false, ++ BFQ_BFQQ_BUDGET_TIMEOUT); ++ ++ /* ++ * Let the request rip immediately, or let a new queue be ++ * selected if bfqq has just been expired. ++ */ ++ __blk_run_queue(bfqd->queue); ++ } ++} ++ ++static void bfq_insert_request(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct bfq_queue *bfqq = RQ_BFQQ(rq), *new_bfqq; ++ ++ assert_spin_locked(bfqd->queue->queue_lock); ++ ++ /* ++ * An unplug may trigger a requeue of a request from the device ++ * driver: make sure we are in process context while trying to ++ * merge two bfq_queues. ++ */ ++ if (!in_interrupt()) { ++ new_bfqq = bfq_setup_cooperator(bfqd, bfqq, rq, true); ++ if (new_bfqq) { ++ if (bic_to_bfqq(RQ_BIC(rq), 1) != bfqq) ++ new_bfqq = bic_to_bfqq(RQ_BIC(rq), 1); ++ /* ++ * Release the request's reference to the old bfqq ++ * and make sure one is taken to the shared queue. ++ */ ++ new_bfqq->allocated[rq_data_dir(rq)]++; ++ bfqq->allocated[rq_data_dir(rq)]--; ++ new_bfqq->ref++; ++ bfq_clear_bfqq_just_created(bfqq); ++ if (bic_to_bfqq(RQ_BIC(rq), 1) == bfqq) ++ bfq_merge_bfqqs(bfqd, RQ_BIC(rq), ++ bfqq, new_bfqq); ++ /* ++ * rq is about to be enqueued into new_bfqq, ++ * release rq reference on bfqq ++ */ ++ bfq_put_queue(bfqq); ++ rq->elv.priv[1] = new_bfqq; ++ bfqq = new_bfqq; ++ } ++ } ++ ++ bfq_add_request(rq); ++ ++ rq->fifo_time = ktime_get_ns() + bfqd->bfq_fifo_expire[rq_is_sync(rq)]; ++ list_add_tail(&rq->queuelist, &bfqq->fifo); ++ ++ bfq_rq_enqueued(bfqd, bfqq, rq); ++} ++ ++static void bfq_update_hw_tag(struct bfq_data *bfqd) ++{ ++ bfqd->max_rq_in_driver = max_t(int, bfqd->max_rq_in_driver, ++ bfqd->rq_in_driver); ++ ++ if (bfqd->hw_tag == 1) ++ return; ++ ++ /* ++ * This sample is valid if the number of outstanding requests ++ * is large enough to allow a queueing behavior. Note that the ++ * sum is not exact, as it's not taking into account deactivated ++ * requests. ++ */ ++ if (bfqd->rq_in_driver + bfqd->queued < BFQ_HW_QUEUE_THRESHOLD) ++ return; ++ ++ if (bfqd->hw_tag_samples++ < BFQ_HW_QUEUE_SAMPLES) ++ return; ++ ++ bfqd->hw_tag = bfqd->max_rq_in_driver > BFQ_HW_QUEUE_THRESHOLD; ++ bfqd->max_rq_in_driver = 0; ++ bfqd->hw_tag_samples = 0; ++} ++ ++static void bfq_completed_request(struct request_queue *q, struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ struct bfq_data *bfqd = bfqq->bfqd; ++ u64 now_ns; ++ u32 delta_us; ++ ++ bfq_log_bfqq(bfqd, bfqq, "completed one req with %u sects left", ++ blk_rq_sectors(rq)); ++ ++ assert_spin_locked(bfqd->queue->queue_lock); ++ bfq_update_hw_tag(bfqd); ++ ++ BUG_ON(!bfqd->rq_in_driver); ++ BUG_ON(!bfqq->dispatched); ++ bfqd->rq_in_driver--; ++ bfqq->dispatched--; ++ bfqg_stats_update_completion(bfqq_group(bfqq), ++ rq_start_time_ns(rq), ++ rq_io_start_time_ns(rq), ++ rq->cmd_flags); ++ ++ if (!bfqq->dispatched && !bfq_bfqq_busy(bfqq)) { ++ BUG_ON(!RB_EMPTY_ROOT(&bfqq->sort_list)); ++ /* ++ * Set budget_timeout (which we overload to store the ++ * time at which the queue remains with no backlog and ++ * no outstanding request; used by the weight-raising ++ * mechanism). ++ */ ++ bfqq->budget_timeout = jiffies; ++ ++ bfq_weights_tree_remove(bfqd, &bfqq->entity, ++ &bfqd->queue_weights_tree); ++ } ++ ++ now_ns = ktime_get_ns(); ++ ++ RQ_BIC(rq)->ttime.last_end_request = now_ns; ++ ++ /* ++ * Using us instead of ns, to get a reasonable precision in ++ * computing rate in next check. ++ */ ++ delta_us = div_u64(now_ns - bfqd->last_completion, NSEC_PER_USEC); ++ ++ bfq_log(bfqd, "rq_completed: delta %uus/%luus max_size %u rate %llu/%llu", ++ delta_us, BFQ_MIN_TT/NSEC_PER_USEC, bfqd->last_rq_max_size, ++ (USEC_PER_SEC* ++ (u64)((bfqd->last_rq_max_size<>BFQ_RATE_SHIFT, ++ (USEC_PER_SEC*(u64)(1UL<<(BFQ_RATE_SHIFT-10)))>>BFQ_RATE_SHIFT); ++ ++ /* ++ * If the request took rather long to complete, and, according ++ * to the maximum request size recorded, this completion latency ++ * implies that the request was certainly served at a very low ++ * rate (less than 1M sectors/sec), then the whole observation ++ * interval that lasts up to this time instant cannot be a ++ * valid time interval for computing a new peak rate. Invoke ++ * bfq_update_rate_reset to have the following three steps ++ * taken: ++ * - close the observation interval at the last (previous) ++ * request dispatch or completion ++ * - compute rate, if possible, for that observation interval ++ * - reset to zero samples, which will trigger a proper ++ * re-initialization of the observation interval on next ++ * dispatch ++ */ ++ if (delta_us > BFQ_MIN_TT/NSEC_PER_USEC && ++ (bfqd->last_rq_max_size<last_completion = now_ns; ++ ++ /* ++ * If we are waiting to discover whether the request pattern ++ * of the task associated with the queue is actually ++ * isochronous, and both requisites for this condition to hold ++ * are now satisfied, then compute soft_rt_next_start (see the ++ * comments on the function bfq_bfqq_softrt_next_start()). We ++ * schedule this delayed check when bfqq expires, if it still ++ * has in-flight requests. ++ */ ++ if (bfq_bfqq_softrt_update(bfqq) && bfqq->dispatched == 0 && ++ RB_EMPTY_ROOT(&bfqq->sort_list)) ++ bfqq->soft_rt_next_start = ++ bfq_bfqq_softrt_next_start(bfqd, bfqq); ++ ++ /* ++ * If this is the in-service queue, check if it needs to be expired, ++ * or if we want to idle in case it has no pending requests. ++ */ ++ if (bfqd->in_service_queue == bfqq) { ++ if (bfqq->dispatched == 0 && bfq_bfqq_must_idle(bfqq)) { ++ bfq_arm_slice_timer(bfqd); ++ goto out; ++ } else if (bfq_may_expire_for_budg_timeout(bfqq)) ++ bfq_bfqq_expire(bfqd, bfqq, false, ++ BFQ_BFQQ_BUDGET_TIMEOUT); ++ else if (RB_EMPTY_ROOT(&bfqq->sort_list) && ++ (bfqq->dispatched == 0 || ++ !bfq_bfqq_may_idle(bfqq))) ++ bfq_bfqq_expire(bfqd, bfqq, false, ++ BFQ_BFQQ_NO_MORE_REQUESTS); ++ } ++ ++ if (!bfqd->rq_in_driver) ++ bfq_schedule_dispatch(bfqd); ++ ++out: ++ return; ++} ++ ++static int __bfq_may_queue(struct bfq_queue *bfqq) ++{ ++ if (bfq_bfqq_wait_request(bfqq) && bfq_bfqq_must_alloc(bfqq)) { ++ bfq_clear_bfqq_must_alloc(bfqq); ++ return ELV_MQUEUE_MUST; ++ } ++ ++ return ELV_MQUEUE_MAY; ++} ++ ++static int bfq_may_queue(struct request_queue *q, unsigned int op) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct task_struct *tsk = current; ++ struct bfq_io_cq *bic; ++ struct bfq_queue *bfqq; ++ ++ /* ++ * Don't force setup of a queue from here, as a call to may_queue ++ * does not necessarily imply that a request actually will be ++ * queued. So just lookup a possibly existing queue, or return ++ * 'may queue' if that fails. ++ */ ++ bic = bfq_bic_lookup(bfqd, tsk->io_context); ++ if (!bic) ++ return ELV_MQUEUE_MAY; ++ ++ bfqq = bic_to_bfqq(bic, op_is_sync(op)); ++ if (bfqq) ++ return __bfq_may_queue(bfqq); ++ ++ return ELV_MQUEUE_MAY; ++} ++ ++/* ++ * Queue lock held here. ++ */ ++static void bfq_put_request(struct request *rq) ++{ ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ ++ if (bfqq) { ++ const int rw = rq_data_dir(rq); ++ ++ BUG_ON(!bfqq->allocated[rw]); ++ bfqq->allocated[rw]--; ++ ++ rq->elv.priv[0] = NULL; ++ rq->elv.priv[1] = NULL; ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "put_request %p, %d", ++ bfqq, bfqq->ref); ++ bfq_put_queue(bfqq); ++ } ++} ++ ++/* ++ * Returns NULL if a new bfqq should be allocated, or the old bfqq if this ++ * was the last process referring to that bfqq. ++ */ ++static struct bfq_queue * ++bfq_split_bfqq(struct bfq_io_cq *bic, struct bfq_queue *bfqq) ++{ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "splitting queue"); ++ ++ put_io_context(bic->icq.ioc); ++ ++ if (bfqq_process_refs(bfqq) == 1) { ++ bfqq->pid = current->pid; ++ bfq_clear_bfqq_coop(bfqq); ++ bfq_clear_bfqq_split_coop(bfqq); ++ return bfqq; ++ } ++ ++ bic_set_bfqq(bic, NULL, 1); ++ ++ bfq_put_cooperator(bfqq); ++ ++ bfq_put_queue(bfqq); ++ return NULL; ++} ++ ++/* ++ * Allocate bfq data structures associated with this request. ++ */ ++static int bfq_set_request(struct request_queue *q, struct request *rq, ++ struct bio *bio, gfp_t gfp_mask) ++{ ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct bfq_io_cq *bic = icq_to_bic(rq->elv.icq); ++ const int rw = rq_data_dir(rq); ++ const int is_sync = rq_is_sync(rq); ++ struct bfq_queue *bfqq; ++ unsigned long flags; ++ bool bfqq_already_existing = false, split = false; ++ ++ spin_lock_irqsave(q->queue_lock, flags); ++ ++ if (!bic) ++ goto queue_fail; ++ ++ bfq_check_ioprio_change(bic, bio); ++ ++ bfq_bic_update_cgroup(bic, bio); ++ ++new_queue: ++ bfqq = bic_to_bfqq(bic, is_sync); ++ if (!bfqq || bfqq == &bfqd->oom_bfqq) { ++ if (bfqq) ++ bfq_put_queue(bfqq); ++ bfqq = bfq_get_queue(bfqd, bio, is_sync, bic); ++ BUG_ON(!hlist_unhashed(&bfqq->burst_list_node)); ++ ++ bic_set_bfqq(bic, bfqq, is_sync); ++ if (split && is_sync) { ++ bfq_log_bfqq(bfqd, bfqq, ++ "set_request: was_in_list %d " ++ "was_in_large_burst %d " ++ "large burst in progress %d", ++ bic->was_in_burst_list, ++ bic->saved_in_large_burst, ++ bfqd->large_burst); ++ ++ if ((bic->was_in_burst_list && bfqd->large_burst) || ++ bic->saved_in_large_burst) { ++ bfq_log_bfqq(bfqd, bfqq, ++ "set_request: marking in " ++ "large burst"); ++ bfq_mark_bfqq_in_large_burst(bfqq); ++ } else { ++ bfq_log_bfqq(bfqd, bfqq, ++ "set_request: clearing in " ++ "large burst"); ++ bfq_clear_bfqq_in_large_burst(bfqq); ++ if (bic->was_in_burst_list) ++ hlist_add_head(&bfqq->burst_list_node, ++ &bfqd->burst_list); ++ } ++ bfqq->split_time = jiffies; ++ } ++ } else { ++ /* If the queue was seeky for too long, break it apart. */ ++ if (bfq_bfqq_coop(bfqq) && bfq_bfqq_split_coop(bfqq)) { ++ bfq_log_bfqq(bfqd, bfqq, "breaking apart bfqq"); ++ ++ /* Update bic before losing reference to bfqq */ ++ if (bfq_bfqq_in_large_burst(bfqq)) ++ bic->saved_in_large_burst = true; ++ ++ bfqq = bfq_split_bfqq(bic, bfqq); ++ split = true; ++ if (!bfqq) ++ goto new_queue; ++ else ++ bfqq_already_existing = true; ++ } ++ } ++ ++ bfqq->allocated[rw]++; ++ bfqq->ref++; ++ bfq_log_bfqq(bfqd, bfqq, "set_request: bfqq %p, %d", bfqq, bfqq->ref); ++ ++ rq->elv.priv[0] = bic; ++ rq->elv.priv[1] = bfqq; ++ ++ /* ++ * If a bfq_queue has only one process reference, it is owned ++ * by only one bfq_io_cq: we can set the bic field of the ++ * bfq_queue to the address of that structure. Also, if the ++ * queue has just been split, mark a flag so that the ++ * information is available to the other scheduler hooks. ++ */ ++ if (likely(bfqq != &bfqd->oom_bfqq) && bfqq_process_refs(bfqq) == 1) { ++ bfqq->bic = bic; ++ if (split) { ++ /* ++ * If the queue has just been split from a shared ++ * queue, restore the idle window and the possible ++ * weight raising period. ++ */ ++ bfq_bfqq_resume_state(bfqq, bfqd, bic, ++ bfqq_already_existing); ++ } ++ } ++ ++ if (unlikely(bfq_bfqq_just_created(bfqq))) ++ bfq_handle_burst(bfqd, bfqq); ++ ++ spin_unlock_irqrestore(q->queue_lock, flags); ++ ++ return 0; ++ ++queue_fail: ++ bfq_schedule_dispatch(bfqd); ++ spin_unlock_irqrestore(q->queue_lock, flags); ++ ++ return 1; ++} ++ ++static void bfq_kick_queue(struct work_struct *work) ++{ ++ struct bfq_data *bfqd = ++ container_of(work, struct bfq_data, unplug_work); ++ struct request_queue *q = bfqd->queue; ++ ++ spin_lock_irq(q->queue_lock); ++ __blk_run_queue(q); ++ spin_unlock_irq(q->queue_lock); ++} ++ ++/* ++ * Handler of the expiration of the timer running if the in-service queue ++ * is idling inside its time slice. ++ */ ++static enum hrtimer_restart bfq_idle_slice_timer(struct hrtimer *timer) ++{ ++ struct bfq_data *bfqd = container_of(timer, struct bfq_data, ++ idle_slice_timer); ++ struct bfq_queue *bfqq; ++ unsigned long flags; ++ enum bfqq_expiration reason; ++ ++ spin_lock_irqsave(bfqd->queue->queue_lock, flags); ++ ++ bfqq = bfqd->in_service_queue; ++ /* ++ * Theoretical race here: the in-service queue can be NULL or ++ * different from the queue that was idling if the timer handler ++ * spins on the queue_lock and a new request arrives for the ++ * current queue and there is a full dispatch cycle that changes ++ * the in-service queue. This can hardly happen, but in the worst ++ * case we just expire a queue too early. ++ */ ++ if (bfqq) { ++ bfq_log_bfqq(bfqd, bfqq, "slice_timer expired"); ++ bfq_clear_bfqq_wait_request(bfqq); ++ ++ if (bfq_bfqq_budget_timeout(bfqq)) ++ /* ++ * Also here the queue can be safely expired ++ * for budget timeout without wasting ++ * guarantees ++ */ ++ reason = BFQ_BFQQ_BUDGET_TIMEOUT; ++ else if (bfqq->queued[0] == 0 && bfqq->queued[1] == 0) ++ /* ++ * The queue may not be empty upon timer expiration, ++ * because we may not disable the timer when the ++ * first request of the in-service queue arrives ++ * during disk idling. ++ */ ++ reason = BFQ_BFQQ_TOO_IDLE; ++ else ++ goto schedule_dispatch; ++ ++ bfq_bfqq_expire(bfqd, bfqq, true, reason); ++ } ++ ++schedule_dispatch: ++ bfq_schedule_dispatch(bfqd); ++ ++ spin_unlock_irqrestore(bfqd->queue->queue_lock, flags); ++ return HRTIMER_NORESTART; ++} ++ ++static void bfq_shutdown_timer_wq(struct bfq_data *bfqd) ++{ ++ hrtimer_cancel(&bfqd->idle_slice_timer); ++ cancel_work_sync(&bfqd->unplug_work); ++} ++ ++static void __bfq_put_async_bfqq(struct bfq_data *bfqd, ++ struct bfq_queue **bfqq_ptr) ++{ ++ struct bfq_group *root_group = bfqd->root_group; ++ struct bfq_queue *bfqq = *bfqq_ptr; ++ ++ bfq_log(bfqd, "put_async_bfqq: %p", bfqq); ++ if (bfqq) { ++ bfq_bfqq_move(bfqd, bfqq, root_group); ++ bfq_log_bfqq(bfqd, bfqq, "put_async_bfqq: putting %p, %d", ++ bfqq, bfqq->ref); ++ bfq_put_queue(bfqq); ++ *bfqq_ptr = NULL; ++ } ++} ++ ++/* ++ * Release all the bfqg references to its async queues. If we are ++ * deallocating the group these queues may still contain requests, so ++ * we reparent them to the root cgroup (i.e., the only one that will ++ * exist for sure until all the requests on a device are gone). ++ */ ++static void bfq_put_async_queues(struct bfq_data *bfqd, struct bfq_group *bfqg) ++{ ++ int i, j; ++ ++ for (i = 0; i < 2; i++) ++ for (j = 0; j < IOPRIO_BE_NR; j++) ++ __bfq_put_async_bfqq(bfqd, &bfqg->async_bfqq[i][j]); ++ ++ __bfq_put_async_bfqq(bfqd, &bfqg->async_idle_bfqq); ++} ++ ++static void bfq_exit_queue(struct elevator_queue *e) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ struct request_queue *q = bfqd->queue; ++ struct bfq_queue *bfqq, *n; ++ ++ bfq_shutdown_timer_wq(bfqd); ++ ++ spin_lock_irq(q->queue_lock); ++ ++ BUG_ON(bfqd->in_service_queue); ++ list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) ++ bfq_deactivate_bfqq(bfqd, bfqq, false, false); ++ ++ spin_unlock_irq(q->queue_lock); ++ ++ bfq_shutdown_timer_wq(bfqd); ++ ++ BUG_ON(hrtimer_active(&bfqd->idle_slice_timer)); ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ blkcg_deactivate_policy(q, &blkcg_policy_bfq); ++#else ++ bfq_put_async_queues(bfqd, bfqd->root_group); ++ kfree(bfqd->root_group); ++#endif ++ ++ kfree(bfqd); ++} ++ ++static void bfq_init_root_group(struct bfq_group *root_group, ++ struct bfq_data *bfqd) ++{ ++ int i; ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ root_group->entity.parent = NULL; ++ root_group->my_entity = NULL; ++ root_group->bfqd = bfqd; ++#endif ++ root_group->rq_pos_tree = RB_ROOT; ++ for (i = 0; i < BFQ_IOPRIO_CLASSES; i++) ++ root_group->sched_data.service_tree[i] = BFQ_SERVICE_TREE_INIT; ++ root_group->sched_data.bfq_class_idle_last_service = jiffies; ++} ++ ++static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) ++{ ++ struct bfq_data *bfqd; ++ struct elevator_queue *eq; ++ ++ eq = elevator_alloc(q, e); ++ if (!eq) ++ return -ENOMEM; ++ ++ bfqd = kzalloc_node(sizeof(*bfqd), GFP_KERNEL, q->node); ++ if (!bfqd) { ++ kobject_put(&eq->kobj); ++ return -ENOMEM; ++ } ++ eq->elevator_data = bfqd; ++ ++ /* ++ * Our fallback bfqq if bfq_find_alloc_queue() runs into OOM issues. ++ * Grab a permanent reference to it, so that the normal code flow ++ * will not attempt to free it. ++ */ ++ bfq_init_bfqq(bfqd, &bfqd->oom_bfqq, NULL, 1, 0); ++ bfqd->oom_bfqq.ref++; ++ bfqd->oom_bfqq.new_ioprio = BFQ_DEFAULT_QUEUE_IOPRIO; ++ bfqd->oom_bfqq.new_ioprio_class = IOPRIO_CLASS_BE; ++ bfqd->oom_bfqq.entity.new_weight = ++ bfq_ioprio_to_weight(bfqd->oom_bfqq.new_ioprio); ++ ++ /* oom_bfqq does not participate to bursts */ ++ bfq_clear_bfqq_just_created(&bfqd->oom_bfqq); ++ /* ++ * Trigger weight initialization, according to ioprio, at the ++ * oom_bfqq's first activation. The oom_bfqq's ioprio and ioprio ++ * class won't be changed any more. ++ */ ++ bfqd->oom_bfqq.entity.prio_changed = 1; ++ ++ bfqd->queue = q; ++ ++ spin_lock_irq(q->queue_lock); ++ q->elevator = eq; ++ spin_unlock_irq(q->queue_lock); ++ ++ bfqd->root_group = bfq_create_group_hierarchy(bfqd, q->node); ++ if (!bfqd->root_group) ++ goto out_free; ++ bfq_init_root_group(bfqd->root_group, bfqd); ++ bfq_init_entity(&bfqd->oom_bfqq.entity, bfqd->root_group); ++ ++ hrtimer_init(&bfqd->idle_slice_timer, CLOCK_MONOTONIC, ++ HRTIMER_MODE_REL); ++ bfqd->idle_slice_timer.function = bfq_idle_slice_timer; ++ ++ bfqd->queue_weights_tree = RB_ROOT; ++ bfqd->group_weights_tree = RB_ROOT; ++ ++ INIT_WORK(&bfqd->unplug_work, bfq_kick_queue); ++ ++ INIT_LIST_HEAD(&bfqd->active_list); ++ INIT_LIST_HEAD(&bfqd->idle_list); ++ INIT_HLIST_HEAD(&bfqd->burst_list); ++ ++ bfqd->hw_tag = -1; ++ ++ bfqd->bfq_max_budget = bfq_default_max_budget; ++ ++ bfqd->bfq_fifo_expire[0] = bfq_fifo_expire[0]; ++ bfqd->bfq_fifo_expire[1] = bfq_fifo_expire[1]; ++ bfqd->bfq_back_max = bfq_back_max; ++ bfqd->bfq_back_penalty = bfq_back_penalty; ++ bfqd->bfq_slice_idle = bfq_slice_idle; ++ bfqd->bfq_timeout = bfq_timeout; ++ ++ bfqd->bfq_requests_within_timer = 120; ++ ++ bfqd->bfq_large_burst_thresh = 8; ++ bfqd->bfq_burst_interval = msecs_to_jiffies(180); ++ ++ bfqd->low_latency = true; ++ ++ /* ++ * Trade-off between responsiveness and fairness. ++ */ ++ bfqd->bfq_wr_coeff = 30; ++ bfqd->bfq_wr_rt_max_time = msecs_to_jiffies(300); ++ bfqd->bfq_wr_max_time = 0; ++ bfqd->bfq_wr_min_idle_time = msecs_to_jiffies(2000); ++ bfqd->bfq_wr_min_inter_arr_async = msecs_to_jiffies(500); ++ bfqd->bfq_wr_max_softrt_rate = 7000; /* ++ * Approximate rate required ++ * to playback or record a ++ * high-definition compressed ++ * video. ++ */ ++ bfqd->wr_busy_queues = 0; ++ ++ /* ++ * Begin by assuming, optimistically, that the device is a ++ * high-speed one, and that its peak rate is equal to 2/3 of ++ * the highest reference rate. ++ */ ++ bfqd->RT_prod = R_fast[blk_queue_nonrot(bfqd->queue)] * ++ T_fast[blk_queue_nonrot(bfqd->queue)]; ++ bfqd->peak_rate = R_fast[blk_queue_nonrot(bfqd->queue)] * 2 / 3; ++ bfqd->device_speed = BFQ_BFQD_FAST; ++ ++ return 0; ++ ++out_free: ++ kfree(bfqd); ++ kobject_put(&eq->kobj); ++ return -ENOMEM; ++} ++ ++static void bfq_slab_kill(void) ++{ ++ kmem_cache_destroy(bfq_pool); ++} ++ ++static int __init bfq_slab_setup(void) ++{ ++ bfq_pool = KMEM_CACHE(bfq_queue, 0); ++ if (!bfq_pool) ++ return -ENOMEM; ++ return 0; ++} ++ ++static ssize_t bfq_var_show(unsigned int var, char *page) ++{ ++ return sprintf(page, "%u\n", var); ++} ++ ++static ssize_t bfq_var_store(unsigned long *var, const char *page, ++ size_t count) ++{ ++ unsigned long new_val; ++ int ret = kstrtoul(page, 10, &new_val); ++ ++ if (ret == 0) ++ *var = new_val; ++ ++ return count; ++} ++ ++static ssize_t bfq_wr_max_time_show(struct elevator_queue *e, char *page) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ ++ return sprintf(page, "%d\n", bfqd->bfq_wr_max_time > 0 ? ++ jiffies_to_msecs(bfqd->bfq_wr_max_time) : ++ jiffies_to_msecs(bfq_wr_duration(bfqd))); ++} ++ ++static ssize_t bfq_weights_show(struct elevator_queue *e, char *page) ++{ ++ struct bfq_queue *bfqq; ++ struct bfq_data *bfqd = e->elevator_data; ++ ssize_t num_char = 0; ++ ++ num_char += sprintf(page + num_char, "Tot reqs queued %d\n\n", ++ bfqd->queued); ++ ++ spin_lock_irq(bfqd->queue->queue_lock); ++ ++ num_char += sprintf(page + num_char, "Active:\n"); ++ list_for_each_entry(bfqq, &bfqd->active_list, bfqq_list) { ++ num_char += sprintf(page + num_char, ++ "pid%d: weight %hu, nr_queued %d %d, ", ++ bfqq->pid, ++ bfqq->entity.weight, ++ bfqq->queued[0], ++ bfqq->queued[1]); ++ num_char += sprintf(page + num_char, ++ "dur %d/%u\n", ++ jiffies_to_msecs( ++ jiffies - ++ bfqq->last_wr_start_finish), ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ } ++ ++ num_char += sprintf(page + num_char, "Idle:\n"); ++ list_for_each_entry(bfqq, &bfqd->idle_list, bfqq_list) { ++ num_char += sprintf(page + num_char, ++ "pid%d: weight %hu, dur %d/%u\n", ++ bfqq->pid, ++ bfqq->entity.weight, ++ jiffies_to_msecs(jiffies - ++ bfqq->last_wr_start_finish), ++ jiffies_to_msecs(bfqq->wr_cur_max_time)); ++ } ++ ++ spin_unlock_irq(bfqd->queue->queue_lock); ++ ++ return num_char; ++} ++ ++#define SHOW_FUNCTION(__FUNC, __VAR, __CONV) \ ++static ssize_t __FUNC(struct elevator_queue *e, char *page) \ ++{ \ ++ struct bfq_data *bfqd = e->elevator_data; \ ++ u64 __data = __VAR; \ ++ if (__CONV == 1) \ ++ __data = jiffies_to_msecs(__data); \ ++ else if (__CONV == 2) \ ++ __data = div_u64(__data, NSEC_PER_MSEC); \ ++ return bfq_var_show(__data, (page)); \ ++} ++SHOW_FUNCTION(bfq_fifo_expire_sync_show, bfqd->bfq_fifo_expire[1], 2); ++SHOW_FUNCTION(bfq_fifo_expire_async_show, bfqd->bfq_fifo_expire[0], 2); ++SHOW_FUNCTION(bfq_back_seek_max_show, bfqd->bfq_back_max, 0); ++SHOW_FUNCTION(bfq_back_seek_penalty_show, bfqd->bfq_back_penalty, 0); ++SHOW_FUNCTION(bfq_slice_idle_show, bfqd->bfq_slice_idle, 2); ++SHOW_FUNCTION(bfq_max_budget_show, bfqd->bfq_user_max_budget, 0); ++SHOW_FUNCTION(bfq_timeout_sync_show, bfqd->bfq_timeout, 1); ++SHOW_FUNCTION(bfq_strict_guarantees_show, bfqd->strict_guarantees, 0); ++SHOW_FUNCTION(bfq_low_latency_show, bfqd->low_latency, 0); ++SHOW_FUNCTION(bfq_wr_coeff_show, bfqd->bfq_wr_coeff, 0); ++SHOW_FUNCTION(bfq_wr_rt_max_time_show, bfqd->bfq_wr_rt_max_time, 1); ++SHOW_FUNCTION(bfq_wr_min_idle_time_show, bfqd->bfq_wr_min_idle_time, 1); ++SHOW_FUNCTION(bfq_wr_min_inter_arr_async_show, bfqd->bfq_wr_min_inter_arr_async, ++ 1); ++SHOW_FUNCTION(bfq_wr_max_softrt_rate_show, bfqd->bfq_wr_max_softrt_rate, 0); ++#undef SHOW_FUNCTION ++ ++#define USEC_SHOW_FUNCTION(__FUNC, __VAR) \ ++static ssize_t __FUNC(struct elevator_queue *e, char *page) \ ++{ \ ++ struct bfq_data *bfqd = e->elevator_data; \ ++ u64 __data = __VAR; \ ++ __data = div_u64(__data, NSEC_PER_USEC); \ ++ return bfq_var_show(__data, (page)); \ ++} ++USEC_SHOW_FUNCTION(bfq_slice_idle_us_show, bfqd->bfq_slice_idle); ++#undef USEC_SHOW_FUNCTION ++ ++#define STORE_FUNCTION(__FUNC, __PTR, MIN, MAX, __CONV) \ ++static ssize_t \ ++__FUNC(struct elevator_queue *e, const char *page, size_t count) \ ++{ \ ++ struct bfq_data *bfqd = e->elevator_data; \ ++ unsigned long uninitialized_var(__data); \ ++ int ret = bfq_var_store(&__data, (page), count); \ ++ if (__data < (MIN)) \ ++ __data = (MIN); \ ++ else if (__data > (MAX)) \ ++ __data = (MAX); \ ++ if (__CONV == 1) \ ++ *(__PTR) = msecs_to_jiffies(__data); \ ++ else if (__CONV == 2) \ ++ *(__PTR) = (u64)__data * NSEC_PER_MSEC; \ ++ else \ ++ *(__PTR) = __data; \ ++ return ret; \ ++} ++STORE_FUNCTION(bfq_fifo_expire_sync_store, &bfqd->bfq_fifo_expire[1], 1, ++ INT_MAX, 2); ++STORE_FUNCTION(bfq_fifo_expire_async_store, &bfqd->bfq_fifo_expire[0], 1, ++ INT_MAX, 2); ++STORE_FUNCTION(bfq_back_seek_max_store, &bfqd->bfq_back_max, 0, INT_MAX, 0); ++STORE_FUNCTION(bfq_back_seek_penalty_store, &bfqd->bfq_back_penalty, 1, ++ INT_MAX, 0); ++STORE_FUNCTION(bfq_slice_idle_store, &bfqd->bfq_slice_idle, 0, INT_MAX, 2); ++STORE_FUNCTION(bfq_wr_coeff_store, &bfqd->bfq_wr_coeff, 1, INT_MAX, 0); ++STORE_FUNCTION(bfq_wr_max_time_store, &bfqd->bfq_wr_max_time, 0, INT_MAX, 1); ++STORE_FUNCTION(bfq_wr_rt_max_time_store, &bfqd->bfq_wr_rt_max_time, 0, INT_MAX, ++ 1); ++STORE_FUNCTION(bfq_wr_min_idle_time_store, &bfqd->bfq_wr_min_idle_time, 0, ++ INT_MAX, 1); ++STORE_FUNCTION(bfq_wr_min_inter_arr_async_store, ++ &bfqd->bfq_wr_min_inter_arr_async, 0, INT_MAX, 1); ++STORE_FUNCTION(bfq_wr_max_softrt_rate_store, &bfqd->bfq_wr_max_softrt_rate, 0, ++ INT_MAX, 0); ++#undef STORE_FUNCTION ++ ++#define USEC_STORE_FUNCTION(__FUNC, __PTR, MIN, MAX) \ ++static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)\ ++{ \ ++ struct bfq_data *bfqd = e->elevator_data; \ ++ unsigned long uninitialized_var(__data); \ ++ int ret = bfq_var_store(&__data, (page), count); \ ++ if (__data < (MIN)) \ ++ __data = (MIN); \ ++ else if (__data > (MAX)) \ ++ __data = (MAX); \ ++ *(__PTR) = (u64)__data * NSEC_PER_USEC; \ ++ return ret; \ ++} ++USEC_STORE_FUNCTION(bfq_slice_idle_us_store, &bfqd->bfq_slice_idle, 0, ++ UINT_MAX); ++#undef USEC_STORE_FUNCTION ++ ++/* do nothing for the moment */ ++static ssize_t bfq_weights_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ return count; ++} ++ ++static ssize_t bfq_max_budget_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ unsigned long uninitialized_var(__data); ++ int ret = bfq_var_store(&__data, (page), count); ++ ++ if (__data == 0) ++ bfqd->bfq_max_budget = bfq_calc_max_budget(bfqd); ++ else { ++ if (__data > INT_MAX) ++ __data = INT_MAX; ++ bfqd->bfq_max_budget = __data; ++ } ++ ++ bfqd->bfq_user_max_budget = __data; ++ ++ return ret; ++} ++ ++/* ++ * Leaving this name to preserve name compatibility with cfq ++ * parameters, but this timeout is used for both sync and async. ++ */ ++static ssize_t bfq_timeout_sync_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ unsigned long uninitialized_var(__data); ++ int ret = bfq_var_store(&__data, (page), count); ++ ++ if (__data < 1) ++ __data = 1; ++ else if (__data > INT_MAX) ++ __data = INT_MAX; ++ ++ bfqd->bfq_timeout = msecs_to_jiffies(__data); ++ if (bfqd->bfq_user_max_budget == 0) ++ bfqd->bfq_max_budget = bfq_calc_max_budget(bfqd); ++ ++ return ret; ++} ++ ++static ssize_t bfq_strict_guarantees_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ unsigned long uninitialized_var(__data); ++ int ret = bfq_var_store(&__data, (page), count); ++ ++ if (__data > 1) ++ __data = 1; ++ if (!bfqd->strict_guarantees && __data == 1 ++ && bfqd->bfq_slice_idle < 8 * NSEC_PER_MSEC) ++ bfqd->bfq_slice_idle = 8 * NSEC_PER_MSEC; ++ ++ bfqd->strict_guarantees = __data; ++ ++ return ret; ++} ++ ++static ssize_t bfq_low_latency_store(struct elevator_queue *e, ++ const char *page, size_t count) ++{ ++ struct bfq_data *bfqd = e->elevator_data; ++ unsigned long uninitialized_var(__data); ++ int ret = bfq_var_store(&__data, (page), count); ++ ++ if (__data > 1) ++ __data = 1; ++ if (__data == 0 && bfqd->low_latency != 0) ++ bfq_end_wr(bfqd); ++ bfqd->low_latency = __data; ++ ++ return ret; ++} ++ ++#define BFQ_ATTR(name) \ ++ __ATTR(name, S_IRUGO|S_IWUSR, bfq_##name##_show, bfq_##name##_store) ++ ++static struct elv_fs_entry bfq_attrs[] = { ++ BFQ_ATTR(fifo_expire_sync), ++ BFQ_ATTR(fifo_expire_async), ++ BFQ_ATTR(back_seek_max), ++ BFQ_ATTR(back_seek_penalty), ++ BFQ_ATTR(slice_idle), ++ BFQ_ATTR(slice_idle_us), ++ BFQ_ATTR(max_budget), ++ BFQ_ATTR(timeout_sync), ++ BFQ_ATTR(strict_guarantees), ++ BFQ_ATTR(low_latency), ++ BFQ_ATTR(wr_coeff), ++ BFQ_ATTR(wr_max_time), ++ BFQ_ATTR(wr_rt_max_time), ++ BFQ_ATTR(wr_min_idle_time), ++ BFQ_ATTR(wr_min_inter_arr_async), ++ BFQ_ATTR(wr_max_softrt_rate), ++ BFQ_ATTR(weights), ++ __ATTR_NULL ++}; ++ ++static struct elevator_type iosched_bfq = { ++ .ops.sq = { ++ .elevator_merge_fn = bfq_merge, ++ .elevator_merged_fn = bfq_merged_request, ++ .elevator_merge_req_fn = bfq_merged_requests, ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ .elevator_bio_merged_fn = bfq_bio_merged, ++#endif ++ .elevator_allow_bio_merge_fn = bfq_allow_bio_merge, ++ .elevator_allow_rq_merge_fn = bfq_allow_rq_merge, ++ .elevator_dispatch_fn = bfq_dispatch_requests, ++ .elevator_add_req_fn = bfq_insert_request, ++ .elevator_activate_req_fn = bfq_activate_request, ++ .elevator_deactivate_req_fn = bfq_deactivate_request, ++ .elevator_completed_req_fn = bfq_completed_request, ++ .elevator_former_req_fn = elv_rb_former_request, ++ .elevator_latter_req_fn = elv_rb_latter_request, ++ .elevator_init_icq_fn = bfq_init_icq, ++ .elevator_exit_icq_fn = bfq_exit_icq, ++ .elevator_set_req_fn = bfq_set_request, ++ .elevator_put_req_fn = bfq_put_request, ++ .elevator_may_queue_fn = bfq_may_queue, ++ .elevator_init_fn = bfq_init_queue, ++ .elevator_exit_fn = bfq_exit_queue, ++ }, ++ .icq_size = sizeof(struct bfq_io_cq), ++ .icq_align = __alignof__(struct bfq_io_cq), ++ .elevator_attrs = bfq_attrs, ++ .elevator_name = "bfq-sq", ++ .elevator_owner = THIS_MODULE, ++}; ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++static struct blkcg_policy blkcg_policy_bfq = { ++ .dfl_cftypes = bfq_blkg_files, ++ .legacy_cftypes = bfq_blkcg_legacy_files, ++ ++ .cpd_alloc_fn = bfq_cpd_alloc, ++ .cpd_init_fn = bfq_cpd_init, ++ .cpd_bind_fn = bfq_cpd_init, ++ .cpd_free_fn = bfq_cpd_free, ++ ++ .pd_alloc_fn = bfq_pd_alloc, ++ .pd_init_fn = bfq_pd_init, ++ .pd_offline_fn = bfq_pd_offline, ++ .pd_free_fn = bfq_pd_free, ++ .pd_reset_stats_fn = bfq_pd_reset_stats, ++}; ++#endif ++ ++static int __init bfq_init(void) ++{ ++ int ret; ++ char msg[60] = "BFQ I/O-scheduler: v8r12"; ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ ret = blkcg_policy_register(&blkcg_policy_bfq); ++ if (ret) ++ return ret; ++#endif ++ ++ ret = -ENOMEM; ++ if (bfq_slab_setup()) ++ goto err_pol_unreg; ++ ++ /* ++ * Times to load large popular applications for the typical ++ * systems installed on the reference devices (see the ++ * comments before the definitions of the next two ++ * arrays). Actually, we use slightly slower values, as the ++ * estimated peak rate tends to be smaller than the actual ++ * peak rate. The reason for this last fact is that estimates ++ * are computed over much shorter time intervals than the long ++ * intervals typically used for benchmarking. Why? First, to ++ * adapt more quickly to variations. Second, because an I/O ++ * scheduler cannot rely on a peak-rate-evaluation workload to ++ * be run for a long time. ++ */ ++ T_slow[0] = msecs_to_jiffies(3500); /* actually 4 sec */ ++ T_slow[1] = msecs_to_jiffies(6000); /* actually 6.5 sec */ ++ T_fast[0] = msecs_to_jiffies(7000); /* actually 8 sec */ ++ T_fast[1] = msecs_to_jiffies(2500); /* actually 3 sec */ ++ ++ /* ++ * Thresholds that determine the switch between speed classes ++ * (see the comments before the definition of the array ++ * device_speed_thresh). These thresholds are biased towards ++ * transitions to the fast class. This is safer than the ++ * opposite bias. In fact, a wrong transition to the slow ++ * class results in short weight-raising periods, because the ++ * speed of the device then tends to be higher that the ++ * reference peak rate. On the opposite end, a wrong ++ * transition to the fast class tends to increase ++ * weight-raising periods, because of the opposite reason. ++ */ ++ device_speed_thresh[0] = (4 * R_slow[0]) / 3; ++ device_speed_thresh[1] = (4 * R_slow[1]) / 3; ++ ++ ret = elv_register(&iosched_bfq); ++ if (ret) ++ goto err_pol_unreg; ++ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ strcat(msg, " (with cgroups support)"); ++#endif ++ pr_info("%s", msg); ++ ++ return 0; ++ ++err_pol_unreg: ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ blkcg_policy_unregister(&blkcg_policy_bfq); ++#endif ++ return ret; ++} ++ ++static void __exit bfq_exit(void) ++{ ++ elv_unregister(&iosched_bfq); ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++ blkcg_policy_unregister(&blkcg_policy_bfq); ++#endif ++ bfq_slab_kill(); ++} ++ ++module_init(bfq_init); ++module_exit(bfq_exit); ++ ++MODULE_AUTHOR("Arianna Avanzini, Fabio Checconi, Paolo Valente"); ++MODULE_LICENSE("GPL"); + +From e24d2e6461479dbd13d58be2dc44b23b5e24487c Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Mon, 19 Dec 2016 17:13:39 +0100 +Subject: [PATCH 07/51] Add config and build bits for bfq-mq-iosched + +Signed-off-by: Paolo Valente +--- + block/Kconfig.iosched | 10 +++++++++ + block/Makefile | 1 + + block/bfq-cgroup-included.c | 4 ++-- + block/bfq-mq-iosched.c | 25 ++++++++++++----------- + block/bfq-sched.c | 50 ++++++++++++++++++++++----------------------- + block/bfq-sq-iosched.c | 24 +++++++++++----------- + block/bfq.h | 36 +++++++++++++++++++++----------- + 8 files changed, 88 insertions(+), 64 deletions(-) + +diff --git a/block/Kconfig.iosched b/block/Kconfig.iosched +index 9e3f4c2f7390..2d94af3d8b0a 100644 +--- a/block/Kconfig.iosched ++++ b/block/Kconfig.iosched +@@ -96,6 +96,16 @@ config DEFAULT_IOSCHED + default "bfq-sq" if DEFAULT_BFQ_SQ + default "noop" if DEFAULT_NOOP + ++config MQ_IOSCHED_BFQ ++ tristate "BFQ-MQ I/O Scheduler" ++ default y ++ ---help--- ++ BFQ I/O scheduler for BLK-MQ. BFQ-MQ distributes bandwidth ++ among all processes according to their weights, regardless of ++ the device parameters and with any workload. It also ++ guarantees a low latency to interactive and soft real-time ++ applications. Details in Documentation/block/bfq-iosched.txt ++ + config MQ_IOSCHED_DEADLINE + tristate "MQ deadline I/O scheduler" + default y +diff --git a/block/Makefile b/block/Makefile +index 59026b425791..a571329c23f0 100644 +--- a/block/Makefile ++++ b/block/Makefile +@@ -25,6 +25,7 @@ obj-$(CONFIG_MQ_IOSCHED_KYBER) += kyber-iosched.o + bfq-y := bfq-iosched.o bfq-wf2q.o bfq-cgroup.o + obj-$(CONFIG_IOSCHED_BFQ) += bfq.o + obj-$(CONFIG_IOSCHED_BFQ_SQ) += bfq-sq-iosched.o ++obj-$(CONFIG_MQ_IOSCHED_BFQ) += bfq-mq-iosched.o + + obj-$(CONFIG_BLOCK_COMPAT) += compat_ioctl.o + obj-$(CONFIG_BLK_CMDLINE_PARSER) += cmdline-parser.o +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index af7c216a3540..9c483b658179 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -15,7 +15,7 @@ + * file. + */ + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + + /* bfqg stats flags */ + enum bfqg_stats_flags { +@@ -1116,7 +1116,7 @@ static struct cftype bfq_blkg_files[] = { + {} /* terminate */ + }; + +-#else /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++#else /* BFQ_GROUP_IOSCHED_ENABLED */ + + static inline void bfqg_stats_update_io_add(struct bfq_group *bfqg, + struct bfq_queue *bfqq, unsigned int op) { } +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 30d019fc67e0..e88e00f1e0a7 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -82,6 +82,7 @@ + #include + #include + #include "blk.h" ++#undef CONFIG_BFQ_GROUP_IOSCHED /* cgroups support not yet functional */ + #include "bfq.h" + + /* Expiration time of sync (0) and async (1) requests, in ns. */ +@@ -387,7 +388,7 @@ static bool bfq_differentiated_weights(struct bfq_data *bfqd) + return (!RB_EMPTY_ROOT(&bfqd->queue_weights_tree) && + (bfqd->queue_weights_tree.rb_node->rb_left || + bfqd->queue_weights_tree.rb_node->rb_right) +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + ) || + (!RB_EMPTY_ROOT(&bfqd->group_weights_tree) && + (bfqd->group_weights_tree.rb_node->rb_left || +@@ -1672,7 +1673,7 @@ static void bfq_merged_request(struct request_queue *q, struct request *req, + } + } + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + static void bfq_bio_merged(struct request_queue *q, struct request *req, + struct bio *bio) + { +@@ -3879,7 +3880,7 @@ static int bfq_dispatch_requests(struct request_queue *q, int force) + */ + static void bfq_put_queue(struct bfq_queue *bfqq) + { +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + struct bfq_group *bfqg = bfqq_group(bfqq); + #endif + +@@ -3909,7 +3910,7 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); + + kmem_cache_free(bfq_pool, bfqq); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + bfqg_put(bfqg); + #endif + } +@@ -4835,7 +4836,7 @@ static void bfq_exit_queue(struct elevator_queue *e) + + BUG_ON(hrtimer_active(&bfqd->idle_slice_timer)); + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_deactivate_policy(q, &blkcg_policy_bfq); + #else + bfq_put_async_queues(bfqd, bfqd->root_group); +@@ -4850,7 +4851,7 @@ static void bfq_init_root_group(struct bfq_group *root_group, + { + int i; + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + root_group->entity.parent = NULL; + root_group->my_entity = NULL; + root_group->bfqd = bfqd; +@@ -5265,7 +5266,7 @@ static struct elevator_type iosched_bfq = { + .elevator_merge_fn = bfq_merge, + .elevator_merged_fn = bfq_merged_request, + .elevator_merge_req_fn = bfq_merged_requests, +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + .elevator_bio_merged_fn = bfq_bio_merged, + #endif + .elevator_allow_bio_merge_fn = bfq_allow_bio_merge, +@@ -5292,7 +5293,7 @@ static struct elevator_type iosched_bfq = { + .elevator_owner = THIS_MODULE, + }; + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + static struct blkcg_policy blkcg_policy_bfq = { + .dfl_cftypes = bfq_blkg_files, + .legacy_cftypes = bfq_blkcg_legacy_files, +@@ -5315,7 +5316,7 @@ static int __init bfq_init(void) + int ret; + char msg[60] = "BFQ I/O-scheduler: v8r12"; + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + ret = blkcg_policy_register(&blkcg_policy_bfq); + if (ret) + return ret; +@@ -5362,7 +5363,7 @@ static int __init bfq_init(void) + if (ret) + goto err_pol_unreg; + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + strcat(msg, " (with cgroups support)"); + #endif + pr_info("%s", msg); +@@ -5370,7 +5371,7 @@ static int __init bfq_init(void) + return 0; + + err_pol_unreg: +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_policy_unregister(&blkcg_policy_bfq); + #endif + return ret; +@@ -5379,7 +5380,7 @@ static int __init bfq_init(void) + static void __exit bfq_exit(void) + { + elv_unregister(&iosched_bfq); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_policy_unregister(&blkcg_policy_bfq); + #endif + bfq_slab_kill(); +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index 5c0f9290a79c..b54a638186e3 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -136,7 +136,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, + "update_next_in_service: chosen this queue"); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = + container_of(next_in_service, +@@ -149,7 +149,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + return parent_sched_may_change; + } + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + /* both next loops stop at one of the child entities of the root group */ + #define for_each_entity(entity) \ + for (; entity ; entity = entity->parent) +@@ -243,7 +243,7 @@ static bool bfq_no_longer_next_in_service(struct bfq_entity *entity) + return false; + } + +-#else /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++#else /* BFQ_GROUP_IOSCHED_ENABLED */ + #define for_each_entity(entity) \ + for (; entity ; entity = NULL) + +@@ -260,7 +260,7 @@ static bool bfq_no_longer_next_in_service(struct bfq_entity *entity) + return true; + } + +-#endif /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++#endif /* BFQ_GROUP_IOSCHED_ENABLED */ + + /* + * Shift for timestamp calculations. This actually limits the maximum +@@ -323,7 +323,7 @@ static void bfq_calc_finish(struct bfq_entity *entity, unsigned long service) + bfq_log_bfqq(bfqq->bfqd, bfqq, + "calc_finish: start %llu, finish %llu, delta %llu", + start, finish, delta); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, entity); +@@ -473,7 +473,7 @@ static void bfq_update_active_node(struct rb_node *node) + bfq_log_bfqq(bfqq->bfqd, bfqq, + "update_active_node: new min_start %llu", + ((entity->min_start>>10)*1000)>>12); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, entity); +@@ -540,7 +540,7 @@ static void bfq_active_insert(struct bfq_service_tree *st, + { + struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); + struct rb_node *node = &entity->rb_node; +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + struct bfq_sched_data *sd = NULL; + struct bfq_group *bfqg = NULL; + struct bfq_data *bfqd = NULL; +@@ -555,7 +555,7 @@ static void bfq_active_insert(struct bfq_service_tree *st, + + bfq_update_active_tree(node); + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + sd = entity->sched_data; + bfqg = container_of(sd, struct bfq_group, sched_data); + BUG_ON(!bfqg); +@@ -563,7 +563,7 @@ static void bfq_active_insert(struct bfq_service_tree *st, + #endif + if (bfqq) + list_add(&bfqq->bfqq_list, &bfqq->bfqd->active_list); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { /* bfq_group */ + BUG_ON(!bfqd); + bfq_weights_tree_add(bfqd, entity, &bfqd->group_weights_tree); +@@ -652,7 +652,7 @@ static void bfq_active_extract(struct bfq_service_tree *st, + { + struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); + struct rb_node *node; +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + struct bfq_sched_data *sd = NULL; + struct bfq_group *bfqg = NULL; + struct bfq_data *bfqd = NULL; +@@ -664,7 +664,7 @@ static void bfq_active_extract(struct bfq_service_tree *st, + if (node) + bfq_update_active_tree(node); + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + sd = entity->sched_data; + bfqg = container_of(sd, struct bfq_group, sched_data); + BUG_ON(!bfqg); +@@ -672,7 +672,7 @@ static void bfq_active_extract(struct bfq_service_tree *st, + #endif + if (bfqq) + list_del(&bfqq->bfqq_list); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { /* bfq_group */ + BUG_ON(!bfqd); + bfq_weights_tree_remove(bfqd, entity, +@@ -809,14 +809,14 @@ __bfq_entity_update_weight_prio(struct bfq_service_tree *old_st, + unsigned int prev_weight, new_weight; + struct bfq_data *bfqd = NULL; + struct rb_root *root; +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + struct bfq_sched_data *sd; + struct bfq_group *bfqg; + #endif + + if (bfqq) + bfqd = bfqq->bfqd; +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + sd = entity->my_sched_data; + bfqg = container_of(sd, struct bfq_group, sched_data); +@@ -907,7 +907,7 @@ __bfq_entity_update_weight_prio(struct bfq_service_tree *old_st, + return new_st; + } + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + static void bfqg_stats_set_start_empty_time(struct bfq_group *bfqg); + #endif + +@@ -936,7 +936,7 @@ static void bfq_bfqq_served(struct bfq_queue *bfqq, int served) + st->vtime += bfq_delta(served, st->wsum); + bfq_forget_idle(st); + } +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + bfqg_stats_set_start_empty_time(bfqq_group(bfqq)); + #endif + st = bfq_entity_service_tree(&bfqq->entity); +@@ -1060,7 +1060,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + bfq_log_bfqq(bfqq->bfqd, bfqq, + "__activate_entity: new queue finish %llu", + ((entity->finish>>10)*1000)>>12); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, entity); +@@ -1078,7 +1078,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + bfq_log_bfqq(bfqq->bfqd, bfqq, + "__activate_entity: queue %seligible in st %p", + entity->start <= st->vtime ? "" : "non ", st); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, entity); +@@ -1153,7 +1153,7 @@ static void __bfq_activate_entity(struct bfq_entity *entity, + + BUG_ON(entity->on_st && bfqq); + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + if (entity->on_st && !bfqq) { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, +@@ -1485,7 +1485,7 @@ static void bfq_deactivate_entity(struct bfq_entity *entity, + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, + "invoking udpdate_next for this queue"); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = + container_of(entity, +@@ -1525,7 +1525,7 @@ static u64 bfq_calc_vtime_jump(struct bfq_service_tree *st) + bfq_log_bfqq(bfqq->bfqd, bfqq, + "calc_vtime_jump: new value %llu", + root_entity->min_start); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = + container_of(root_entity, struct bfq_group, +@@ -1661,7 +1661,7 @@ __bfq_lookup_next_entity(struct bfq_service_tree *st, bool in_service + "__lookup_next: start %llu vtime %llu st %p", + ((entity->start>>10)*1000)>>12, + ((new_vtime>>10)*1000)>>12, st); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, entity); +@@ -1735,7 +1735,7 @@ static struct bfq_entity *bfq_lookup_next_entity(struct bfq_sched_data *sd) + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, "chosen from st %p %d", + st + class_idx, class_idx); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, entity); +@@ -1777,7 +1777,7 @@ static struct bfq_queue *bfq_get_next_queue(struct bfq_data *bfqd) + */ + sd = &bfqd->root_group->sched_data; + for (; sd ; sd = entity->my_sched_data) { +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + if (entity) { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, entity); +@@ -1867,7 +1867,7 @@ static struct bfq_queue *bfq_get_next_queue(struct bfq_data *bfqd) + bfq_log_bfqq(bfqd, bfqq, + "get_next_queue: this queue, finish %llu", + (((entity->finish>>10)*1000)>>10)>>2); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, entity); +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 30d019fc67e0..25da0d1c0622 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -387,7 +387,7 @@ static bool bfq_differentiated_weights(struct bfq_data *bfqd) + return (!RB_EMPTY_ROOT(&bfqd->queue_weights_tree) && + (bfqd->queue_weights_tree.rb_node->rb_left || + bfqd->queue_weights_tree.rb_node->rb_right) +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + ) || + (!RB_EMPTY_ROOT(&bfqd->group_weights_tree) && + (bfqd->group_weights_tree.rb_node->rb_left || +@@ -1672,7 +1672,7 @@ static void bfq_merged_request(struct request_queue *q, struct request *req, + } + } + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + static void bfq_bio_merged(struct request_queue *q, struct request *req, + struct bio *bio) + { +@@ -3879,7 +3879,7 @@ static int bfq_dispatch_requests(struct request_queue *q, int force) + */ + static void bfq_put_queue(struct bfq_queue *bfqq) + { +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + struct bfq_group *bfqg = bfqq_group(bfqq); + #endif + +@@ -3909,7 +3909,7 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); + + kmem_cache_free(bfq_pool, bfqq); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + bfqg_put(bfqg); + #endif + } +@@ -4835,7 +4835,7 @@ static void bfq_exit_queue(struct elevator_queue *e) + + BUG_ON(hrtimer_active(&bfqd->idle_slice_timer)); + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_deactivate_policy(q, &blkcg_policy_bfq); + #else + bfq_put_async_queues(bfqd, bfqd->root_group); +@@ -4850,7 +4850,7 @@ static void bfq_init_root_group(struct bfq_group *root_group, + { + int i; + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + root_group->entity.parent = NULL; + root_group->my_entity = NULL; + root_group->bfqd = bfqd; +@@ -5265,7 +5265,7 @@ static struct elevator_type iosched_bfq = { + .elevator_merge_fn = bfq_merge, + .elevator_merged_fn = bfq_merged_request, + .elevator_merge_req_fn = bfq_merged_requests, +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + .elevator_bio_merged_fn = bfq_bio_merged, + #endif + .elevator_allow_bio_merge_fn = bfq_allow_bio_merge, +@@ -5292,7 +5292,7 @@ static struct elevator_type iosched_bfq = { + .elevator_owner = THIS_MODULE, + }; + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + static struct blkcg_policy blkcg_policy_bfq = { + .dfl_cftypes = bfq_blkg_files, + .legacy_cftypes = bfq_blkcg_legacy_files, +@@ -5315,7 +5315,7 @@ static int __init bfq_init(void) + int ret; + char msg[60] = "BFQ I/O-scheduler: v8r12"; + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + ret = blkcg_policy_register(&blkcg_policy_bfq); + if (ret) + return ret; +@@ -5362,7 +5362,7 @@ static int __init bfq_init(void) + if (ret) + goto err_pol_unreg; + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + strcat(msg, " (with cgroups support)"); + #endif + pr_info("%s", msg); +@@ -5370,7 +5370,7 @@ static int __init bfq_init(void) + return 0; + + err_pol_unreg: +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_policy_unregister(&blkcg_policy_bfq); + #endif + return ret; +@@ -5379,7 +5379,7 @@ static int __init bfq_init(void) + static void __exit bfq_exit(void) + { + elv_unregister(&iosched_bfq); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_policy_unregister(&blkcg_policy_bfq); + #endif + bfq_slab_kill(); +diff --git a/block/bfq.h b/block/bfq.h +index 34fc4697fd89..53954d1b87f8 100644 +--- a/block/bfq.h ++++ b/block/bfq.h +@@ -19,6 +19,18 @@ + #include + #include + ++/* ++ * Define an alternative macro to compile cgroups support. This is one ++ * of the steps needed to let bfq-mq share the files bfq-sched.c and ++ * bfq-cgroup.c with bfq-sq. For bfq-mq, the macro ++ * BFQ_GROUP_IOSCHED_ENABLED will be defined as a function of whether ++ * the configuration option CONFIG_BFQ_MQ_GROUP_IOSCHED, and not ++ * CONFIG_BFQ_GROUP_IOSCHED, is defined. ++ */ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#define BFQ_GROUP_IOSCHED_ENABLED ++#endif ++ + #define BFQ_IOPRIO_CLASSES 3 + #define BFQ_CL_IDLE_TIMEOUT (HZ/5) + +@@ -344,7 +356,7 @@ struct bfq_io_cq { + struct bfq_ttime ttime; + /* per (request_queue, blkcg) ioprio */ + int ioprio; +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + uint64_t blkcg_serial_nr; /* the current blkcg serial */ + #endif + +@@ -671,7 +683,7 @@ static const char *checked_dev_name(const struct device *dev) + return nodev; + } + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); + static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + +@@ -696,7 +708,7 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + __pbuf, ##args); \ + } while (0) + +-#else /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++#else /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ + pr_crit("%s bfq%d%c " fmt "\n", \ +@@ -705,7 +717,7 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + ##args) + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) + +-#endif /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++#endif /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log(bfqd, fmt, args...) \ + pr_crit("%s bfq " fmt "\n", \ +@@ -713,7 +725,7 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + ##args) + + #else /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); + static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + +@@ -735,7 +747,7 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + blk_add_trace_msg((bfqd)->queue, "%s " fmt, __pbuf, ##args); \ + } while (0) + +-#else /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++#else /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ + blk_add_trace_msg((bfqd)->queue, "bfq%d%c " fmt, (bfqq)->pid, \ +@@ -743,7 +755,7 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + ##args) + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) + +-#endif /* CONFIG_BFQ_SQ_GROUP_IOSCHED */ ++#endif /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log(bfqd, fmt, args...) \ + blk_add_trace_msg((bfqd)->queue, "bfq " fmt, ##args) +@@ -763,7 +775,7 @@ enum bfqq_expiration { + + + struct bfqg_stats { +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + /* number of ios merged */ + struct blkg_rwstat merged; + /* total time spent on device in ns, may not be accurate w/ queueing */ +@@ -794,7 +806,7 @@ struct bfqg_stats { + #endif + }; + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + /* + * struct bfq_group_data - per-blkcg storage for the blkio subsystem. + * +@@ -895,7 +907,7 @@ bfq_entity_service_tree(struct bfq_entity *entity) + bfq_log_bfqq(bfqq->bfqd, bfqq, + "entity_service_tree %p %d", + sched_data->service_tree + idx, idx); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = + container_of(entity, struct bfq_group, entity); +@@ -924,7 +936,7 @@ static struct bfq_data *bic_to_bfqd(struct bfq_io_cq *bic) + return bic->icq.q->elevator->elevator_data; + } + +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + + static struct bfq_group *bfq_bfqq_to_bfqg(struct bfq_queue *bfqq) + { +@@ -953,7 +965,7 @@ static struct bfq_queue *bfq_get_queue(struct bfq_data *bfqd, + struct bfq_io_cq *bic); + static void bfq_end_wr_async_queues(struct bfq_data *bfqd, + struct bfq_group *bfqg); +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + static void bfq_put_async_queues(struct bfq_data *bfqd, struct bfq_group *bfqg); + #endif + static void bfq_exit_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq); + +From add91dbd756cf8ca3aa3add9a19eef742d5fca6b Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 20 Jan 2017 09:18:25 +0100 +Subject: [PATCH 08/51] Increase max policies for io controller + +To let bfq-mq policy be plugged too (however cgroups +suppport is not yet functional in bfq-mq). + +Signed-off-by: Paolo Valente +--- + include/linux/blkdev.h | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h +index bf000c58644b..10f892ca585d 100644 +--- a/include/linux/blkdev.h ++++ b/include/linux/blkdev.h +@@ -54,7 +54,7 @@ struct blk_stat_callback; + * Maximum number of blkcg policies allowed to be registered concurrently. + * Defined here to simplify include dependency. + */ +-#define BLKCG_MAX_POLS 4 ++#define BLKCG_MAX_POLS 5 + + typedef void (rq_end_io_fn)(struct request *, blk_status_t); + + +From 2c39a1d9ab4516d44e01e96f19f578b927e7f2e9 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Mon, 19 Dec 2016 18:11:33 +0100 +Subject: [PATCH 09/51] Copy header file bfq.h as bfq-mq.h + +This commit introduces the header file bfq-mq.h, that will play +for bfq-mq-iosched.c the same role that bfq.h plays for bfq-iosched.c. + +For the moment, the file bfq-mq.h is just a copy of bfq.h. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 2 +- + block/bfq-mq.h | 973 +++++++++++++++++++++++++++++++++++++++++++++++++ + 2 files changed, 974 insertions(+), 1 deletion(-) + create mode 100644 block/bfq-mq.h + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index e88e00f1e0a7..d1125aee658c 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -83,7 +83,7 @@ + #include + #include "blk.h" + #undef CONFIG_BFQ_GROUP_IOSCHED /* cgroups support not yet functional */ +-#include "bfq.h" ++#include "bfq-mq.h" + + /* Expiration time of sync (0) and async (1) requests, in ns. */ + static const u64 bfq_fifo_expire[2] = { NSEC_PER_SEC / 4, NSEC_PER_SEC / 8 }; +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +new file mode 100644 +index 000000000000..53954d1b87f8 +--- /dev/null ++++ b/block/bfq-mq.h +@@ -0,0 +1,973 @@ ++/* ++ * BFQ v8r12 for 4.11.0: data structures and common functions prototypes. ++ * ++ * Based on ideas and code from CFQ: ++ * Copyright (C) 2003 Jens Axboe ++ * ++ * Copyright (C) 2008 Fabio Checconi ++ * Paolo Valente ++ * ++ * Copyright (C) 2015 Paolo Valente ++ * ++ * Copyright (C) 2017 Paolo Valente ++ */ ++ ++#ifndef _BFQ_H ++#define _BFQ_H ++ ++#include ++#include ++#include ++ ++/* ++ * Define an alternative macro to compile cgroups support. This is one ++ * of the steps needed to let bfq-mq share the files bfq-sched.c and ++ * bfq-cgroup.c with bfq-sq. For bfq-mq, the macro ++ * BFQ_GROUP_IOSCHED_ENABLED will be defined as a function of whether ++ * the configuration option CONFIG_BFQ_MQ_GROUP_IOSCHED, and not ++ * CONFIG_BFQ_GROUP_IOSCHED, is defined. ++ */ ++#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++#define BFQ_GROUP_IOSCHED_ENABLED ++#endif ++ ++#define BFQ_IOPRIO_CLASSES 3 ++#define BFQ_CL_IDLE_TIMEOUT (HZ/5) ++ ++#define BFQ_MIN_WEIGHT 1 ++#define BFQ_MAX_WEIGHT 1000 ++#define BFQ_WEIGHT_CONVERSION_COEFF 10 ++ ++#define BFQ_DEFAULT_QUEUE_IOPRIO 4 ++ ++#define BFQ_WEIGHT_LEGACY_DFL 100 ++#define BFQ_DEFAULT_GRP_IOPRIO 0 ++#define BFQ_DEFAULT_GRP_CLASS IOPRIO_CLASS_BE ++ ++/* ++ * Soft real-time applications are extremely more latency sensitive ++ * than interactive ones. Over-raise the weight of the former to ++ * privilege them against the latter. ++ */ ++#define BFQ_SOFTRT_WEIGHT_FACTOR 100 ++ ++struct bfq_entity; ++ ++/** ++ * struct bfq_service_tree - per ioprio_class service tree. ++ * ++ * Each service tree represents a B-WF2Q+ scheduler on its own. Each ++ * ioprio_class has its own independent scheduler, and so its own ++ * bfq_service_tree. All the fields are protected by the queue lock ++ * of the containing bfqd. ++ */ ++struct bfq_service_tree { ++ /* tree for active entities (i.e., those backlogged) */ ++ struct rb_root active; ++ /* tree for idle entities (i.e., not backlogged, with V <= F_i)*/ ++ struct rb_root idle; ++ ++ struct bfq_entity *first_idle; /* idle entity with minimum F_i */ ++ struct bfq_entity *last_idle; /* idle entity with maximum F_i */ ++ ++ u64 vtime; /* scheduler virtual time */ ++ /* scheduler weight sum; active and idle entities contribute to it */ ++ unsigned long wsum; ++}; ++ ++/** ++ * struct bfq_sched_data - multi-class scheduler. ++ * ++ * bfq_sched_data is the basic scheduler queue. It supports three ++ * ioprio_classes, and can be used either as a toplevel queue or as an ++ * intermediate queue in a hierarchical setup. ++ * ++ * The supported ioprio_classes are the same as in CFQ, in descending ++ * priority order, IOPRIO_CLASS_RT, IOPRIO_CLASS_BE, IOPRIO_CLASS_IDLE. ++ * Requests from higher priority queues are served before all the ++ * requests from lower priority queues; among requests of the same ++ * queue requests are served according to B-WF2Q+. ++ * ++ * The schedule is implemented by the service trees, plus the field ++ * @next_in_service, which points to the entity on the active trees ++ * that will be served next, if 1) no changes in the schedule occurs ++ * before the current in-service entity is expired, 2) the in-service ++ * queue becomes idle when it expires, and 3) if the entity pointed by ++ * in_service_entity is not a queue, then the in-service child entity ++ * of the entity pointed by in_service_entity becomes idle on ++ * expiration. This peculiar definition allows for the following ++ * optimization, not yet exploited: while a given entity is still in ++ * service, we already know which is the best candidate for next ++ * service among the other active entitities in the same parent ++ * entity. We can then quickly compare the timestamps of the ++ * in-service entity with those of such best candidate. ++ * ++ * All the fields are protected by the queue lock of the containing ++ * bfqd. ++ */ ++struct bfq_sched_data { ++ struct bfq_entity *in_service_entity; /* entity in service */ ++ /* head-of-the-line entity in the scheduler (see comments above) */ ++ struct bfq_entity *next_in_service; ++ /* array of service trees, one per ioprio_class */ ++ struct bfq_service_tree service_tree[BFQ_IOPRIO_CLASSES]; ++ /* last time CLASS_IDLE was served */ ++ unsigned long bfq_class_idle_last_service; ++ ++}; ++ ++/** ++ * struct bfq_weight_counter - counter of the number of all active entities ++ * with a given weight. ++ */ ++struct bfq_weight_counter { ++ unsigned int weight; /* weight of the entities this counter refers to */ ++ unsigned int num_active; /* nr of active entities with this weight */ ++ /* ++ * Weights tree member (see bfq_data's @queue_weights_tree and ++ * @group_weights_tree) ++ */ ++ struct rb_node weights_node; ++}; ++ ++/** ++ * struct bfq_entity - schedulable entity. ++ * ++ * A bfq_entity is used to represent either a bfq_queue (leaf node in the ++ * cgroup hierarchy) or a bfq_group into the upper level scheduler. Each ++ * entity belongs to the sched_data of the parent group in the cgroup ++ * hierarchy. Non-leaf entities have also their own sched_data, stored ++ * in @my_sched_data. ++ * ++ * Each entity stores independently its priority values; this would ++ * allow different weights on different devices, but this ++ * functionality is not exported to userspace by now. Priorities and ++ * weights are updated lazily, first storing the new values into the ++ * new_* fields, then setting the @prio_changed flag. As soon as ++ * there is a transition in the entity state that allows the priority ++ * update to take place the effective and the requested priority ++ * values are synchronized. ++ * ++ * Unless cgroups are used, the weight value is calculated from the ++ * ioprio to export the same interface as CFQ. When dealing with ++ * ``well-behaved'' queues (i.e., queues that do not spend too much ++ * time to consume their budget and have true sequential behavior, and ++ * when there are no external factors breaking anticipation) the ++ * relative weights at each level of the cgroups hierarchy should be ++ * guaranteed. All the fields are protected by the queue lock of the ++ * containing bfqd. ++ */ ++struct bfq_entity { ++ struct rb_node rb_node; /* service_tree member */ ++ /* pointer to the weight counter associated with this entity */ ++ struct bfq_weight_counter *weight_counter; ++ ++ /* ++ * Flag, true if the entity is on a tree (either the active or ++ * the idle one of its service_tree) or is in service. ++ */ ++ bool on_st; ++ ++ u64 finish; /* B-WF2Q+ finish timestamp (aka F_i) */ ++ u64 start; /* B-WF2Q+ start timestamp (aka S_i) */ ++ ++ /* tree the entity is enqueued into; %NULL if not on a tree */ ++ struct rb_root *tree; ++ ++ /* ++ * minimum start time of the (active) subtree rooted at this ++ * entity; used for O(log N) lookups into active trees ++ */ ++ u64 min_start; ++ ++ /* amount of service received during the last service slot */ ++ int service; ++ ++ /* budget, used also to calculate F_i: F_i = S_i + @budget / @weight */ ++ int budget; ++ ++ unsigned int weight; /* weight of the queue */ ++ unsigned int new_weight; /* next weight if a change is in progress */ ++ ++ /* original weight, used to implement weight boosting */ ++ unsigned int orig_weight; ++ ++ /* parent entity, for hierarchical scheduling */ ++ struct bfq_entity *parent; ++ ++ /* ++ * For non-leaf nodes in the hierarchy, the associated ++ * scheduler queue, %NULL on leaf nodes. ++ */ ++ struct bfq_sched_data *my_sched_data; ++ /* the scheduler queue this entity belongs to */ ++ struct bfq_sched_data *sched_data; ++ ++ /* flag, set to request a weight, ioprio or ioprio_class change */ ++ int prio_changed; ++}; ++ ++struct bfq_group; ++ ++/** ++ * struct bfq_queue - leaf schedulable entity. ++ * ++ * A bfq_queue is a leaf request queue; it can be associated with an ++ * io_context or more, if it is async or shared between cooperating ++ * processes. @cgroup holds a reference to the cgroup, to be sure that it ++ * does not disappear while a bfqq still references it (mostly to avoid ++ * races between request issuing and task migration followed by cgroup ++ * destruction). ++ * All the fields are protected by the queue lock of the containing bfqd. ++ */ ++struct bfq_queue { ++ /* reference counter */ ++ int ref; ++ /* parent bfq_data */ ++ struct bfq_data *bfqd; ++ ++ /* current ioprio and ioprio class */ ++ unsigned short ioprio, ioprio_class; ++ /* next ioprio and ioprio class if a change is in progress */ ++ unsigned short new_ioprio, new_ioprio_class; ++ ++ /* ++ * Shared bfq_queue if queue is cooperating with one or more ++ * other queues. ++ */ ++ struct bfq_queue *new_bfqq; ++ /* request-position tree member (see bfq_group's @rq_pos_tree) */ ++ struct rb_node pos_node; ++ /* request-position tree root (see bfq_group's @rq_pos_tree) */ ++ struct rb_root *pos_root; ++ ++ /* sorted list of pending requests */ ++ struct rb_root sort_list; ++ /* if fifo isn't expired, next request to serve */ ++ struct request *next_rq; ++ /* number of sync and async requests queued */ ++ int queued[2]; ++ /* number of sync and async requests currently allocated */ ++ int allocated[2]; ++ /* number of pending metadata requests */ ++ int meta_pending; ++ /* fifo list of requests in sort_list */ ++ struct list_head fifo; ++ ++ /* entity representing this queue in the scheduler */ ++ struct bfq_entity entity; ++ ++ /* maximum budget allowed from the feedback mechanism */ ++ int max_budget; ++ /* budget expiration (in jiffies) */ ++ unsigned long budget_timeout; ++ ++ /* number of requests on the dispatch list or inside driver */ ++ int dispatched; ++ ++ unsigned int flags; /* status flags.*/ ++ ++ /* node for active/idle bfqq list inside parent bfqd */ ++ struct list_head bfqq_list; ++ ++ /* bit vector: a 1 for each seeky requests in history */ ++ u32 seek_history; ++ ++ /* node for the device's burst list */ ++ struct hlist_node burst_list_node; ++ ++ /* position of the last request enqueued */ ++ sector_t last_request_pos; ++ ++ /* Number of consecutive pairs of request completion and ++ * arrival, such that the queue becomes idle after the ++ * completion, but the next request arrives within an idle ++ * time slice; used only if the queue's IO_bound flag has been ++ * cleared. ++ */ ++ unsigned int requests_within_timer; ++ ++ /* pid of the process owning the queue, used for logging purposes */ ++ pid_t pid; ++ ++ /* ++ * Pointer to the bfq_io_cq owning the bfq_queue, set to %NULL ++ * if the queue is shared. ++ */ ++ struct bfq_io_cq *bic; ++ ++ /* current maximum weight-raising time for this queue */ ++ unsigned long wr_cur_max_time; ++ /* ++ * Minimum time instant such that, only if a new request is ++ * enqueued after this time instant in an idle @bfq_queue with ++ * no outstanding requests, then the task associated with the ++ * queue it is deemed as soft real-time (see the comments on ++ * the function bfq_bfqq_softrt_next_start()) ++ */ ++ unsigned long soft_rt_next_start; ++ /* ++ * Start time of the current weight-raising period if ++ * the @bfq-queue is being weight-raised, otherwise ++ * finish time of the last weight-raising period. ++ */ ++ unsigned long last_wr_start_finish; ++ /* factor by which the weight of this queue is multiplied */ ++ unsigned int wr_coeff; ++ /* ++ * Time of the last transition of the @bfq_queue from idle to ++ * backlogged. ++ */ ++ unsigned long last_idle_bklogged; ++ /* ++ * Cumulative service received from the @bfq_queue since the ++ * last transition from idle to backlogged. ++ */ ++ unsigned long service_from_backlogged; ++ /* ++ * Value of wr start time when switching to soft rt ++ */ ++ unsigned long wr_start_at_switch_to_srt; ++ ++ unsigned long split_time; /* time of last split */ ++}; ++ ++/** ++ * struct bfq_ttime - per process thinktime stats. ++ */ ++struct bfq_ttime { ++ u64 last_end_request; /* completion time of last request */ ++ ++ u64 ttime_total; /* total process thinktime */ ++ unsigned long ttime_samples; /* number of thinktime samples */ ++ u64 ttime_mean; /* average process thinktime */ ++ ++}; ++ ++/** ++ * struct bfq_io_cq - per (request_queue, io_context) structure. ++ */ ++struct bfq_io_cq { ++ /* associated io_cq structure */ ++ struct io_cq icq; /* must be the first member */ ++ /* array of two process queues, the sync and the async */ ++ struct bfq_queue *bfqq[2]; ++ /* associated @bfq_ttime struct */ ++ struct bfq_ttime ttime; ++ /* per (request_queue, blkcg) ioprio */ ++ int ioprio; ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ uint64_t blkcg_serial_nr; /* the current blkcg serial */ ++#endif ++ ++ /* ++ * Snapshot of the has_short_time flag before merging; taken ++ * to remember its value while the queue is merged, so as to ++ * be able to restore it in case of split. ++ */ ++ bool saved_has_short_ttime; ++ /* ++ * Same purpose as the previous two fields for the I/O bound ++ * classification of a queue. ++ */ ++ bool saved_IO_bound; ++ ++ /* ++ * Same purpose as the previous fields for the value of the ++ * field keeping the queue's belonging to a large burst ++ */ ++ bool saved_in_large_burst; ++ /* ++ * True if the queue belonged to a burst list before its merge ++ * with another cooperating queue. ++ */ ++ bool was_in_burst_list; ++ ++ /* ++ * Similar to previous fields: save wr information. ++ */ ++ unsigned long saved_wr_coeff; ++ unsigned long saved_last_wr_start_finish; ++ unsigned long saved_wr_start_at_switch_to_srt; ++ unsigned int saved_wr_cur_max_time; ++}; ++ ++enum bfq_device_speed { ++ BFQ_BFQD_FAST, ++ BFQ_BFQD_SLOW, ++}; ++ ++/** ++ * struct bfq_data - per-device data structure. ++ * ++ * All the fields are protected by the @queue lock. ++ */ ++struct bfq_data { ++ /* request queue for the device */ ++ struct request_queue *queue; ++ ++ /* root bfq_group for the device */ ++ struct bfq_group *root_group; ++ ++ /* ++ * rbtree of weight counters of @bfq_queues, sorted by ++ * weight. Used to keep track of whether all @bfq_queues have ++ * the same weight. The tree contains one counter for each ++ * distinct weight associated to some active and not ++ * weight-raised @bfq_queue (see the comments to the functions ++ * bfq_weights_tree_[add|remove] for further details). ++ */ ++ struct rb_root queue_weights_tree; ++ /* ++ * rbtree of non-queue @bfq_entity weight counters, sorted by ++ * weight. Used to keep track of whether all @bfq_groups have ++ * the same weight. The tree contains one counter for each ++ * distinct weight associated to some active @bfq_group (see ++ * the comments to the functions bfq_weights_tree_[add|remove] ++ * for further details). ++ */ ++ struct rb_root group_weights_tree; ++ ++ /* ++ * Number of bfq_queues containing requests (including the ++ * queue in service, even if it is idling). ++ */ ++ int busy_queues; ++ /* number of weight-raised busy @bfq_queues */ ++ int wr_busy_queues; ++ /* number of queued requests */ ++ int queued; ++ /* number of requests dispatched and waiting for completion */ ++ int rq_in_driver; ++ ++ /* ++ * Maximum number of requests in driver in the last ++ * @hw_tag_samples completed requests. ++ */ ++ int max_rq_in_driver; ++ /* number of samples used to calculate hw_tag */ ++ int hw_tag_samples; ++ /* flag set to one if the driver is showing a queueing behavior */ ++ int hw_tag; ++ ++ /* number of budgets assigned */ ++ int budgets_assigned; ++ ++ /* ++ * Timer set when idling (waiting) for the next request from ++ * the queue in service. ++ */ ++ struct hrtimer idle_slice_timer; ++ /* delayed work to restart dispatching on the request queue */ ++ struct work_struct unplug_work; ++ ++ /* bfq_queue in service */ ++ struct bfq_queue *in_service_queue; ++ /* bfq_io_cq (bic) associated with the @in_service_queue */ ++ struct bfq_io_cq *in_service_bic; ++ ++ /* on-disk position of the last served request */ ++ sector_t last_position; ++ ++ /* time of last request completion (ns) */ ++ u64 last_completion; ++ ++ /* time of first rq dispatch in current observation interval (ns) */ ++ u64 first_dispatch; ++ /* time of last rq dispatch in current observation interval (ns) */ ++ u64 last_dispatch; ++ ++ /* beginning of the last budget */ ++ ktime_t last_budget_start; ++ /* beginning of the last idle slice */ ++ ktime_t last_idling_start; ++ ++ /* number of samples in current observation interval */ ++ int peak_rate_samples; ++ /* num of samples of seq dispatches in current observation interval */ ++ u32 sequential_samples; ++ /* total num of sectors transferred in current observation interval */ ++ u64 tot_sectors_dispatched; ++ /* max rq size seen during current observation interval (sectors) */ ++ u32 last_rq_max_size; ++ /* time elapsed from first dispatch in current observ. interval (us) */ ++ u64 delta_from_first; ++ /* current estimate of device peak rate */ ++ u32 peak_rate; ++ ++ /* maximum budget allotted to a bfq_queue before rescheduling */ ++ int bfq_max_budget; ++ ++ /* list of all the bfq_queues active on the device */ ++ struct list_head active_list; ++ /* list of all the bfq_queues idle on the device */ ++ struct list_head idle_list; ++ ++ /* ++ * Timeout for async/sync requests; when it fires, requests ++ * are served in fifo order. ++ */ ++ u64 bfq_fifo_expire[2]; ++ /* weight of backward seeks wrt forward ones */ ++ unsigned int bfq_back_penalty; ++ /* maximum allowed backward seek */ ++ unsigned int bfq_back_max; ++ /* maximum idling time */ ++ u32 bfq_slice_idle; ++ ++ /* user-configured max budget value (0 for auto-tuning) */ ++ int bfq_user_max_budget; ++ /* ++ * Timeout for bfq_queues to consume their budget; used to ++ * prevent seeky queues from imposing long latencies to ++ * sequential or quasi-sequential ones (this also implies that ++ * seeky queues cannot receive guarantees in the service ++ * domain; after a timeout they are charged for the time they ++ * have been in service, to preserve fairness among them, but ++ * without service-domain guarantees). ++ */ ++ unsigned int bfq_timeout; ++ ++ /* ++ * Number of consecutive requests that must be issued within ++ * the idle time slice to set again idling to a queue which ++ * was marked as non-I/O-bound (see the definition of the ++ * IO_bound flag for further details). ++ */ ++ unsigned int bfq_requests_within_timer; ++ ++ /* ++ * Force device idling whenever needed to provide accurate ++ * service guarantees, without caring about throughput ++ * issues. CAVEAT: this may even increase latencies, in case ++ * of useless idling for processes that did stop doing I/O. ++ */ ++ bool strict_guarantees; ++ ++ /* ++ * Last time at which a queue entered the current burst of ++ * queues being activated shortly after each other; for more ++ * details about this and the following parameters related to ++ * a burst of activations, see the comments on the function ++ * bfq_handle_burst. ++ */ ++ unsigned long last_ins_in_burst; ++ /* ++ * Reference time interval used to decide whether a queue has ++ * been activated shortly after @last_ins_in_burst. ++ */ ++ unsigned long bfq_burst_interval; ++ /* number of queues in the current burst of queue activations */ ++ int burst_size; ++ ++ /* common parent entity for the queues in the burst */ ++ struct bfq_entity *burst_parent_entity; ++ /* Maximum burst size above which the current queue-activation ++ * burst is deemed as 'large'. ++ */ ++ unsigned long bfq_large_burst_thresh; ++ /* true if a large queue-activation burst is in progress */ ++ bool large_burst; ++ /* ++ * Head of the burst list (as for the above fields, more ++ * details in the comments on the function bfq_handle_burst). ++ */ ++ struct hlist_head burst_list; ++ ++ /* if set to true, low-latency heuristics are enabled */ ++ bool low_latency; ++ /* ++ * Maximum factor by which the weight of a weight-raised queue ++ * is multiplied. ++ */ ++ unsigned int bfq_wr_coeff; ++ /* maximum duration of a weight-raising period (jiffies) */ ++ unsigned int bfq_wr_max_time; ++ ++ /* Maximum weight-raising duration for soft real-time processes */ ++ unsigned int bfq_wr_rt_max_time; ++ /* ++ * Minimum idle period after which weight-raising may be ++ * reactivated for a queue (in jiffies). ++ */ ++ unsigned int bfq_wr_min_idle_time; ++ /* ++ * Minimum period between request arrivals after which ++ * weight-raising may be reactivated for an already busy async ++ * queue (in jiffies). ++ */ ++ unsigned long bfq_wr_min_inter_arr_async; ++ ++ /* Max service-rate for a soft real-time queue, in sectors/sec */ ++ unsigned int bfq_wr_max_softrt_rate; ++ /* ++ * Cached value of the product R*T, used for computing the ++ * maximum duration of weight raising automatically. ++ */ ++ u64 RT_prod; ++ /* device-speed class for the low-latency heuristic */ ++ enum bfq_device_speed device_speed; ++ ++ /* fallback dummy bfqq for extreme OOM conditions */ ++ struct bfq_queue oom_bfqq; ++}; ++ ++enum bfqq_state_flags { ++ BFQ_BFQQ_FLAG_just_created = 0, /* queue just allocated */ ++ BFQ_BFQQ_FLAG_busy, /* has requests or is in service */ ++ BFQ_BFQQ_FLAG_wait_request, /* waiting for a request */ ++ BFQ_BFQQ_FLAG_non_blocking_wait_rq, /* ++ * waiting for a request ++ * without idling the device ++ */ ++ BFQ_BFQQ_FLAG_must_alloc, /* must be allowed rq alloc */ ++ BFQ_BFQQ_FLAG_fifo_expire, /* FIFO checked in this slice */ ++ BFQ_BFQQ_FLAG_has_short_ttime, /* queue has a short think time */ ++ BFQ_BFQQ_FLAG_sync, /* synchronous queue */ ++ BFQ_BFQQ_FLAG_IO_bound, /* ++ * bfqq has timed-out at least once ++ * having consumed at most 2/10 of ++ * its budget ++ */ ++ BFQ_BFQQ_FLAG_in_large_burst, /* ++ * bfqq activated in a large burst, ++ * see comments to bfq_handle_burst. ++ */ ++ BFQ_BFQQ_FLAG_softrt_update, /* ++ * may need softrt-next-start ++ * update ++ */ ++ BFQ_BFQQ_FLAG_coop, /* bfqq is shared */ ++ BFQ_BFQQ_FLAG_split_coop /* shared bfqq will be split */ ++}; ++ ++#define BFQ_BFQQ_FNS(name) \ ++static void bfq_mark_bfqq_##name(struct bfq_queue *bfqq) \ ++{ \ ++ (bfqq)->flags |= (1 << BFQ_BFQQ_FLAG_##name); \ ++} \ ++static void bfq_clear_bfqq_##name(struct bfq_queue *bfqq) \ ++{ \ ++ (bfqq)->flags &= ~(1 << BFQ_BFQQ_FLAG_##name); \ ++} \ ++static int bfq_bfqq_##name(const struct bfq_queue *bfqq) \ ++{ \ ++ return ((bfqq)->flags & (1 << BFQ_BFQQ_FLAG_##name)) != 0; \ ++} ++ ++BFQ_BFQQ_FNS(just_created); ++BFQ_BFQQ_FNS(busy); ++BFQ_BFQQ_FNS(wait_request); ++BFQ_BFQQ_FNS(non_blocking_wait_rq); ++BFQ_BFQQ_FNS(must_alloc); ++BFQ_BFQQ_FNS(fifo_expire); ++BFQ_BFQQ_FNS(has_short_ttime); ++BFQ_BFQQ_FNS(sync); ++BFQ_BFQQ_FNS(IO_bound); ++BFQ_BFQQ_FNS(in_large_burst); ++BFQ_BFQQ_FNS(coop); ++BFQ_BFQQ_FNS(split_coop); ++BFQ_BFQQ_FNS(softrt_update); ++#undef BFQ_BFQQ_FNS ++ ++/* Logging facilities. */ ++#ifdef CONFIG_BFQ_REDIRECT_TO_CONSOLE ++ ++static const char *checked_dev_name(const struct device *dev) ++{ ++ static const char nodev[] = "nodev"; ++ ++ if (dev) ++ return dev_name(dev); ++ ++ return nodev; ++} ++ ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); ++static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ ++ char __pbuf[128]; \ ++ \ ++ assert_spin_locked((bfqd)->queue->queue_lock); \ ++ blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ ++ pr_crit("%s bfq%d%c %s " fmt "\n", \ ++ checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ ++ (bfqq)->pid, \ ++ bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ ++ __pbuf, ##args); \ ++} while (0) ++ ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ ++ char __pbuf[128]; \ ++ \ ++ blkg_path(bfqg_to_blkg(bfqg), __pbuf, sizeof(__pbuf)); \ ++ pr_crit("%s %s " fmt "\n", \ ++ checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ ++ __pbuf, ##args); \ ++} while (0) ++ ++#else /* BFQ_GROUP_IOSCHED_ENABLED */ ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ ++ pr_crit("%s bfq%d%c " fmt "\n", \ ++ checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ ++ (bfqq)->pid, bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ ++ ##args) ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) ++ ++#endif /* BFQ_GROUP_IOSCHED_ENABLED */ ++ ++#define bfq_log(bfqd, fmt, args...) \ ++ pr_crit("%s bfq " fmt "\n", \ ++ checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ ++ ##args) ++ ++#else /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); ++static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ ++ char __pbuf[128]; \ ++ \ ++ assert_spin_locked((bfqd)->queue->queue_lock); \ ++ blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ ++ blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s " fmt, \ ++ (bfqq)->pid, \ ++ bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ ++ __pbuf, ##args); \ ++} while (0) ++ ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ ++ char __pbuf[128]; \ ++ \ ++ blkg_path(bfqg_to_blkg(bfqg), __pbuf, sizeof(__pbuf)); \ ++ blk_add_trace_msg((bfqd)->queue, "%s " fmt, __pbuf, ##args); \ ++} while (0) ++ ++#else /* BFQ_GROUP_IOSCHED_ENABLED */ ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ ++ blk_add_trace_msg((bfqd)->queue, "bfq%d%c " fmt, (bfqq)->pid, \ ++ bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ ++ ##args) ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) ++ ++#endif /* BFQ_GROUP_IOSCHED_ENABLED */ ++ ++#define bfq_log(bfqd, fmt, args...) \ ++ blk_add_trace_msg((bfqd)->queue, "bfq " fmt, ##args) ++#endif /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ ++ ++/* Expiration reasons. */ ++enum bfqq_expiration { ++ BFQ_BFQQ_TOO_IDLE = 0, /* ++ * queue has been idling for ++ * too long ++ */ ++ BFQ_BFQQ_BUDGET_TIMEOUT, /* budget took too long to be used */ ++ BFQ_BFQQ_BUDGET_EXHAUSTED, /* budget consumed */ ++ BFQ_BFQQ_NO_MORE_REQUESTS, /* the queue has no more requests */ ++ BFQ_BFQQ_PREEMPTED /* preemption in progress */ ++}; ++ ++ ++struct bfqg_stats { ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ /* number of ios merged */ ++ struct blkg_rwstat merged; ++ /* total time spent on device in ns, may not be accurate w/ queueing */ ++ struct blkg_rwstat service_time; ++ /* total time spent waiting in scheduler queue in ns */ ++ struct blkg_rwstat wait_time; ++ /* number of IOs queued up */ ++ struct blkg_rwstat queued; ++ /* total disk time and nr sectors dispatched by this group */ ++ struct blkg_stat time; ++ /* sum of number of ios queued across all samples */ ++ struct blkg_stat avg_queue_size_sum; ++ /* count of samples taken for average */ ++ struct blkg_stat avg_queue_size_samples; ++ /* how many times this group has been removed from service tree */ ++ struct blkg_stat dequeue; ++ /* total time spent waiting for it to be assigned a timeslice. */ ++ struct blkg_stat group_wait_time; ++ /* time spent idling for this blkcg_gq */ ++ struct blkg_stat idle_time; ++ /* total time with empty current active q with other requests queued */ ++ struct blkg_stat empty_time; ++ /* fields after this shouldn't be cleared on stat reset */ ++ uint64_t start_group_wait_time; ++ uint64_t start_idle_time; ++ uint64_t start_empty_time; ++ uint16_t flags; ++#endif ++}; ++ ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++/* ++ * struct bfq_group_data - per-blkcg storage for the blkio subsystem. ++ * ++ * @ps: @blkcg_policy_storage that this structure inherits ++ * @weight: weight of the bfq_group ++ */ ++struct bfq_group_data { ++ /* must be the first member */ ++ struct blkcg_policy_data pd; ++ ++ unsigned int weight; ++}; ++ ++/** ++ * struct bfq_group - per (device, cgroup) data structure. ++ * @entity: schedulable entity to insert into the parent group sched_data. ++ * @sched_data: own sched_data, to contain child entities (they may be ++ * both bfq_queues and bfq_groups). ++ * @bfqd: the bfq_data for the device this group acts upon. ++ * @async_bfqq: array of async queues for all the tasks belonging to ++ * the group, one queue per ioprio value per ioprio_class, ++ * except for the idle class that has only one queue. ++ * @async_idle_bfqq: async queue for the idle class (ioprio is ignored). ++ * @my_entity: pointer to @entity, %NULL for the toplevel group; used ++ * to avoid too many special cases during group creation/ ++ * migration. ++ * @active_entities: number of active entities belonging to the group; ++ * unused for the root group. Used to know whether there ++ * are groups with more than one active @bfq_entity ++ * (see the comments to the function ++ * bfq_bfqq_may_idle()). ++ * @rq_pos_tree: rbtree sorted by next_request position, used when ++ * determining if two or more queues have interleaving ++ * requests (see bfq_find_close_cooperator()). ++ * ++ * Each (device, cgroup) pair has its own bfq_group, i.e., for each cgroup ++ * there is a set of bfq_groups, each one collecting the lower-level ++ * entities belonging to the group that are acting on the same device. ++ * ++ * Locking works as follows: ++ * o @bfqd is protected by the queue lock, RCU is used to access it ++ * from the readers. ++ * o All the other fields are protected by the @bfqd queue lock. ++ */ ++struct bfq_group { ++ /* must be the first member */ ++ struct blkg_policy_data pd; ++ ++ struct bfq_entity entity; ++ struct bfq_sched_data sched_data; ++ ++ void *bfqd; ++ ++ struct bfq_queue *async_bfqq[2][IOPRIO_BE_NR]; ++ struct bfq_queue *async_idle_bfqq; ++ ++ struct bfq_entity *my_entity; ++ ++ int active_entities; ++ ++ struct rb_root rq_pos_tree; ++ ++ struct bfqg_stats stats; ++}; ++ ++#else ++struct bfq_group { ++ struct bfq_sched_data sched_data; ++ ++ struct bfq_queue *async_bfqq[2][IOPRIO_BE_NR]; ++ struct bfq_queue *async_idle_bfqq; ++ ++ struct rb_root rq_pos_tree; ++}; ++#endif ++ ++static struct bfq_queue *bfq_entity_to_bfqq(struct bfq_entity *entity); ++ ++static unsigned int bfq_class_idx(struct bfq_entity *entity) ++{ ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ ++ return bfqq ? bfqq->ioprio_class - 1 : ++ BFQ_DEFAULT_GRP_CLASS - 1; ++} ++ ++static struct bfq_service_tree * ++bfq_entity_service_tree(struct bfq_entity *entity) ++{ ++ struct bfq_sched_data *sched_data = entity->sched_data; ++ struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); ++ unsigned int idx = bfq_class_idx(entity); ++ ++ BUG_ON(idx >= BFQ_IOPRIO_CLASSES); ++ BUG_ON(sched_data == NULL); ++ ++ if (bfqq) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "entity_service_tree %p %d", ++ sched_data->service_tree + idx, idx); ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ else { ++ struct bfq_group *bfqg = ++ container_of(entity, struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, ++ "entity_service_tree %p %d", ++ sched_data->service_tree + idx, idx); ++ } ++#endif ++ return sched_data->service_tree + idx; ++} ++ ++static struct bfq_queue *bic_to_bfqq(struct bfq_io_cq *bic, bool is_sync) ++{ ++ return bic->bfqq[is_sync]; ++} ++ ++static void bic_set_bfqq(struct bfq_io_cq *bic, struct bfq_queue *bfqq, ++ bool is_sync) ++{ ++ bic->bfqq[is_sync] = bfqq; ++} ++ ++static struct bfq_data *bic_to_bfqd(struct bfq_io_cq *bic) ++{ ++ return bic->icq.q->elevator->elevator_data; ++} ++ ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ ++static struct bfq_group *bfq_bfqq_to_bfqg(struct bfq_queue *bfqq) ++{ ++ struct bfq_entity *group_entity = bfqq->entity.parent; ++ ++ if (!group_entity) ++ group_entity = &bfqq->bfqd->root_group->entity; ++ ++ return container_of(group_entity, struct bfq_group, entity); ++} ++ ++#else ++ ++static struct bfq_group *bfq_bfqq_to_bfqg(struct bfq_queue *bfqq) ++{ ++ return bfqq->bfqd->root_group; ++} ++ ++#endif ++ ++static void bfq_check_ioprio_change(struct bfq_io_cq *bic, struct bio *bio); ++static void bfq_put_queue(struct bfq_queue *bfqq); ++static void bfq_dispatch_insert(struct request_queue *q, struct request *rq); ++static struct bfq_queue *bfq_get_queue(struct bfq_data *bfqd, ++ struct bio *bio, bool is_sync, ++ struct bfq_io_cq *bic); ++static void bfq_end_wr_async_queues(struct bfq_data *bfqd, ++ struct bfq_group *bfqg); ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++static void bfq_put_async_queues(struct bfq_data *bfqd, struct bfq_group *bfqg); ++#endif ++static void bfq_exit_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq); ++ ++#endif /* _BFQ_H */ + +From 0bd96428e086fd28800efdf5f0a5f62869af6e30 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Sat, 21 Jan 2017 12:41:14 +0100 +Subject: [PATCH 10/51] Move thinktime from bic to bfqq + +Prep change to make it possible to protect this field with a +scheduler lock. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 28 ++++++++++++++-------------- + block/bfq-mq.h | 30 ++++++++++++++++-------------- + 2 files changed, 30 insertions(+), 28 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index d1125aee658c..65f5dfb79417 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -698,6 +698,7 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + if (unlikely(busy)) + old_wr_coeff = bfqq->wr_coeff; + ++ bfqq->ttime = bic->saved_ttime; + bfqq->wr_coeff = bic->saved_wr_coeff; + bfqq->wr_start_at_switch_to_srt = bic->saved_wr_start_at_switch_to_srt; + BUG_ON(time_is_after_jiffies(bfqq->wr_start_at_switch_to_srt)); +@@ -1287,7 +1288,7 @@ static void bfq_bfqq_handle_idle_busy_switch(struct bfq_data *bfqd, + * details on the usage of the next variable. + */ + arrived_in_time = ktime_get_ns() <= +- RQ_BIC(rq)->ttime.last_end_request + ++ bfqq->ttime.last_end_request + + bfqd->bfq_slice_idle * 3; + + bfq_log_bfqq(bfqd, bfqq, +@@ -2048,6 +2049,7 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq) + if (!bic) + return; + ++ bic->saved_ttime = bfqq->ttime; + bic->saved_has_short_ttime = bfq_bfqq_has_short_ttime(bfqq); + bic->saved_IO_bound = bfq_bfqq_IO_bound(bfqq); + bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq); +@@ -3948,11 +3950,6 @@ static void bfq_exit_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfq_put_queue(bfqq); /* release process reference */ + } + +-static void bfq_init_icq(struct io_cq *icq) +-{ +- icq_to_bic(icq)->ttime.last_end_request = ktime_get_ns() - (1ULL<<32); +-} +- + static void bfq_exit_icq(struct io_cq *icq) + { + struct bfq_io_cq *bic = icq_to_bic(icq); +@@ -4084,6 +4081,9 @@ static void bfq_init_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, + bfq_mark_bfqq_just_created(bfqq); + } else + bfq_clear_bfqq_sync(bfqq); ++ ++ bfqq->ttime.last_end_request = ktime_get_ns() - (1ULL<<32); ++ + bfq_mark_bfqq_IO_bound(bfqq); + + /* Tentative initial value to trade off between thr and lat */ +@@ -4191,14 +4191,14 @@ static struct bfq_queue *bfq_get_queue(struct bfq_data *bfqd, + } + + static void bfq_update_io_thinktime(struct bfq_data *bfqd, +- struct bfq_io_cq *bic) ++ struct bfq_queue *bfqq) + { +- struct bfq_ttime *ttime = &bic->ttime; +- u64 elapsed = ktime_get_ns() - bic->ttime.last_end_request; ++ struct bfq_ttime *ttime = &bfqq->ttime; ++ u64 elapsed = ktime_get_ns() - bfqq->ttime.last_end_request; + + elapsed = min_t(u64, elapsed, 2 * bfqd->bfq_slice_idle); + +- ttime->ttime_samples = (7*bic->ttime.ttime_samples + 256) / 8; ++ ttime->ttime_samples = (7*bfqq->ttime.ttime_samples + 256) / 8; + ttime->ttime_total = div_u64(7*ttime->ttime_total + 256*elapsed, 8); + ttime->ttime_mean = div64_ul(ttime->ttime_total + 128, + ttime->ttime_samples); +@@ -4240,8 +4240,8 @@ static void bfq_update_has_short_ttime(struct bfq_data *bfqd, + * decide whether to mark as has_short_ttime + */ + if (atomic_read(&bic->icq.ioc->active_ref) == 0 || +- (bfq_sample_valid(bic->ttime.ttime_samples) && +- bic->ttime.ttime_mean > bfqd->bfq_slice_idle)) ++ (bfq_sample_valid(bfqq->ttime.ttime_samples) && ++ bfqq->ttime.ttime_mean > bfqd->bfq_slice_idle)) + has_short_ttime = false; + + bfq_log_bfqq(bfqd, bfqq, "update_has_short_ttime: has_short_ttime %d", +@@ -4265,7 +4265,7 @@ static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, + if (rq->cmd_flags & REQ_META) + bfqq->meta_pending++; + +- bfq_update_io_thinktime(bfqd, bic); ++ bfq_update_io_thinktime(bfqd, bfqq); + bfq_update_has_short_ttime(bfqd, bfqq, bic); + bfq_update_io_seektime(bfqd, bfqq, rq); + +@@ -4436,7 +4436,7 @@ static void bfq_completed_request(struct request_queue *q, struct request *rq) + + now_ns = ktime_get_ns(); + +- RQ_BIC(rq)->ttime.last_end_request = now_ns; ++ bfqq->ttime.last_end_request = now_ns; + + /* + * Using us instead of ns, to get a reasonable precision in +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 53954d1b87f8..0f51f270469c 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -210,6 +210,18 @@ struct bfq_entity { + struct bfq_group; + + /** ++ * struct bfq_ttime - per process thinktime stats. ++ */ ++struct bfq_ttime { ++ u64 last_end_request; /* completion time of last request */ ++ ++ u64 ttime_total; /* total process thinktime */ ++ unsigned long ttime_samples; /* number of thinktime samples */ ++ u64 ttime_mean; /* average process thinktime */ ++ ++}; ++ ++/** + * struct bfq_queue - leaf schedulable entity. + * + * A bfq_queue is a leaf request queue; it can be associated with an +@@ -270,6 +282,9 @@ struct bfq_queue { + /* node for active/idle bfqq list inside parent bfqd */ + struct list_head bfqq_list; + ++ /* associated @bfq_ttime struct */ ++ struct bfq_ttime ttime; ++ + /* bit vector: a 1 for each seeky requests in history */ + u32 seek_history; + +@@ -333,18 +348,6 @@ struct bfq_queue { + }; + + /** +- * struct bfq_ttime - per process thinktime stats. +- */ +-struct bfq_ttime { +- u64 last_end_request; /* completion time of last request */ +- +- u64 ttime_total; /* total process thinktime */ +- unsigned long ttime_samples; /* number of thinktime samples */ +- u64 ttime_mean; /* average process thinktime */ +- +-}; +- +-/** + * struct bfq_io_cq - per (request_queue, io_context) structure. + */ + struct bfq_io_cq { +@@ -352,8 +355,6 @@ struct bfq_io_cq { + struct io_cq icq; /* must be the first member */ + /* array of two process queues, the sync and the async */ + struct bfq_queue *bfqq[2]; +- /* associated @bfq_ttime struct */ +- struct bfq_ttime ttime; + /* per (request_queue, blkcg) ioprio */ + int ioprio; + #ifdef BFQ_GROUP_IOSCHED_ENABLED +@@ -390,6 +391,7 @@ struct bfq_io_cq { + unsigned long saved_last_wr_start_finish; + unsigned long saved_wr_start_at_switch_to_srt; + unsigned int saved_wr_cur_max_time; ++ struct bfq_ttime saved_ttime; + }; + + enum bfq_device_speed { + +From 351a9aea7c0c9c30edacdbf2a3c0d089470de1e8 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 18 Jan 2017 11:42:22 +0100 +Subject: [PATCH 11/51] Embed bfq-ioc.c and add locking on request queue + +The version of bfq-ioc.c for bfq-iosched.c is not correct any more for +bfq-mq, because, in bfq-mq, the request queue lock is not being held +when bfq_bic_lookup is invoked. That function must then take that look +on its own. This commit removes the inclusion of bfq-ioc.c, copies the +content of bfq-ioc.c into bfq-mq-iosched.c, and adds the grabbing of +the lock. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 39 ++++++++++++++++++++++++++++++++++++--- + 1 file changed, 36 insertions(+), 3 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 65f5dfb79417..756a618d5902 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -195,7 +195,39 @@ static int device_speed_thresh[2]; + + static void bfq_schedule_dispatch(struct bfq_data *bfqd); + +-#include "bfq-ioc.c" ++/** ++ * icq_to_bic - convert iocontext queue structure to bfq_io_cq. ++ * @icq: the iocontext queue. ++ */ ++static struct bfq_io_cq *icq_to_bic(struct io_cq *icq) ++{ ++ /* bic->icq is the first member, %NULL will convert to %NULL */ ++ return container_of(icq, struct bfq_io_cq, icq); ++} ++ ++/** ++ * bfq_bic_lookup - search into @ioc a bic associated to @bfqd. ++ * @bfqd: the lookup key. ++ * @ioc: the io_context of the process doing I/O. ++ * @q: the request queue. ++ */ ++static struct bfq_io_cq *bfq_bic_lookup(struct bfq_data *bfqd, ++ struct io_context *ioc, ++ struct request_queue *q) ++{ ++ if (ioc) { ++ struct bfq_io_cq *icq; ++ ++ spin_lock_irq(q->queue_lock); ++ icq = icq_to_bic(ioc_lookup_icq(ioc, q)); ++ spin_unlock_irq(q->queue_lock); ++ ++ return icq; ++ } ++ ++ return NULL; ++} ++ + #include "bfq-sched.c" + #include "bfq-cgroup-included.c" + +@@ -1520,13 +1552,14 @@ static void bfq_add_request(struct request *rq) + } + + static struct request *bfq_find_rq_fmerge(struct bfq_data *bfqd, +- struct bio *bio) ++ struct bio *bio, ++ struct request_queue *q) + { + struct task_struct *tsk = current; + struct bfq_io_cq *bic; + struct bfq_queue *bfqq; + +- bic = bfq_bic_lookup(bfqd, tsk->io_context); ++ bic = bfq_bic_lookup(bfqd, tsk->io_context, q); + if (!bic) + return NULL; + + +From ed0d64e27b2308813a2a846139e405e0479f0849 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Tue, 20 Dec 2016 09:07:19 +0100 +Subject: [PATCH 12/51] Modify interface and operation to comply with + blk-mq-sched + +As for modifications of the operation, the major changes are the introduction +of a scheduler lock, and the moving to deferred work of the body of the hook +exit_icq. The latter change has been made to avoid deadlocks caused by the +combination of the following facts: 1) such a body takes the scheduler lock, +and, if not deferred, 2) it does so from inside the exit_icq hook, which is +invoked with the queue lock held, and 3) there is at least one code path, +namely that starting from bfq_bio_merge, which takes these locks in the +opposite order. + +Signed-off-by: Paolo Valente +--- + block/bfq-cgroup-included.c | 4 - + block/bfq-mq-iosched.c | 695 ++++++++++++++++++++++++-------------------- + block/bfq-mq.h | 35 +-- + 3 files changed, 394 insertions(+), 340 deletions(-) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index 9c483b658179..8a73de76f32b 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -472,8 +472,6 @@ static struct bfq_group *bfq_find_set_group(struct bfq_data *bfqd, + struct bfq_group *bfqg, *parent; + struct bfq_entity *entity; + +- assert_spin_locked(bfqd->queue->queue_lock); +- + bfqg = bfq_lookup_bfqg(bfqd, blkcg); + + if (unlikely(!bfqg)) +@@ -602,8 +600,6 @@ static struct bfq_group *__bfq_bic_change_cgroup(struct bfq_data *bfqd, + struct bfq_group *bfqg; + struct bfq_entity *entity; + +- lockdep_assert_held(bfqd->queue->queue_lock); +- + bfqg = bfq_find_set_group(bfqd, blkcg); + + if (unlikely(!bfqg)) +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 756a618d5902..c963d92a32c2 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -81,7 +81,13 @@ + #include + #include + #include ++#include ++#include ++ + #include "blk.h" ++#include "blk-mq.h" ++#include "blk-mq-tag.h" ++#include "blk-mq-sched.h" + #undef CONFIG_BFQ_GROUP_IOSCHED /* cgroups support not yet functional */ + #include "bfq-mq.h" + +@@ -193,8 +199,6 @@ static int device_speed_thresh[2]; + #define RQ_BIC(rq) ((struct bfq_io_cq *) (rq)->elv.priv[0]) + #define RQ_BFQQ(rq) ((rq)->elv.priv[1]) + +-static void bfq_schedule_dispatch(struct bfq_data *bfqd); +- + /** + * icq_to_bic - convert iocontext queue structure to bfq_io_cq. + * @icq: the iocontext queue. +@@ -216,11 +220,12 @@ static struct bfq_io_cq *bfq_bic_lookup(struct bfq_data *bfqd, + struct request_queue *q) + { + if (ioc) { ++ unsigned long flags; + struct bfq_io_cq *icq; + +- spin_lock_irq(q->queue_lock); ++ spin_lock_irqsave(q->queue_lock, flags); + icq = icq_to_bic(ioc_lookup_icq(ioc, q)); +- spin_unlock_irq(q->queue_lock); ++ spin_unlock_irqrestore(q->queue_lock, flags); + + return icq; + } +@@ -244,7 +249,7 @@ static void bfq_schedule_dispatch(struct bfq_data *bfqd) + { + if (bfqd->queued != 0) { + bfq_log(bfqd, "schedule dispatch"); +- kblockd_schedule_work(&bfqd->unplug_work); ++ blk_mq_run_hw_queues(bfqd->queue, true); + } + } + +@@ -768,9 +773,7 @@ static int bfqq_process_refs(struct bfq_queue *bfqq) + { + int process_refs, io_refs; + +- lockdep_assert_held(bfqq->bfqd->queue->queue_lock); +- +- io_refs = bfqq->allocated[READ] + bfqq->allocated[WRITE]; ++ io_refs = bfqq->allocated; + process_refs = bfqq->ref - io_refs - bfqq->entity.on_st; + BUG_ON(process_refs < 0); + return process_refs; +@@ -1584,6 +1587,7 @@ static sector_t get_sdist(sector_t last_pos, struct request *rq) + return sdist; + } + ++#if 0 /* Still not clear if we can do without next two functions */ + static void bfq_activate_request(struct request_queue *q, struct request *rq) + { + struct bfq_data *bfqd = q->elevator->elevator_data; +@@ -1597,8 +1601,10 @@ static void bfq_deactivate_request(struct request_queue *q, struct request *rq) + BUG_ON(bfqd->rq_in_driver == 0); + bfqd->rq_in_driver--; + } ++#endif + +-static void bfq_remove_request(struct request *rq) ++static void bfq_remove_request(struct request_queue *q, ++ struct request *rq) + { + struct bfq_queue *bfqq = RQ_BFQQ(rq); + struct bfq_data *bfqd = bfqq->bfqd; +@@ -1619,6 +1625,10 @@ static void bfq_remove_request(struct request *rq) + bfqd->queued--; + elv_rb_del(&bfqq->sort_list, rq); + ++ elv_rqhash_del(q, rq); ++ if (q->last_merge == rq) ++ q->last_merge = NULL; ++ + if (RB_EMPTY_ROOT(&bfqq->sort_list)) { + bfqq->next_rq = NULL; + +@@ -1659,13 +1669,36 @@ static void bfq_remove_request(struct request *rq) + bfqg_stats_update_io_remove(bfqq_group(bfqq), rq->cmd_flags); + } + +-static enum elv_merge bfq_merge(struct request_queue *q, struct request **req, +- struct bio *bio) ++static bool bfq_bio_merge(struct blk_mq_hw_ctx *hctx, struct bio *bio) ++{ ++ struct request_queue *q = hctx->queue; ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct request *free = NULL; ++ bool ret; ++ ++ spin_lock_irq(&bfqd->lock); ++ ret = blk_mq_sched_try_merge(q, bio, &free); ++ ++ /* ++ * XXX Not yet freeing without lock held, to avoid an ++ * inconsistency with respect to the lock-protected invocation ++ * of blk_mq_sched_try_insert_merge in bfq_bio_merge. Waiting ++ * for clarifications from Jens. ++ */ ++ if (free) ++ blk_mq_free_request(free); ++ spin_unlock_irq(&bfqd->lock); ++ ++ return ret; ++} ++ ++static int bfq_request_merge(struct request_queue *q, struct request **req, ++ struct bio *bio) + { + struct bfq_data *bfqd = q->elevator->elevator_data; + struct request *__rq; + +- __rq = bfq_find_rq_fmerge(bfqd, bio); ++ __rq = bfq_find_rq_fmerge(bfqd, bio, q); + if (__rq && elv_bio_merge_ok(__rq, bio)) { + *req = __rq; + return ELEVATOR_FRONT_MERGE; +@@ -1674,7 +1707,7 @@ static enum elv_merge bfq_merge(struct request_queue *q, struct request **req, + return ELEVATOR_NO_MERGE; + } + +-static void bfq_merged_request(struct request_queue *q, struct request *req, ++static void bfq_request_merged(struct request_queue *q, struct request *req, + enum elv_merge type) + { + if (type == ELEVATOR_FRONT_MERGE && +@@ -1689,6 +1722,8 @@ static void bfq_merged_request(struct request_queue *q, struct request *req, + /* Reposition request in its sort_list */ + elv_rb_del(&bfqq->sort_list, req); + elv_rb_add(&bfqq->sort_list, req); ++ ++ spin_lock_irq(&bfqd->lock); + /* Choose next request to be served for bfqq */ + prev = bfqq->next_rq; + next_rq = bfq_choose_req(bfqd, bfqq->next_rq, req, +@@ -1704,22 +1739,19 @@ static void bfq_merged_request(struct request_queue *q, struct request *req, + bfq_updated_next_req(bfqd, bfqq); + bfq_pos_tree_add_move(bfqd, bfqq); + } ++ spin_unlock_irq(&bfqd->lock); + } + } + +-#ifdef BFQ_GROUP_IOSCHED_ENABLED +-static void bfq_bio_merged(struct request_queue *q, struct request *req, +- struct bio *bio) +-{ +- bfqg_stats_update_io_merged(bfqq_group(RQ_BFQQ(req)), bio->bi_opf); +-} +-#endif +- +-static void bfq_merged_requests(struct request_queue *q, struct request *rq, ++static void bfq_requests_merged(struct request_queue *q, struct request *rq, + struct request *next) + { + struct bfq_queue *bfqq = RQ_BFQQ(rq), *next_bfqq = RQ_BFQQ(next); + ++ if (!RB_EMPTY_NODE(&rq->rb_node)) ++ goto end; ++ spin_lock_irq(&bfqq->bfqd->lock); ++ + /* + * If next and rq belong to the same bfq_queue and next is older + * than rq, then reposition rq in the fifo (by substituting next +@@ -1740,7 +1772,10 @@ static void bfq_merged_requests(struct request_queue *q, struct request *rq, + if (bfqq->next_rq == next) + bfqq->next_rq = rq; + +- bfq_remove_request(next); ++ bfq_remove_request(q, next); ++ ++ spin_unlock_irq(&bfqq->bfqd->lock); ++end: + bfqg_stats_update_io_merged(bfqq_group(bfqq), next->cmd_flags); + } + +@@ -1786,7 +1821,7 @@ static void bfq_end_wr(struct bfq_data *bfqd) + { + struct bfq_queue *bfqq; + +- spin_lock_irq(bfqd->queue->queue_lock); ++ spin_lock_irq(&bfqd->lock); + + list_for_each_entry(bfqq, &bfqd->active_list, bfqq_list) + bfq_bfqq_end_wr(bfqq); +@@ -1794,7 +1829,7 @@ static void bfq_end_wr(struct bfq_data *bfqd) + bfq_bfqq_end_wr(bfqq); + bfq_end_wr_async(bfqd); + +- spin_unlock_irq(bfqd->queue->queue_lock); ++ spin_unlock_irq(&bfqd->lock); + } + + static sector_t bfq_io_struct_pos(void *io_struct, bool request) +@@ -2184,8 +2219,8 @@ bfq_merge_bfqqs(struct bfq_data *bfqd, struct bfq_io_cq *bic, + bfq_put_queue(bfqq); + } + +-static int bfq_allow_bio_merge(struct request_queue *q, struct request *rq, +- struct bio *bio) ++static bool bfq_allow_bio_merge(struct request_queue *q, struct request *rq, ++ struct bio *bio) + { + struct bfq_data *bfqd = q->elevator->elevator_data; + bool is_sync = op_is_sync(bio->bi_opf); +@@ -2203,7 +2238,7 @@ static int bfq_allow_bio_merge(struct request_queue *q, struct request *rq, + * merge only if rq is queued there. + * Queue lock is held here. + */ +- bic = bfq_bic_lookup(bfqd, current->io_context); ++ bic = bfq_bic_lookup(bfqd, current->io_context, q); + if (!bic) + return false; + +@@ -2228,12 +2263,6 @@ static int bfq_allow_bio_merge(struct request_queue *q, struct request *rq, + return bfqq == RQ_BFQQ(rq); + } + +-static int bfq_allow_rq_merge(struct request_queue *q, struct request *rq, +- struct request *next) +-{ +- return RQ_BFQQ(rq) == RQ_BFQQ(next); +-} +- + /* + * Set the maximum time for the in-service queue to consume its + * budget. This prevents seeky processes from lowering the throughput. +@@ -2264,7 +2293,6 @@ static void __bfq_set_in_service_queue(struct bfq_data *bfqd, + { + if (bfqq) { + bfqg_stats_update_avg_queue_size(bfqq_group(bfqq)); +- bfq_mark_bfqq_must_alloc(bfqq); + bfq_clear_bfqq_fifo_expire(bfqq); + + bfqd->budgets_assigned = (bfqd->budgets_assigned*7 + 256) / 8; +@@ -2703,27 +2731,28 @@ static void bfq_update_peak_rate(struct bfq_data *bfqd, struct request *rq) + } + + /* +- * Move request from internal lists to the dispatch list of the request queue ++ * Remove request from internal lists. + */ +-static void bfq_dispatch_insert(struct request_queue *q, struct request *rq) ++static void bfq_dispatch_remove(struct request_queue *q, struct request *rq) + { + struct bfq_queue *bfqq = RQ_BFQQ(rq); + + /* +- * For consistency, the next instruction should have been executed +- * after removing the request from the queue and dispatching it. +- * We execute instead this instruction before bfq_remove_request() +- * (and hence introduce a temporary inconsistency), for efficiency. +- * In fact, in a forced_dispatch, this prevents two counters related +- * to bfqq->dispatched to risk to be uselessly decremented if bfqq +- * is not in service, and then to be incremented again after +- * incrementing bfqq->dispatched. ++ * For consistency, the next instruction should have been ++ * executed after removing the request from the queue and ++ * dispatching it. We execute instead this instruction before ++ * bfq_remove_request() (and hence introduce a temporary ++ * inconsistency), for efficiency. In fact, should this ++ * dispatch occur for a non in-service bfqq, this anticipated ++ * increment prevents two counters related to bfqq->dispatched ++ * from risking to be, first, uselessly decremented, and then ++ * incremented again when the (new) value of bfqq->dispatched ++ * happens to be taken into account. + */ + bfqq->dispatched++; + bfq_update_peak_rate(q->elevator->elevator_data, rq); + +- bfq_remove_request(rq); +- elv_dispatch_sort(q, rq); ++ bfq_remove_request(q, rq); + } + + static void __bfq_bfqq_expire(struct bfq_data *bfqd, struct bfq_queue *bfqq) +@@ -3605,7 +3634,7 @@ static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) + bfq_log_bfqq(bfqd, bfqq, "select_queue: already in-service queue"); + + if (bfq_may_expire_for_budg_timeout(bfqq) && +- !hrtimer_active(&bfqd->idle_slice_timer) && ++ !bfq_bfqq_wait_request(bfqq) && + !bfq_bfqq_must_idle(bfqq)) + goto expire; + +@@ -3641,7 +3670,6 @@ static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) + * arrives. + */ + if (bfq_bfqq_wait_request(bfqq)) { +- BUG_ON(!hrtimer_active(&bfqd->idle_slice_timer)); + /* + * If we get here: 1) at least a new request + * has arrived but we have not disabled the +@@ -3668,7 +3696,7 @@ static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) + * for a new request, or has requests waiting for a completion and + * may idle after their completion, then keep it anyway. + */ +- if (hrtimer_active(&bfqd->idle_slice_timer) || ++ if (bfq_bfqq_wait_request(bfqq) || + (bfqq->dispatched != 0 && bfq_bfqq_may_idle(bfqq))) { + bfqq = NULL; + goto keep_queue; +@@ -3753,13 +3781,11 @@ static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq) + } + + /* +- * Dispatch one request from bfqq, moving it to the request queue +- * dispatch list. ++ * Dispatch next request from bfqq. + */ +-static int bfq_dispatch_request(struct bfq_data *bfqd, +- struct bfq_queue *bfqq) ++static struct request *bfq_dispatch_rq_from_bfqq(struct bfq_data *bfqd, ++ struct bfq_queue *bfqq) + { +- int dispatched = 0; + struct request *rq = bfqq->next_rq; + unsigned long service_to_charge; + +@@ -3775,7 +3801,7 @@ static int bfq_dispatch_request(struct bfq_data *bfqd, + + BUG_ON(bfqq->entity.budget < bfqq->entity.service); + +- bfq_dispatch_insert(bfqd->queue, rq); ++ bfq_dispatch_remove(bfqd->queue, rq); + + /* + * If weight raising has to terminate for bfqq, then next +@@ -3791,86 +3817,61 @@ static int bfq_dispatch_request(struct bfq_data *bfqd, + bfq_update_wr_data(bfqd, bfqq); + + bfq_log_bfqq(bfqd, bfqq, +- "dispatched %u sec req (%llu), budg left %d", ++ "dispatched %u sec req (%llu), budg left %d, new disp_nr %d", + blk_rq_sectors(rq), + (unsigned long long) blk_rq_pos(rq), +- bfq_bfqq_budget_left(bfqq)); +- +- dispatched++; ++ bfq_bfqq_budget_left(bfqq), ++ bfqq->dispatched); + + if (!bfqd->in_service_bic) { + atomic_long_inc(&RQ_BIC(rq)->icq.ioc->refcount); + bfqd->in_service_bic = RQ_BIC(rq); + } + ++ /* ++ * Expire bfqq, pretending that its budget expired, if bfqq ++ * belongs to CLASS_IDLE and other queues are waiting for ++ * service. ++ */ + if (bfqd->busy_queues > 1 && bfq_class_idle(bfqq)) + goto expire; + +- return dispatched; ++ return rq; + + expire: + bfq_bfqq_expire(bfqd, bfqq, false, BFQ_BFQQ_BUDGET_EXHAUSTED); +- return dispatched; +-} +- +-static int __bfq_forced_dispatch_bfqq(struct bfq_queue *bfqq) +-{ +- int dispatched = 0; +- +- while (bfqq->next_rq) { +- bfq_dispatch_insert(bfqq->bfqd->queue, bfqq->next_rq); +- dispatched++; +- } +- +- BUG_ON(!list_empty(&bfqq->fifo)); +- return dispatched; ++ return rq; + } + +-/* +- * Drain our current requests. +- * Used for barriers and when switching io schedulers on-the-fly. +- */ +-static int bfq_forced_dispatch(struct bfq_data *bfqd) ++static bool bfq_has_work(struct blk_mq_hw_ctx *hctx) + { +- struct bfq_queue *bfqq, *n; +- struct bfq_service_tree *st; +- int dispatched = 0; +- +- bfqq = bfqd->in_service_queue; +- if (bfqq) +- __bfq_bfqq_expire(bfqd, bfqq); ++ struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; + + /* +- * Loop through classes, and be careful to leave the scheduler +- * in a consistent state, as feedback mechanisms and vtime +- * updates cannot be disabled during the process. ++ * Avoiding lock: a race on bfqd->busy_queues should cause at ++ * most a call to dispatch for nothing + */ +- list_for_each_entry_safe(bfqq, n, &bfqd->active_list, bfqq_list) { +- st = bfq_entity_service_tree(&bfqq->entity); +- +- dispatched += __bfq_forced_dispatch_bfqq(bfqq); +- +- bfqq->max_budget = bfq_max_budget(bfqd); +- bfq_forget_idle(st); +- } +- +- BUG_ON(bfqd->busy_queues != 0); +- +- return dispatched; ++ return !list_empty_careful(&bfqd->dispatch) || ++ bfqd->busy_queues > 0; + } + +-static int bfq_dispatch_requests(struct request_queue *q, int force) ++static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + { +- struct bfq_data *bfqd = q->elevator->elevator_data; +- struct bfq_queue *bfqq; ++ struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; ++ struct request *rq = NULL; ++ struct bfq_queue *bfqq = NULL; ++ ++ if (!list_empty(&bfqd->dispatch)) { ++ rq = list_first_entry(&bfqd->dispatch, struct request, ++ queuelist); ++ list_del_init(&rq->queuelist); ++ goto exit; ++ } + + bfq_log(bfqd, "dispatch requests: %d busy queues", bfqd->busy_queues); + + if (bfqd->busy_queues == 0) +- return 0; +- +- if (unlikely(force)) +- return bfq_forced_dispatch(bfqd); ++ goto exit; + + /* + * Force device to serve one request at a time if +@@ -3885,25 +3886,39 @@ static int bfq_dispatch_requests(struct request_queue *q, int force) + * throughput. + */ + if (bfqd->strict_guarantees && bfqd->rq_in_driver > 0) +- return 0; ++ goto exit; + + bfqq = bfq_select_queue(bfqd); + if (!bfqq) +- return 0; ++ goto exit; + + BUG_ON(bfqq->entity.budget < bfqq->entity.service); + + BUG_ON(bfq_bfqq_wait_request(bfqq)); + +- if (!bfq_dispatch_request(bfqd, bfqq)) +- return 0; +- +- bfq_log_bfqq(bfqd, bfqq, "dispatched %s request", +- bfq_bfqq_sync(bfqq) ? "sync" : "async"); ++ rq = bfq_dispatch_rq_from_bfqq(bfqd, bfqq); + + BUG_ON(bfqq->next_rq == NULL && + bfqq->entity.budget < bfqq->entity.service); +- return 1; ++exit: ++ if (rq) { ++ rq->rq_flags |= RQF_STARTED; ++ bfqd->rq_in_driver++; ++ } ++ ++ return rq; ++} ++ ++static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) ++{ ++ struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; ++ struct request *rq; ++ ++ spin_lock_irq(&bfqd->lock); ++ rq = __bfq_dispatch_request(hctx); ++ spin_unlock_irq(&bfqd->lock); ++ ++ return rq; + } + + /* +@@ -3921,13 +3936,14 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + + BUG_ON(bfqq->ref <= 0); + +- bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p %d", bfqq, bfqq->ref); ++ if (bfqq->bfqd) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p %d", bfqq, bfqq->ref); ++ + bfqq->ref--; + if (bfqq->ref) + return; + + BUG_ON(rb_first(&bfqq->sort_list)); +- BUG_ON(bfqq->allocated[READ] + bfqq->allocated[WRITE] != 0); + BUG_ON(bfqq->entity.tree); + BUG_ON(bfq_bfqq_busy(bfqq)); + +@@ -3942,7 +3958,8 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + */ + hlist_del_init(&bfqq->burst_list_node); + +- bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); ++ if (bfqq->bfqd) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); + + kmem_cache_free(bfq_pool, bfqq); + #ifdef BFQ_GROUP_IOSCHED_ENABLED +@@ -3983,29 +4000,52 @@ static void bfq_exit_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfq_put_queue(bfqq); /* release process reference */ + } + +-static void bfq_exit_icq(struct io_cq *icq) ++static void bfq_exit_icq_bfqq(struct bfq_io_cq *bic, bool is_sync) + { +- struct bfq_io_cq *bic = icq_to_bic(icq); +- struct bfq_data *bfqd = bic_to_bfqd(bic); ++ struct bfq_queue *bfqq = bic_to_bfqq(bic, is_sync); ++ struct bfq_data *bfqd; + +- if (bic_to_bfqq(bic, false)) { +- bfq_exit_bfqq(bfqd, bic_to_bfqq(bic, false)); +- bic_set_bfqq(bic, NULL, false); +- } ++ if (bfqq) ++ bfqd = bfqq->bfqd; /* NULL if scheduler already exited */ + +- if (bic_to_bfqq(bic, true)) { ++ if (bfqq && bfqd) { ++ spin_lock_irq(&bfqd->lock); + /* + * If the bic is using a shared queue, put the reference + * taken on the io_context when the bic started using a + * shared bfq_queue. + */ +- if (bfq_bfqq_coop(bic_to_bfqq(bic, true))) +- put_io_context(icq->ioc); +- bfq_exit_bfqq(bfqd, bic_to_bfqq(bic, true)); +- bic_set_bfqq(bic, NULL, true); ++ if (is_sync && bfq_bfqq_coop(bfqq)) ++ put_io_context(bic->icq.ioc); ++ bfq_exit_bfqq(bfqd, bfqq); ++ bic_set_bfqq(bic, NULL, is_sync); ++ spin_unlock_irq(&bfqd->lock); + } + } + ++static void bfq_exit_icq_body(struct work_struct *work) ++{ ++ struct bfq_io_cq *bic = ++ container_of(work, struct bfq_io_cq, exit_icq_work); ++ ++ bfq_exit_icq_bfqq(bic, true); ++ bfq_exit_icq_bfqq(bic, false); ++} ++ ++static void bfq_init_icq(struct io_cq *icq) ++{ ++ struct bfq_io_cq *bic = icq_to_bic(icq); ++ ++ INIT_WORK(&bic->exit_icq_work, bfq_exit_icq_body); ++} ++ ++static void bfq_exit_icq(struct io_cq *icq) ++{ ++ struct bfq_io_cq *bic = icq_to_bic(icq); ++ ++ kblockd_schedule_work(&bic->exit_icq_work); ++} ++ + /* + * Update the entity prio values; note that the new values will not + * be used until the next (re)activation. +@@ -4015,6 +4055,10 @@ static void bfq_set_next_ioprio_data(struct bfq_queue *bfqq, + { + struct task_struct *tsk = current; + int ioprio_class; ++ struct bfq_data *bfqd = bfqq->bfqd; ++ ++ if (!bfqd) ++ return; + + ioprio_class = IOPRIO_PRIO_CLASS(bic->ioprio); + switch (ioprio_class) { +@@ -4095,6 +4139,8 @@ static void bfq_init_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, + INIT_HLIST_NODE(&bfqq->burst_list_node); + BUG_ON(!hlist_unhashed(&bfqq->burst_list_node)); + ++ spin_lock_init(&bfqq->lock); ++ + bfqq->ref = 0; + bfqq->bfqd = bfqd; + +@@ -4351,22 +4397,13 @@ static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, + if (budget_timeout) + bfq_bfqq_expire(bfqd, bfqq, false, + BFQ_BFQQ_BUDGET_TIMEOUT); +- +- /* +- * Let the request rip immediately, or let a new queue be +- * selected if bfqq has just been expired. +- */ +- __blk_run_queue(bfqd->queue); + } + } + +-static void bfq_insert_request(struct request_queue *q, struct request *rq) ++static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + { +- struct bfq_data *bfqd = q->elevator->elevator_data; + struct bfq_queue *bfqq = RQ_BFQQ(rq), *new_bfqq; + +- assert_spin_locked(bfqd->queue->queue_lock); +- + /* + * An unplug may trigger a requeue of a request from the device + * driver: make sure we are in process context while trying to +@@ -4381,8 +4418,8 @@ static void bfq_insert_request(struct request_queue *q, struct request *rq) + * Release the request's reference to the old bfqq + * and make sure one is taken to the shared queue. + */ +- new_bfqq->allocated[rq_data_dir(rq)]++; +- bfqq->allocated[rq_data_dir(rq)]--; ++ new_bfqq->allocated++; ++ bfqq->allocated--; + new_bfqq->ref++; + bfq_clear_bfqq_just_created(bfqq); + if (bic_to_bfqq(RQ_BIC(rq), 1) == bfqq) +@@ -4406,6 +4443,55 @@ static void bfq_insert_request(struct request_queue *q, struct request *rq) + bfq_rq_enqueued(bfqd, bfqq, rq); + } + ++static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, ++ bool at_head) ++{ ++ struct request_queue *q = hctx->queue; ++ struct bfq_data *bfqd = q->elevator->elevator_data; ++ ++ spin_lock_irq(&bfqd->lock); ++ if (blk_mq_sched_try_insert_merge(q, rq)) ++ goto done; ++ spin_unlock_irq(&bfqd->lock); ++ ++ blk_mq_sched_request_inserted(rq); ++ ++ spin_lock_irq(&bfqd->lock); ++ if (at_head || blk_rq_is_passthrough(rq)) { ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ ++ if (at_head) ++ list_add(&rq->queuelist, &bfqd->dispatch); ++ else ++ list_add_tail(&rq->queuelist, &bfqd->dispatch); ++ ++ if (bfqq) ++ bfqq->dispatched++; ++ } else { ++ __bfq_insert_request(bfqd, rq); ++ ++ if (rq_mergeable(rq)) { ++ elv_rqhash_add(q, rq); ++ if (!q->last_merge) ++ q->last_merge = rq; ++ } ++ } ++done: ++ spin_unlock_irq(&bfqd->lock); ++} ++ ++static void bfq_insert_requests(struct blk_mq_hw_ctx *hctx, ++ struct list_head *list, bool at_head) ++{ ++ while (!list_empty(list)) { ++ struct request *rq; ++ ++ rq = list_first_entry(list, struct request, queuelist); ++ list_del_init(&rq->queuelist); ++ bfq_insert_request(hctx, rq, at_head); ++ } ++} ++ + static void bfq_update_hw_tag(struct bfq_data *bfqd) + { + bfqd->max_rq_in_driver = max_t(int, bfqd->max_rq_in_driver, +@@ -4431,27 +4517,17 @@ static void bfq_update_hw_tag(struct bfq_data *bfqd) + bfqd->hw_tag_samples = 0; + } + +-static void bfq_completed_request(struct request_queue *q, struct request *rq) ++static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd) + { +- struct bfq_queue *bfqq = RQ_BFQQ(rq); +- struct bfq_data *bfqd = bfqq->bfqd; + u64 now_ns; + u32 delta_us; + +- bfq_log_bfqq(bfqd, bfqq, "completed one req with %u sects left", +- blk_rq_sectors(rq)); +- +- assert_spin_locked(bfqd->queue->queue_lock); + bfq_update_hw_tag(bfqd); + + BUG_ON(!bfqd->rq_in_driver); + BUG_ON(!bfqq->dispatched); + bfqd->rq_in_driver--; + bfqq->dispatched--; +- bfqg_stats_update_completion(bfqq_group(bfqq), +- rq_start_time_ns(rq), +- rq_io_start_time_ns(rq), +- rq->cmd_flags); + + if (!bfqq->dispatched && !bfq_bfqq_busy(bfqq)) { + BUG_ON(!RB_EMPTY_ROOT(&bfqq->sort_list)); +@@ -4477,7 +4553,8 @@ static void bfq_completed_request(struct request_queue *q, struct request *rq) + */ + delta_us = div_u64(now_ns - bfqd->last_completion, NSEC_PER_USEC); + +- bfq_log(bfqd, "rq_completed: delta %uus/%luus max_size %u rate %llu/%llu", ++ bfq_log_bfqq(bfqd, bfqq, ++ "rq_completed: delta %uus/%luus max_size %u rate %llu/%llu", + delta_us, BFQ_MIN_TT/NSEC_PER_USEC, bfqd->last_rq_max_size, + (USEC_PER_SEC* + (u64)((bfqd->last_rq_max_size<in_service_queue == bfqq) { + if (bfqq->dispatched == 0 && bfq_bfqq_must_idle(bfqq)) { + bfq_arm_slice_timer(bfqd); +- goto out; ++ return; + } else if (bfq_may_expire_for_budg_timeout(bfqq)) + bfq_bfqq_expire(bfqd, bfqq, false, + BFQ_BFQQ_BUDGET_TIMEOUT); +@@ -4537,68 +4614,55 @@ static void bfq_completed_request(struct request_queue *q, struct request *rq) + bfq_bfqq_expire(bfqd, bfqq, false, + BFQ_BFQQ_NO_MORE_REQUESTS); + } +- +- if (!bfqd->rq_in_driver) +- bfq_schedule_dispatch(bfqd); +- +-out: +- return; + } + +-static int __bfq_may_queue(struct bfq_queue *bfqq) ++static void bfq_put_rq_priv_body(struct bfq_queue *bfqq) + { +- if (bfq_bfqq_wait_request(bfqq) && bfq_bfqq_must_alloc(bfqq)) { +- bfq_clear_bfqq_must_alloc(bfqq); +- return ELV_MQUEUE_MUST; +- } ++ bfqq->allocated--; + +- return ELV_MQUEUE_MAY; ++ bfq_put_queue(bfqq); + } + +-static int bfq_may_queue(struct request_queue *q, unsigned int op) ++static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + { +- struct bfq_data *bfqd = q->elevator->elevator_data; +- struct task_struct *tsk = current; +- struct bfq_io_cq *bic; +- struct bfq_queue *bfqq; +- +- /* +- * Don't force setup of a queue from here, as a call to may_queue +- * does not necessarily imply that a request actually will be +- * queued. So just lookup a possibly existing queue, or return +- * 'may queue' if that fails. +- */ +- bic = bfq_bic_lookup(bfqd, tsk->io_context); +- if (!bic) +- return ELV_MQUEUE_MAY; +- +- bfqq = bic_to_bfqq(bic, op_is_sync(op)); +- if (bfqq) +- return __bfq_may_queue(bfqq); ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ struct bfq_data *bfqd = bfqq->bfqd; + +- return ELV_MQUEUE_MAY; +-} ++ if (rq->rq_flags & RQF_STARTED) ++ bfqg_stats_update_completion(bfqq_group(bfqq), ++ rq_start_time_ns(rq), ++ rq_io_start_time_ns(rq), ++ rq->cmd_flags); + +-/* +- * Queue lock held here. +- */ +-static void bfq_put_request(struct request *rq) +-{ +- struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ if (likely(rq->rq_flags & RQF_STARTED)) { ++ unsigned long flags; + +- if (bfqq) { +- const int rw = rq_data_dir(rq); ++ spin_lock_irqsave(&bfqd->lock, flags); + +- BUG_ON(!bfqq->allocated[rw]); +- bfqq->allocated[rw]--; ++ bfq_completed_request(bfqq, bfqd); ++ bfq_put_rq_priv_body(bfqq); + +- rq->elv.priv[0] = NULL; +- rq->elv.priv[1] = NULL; ++ spin_unlock_irqrestore(&bfqd->lock, flags); ++ } else { ++ /* ++ * Request rq may be still/already in the scheduler, ++ * in which case we need to remove it. And we cannot ++ * defer such a check and removal, to avoid ++ * inconsistencies in the time interval from the end ++ * of this function to the start of the deferred work. ++ * Fortunately, this situation occurs only in process ++ * context, so taking the scheduler lock does not ++ * cause any deadlock, even if other locks are already ++ * (correctly) held by this process. ++ */ + +- bfq_log_bfqq(bfqq->bfqd, bfqq, "put_request %p, %d", +- bfqq, bfqq->ref); +- bfq_put_queue(bfqq); ++ if (!RB_EMPTY_NODE(&rq->rb_node)) ++ bfq_remove_request(q, rq); ++ bfq_put_rq_priv_body(bfqq); + } ++ ++ rq->elv.priv[0] = NULL; ++ rq->elv.priv[1] = NULL; + } + + /* +@@ -4630,18 +4694,16 @@ bfq_split_bfqq(struct bfq_io_cq *bic, struct bfq_queue *bfqq) + /* + * Allocate bfq data structures associated with this request. + */ +-static int bfq_set_request(struct request_queue *q, struct request *rq, +- struct bio *bio, gfp_t gfp_mask) ++static int bfq_get_rq_private(struct request_queue *q, struct request *rq, ++ struct bio *bio) + { + struct bfq_data *bfqd = q->elevator->elevator_data; + struct bfq_io_cq *bic = icq_to_bic(rq->elv.icq); +- const int rw = rq_data_dir(rq); + const int is_sync = rq_is_sync(rq); + struct bfq_queue *bfqq; +- unsigned long flags; + bool bfqq_already_existing = false, split = false; + +- spin_lock_irqsave(q->queue_lock, flags); ++ spin_lock_irq(&bfqd->lock); + + if (!bic) + goto queue_fail; +@@ -4661,7 +4723,7 @@ static int bfq_set_request(struct request_queue *q, struct request *rq, + bic_set_bfqq(bic, bfqq, is_sync); + if (split && is_sync) { + bfq_log_bfqq(bfqd, bfqq, +- "set_request: was_in_list %d " ++ "get_request: was_in_list %d " + "was_in_large_burst %d " + "large burst in progress %d", + bic->was_in_burst_list, +@@ -4671,12 +4733,12 @@ static int bfq_set_request(struct request_queue *q, struct request *rq, + if ((bic->was_in_burst_list && bfqd->large_burst) || + bic->saved_in_large_burst) { + bfq_log_bfqq(bfqd, bfqq, +- "set_request: marking in " ++ "get_request: marking in " + "large burst"); + bfq_mark_bfqq_in_large_burst(bfqq); + } else { + bfq_log_bfqq(bfqd, bfqq, +- "set_request: clearing in " ++ "get_request: clearing in " + "large burst"); + bfq_clear_bfqq_in_large_burst(bfqq); + if (bic->was_in_burst_list) +@@ -4703,9 +4765,12 @@ static int bfq_set_request(struct request_queue *q, struct request *rq, + } + } + +- bfqq->allocated[rw]++; ++ bfqq->allocated++; ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "get_request: new allocated %d", bfqq->allocated); ++ + bfqq->ref++; +- bfq_log_bfqq(bfqd, bfqq, "set_request: bfqq %p, %d", bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqd, bfqq, "get_request: bfqq %p, %d", bfqq, bfqq->ref); + + rq->elv.priv[0] = bic; + rq->elv.priv[1] = bfqq; +@@ -4733,26 +4798,53 @@ static int bfq_set_request(struct request_queue *q, struct request *rq, + if (unlikely(bfq_bfqq_just_created(bfqq))) + bfq_handle_burst(bfqd, bfqq); + +- spin_unlock_irqrestore(q->queue_lock, flags); ++ spin_unlock_irq(&bfqd->lock); + + return 0; + + queue_fail: +- bfq_schedule_dispatch(bfqd); +- spin_unlock_irqrestore(q->queue_lock, flags); ++ spin_unlock_irq(&bfqd->lock); + + return 1; + } + +-static void bfq_kick_queue(struct work_struct *work) ++static void bfq_idle_slice_timer_body(struct bfq_queue *bfqq) + { +- struct bfq_data *bfqd = +- container_of(work, struct bfq_data, unplug_work); +- struct request_queue *q = bfqd->queue; ++ struct bfq_data *bfqd = bfqq->bfqd; ++ enum bfqq_expiration reason; ++ unsigned long flags; ++ ++ spin_lock_irqsave(&bfqd->lock, flags); ++ bfq_clear_bfqq_wait_request(bfqq); + +- spin_lock_irq(q->queue_lock); +- __blk_run_queue(q); +- spin_unlock_irq(q->queue_lock); ++ if (bfqq != bfqd->in_service_queue) { ++ spin_unlock_irqrestore(&bfqd->lock, flags); ++ return; ++ } ++ ++ if (bfq_bfqq_budget_timeout(bfqq)) ++ /* ++ * Also here the queue can be safely expired ++ * for budget timeout without wasting ++ * guarantees ++ */ ++ reason = BFQ_BFQQ_BUDGET_TIMEOUT; ++ else if (bfqq->queued[0] == 0 && bfqq->queued[1] == 0) ++ /* ++ * The queue may not be empty upon timer expiration, ++ * because we may not disable the timer when the ++ * first request of the in-service queue arrives ++ * during disk idling. ++ */ ++ reason = BFQ_BFQQ_TOO_IDLE; ++ else ++ goto schedule_dispatch; ++ ++ bfq_bfqq_expire(bfqd, bfqq, true, reason); ++ ++schedule_dispatch: ++ spin_unlock_irqrestore(&bfqd->lock, flags); ++ bfq_schedule_dispatch(bfqd); + } + + /* +@@ -4763,59 +4855,22 @@ static enum hrtimer_restart bfq_idle_slice_timer(struct hrtimer *timer) + { + struct bfq_data *bfqd = container_of(timer, struct bfq_data, + idle_slice_timer); +- struct bfq_queue *bfqq; +- unsigned long flags; +- enum bfqq_expiration reason; +- +- spin_lock_irqsave(bfqd->queue->queue_lock, flags); ++ struct bfq_queue *bfqq = bfqd->in_service_queue; + +- bfqq = bfqd->in_service_queue; + /* + * Theoretical race here: the in-service queue can be NULL or +- * different from the queue that was idling if the timer handler +- * spins on the queue_lock and a new request arrives for the +- * current queue and there is a full dispatch cycle that changes +- * the in-service queue. This can hardly happen, but in the worst +- * case we just expire a queue too early. ++ * different from the queue that was idling if a new request ++ * arrives for the current queue and there is a full dispatch ++ * cycle that changes the in-service queue. This can hardly ++ * happen, but in the worst case we just expire a queue too ++ * early. + */ +- if (bfqq) { +- bfq_log_bfqq(bfqd, bfqq, "slice_timer expired"); +- bfq_clear_bfqq_wait_request(bfqq); +- +- if (bfq_bfqq_budget_timeout(bfqq)) +- /* +- * Also here the queue can be safely expired +- * for budget timeout without wasting +- * guarantees +- */ +- reason = BFQ_BFQQ_BUDGET_TIMEOUT; +- else if (bfqq->queued[0] == 0 && bfqq->queued[1] == 0) +- /* +- * The queue may not be empty upon timer expiration, +- * because we may not disable the timer when the +- * first request of the in-service queue arrives +- * during disk idling. +- */ +- reason = BFQ_BFQQ_TOO_IDLE; +- else +- goto schedule_dispatch; +- +- bfq_bfqq_expire(bfqd, bfqq, true, reason); +- } +- +-schedule_dispatch: +- bfq_schedule_dispatch(bfqd); ++ if (bfqq) ++ bfq_idle_slice_timer_body(bfqq); + +- spin_unlock_irqrestore(bfqd->queue->queue_lock, flags); + return HRTIMER_NORESTART; + } + +-static void bfq_shutdown_timer_wq(struct bfq_data *bfqd) +-{ +- hrtimer_cancel(&bfqd->idle_slice_timer); +- cancel_work_sync(&bfqd->unplug_work); +-} +- + static void __bfq_put_async_bfqq(struct bfq_data *bfqd, + struct bfq_queue **bfqq_ptr) + { +@@ -4852,28 +4907,40 @@ static void bfq_put_async_queues(struct bfq_data *bfqd, struct bfq_group *bfqg) + static void bfq_exit_queue(struct elevator_queue *e) + { + struct bfq_data *bfqd = e->elevator_data; +- struct request_queue *q = bfqd->queue; + struct bfq_queue *bfqq, *n; + +- bfq_shutdown_timer_wq(bfqd); +- +- spin_lock_irq(q->queue_lock); ++ hrtimer_cancel(&bfqd->idle_slice_timer); + + BUG_ON(bfqd->in_service_queue); +- list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) +- bfq_deactivate_bfqq(bfqd, bfqq, false, false); + +- spin_unlock_irq(q->queue_lock); ++ list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) { ++ if (bfqq->bic) /* bfqqs without bic are handled below */ ++ cancel_work_sync(&bfqq->bic->exit_icq_work); ++ } ++ ++ spin_lock_irq(&bfqd->lock); ++ list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) { ++ bfq_deactivate_bfqq(bfqd, bfqq, false, false); ++ /* ++ * Make sure that deferred exit_icq_work completes ++ * without errors for bfq_queues without bic ++ */ ++ if (!bfqq->bic) ++ bfqq->bfqd = NULL; ++ } ++ spin_unlock_irq(&bfqd->lock); + +- bfq_shutdown_timer_wq(bfqd); ++ hrtimer_cancel(&bfqd->idle_slice_timer); + + BUG_ON(hrtimer_active(&bfqd->idle_slice_timer)); + + #ifdef BFQ_GROUP_IOSCHED_ENABLED +- blkcg_deactivate_policy(q, &blkcg_policy_bfq); ++ blkcg_deactivate_policy(bfqd->queue, &blkcg_policy_bfq); + #else ++ spin_lock_irq(&bfqd->lock); + bfq_put_async_queues(bfqd, bfqd->root_group); + kfree(bfqd->root_group); ++ spin_unlock_irq(&bfqd->lock); + #endif + + kfree(bfqd); +@@ -4934,10 +5001,6 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) + + bfqd->queue = q; + +- spin_lock_irq(q->queue_lock); +- q->elevator = eq; +- spin_unlock_irq(q->queue_lock); +- + bfqd->root_group = bfq_create_group_hierarchy(bfqd, q->node); + if (!bfqd->root_group) + goto out_free; +@@ -4951,8 +5014,6 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) + bfqd->queue_weights_tree = RB_ROOT; + bfqd->group_weights_tree = RB_ROOT; + +- INIT_WORK(&bfqd->unplug_work, bfq_kick_queue); +- + INIT_LIST_HEAD(&bfqd->active_list); + INIT_LIST_HEAD(&bfqd->idle_list); + INIT_HLIST_HEAD(&bfqd->burst_list); +@@ -5001,6 +5062,11 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) + bfqd->peak_rate = R_fast[blk_queue_nonrot(bfqd->queue)] * 2 / 3; + bfqd->device_speed = BFQ_BFQD_FAST; + ++ spin_lock_init(&bfqd->lock); ++ INIT_LIST_HEAD(&bfqd->dispatch); ++ ++ q->elevator = eq; ++ + return 0; + + out_free: +@@ -5057,7 +5123,7 @@ static ssize_t bfq_weights_show(struct elevator_queue *e, char *page) + num_char += sprintf(page + num_char, "Tot reqs queued %d\n\n", + bfqd->queued); + +- spin_lock_irq(bfqd->queue->queue_lock); ++ spin_lock_irq(&bfqd->lock); + + num_char += sprintf(page + num_char, "Active:\n"); + list_for_each_entry(bfqq, &bfqd->active_list, bfqq_list) { +@@ -5086,7 +5152,7 @@ static ssize_t bfq_weights_show(struct elevator_queue *e, char *page) + jiffies_to_msecs(bfqq->wr_cur_max_time)); + } + +- spin_unlock_irq(bfqd->queue->queue_lock); ++ spin_unlock_irq(&bfqd->lock); + + return num_char; + } +@@ -5294,35 +5360,31 @@ static struct elv_fs_entry bfq_attrs[] = { + __ATTR_NULL + }; + +-static struct elevator_type iosched_bfq = { +- .ops.sq = { +- .elevator_merge_fn = bfq_merge, +- .elevator_merged_fn = bfq_merged_request, +- .elevator_merge_req_fn = bfq_merged_requests, +-#ifdef BFQ_GROUP_IOSCHED_ENABLED +- .elevator_bio_merged_fn = bfq_bio_merged, +-#endif +- .elevator_allow_bio_merge_fn = bfq_allow_bio_merge, +- .elevator_allow_rq_merge_fn = bfq_allow_rq_merge, +- .elevator_dispatch_fn = bfq_dispatch_requests, +- .elevator_add_req_fn = bfq_insert_request, +- .elevator_activate_req_fn = bfq_activate_request, +- .elevator_deactivate_req_fn = bfq_deactivate_request, +- .elevator_completed_req_fn = bfq_completed_request, +- .elevator_former_req_fn = elv_rb_former_request, +- .elevator_latter_req_fn = elv_rb_latter_request, +- .elevator_init_icq_fn = bfq_init_icq, +- .elevator_exit_icq_fn = bfq_exit_icq, +- .elevator_set_req_fn = bfq_set_request, +- .elevator_put_req_fn = bfq_put_request, +- .elevator_may_queue_fn = bfq_may_queue, +- .elevator_init_fn = bfq_init_queue, +- .elevator_exit_fn = bfq_exit_queue, ++static struct elevator_type iosched_bfq_mq = { ++ .ops.mq = { ++ .get_rq_priv = bfq_get_rq_private, ++ .put_rq_priv = bfq_put_rq_private, ++ .init_icq = bfq_init_icq, ++ .exit_icq = bfq_exit_icq, ++ .insert_requests = bfq_insert_requests, ++ .dispatch_request = bfq_dispatch_request, ++ .next_request = elv_rb_latter_request, ++ .former_request = elv_rb_former_request, ++ .allow_merge = bfq_allow_bio_merge, ++ .bio_merge = bfq_bio_merge, ++ .request_merge = bfq_request_merge, ++ .requests_merged = bfq_requests_merged, ++ .request_merged = bfq_request_merged, ++ .has_work = bfq_has_work, ++ .init_sched = bfq_init_queue, ++ .exit_sched = bfq_exit_queue, + }, ++ ++ .uses_mq = true, + .icq_size = sizeof(struct bfq_io_cq), + .icq_align = __alignof__(struct bfq_io_cq), + .elevator_attrs = bfq_attrs, +- .elevator_name = "bfq-sq", ++ .elevator_name = "bfq-mq", + .elevator_owner = THIS_MODULE, + }; + +@@ -5392,7 +5454,7 @@ static int __init bfq_init(void) + device_speed_thresh[0] = (4 * R_slow[0]) / 3; + device_speed_thresh[1] = (4 * R_slow[1]) / 3; + +- ret = elv_register(&iosched_bfq); ++ ret = elv_register(&iosched_bfq_mq); + if (ret) + goto err_pol_unreg; + +@@ -5412,8 +5474,8 @@ static int __init bfq_init(void) + + static void __exit bfq_exit(void) + { +- elv_unregister(&iosched_bfq); +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ elv_unregister(&iosched_bfq_mq); ++#ifdef CONFIG_BFQ_GROUP_ENABLED + blkcg_policy_unregister(&blkcg_policy_bfq); + #endif + bfq_slab_kill(); +@@ -5422,5 +5484,6 @@ static void __exit bfq_exit(void) + module_init(bfq_init); + module_exit(bfq_exit); + +-MODULE_AUTHOR("Arianna Avanzini, Fabio Checconi, Paolo Valente"); ++MODULE_AUTHOR("Paolo Valente"); + MODULE_LICENSE("GPL"); ++MODULE_DESCRIPTION("MQ Budget Fair Queueing I/O Scheduler"); +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 0f51f270469c..c3fcd5ebd735 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -19,15 +19,8 @@ + #include + #include + +-/* +- * Define an alternative macro to compile cgroups support. This is one +- * of the steps needed to let bfq-mq share the files bfq-sched.c and +- * bfq-cgroup.c with bfq-sq. For bfq-mq, the macro +- * BFQ_GROUP_IOSCHED_ENABLED will be defined as a function of whether +- * the configuration option CONFIG_BFQ_MQ_GROUP_IOSCHED, and not +- * CONFIG_BFQ_GROUP_IOSCHED, is defined. +- */ +-#ifdef CONFIG_BFQ_SQ_GROUP_IOSCHED ++/* see comments on CONFIG_BFQ_GROUP_IOSCHED in bfq.h */ ++#ifdef CONFIG_BFQ_MQ_GROUP_IOSCHED + #define BFQ_GROUP_IOSCHED_ENABLED + #endif + +@@ -259,8 +252,8 @@ struct bfq_queue { + struct request *next_rq; + /* number of sync and async requests queued */ + int queued[2]; +- /* number of sync and async requests currently allocated */ +- int allocated[2]; ++ /* number of requests currently allocated */ ++ int allocated; + /* number of pending metadata requests */ + int meta_pending; + /* fifo list of requests in sort_list */ +@@ -345,6 +338,8 @@ struct bfq_queue { + unsigned long wr_start_at_switch_to_srt; + + unsigned long split_time; /* time of last split */ ++ ++ spinlock_t lock; + }; + + /** +@@ -361,6 +356,9 @@ struct bfq_io_cq { + uint64_t blkcg_serial_nr; /* the current blkcg serial */ + #endif + ++ /* delayed work to exec the body of the the exit_icq handler */ ++ struct work_struct exit_icq_work; ++ + /* + * Snapshot of the has_short_time flag before merging; taken + * to remember its value while the queue is merged, so as to +@@ -402,11 +400,13 @@ enum bfq_device_speed { + /** + * struct bfq_data - per-device data structure. + * +- * All the fields are protected by the @queue lock. ++ * All the fields are protected by @lock. + */ + struct bfq_data { +- /* request queue for the device */ ++ /* device request queue */ + struct request_queue *queue; ++ /* dispatch queue */ ++ struct list_head dispatch; + + /* root bfq_group for the device */ + struct bfq_group *root_group; +@@ -460,8 +460,6 @@ struct bfq_data { + * the queue in service. + */ + struct hrtimer idle_slice_timer; +- /* delayed work to restart dispatching on the request queue */ +- struct work_struct unplug_work; + + /* bfq_queue in service */ + struct bfq_queue *in_service_queue; +@@ -612,6 +610,8 @@ struct bfq_data { + + /* fallback dummy bfqq for extreme OOM conditions */ + struct bfq_queue oom_bfqq; ++ ++ spinlock_t lock; + }; + + enum bfqq_state_flags { +@@ -622,7 +622,6 @@ enum bfqq_state_flags { + * waiting for a request + * without idling the device + */ +- BFQ_BFQQ_FLAG_must_alloc, /* must be allowed rq alloc */ + BFQ_BFQQ_FLAG_fifo_expire, /* FIFO checked in this slice */ + BFQ_BFQQ_FLAG_has_short_ttime, /* queue has a short think time */ + BFQ_BFQQ_FLAG_sync, /* synchronous queue */ +@@ -661,7 +660,6 @@ BFQ_BFQQ_FNS(just_created); + BFQ_BFQQ_FNS(busy); + BFQ_BFQQ_FNS(wait_request); + BFQ_BFQQ_FNS(non_blocking_wait_rq); +-BFQ_BFQQ_FNS(must_alloc); + BFQ_BFQQ_FNS(fifo_expire); + BFQ_BFQQ_FNS(has_short_ttime); + BFQ_BFQQ_FNS(sync); +@@ -692,7 +690,6 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ + char __pbuf[128]; \ + \ +- assert_spin_locked((bfqd)->queue->queue_lock); \ + blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ + pr_crit("%s bfq%d%c %s " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ +@@ -734,7 +731,6 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ + char __pbuf[128]; \ + \ +- assert_spin_locked((bfqd)->queue->queue_lock); \ + blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ + blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s " fmt, \ + (bfqq)->pid, \ +@@ -961,7 +957,6 @@ static struct bfq_group *bfq_bfqq_to_bfqg(struct bfq_queue *bfqq) + + static void bfq_check_ioprio_change(struct bfq_io_cq *bic, struct bio *bio); + static void bfq_put_queue(struct bfq_queue *bfqq); +-static void bfq_dispatch_insert(struct request_queue *q, struct request *rq); + static struct bfq_queue *bfq_get_queue(struct bfq_data *bfqd, + struct bio *bio, bool is_sync, + struct bfq_io_cq *bic); + +From bde5235de2241502c1c00337bd51c96d9b60b6df Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 3 Mar 2017 08:52:40 +0100 +Subject: [PATCH 13/51] Add checks and extra log messages - Part I + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 112 +++++++++++++++++++++++++++++++++++++++++++++++-- + 1 file changed, 109 insertions(+), 3 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index c963d92a32c2..40eadb3f7073 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -773,6 +773,8 @@ static int bfqq_process_refs(struct bfq_queue *bfqq) + { + int process_refs, io_refs; + ++ lockdep_assert_held(&bfqq->bfqd->lock); ++ + io_refs = bfqq->allocated; + process_refs = bfqq->ref - io_refs - bfqq->entity.on_st; + BUG_ON(process_refs < 0); +@@ -1483,6 +1485,8 @@ static void bfq_add_request(struct request *rq) + bfqq->queued[rq_is_sync(rq)]++; + bfqd->queued++; + ++ BUG_ON(!RQ_BFQQ(rq)); ++ BUG_ON(RQ_BFQQ(rq) != bfqq); + elv_rb_add(&bfqq->sort_list, rq); + + /* +@@ -1491,6 +1495,8 @@ static void bfq_add_request(struct request *rq) + prev = bfqq->next_rq; + next_rq = bfq_choose_req(bfqd, bfqq->next_rq, rq, bfqd->last_position); + BUG_ON(!next_rq); ++ BUG_ON(!RQ_BFQQ(next_rq)); ++ BUG_ON(RQ_BFQQ(next_rq) != bfqq); + bfqq->next_rq = next_rq; + + /* +@@ -1615,6 +1621,19 @@ static void bfq_remove_request(struct request_queue *q, + + if (bfqq->next_rq == rq) { + bfqq->next_rq = bfq_find_next_rq(bfqd, bfqq, rq); ++ if (bfqq->next_rq && !RQ_BFQQ(bfqq->next_rq)) { ++ pr_crit("no bfqq! for next rq %p bfqq %p\n", ++ bfqq->next_rq, bfqq); ++ } ++ ++ BUG_ON(bfqq->next_rq && !RQ_BFQQ(bfqq->next_rq)); ++ if (bfqq->next_rq && RQ_BFQQ(bfqq->next_rq) != bfqq) { ++ pr_crit( ++ "wrong bfqq! for next rq %p, rq_bfqq %p bfqq %p\n", ++ bfqq->next_rq, RQ_BFQQ(bfqq->next_rq), bfqq); ++ } ++ BUG_ON(bfqq->next_rq && RQ_BFQQ(bfqq->next_rq) != bfqq); ++ + bfq_updated_next_req(bfqd, bfqq); + } + +@@ -1701,6 +1720,8 @@ static int bfq_request_merge(struct request_queue *q, struct request **req, + __rq = bfq_find_rq_fmerge(bfqd, bio, q); + if (__rq && elv_bio_merge_ok(__rq, bio)) { + *req = __rq; ++ bfq_log(bfqd, "request_merge: req %p", __rq); ++ + return ELEVATOR_FRONT_MERGE; + } + +@@ -1721,6 +1742,8 @@ static void bfq_request_merged(struct request_queue *q, struct request *req, + + /* Reposition request in its sort_list */ + elv_rb_del(&bfqq->sort_list, req); ++ BUG_ON(!RQ_BFQQ(req)); ++ BUG_ON(RQ_BFQQ(req) != bfqq); + elv_rb_add(&bfqq->sort_list, req); + + spin_lock_irq(&bfqd->lock); +@@ -1729,7 +1752,13 @@ static void bfq_request_merged(struct request_queue *q, struct request *req, + next_rq = bfq_choose_req(bfqd, bfqq->next_rq, req, + bfqd->last_position); + BUG_ON(!next_rq); ++ + bfqq->next_rq = next_rq; ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "requests_merged: req %p prev %p next_rq %p bfqq %p", ++ req, prev, next_rq, bfqq); ++ + /* + * If next_rq changes, update both the queue's budget to + * fit the new request and the queue's position in its +@@ -1748,8 +1777,16 @@ static void bfq_requests_merged(struct request_queue *q, struct request *rq, + { + struct bfq_queue *bfqq = RQ_BFQQ(rq), *next_bfqq = RQ_BFQQ(next); + ++ BUG_ON(!RQ_BFQQ(rq)); ++ BUG_ON(!RQ_BFQQ(next)); ++ + if (!RB_EMPTY_NODE(&rq->rb_node)) + goto end; ++ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "requests_merged: rq %p next %p bfqq %p next_bfqq %p", ++ rq, next, bfqq, next_bfqq); ++ + spin_lock_irq(&bfqq->bfqd->lock); + + /* +@@ -3847,6 +3884,9 @@ static bool bfq_has_work(struct blk_mq_hw_ctx *hctx) + { + struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; + ++ bfq_log(bfqd, "has_work, dispatch_non_empty %d busy_queues %d", ++ !list_empty_careful(&bfqd->dispatch), bfqd->busy_queues > 0); ++ + /* + * Avoiding lock: a race on bfqd->busy_queues should cause at + * most a call to dispatch for nothing +@@ -3865,6 +3905,8 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + rq = list_first_entry(&bfqd->dispatch, struct request, + queuelist); + list_del_init(&rq->queuelist); ++ bfq_log(bfqd, ++ "dispatch requests: picked %p from dispatch list", rq); + goto exit; + } + +@@ -3904,7 +3946,20 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + if (rq) { + rq->rq_flags |= RQF_STARTED; + bfqd->rq_in_driver++; +- } ++ if (bfqq) ++ bfq_log_bfqq(bfqd, bfqq, ++ "dispatched %s request %p, rq_in_driver %d", ++ bfq_bfqq_sync(bfqq) ? "sync" : "async", ++ rq, ++ bfqd->rq_in_driver); ++ else ++ bfq_log(bfqd, ++ "dispatched request %p from dispatch list, rq_in_driver %d", ++ rq, bfqd->rq_in_driver); ++ } else ++ bfq_log(bfqd, ++ "returned NULL request, rq_in_driver %d", ++ bfqd->rq_in_driver); + + return rq; + } +@@ -3944,6 +3999,7 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + return; + + BUG_ON(rb_first(&bfqq->sort_list)); ++ BUG_ON(bfqq->allocated != 0); + BUG_ON(bfqq->entity.tree); + BUG_ON(bfq_bfqq_busy(bfqq)); + +@@ -4043,6 +4099,7 @@ static void bfq_exit_icq(struct io_cq *icq) + { + struct bfq_io_cq *bic = icq_to_bic(icq); + ++ BUG_ON(!bic); + kblockd_schedule_work(&bic->exit_icq_work); + } + +@@ -4057,6 +4114,7 @@ static void bfq_set_next_ioprio_data(struct bfq_queue *bfqq, + int ioprio_class; + struct bfq_data *bfqd = bfqq->bfqd; + ++ WARN_ON(!bfqd); + if (!bfqd) + return; + +@@ -4404,6 +4462,10 @@ static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + { + struct bfq_queue *bfqq = RQ_BFQQ(rq), *new_bfqq; + ++ assert_spin_locked(&bfqd->lock); ++ ++ bfq_log_bfqq(bfqd, bfqq, "__insert_req: rq %p bfqq %p", rq, bfqq); ++ + /* + * An unplug may trigger a requeue of a request from the device + * driver: make sure we are in process context while trying to +@@ -4420,6 +4482,12 @@ static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + */ + new_bfqq->allocated++; + bfqq->allocated--; ++ bfq_log_bfqq(bfqd, bfqq, ++ "insert_request: new allocated %d", bfqq->allocated); ++ bfq_log_bfqq(bfqd, new_bfqq, ++ "insert_request: new_bfqq new allocated %d", ++ bfqq->allocated); ++ + new_bfqq->ref++; + bfq_clear_bfqq_just_created(bfqq); + if (bic_to_bfqq(RQ_BIC(rq), 1) == bfqq) +@@ -4529,6 +4597,10 @@ static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd) + bfqd->rq_in_driver--; + bfqq->dispatched--; + ++ bfq_log_bfqq(bfqd, bfqq, ++ "completed_requests: new disp %d, new rq_in_driver %d", ++ bfqq->dispatched, bfqd->rq_in_driver); ++ + if (!bfqq->dispatched && !bfq_bfqq_busy(bfqq)) { + BUG_ON(!RB_EMPTY_ROOT(&bfqq->sort_list)); + /* +@@ -4618,6 +4690,9 @@ static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd) + + static void bfq_put_rq_priv_body(struct bfq_queue *bfqq) + { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "put_request_body: allocated %d", bfqq->allocated); ++ BUG_ON(!bfqq->allocated); + bfqq->allocated--; + + bfq_put_queue(bfqq); +@@ -4625,8 +4700,27 @@ static void bfq_put_rq_priv_body(struct bfq_queue *bfqq) + + static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + { +- struct bfq_queue *bfqq = RQ_BFQQ(rq); +- struct bfq_data *bfqd = bfqq->bfqd; ++ struct bfq_queue *bfqq; ++ struct bfq_data *bfqd; ++ struct bfq_io_cq *bic; ++ ++ BUG_ON(!rq); ++ bfqq = RQ_BFQQ(rq); ++ BUG_ON(!bfqq); ++ ++ bic = RQ_BIC(rq); ++ BUG_ON(!bic); ++ ++ bfqd = bfqq->bfqd; ++ BUG_ON(!bfqd); ++ ++ BUG_ON(rq->rq_flags & RQF_QUEUED); ++ BUG_ON(!(rq->rq_flags & RQF_ELVPRIV)); ++ ++ bfq_log_bfqq(bfqd, bfqq, ++ "putting rq %p with %u sects left, STARTED %d", ++ rq, blk_rq_sectors(rq), ++ rq->rq_flags & RQF_STARTED); + + if (rq->rq_flags & RQF_STARTED) + bfqg_stats_update_completion(bfqq_group(bfqq), +@@ -4634,6 +4728,8 @@ static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + rq_io_start_time_ns(rq), + rq->cmd_flags); + ++ BUG_ON(blk_rq_sectors(rq) == 0 && !(rq->rq_flags & RQF_STARTED)); ++ + if (likely(rq->rq_flags & RQF_STARTED)) { + unsigned long flags; + +@@ -4655,7 +4751,9 @@ static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + * cause any deadlock, even if other locks are already + * (correctly) held by this process. + */ ++ BUG_ON(in_interrupt()); + ++ assert_spin_locked(&bfqd->lock); + if (!RB_EMPTY_NODE(&rq->rb_node)) + bfq_remove_request(q, rq); + bfq_put_rq_priv_body(bfqq); +@@ -4814,7 +4912,9 @@ static void bfq_idle_slice_timer_body(struct bfq_queue *bfqq) + enum bfqq_expiration reason; + unsigned long flags; + ++ BUG_ON(!bfqd); + spin_lock_irqsave(&bfqd->lock, flags); ++ bfq_log_bfqq(bfqd, bfqq, "handling slice_timer expiration"); + bfq_clear_bfqq_wait_request(bfqq); + + if (bfqq != bfqd->in_service_queue) { +@@ -4857,6 +4957,8 @@ static enum hrtimer_restart bfq_idle_slice_timer(struct hrtimer *timer) + idle_slice_timer); + struct bfq_queue *bfqq = bfqd->in_service_queue; + ++ bfq_log(bfqd, "slice_timer expired"); ++ + /* + * Theoretical race here: the in-service queue can be NULL or + * different from the queue that was idling if a new request +@@ -4909,9 +5011,12 @@ static void bfq_exit_queue(struct elevator_queue *e) + struct bfq_data *bfqd = e->elevator_data; + struct bfq_queue *bfqq, *n; + ++ bfq_log(bfqd, "exit_queue: starting ..."); ++ + hrtimer_cancel(&bfqd->idle_slice_timer); + + BUG_ON(bfqd->in_service_queue); ++ BUG_ON(!list_empty(&bfqd->active_list)); + + list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) { + if (bfqq->bic) /* bfqqs without bic are handled below */ +@@ -4943,6 +5048,7 @@ static void bfq_exit_queue(struct elevator_queue *e) + spin_unlock_irq(&bfqd->lock); + #endif + ++ bfq_log(bfqd, "exit_queue: finished ..."); + kfree(bfqd); + } + + +From 7f59486861e368d25f59d4136cf8e51a75b7edf9 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 9 Feb 2017 10:36:27 +0100 +Subject: [PATCH 14/51] Add lock check in bfq_allow_bio_merge + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 1 + + 1 file changed, 1 insertion(+) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 40eadb3f7073..21b876aeba16 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -2279,6 +2279,7 @@ static bool bfq_allow_bio_merge(struct request_queue *q, struct request *rq, + if (!bic) + return false; + ++ assert_spin_locked(&bfqd->lock); + bfqq = bic_to_bfqq(bic, is_sync); + /* + * We take advantage of this function to perform an early merge + +From a2dd19a4d95cf401268c144c79ce549c7fc4bbca Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Tue, 7 Feb 2017 15:14:29 +0100 +Subject: [PATCH 15/51] bfq-mq: execute exit_icq operations immediately + +Exploting Omar's patch that removes the taking of the queue lock in +put_io_context_active, this patch moves back the operation of the bfq_exit_icq +hook from a deferred work to the body of the function. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 34 +++------------------------------- + block/bfq-mq.h | 3 --- + 2 files changed, 3 insertions(+), 34 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 21b876aeba16..1deb79a47181 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4080,28 +4080,13 @@ static void bfq_exit_icq_bfqq(struct bfq_io_cq *bic, bool is_sync) + } + } + +-static void bfq_exit_icq_body(struct work_struct *work) +-{ +- struct bfq_io_cq *bic = +- container_of(work, struct bfq_io_cq, exit_icq_work); +- +- bfq_exit_icq_bfqq(bic, true); +- bfq_exit_icq_bfqq(bic, false); +-} +- +-static void bfq_init_icq(struct io_cq *icq) +-{ +- struct bfq_io_cq *bic = icq_to_bic(icq); +- +- INIT_WORK(&bic->exit_icq_work, bfq_exit_icq_body); +-} +- + static void bfq_exit_icq(struct io_cq *icq) + { + struct bfq_io_cq *bic = icq_to_bic(icq); + + BUG_ON(!bic); +- kblockd_schedule_work(&bic->exit_icq_work); ++ bfq_exit_icq_bfqq(bic, true); ++ bfq_exit_icq_bfqq(bic, false); + } + + /* +@@ -5019,21 +5004,9 @@ static void bfq_exit_queue(struct elevator_queue *e) + BUG_ON(bfqd->in_service_queue); + BUG_ON(!list_empty(&bfqd->active_list)); + +- list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) { +- if (bfqq->bic) /* bfqqs without bic are handled below */ +- cancel_work_sync(&bfqq->bic->exit_icq_work); +- } +- + spin_lock_irq(&bfqd->lock); +- list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) { ++ list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) + bfq_deactivate_bfqq(bfqd, bfqq, false, false); +- /* +- * Make sure that deferred exit_icq_work completes +- * without errors for bfq_queues without bic +- */ +- if (!bfqq->bic) +- bfqq->bfqd = NULL; +- } + spin_unlock_irq(&bfqd->lock); + + hrtimer_cancel(&bfqd->idle_slice_timer); +@@ -5471,7 +5444,6 @@ static struct elevator_type iosched_bfq_mq = { + .ops.mq = { + .get_rq_priv = bfq_get_rq_private, + .put_rq_priv = bfq_put_rq_private, +- .init_icq = bfq_init_icq, + .exit_icq = bfq_exit_icq, + .insert_requests = bfq_insert_requests, + .dispatch_request = bfq_dispatch_request, +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index c3fcd5ebd735..23744b246db6 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -356,9 +356,6 @@ struct bfq_io_cq { + uint64_t blkcg_serial_nr; /* the current blkcg serial */ + #endif + +- /* delayed work to exec the body of the the exit_icq handler */ +- struct work_struct exit_icq_work; +- + /* + * Snapshot of the has_short_time flag before merging; taken + * to remember its value while the queue is merged, so as to + +From ab7e78a0ff095101de74e700f8743295a500bb20 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Tue, 21 Feb 2017 10:26:22 +0100 +Subject: [PATCH 16/51] Unnest request-queue and ioc locks from scheduler locks + +In some bio-merging functions, the request-queue lock needs to be +taken, to lookup for the bic associated with the process that issued +the bio that may need to be merged. In addition, put_io_context must +be invoked in some other functions, and put_io_context may cause the +lock of the involved ioc to be taken. In both cases, these extra +request-queue or ioc locks are taken, or might be taken, while the +scheduler lock is being held. In this respect, there are other code +paths, in part external to bfq-mq, in which the same locks are taken +(nested) in the opposite order, i.e., it is the scheduler lock to be +taken while the request-queue or the ioc lock is being held. This +leads to circular deadlocks. + +This commit addresses this issue by modifying the logic of the above +functions, so as to let the lookup and put_io_context be performed, +and thus the extra locks be taken, outside the critical sections +protected by the scheduler lock. + +Signed-off-by: Paolo Valente +--- + block/bfq-cgroup-included.c | 9 ++ + block/bfq-mq-iosched.c | 264 ++++++++++++++++++++++++++++---------------- + block/bfq-mq.h | 25 ++++- + block/bfq-sched.c | 11 ++ + 4 files changed, 213 insertions(+), 96 deletions(-) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index 8a73de76f32b..cf59eeb7f08e 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -716,6 +716,9 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) + struct bfq_group *bfqg; + struct bfq_data *bfqd; + struct bfq_entity *entity; ++#ifdef BFQ_MQ ++ unsigned long flags; ++#endif + int i; + + BUG_ON(!pd); +@@ -729,6 +732,9 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) + if (!entity) /* root group */ + return; + ++#ifdef BFQ_MQ ++ spin_lock_irqsave(&bfqd->lock, flags); ++#endif + /* + * Empty all service_trees belonging to this group before + * deactivating the group itself. +@@ -766,6 +772,9 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) + __bfq_deactivate_entity(entity, false); + bfq_put_async_queues(bfqd, bfqg); + ++#ifdef BFQ_MQ ++ bfq_unlock_put_ioc_restore(bfqd, flags); ++#endif + /* + * @blkg is going offline and will be ignored by + * blkg_[rw]stat_recursive_sum(). Transfer stats to the parent so +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 1deb79a47181..69ef3761c95d 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -233,6 +233,7 @@ static struct bfq_io_cq *bfq_bic_lookup(struct bfq_data *bfqd, + return NULL; + } + ++#define BFQ_MQ + #include "bfq-sched.c" + #include "bfq-cgroup-included.c" + +@@ -1564,15 +1565,9 @@ static struct request *bfq_find_rq_fmerge(struct bfq_data *bfqd, + struct bio *bio, + struct request_queue *q) + { +- struct task_struct *tsk = current; +- struct bfq_io_cq *bic; +- struct bfq_queue *bfqq; ++ struct bfq_queue *bfqq = bfqd->bio_bfqq; + +- bic = bfq_bic_lookup(bfqd, tsk->io_context, q); +- if (!bic) +- return NULL; + +- bfqq = bic_to_bfqq(bic, op_is_sync(bio->bi_opf)); + if (bfqq) + return elv_rb_find(&bfqq->sort_list, bio_end_sector(bio)); + +@@ -1693,9 +1688,26 @@ static bool bfq_bio_merge(struct blk_mq_hw_ctx *hctx, struct bio *bio) + struct request_queue *q = hctx->queue; + struct bfq_data *bfqd = q->elevator->elevator_data; + struct request *free = NULL; ++ /* ++ * bfq_bic_lookup grabs the queue_lock: invoke it now and ++ * store its return value for later use, to avoid nesting ++ * queue_lock inside the bfqd->lock. We assume that the bic ++ * returned by bfq_bic_lookup does not go away before ++ * bfqd->lock is taken. ++ */ ++ struct bfq_io_cq *bic = bfq_bic_lookup(bfqd, current->io_context, q); + bool ret; + + spin_lock_irq(&bfqd->lock); ++ ++ if (bic) ++ bfqd->bio_bfqq = bic_to_bfqq(bic, op_is_sync(bio->bi_opf)); ++ else ++ bfqd->bio_bfqq = NULL; ++ bfqd->bio_bic = bic; ++ /* Set next flag just for testing purposes */ ++ bfqd->bio_bfqq_set = true; ++ + ret = blk_mq_sched_try_merge(q, bio, &free); + + /* +@@ -1706,6 +1718,7 @@ static bool bfq_bio_merge(struct blk_mq_hw_ctx *hctx, struct bio *bio) + */ + if (free) + blk_mq_free_request(free); ++ bfqd->bio_bfqq_set = false; + spin_unlock_irq(&bfqd->lock); + + return ret; +@@ -2261,8 +2274,7 @@ static bool bfq_allow_bio_merge(struct request_queue *q, struct request *rq, + { + struct bfq_data *bfqd = q->elevator->elevator_data; + bool is_sync = op_is_sync(bio->bi_opf); +- struct bfq_io_cq *bic; +- struct bfq_queue *bfqq, *new_bfqq; ++ struct bfq_queue *bfqq = bfqd->bio_bfqq, *new_bfqq; + + /* + * Disallow merge of a sync bio into an async request. +@@ -2273,31 +2285,40 @@ static bool bfq_allow_bio_merge(struct request_queue *q, struct request *rq, + /* + * Lookup the bfqq that this bio will be queued with. Allow + * merge only if rq is queued there. +- * Queue lock is held here. + */ +- bic = bfq_bic_lookup(bfqd, current->io_context, q); +- if (!bic) ++ if (!bfqq) + return false; + +- assert_spin_locked(&bfqd->lock); +- bfqq = bic_to_bfqq(bic, is_sync); + /* + * We take advantage of this function to perform an early merge + * of the queues of possible cooperating processes. + */ +- if (bfqq) { +- new_bfqq = bfq_setup_cooperator(bfqd, bfqq, bio, false); +- if (new_bfqq) { +- bfq_merge_bfqqs(bfqd, bic, bfqq, new_bfqq); +- /* +- * If we get here, the bio will be queued in the +- * shared queue, i.e., new_bfqq, so use new_bfqq +- * to decide whether bio and rq can be merged. +- */ +- bfqq = new_bfqq; +- } +- } ++ new_bfqq = bfq_setup_cooperator(bfqd, bfqq, bio, false); ++ if (new_bfqq) { ++ /* ++ * bic still points to bfqq, then it has not yet been ++ * redirected to some other bfq_queue, and a queue ++ * merge beween bfqq and new_bfqq can be safely ++ * fulfillled, i.e., bic can be redirected to new_bfqq ++ * and bfqq can be put. ++ */ ++ bfq_merge_bfqqs(bfqd, bfqd->bio_bic, bfqq, ++ new_bfqq); ++ /* ++ * If we get here, bio will be queued into new_queue, ++ * so use new_bfqq to decide whether bio and rq can be ++ * merged. ++ */ ++ bfqq = new_bfqq; + ++ /* ++ * Change also bqfd->bio_bfqq, as ++ * bfqd->bio_bic now points to new_bfqq, and ++ * this function may be invoked again (and then may ++ * use again bqfd->bio_bfqq). ++ */ ++ bfqd->bio_bfqq = bfqq; ++ } + return bfqq == RQ_BFQQ(rq); + } + +@@ -3965,14 +3986,43 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + return rq; + } + ++/* ++ * Next two functions release bfqd->lock and put the io context ++ * pointed by bfqd->ioc_to_put. This delayed put is used to not risk ++ * to take an ioc->lock while the scheduler lock is being held. ++ */ ++static void bfq_unlock_put_ioc(struct bfq_data *bfqd) ++{ ++ struct io_context *ioc_to_put = bfqd->ioc_to_put; ++ ++ bfqd->ioc_to_put = NULL; ++ spin_unlock_irq(&bfqd->lock); ++ ++ if (ioc_to_put) ++ put_io_context(ioc_to_put); ++} ++ ++static void bfq_unlock_put_ioc_restore(struct bfq_data *bfqd, ++ unsigned long flags) ++{ ++ struct io_context *ioc_to_put = bfqd->ioc_to_put; ++ ++ bfqd->ioc_to_put = NULL; ++ spin_unlock_irqrestore(&bfqd->lock, flags); ++ ++ if (ioc_to_put) ++ put_io_context(ioc_to_put); ++} ++ + static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + { + struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; + struct request *rq; + + spin_lock_irq(&bfqd->lock); ++ + rq = __bfq_dispatch_request(hctx); +- spin_unlock_irq(&bfqd->lock); ++ bfq_unlock_put_ioc(bfqd); + + return rq; + } +@@ -3981,7 +4031,7 @@ static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + * Task holds one reference to the queue, dropped when task exits. Each rq + * in-flight on this queue also holds a reference, dropped when rq is freed. + * +- * Queue lock must be held here. Recall not to use bfqq after calling ++ * Scheduler lock must be held here. Recall not to use bfqq after calling + * this function on it. + */ + static void bfq_put_queue(struct bfq_queue *bfqq) +@@ -4066,17 +4116,23 @@ static void bfq_exit_icq_bfqq(struct bfq_io_cq *bic, bool is_sync) + bfqd = bfqq->bfqd; /* NULL if scheduler already exited */ + + if (bfqq && bfqd) { +- spin_lock_irq(&bfqd->lock); ++ unsigned long flags; ++ ++ spin_lock_irqsave(&bfqd->lock, flags); + /* +- * If the bic is using a shared queue, put the reference +- * taken on the io_context when the bic started using a +- * shared bfq_queue. ++ * If the bic is using a shared queue, put the ++ * reference taken on the io_context when the bic ++ * started using a shared bfq_queue. This put cannot ++ * make ioc->ref_count reach 0, then no ioc->lock ++ * risks to be taken (leading to possible deadlock ++ * scenarios). + */ + if (is_sync && bfq_bfqq_coop(bfqq)) + put_io_context(bic->icq.ioc); ++ + bfq_exit_bfqq(bfqd, bfqq); + bic_set_bfqq(bic, NULL, is_sync); +- spin_unlock_irq(&bfqd->lock); ++ bfq_unlock_put_ioc_restore(bfqd, flags); + } + } + +@@ -4183,8 +4239,6 @@ static void bfq_init_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, + INIT_HLIST_NODE(&bfqq->burst_list_node); + BUG_ON(!hlist_unhashed(&bfqq->burst_list_node)); + +- spin_lock_init(&bfqq->lock); +- + bfqq->ref = 0; + bfqq->bfqd = bfqd; + +@@ -4476,6 +4530,14 @@ static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + + new_bfqq->ref++; + bfq_clear_bfqq_just_created(bfqq); ++ /* ++ * If the bic associated with the process ++ * issuing this request still points to bfqq ++ * (and thus has not been already redirected ++ * to new_bfqq or even some other bfq_queue), ++ * then complete the merge and redirect it to ++ * new_bfqq. ++ */ + if (bic_to_bfqq(RQ_BIC(rq), 1) == bfqq) + bfq_merge_bfqqs(bfqd, RQ_BIC(rq), + bfqq, new_bfqq); +@@ -4498,14 +4560,17 @@ static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + } + + static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, +- bool at_head) ++ bool at_head) + { + struct request_queue *q = hctx->queue; + struct bfq_data *bfqd = q->elevator->elevator_data; + + spin_lock_irq(&bfqd->lock); +- if (blk_mq_sched_try_insert_merge(q, rq)) +- goto done; ++ if (blk_mq_sched_try_insert_merge(q, rq)) { ++ spin_unlock_irq(&bfqd->lock); ++ return; ++ } ++ + spin_unlock_irq(&bfqd->lock); + + blk_mq_sched_request_inserted(rq); +@@ -4530,8 +4595,8 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + q->last_merge = rq; + } + } +-done: +- spin_unlock_irq(&bfqd->lock); ++ ++ bfq_unlock_put_ioc(bfqd); + } + + static void bfq_insert_requests(struct blk_mq_hw_ctx *hctx, +@@ -4724,7 +4789,7 @@ static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + bfq_completed_request(bfqq, bfqd); + bfq_put_rq_priv_body(bfqq); + +- spin_unlock_irqrestore(&bfqd->lock, flags); ++ bfq_unlock_put_ioc_restore(bfqd, flags); + } else { + /* + * Request rq may be still/already in the scheduler, +@@ -4732,10 +4797,10 @@ static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + * defer such a check and removal, to avoid + * inconsistencies in the time interval from the end + * of this function to the start of the deferred work. +- * Fortunately, this situation occurs only in process +- * context, so taking the scheduler lock does not +- * cause any deadlock, even if other locks are already +- * (correctly) held by this process. ++ * This situation seems to occur only in process ++ * context, as a consequence of a merge. In the ++ * current version of the code, this implies that the ++ * lock is held. + */ + BUG_ON(in_interrupt()); + +@@ -4758,8 +4823,6 @@ bfq_split_bfqq(struct bfq_io_cq *bic, struct bfq_queue *bfqq) + { + bfq_log_bfqq(bfqq->bfqd, bfqq, "splitting queue"); + +- put_io_context(bic->icq.ioc); +- + if (bfqq_process_refs(bfqq) == 1) { + bfqq->pid = current->pid; + bfq_clear_bfqq_coop(bfqq); +@@ -4775,6 +4838,41 @@ bfq_split_bfqq(struct bfq_io_cq *bic, struct bfq_queue *bfqq) + return NULL; + } + ++static struct bfq_queue *bfq_get_bfqq_handle_split(struct bfq_data *bfqd, ++ struct bfq_io_cq *bic, ++ struct bio *bio, ++ bool split, bool is_sync, ++ bool *new_queue) ++{ ++ struct bfq_queue *bfqq = bic_to_bfqq(bic, is_sync); ++ ++ if (likely(bfqq && bfqq != &bfqd->oom_bfqq)) ++ return bfqq; ++ ++ if (new_queue) ++ *new_queue = true; ++ ++ if (bfqq) ++ bfq_put_queue(bfqq); ++ bfqq = bfq_get_queue(bfqd, bio, is_sync, bic); ++ ++ bic_set_bfqq(bic, bfqq, is_sync); ++ if (split && is_sync) { ++ if ((bic->was_in_burst_list && bfqd->large_burst) || ++ bic->saved_in_large_burst) ++ bfq_mark_bfqq_in_large_burst(bfqq); ++ else { ++ bfq_clear_bfqq_in_large_burst(bfqq); ++ if (bic->was_in_burst_list) ++ hlist_add_head(&bfqq->burst_list_node, ++ &bfqd->burst_list); ++ } ++ bfqq->split_time = jiffies; ++ } ++ ++ return bfqq; ++} ++ + /* + * Allocate bfq data structures associated with this request. + */ +@@ -4786,6 +4884,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + const int is_sync = rq_is_sync(rq); + struct bfq_queue *bfqq; + bool bfqq_already_existing = false, split = false; ++ bool new_queue = false; + + spin_lock_irq(&bfqd->lock); + +@@ -4796,42 +4895,10 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + + bfq_bic_update_cgroup(bic, bio); + +-new_queue: +- bfqq = bic_to_bfqq(bic, is_sync); +- if (!bfqq || bfqq == &bfqd->oom_bfqq) { +- if (bfqq) +- bfq_put_queue(bfqq); +- bfqq = bfq_get_queue(bfqd, bio, is_sync, bic); +- BUG_ON(!hlist_unhashed(&bfqq->burst_list_node)); ++ bfqq = bfq_get_bfqq_handle_split(bfqd, bic, bio, false, is_sync, ++ &new_queue); + +- bic_set_bfqq(bic, bfqq, is_sync); +- if (split && is_sync) { +- bfq_log_bfqq(bfqd, bfqq, +- "get_request: was_in_list %d " +- "was_in_large_burst %d " +- "large burst in progress %d", +- bic->was_in_burst_list, +- bic->saved_in_large_burst, +- bfqd->large_burst); +- +- if ((bic->was_in_burst_list && bfqd->large_burst) || +- bic->saved_in_large_burst) { +- bfq_log_bfqq(bfqd, bfqq, +- "get_request: marking in " +- "large burst"); +- bfq_mark_bfqq_in_large_burst(bfqq); +- } else { +- bfq_log_bfqq(bfqd, bfqq, +- "get_request: clearing in " +- "large burst"); +- bfq_clear_bfqq_in_large_burst(bfqq); +- if (bic->was_in_burst_list) +- hlist_add_head(&bfqq->burst_list_node, +- &bfqd->burst_list); +- } +- bfqq->split_time = jiffies; +- } +- } else { ++ if (unlikely(!new_queue)) { + /* If the queue was seeky for too long, break it apart. */ + if (bfq_bfqq_coop(bfqq) && bfq_bfqq_split_coop(bfqq)) { + bfq_log_bfqq(bfqd, bfqq, "breaking apart bfqq"); +@@ -4841,9 +4908,19 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + bic->saved_in_large_burst = true; + + bfqq = bfq_split_bfqq(bic, bfqq); +- split = true; ++ /* ++ * A reference to bic->icq.ioc needs to be ++ * released after a queue split. Do not do it ++ * immediately, to not risk to possibly take ++ * an ioc->lock while holding the scheduler ++ * lock. ++ */ ++ bfqd->ioc_to_put = bic->icq.ioc; ++ + if (!bfqq) +- goto new_queue; ++ bfqq = bfq_get_bfqq_handle_split(bfqd, bic, bio, ++ true, is_sync, ++ NULL); + else + bfqq_already_existing = true; + } +@@ -4861,18 +4938,17 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + + /* + * If a bfq_queue has only one process reference, it is owned +- * by only one bfq_io_cq: we can set the bic field of the +- * bfq_queue to the address of that structure. Also, if the +- * queue has just been split, mark a flag so that the +- * information is available to the other scheduler hooks. ++ * by only this bic: we can then set bfqq->bic = bic. in ++ * addition, if the queue has also just been split, we have to ++ * resume its state. + */ + if (likely(bfqq != &bfqd->oom_bfqq) && bfqq_process_refs(bfqq) == 1) { + bfqq->bic = bic; +- if (split) { ++ if (bfqd->ioc_to_put) { /* if true, then there has been a split */ + /* +- * If the queue has just been split from a shared +- * queue, restore the idle window and the possible +- * weight raising period. ++ * The queue has just been split from a shared ++ * queue: restore the idle window and the ++ * possible weight raising period. + */ + bfq_bfqq_resume_state(bfqq, bfqd, bic, + bfqq_already_existing); +@@ -4882,7 +4958,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + if (unlikely(bfq_bfqq_just_created(bfqq))) + bfq_handle_burst(bfqd, bfqq); + +- spin_unlock_irq(&bfqd->lock); ++ bfq_unlock_put_ioc(bfqd); + + return 0; + +@@ -4929,7 +5005,7 @@ static void bfq_idle_slice_timer_body(struct bfq_queue *bfqq) + bfq_bfqq_expire(bfqd, bfqq, true, reason); + + schedule_dispatch: +- spin_unlock_irqrestore(&bfqd->lock, flags); ++ bfq_unlock_put_ioc_restore(bfqd, flags); + bfq_schedule_dispatch(bfqd); + } + +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 23744b246db6..bd83f1c02573 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -338,8 +338,6 @@ struct bfq_queue { + unsigned long wr_start_at_switch_to_srt; + + unsigned long split_time; /* time of last split */ +- +- spinlock_t lock; + }; + + /** +@@ -609,6 +607,29 @@ struct bfq_data { + struct bfq_queue oom_bfqq; + + spinlock_t lock; ++ ++ /* ++ * bic associated with the task issuing current bio for ++ * merging. This and the next field are used as a support to ++ * be able to perform the bic lookup, needed by bio-merge ++ * functions, before the scheduler lock is taken, and thus ++ * avoid taking the request-queue lock while the scheduler ++ * lock is being held. ++ */ ++ struct bfq_io_cq *bio_bic; ++ /* bfqq associated with the task issuing current bio for merging */ ++ struct bfq_queue *bio_bfqq; ++ /* Extra flag used only for TESTING */ ++ bool bio_bfqq_set; ++ ++ /* ++ * io context to put right after bfqd->lock is released. This ++ * filed is used to perform put_io_context, when needed, to ++ * after the scheduler lock has been released, and thus ++ * prevent an ioc->lock from being possibly taken while the ++ * scheduler lock is being held. ++ */ ++ struct io_context *ioc_to_put; + }; + + enum bfqq_state_flags { +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index b54a638186e3..a5c8b4acd33c 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -1905,7 +1905,18 @@ static void __bfq_bfqd_reset_in_service(struct bfq_data *bfqd) + struct bfq_entity *entity = in_serv_entity; + + if (bfqd->in_service_bic) { ++#ifdef BFQ_MQ ++ /* ++ * Schedule the release of a reference to ++ * bfqd->in_service_bic->icq.ioc to right after the ++ * scheduler lock is released. This ioc is not ++ * released immediately, to not risk to possibly take ++ * an ioc->lock while holding the scheduler lock. ++ */ ++ bfqd->ioc_to_put = bfqd->in_service_bic->icq.ioc; ++#else + put_io_context(bfqd->in_service_bic->icq.ioc); ++#endif + bfqd->in_service_bic = NULL; + } + + +From 84cc7140cb4f0574710625f51abbb076a1dd2920 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 3 Mar 2017 09:31:14 +0100 +Subject: [PATCH 17/51] Add checks and extra log messages - Part II + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 42 ++++++++++++++++++++++++++++++++++++++++-- + block/bfq-sched.c | 1 + + 2 files changed, 41 insertions(+), 2 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 69ef3761c95d..5707d42b160d 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -1567,6 +1567,7 @@ static struct request *bfq_find_rq_fmerge(struct bfq_data *bfqd, + { + struct bfq_queue *bfqq = bfqd->bio_bfqq; + ++ BUG_ON(!bfqd->bio_bfqq_set); + + if (bfqq) + return elv_rb_find(&bfqq->sort_list, bio_end_sector(bio)); +@@ -1719,6 +1720,7 @@ static bool bfq_bio_merge(struct blk_mq_hw_ctx *hctx, struct bio *bio) + if (free) + blk_mq_free_request(free); + bfqd->bio_bfqq_set = false; ++ BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + + return ret; +@@ -1781,6 +1783,7 @@ static void bfq_request_merged(struct request_queue *q, struct request *req, + bfq_updated_next_req(bfqd, bfqq); + bfq_pos_tree_add_move(bfqd, bfqq); + } ++ BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + } + } +@@ -1824,6 +1827,7 @@ static void bfq_requests_merged(struct request_queue *q, struct request *rq, + + bfq_remove_request(q, next); + ++ BUG_ON(bfqq->bfqd->ioc_to_put); + spin_unlock_irq(&bfqq->bfqd->lock); + end: + bfqg_stats_update_io_merged(bfqq_group(bfqq), next->cmd_flags); +@@ -2195,9 +2199,11 @@ bfq_merge_bfqqs(struct bfq_data *bfqd, struct bfq_io_cq *bic, + { + bfq_log_bfqq(bfqd, bfqq, "merging with queue %lu", + (unsigned long) new_bfqq->pid); ++ BUG_ON(bfqq->bic && bfqq->bic == new_bfqq->bic); + /* Save weight raising and idle window of the merged queues */ + bfq_bfqq_save_state(bfqq); + bfq_bfqq_save_state(new_bfqq); ++ + if (bfq_bfqq_IO_bound(bfqq)) + bfq_mark_bfqq_IO_bound(new_bfqq); + bfq_clear_bfqq_IO_bound(bfqq); +@@ -2276,6 +2282,7 @@ static bool bfq_allow_bio_merge(struct request_queue *q, struct request *rq, + bool is_sync = op_is_sync(bio->bi_opf); + struct bfq_queue *bfqq = bfqd->bio_bfqq, *new_bfqq; + ++ assert_spin_locked(&bfqd->lock); + /* + * Disallow merge of a sync bio into an async request. + */ +@@ -2286,6 +2293,7 @@ static bool bfq_allow_bio_merge(struct request_queue *q, struct request *rq, + * Lookup the bfqq that this bio will be queued with. Allow + * merge only if rq is queued there. + */ ++ BUG_ON(!bfqd->bio_bfqq_set); + if (!bfqq) + return false; + +@@ -2294,6 +2302,7 @@ static bool bfq_allow_bio_merge(struct request_queue *q, struct request *rq, + * of the queues of possible cooperating processes. + */ + new_bfqq = bfq_setup_cooperator(bfqd, bfqq, bio, false); ++ BUG_ON(new_bfqq == bfqq); + if (new_bfqq) { + /* + * bic still points to bfqq, then it has not yet been +@@ -4040,6 +4049,8 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + struct bfq_group *bfqg = bfqq_group(bfqq); + #endif + ++ assert_spin_locked(&bfqq->bfqd->lock); ++ + BUG_ON(bfqq->ref <= 0); + + if (bfqq->bfqd) +@@ -4119,6 +4130,7 @@ static void bfq_exit_icq_bfqq(struct bfq_io_cq *bic, bool is_sync) + unsigned long flags; + + spin_lock_irqsave(&bfqd->lock, flags); ++ BUG_ON(bfqd->ioc_to_put); + /* + * If the bic is using a shared queue, put the + * reference taken on the io_context when the bic +@@ -4567,10 +4579,12 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + + spin_lock_irq(&bfqd->lock); + if (blk_mq_sched_try_insert_merge(q, rq)) { ++ BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + return; + } + ++ BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + + blk_mq_sched_request_inserted(rq); +@@ -4785,6 +4799,7 @@ static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + unsigned long flags; + + spin_lock_irqsave(&bfqd->lock, flags); ++ BUG_ON(bfqd->ioc_to_put); + + bfq_completed_request(bfqq, bfqd); + bfq_put_rq_priv_body(bfqq); +@@ -4855,13 +4870,28 @@ static struct bfq_queue *bfq_get_bfqq_handle_split(struct bfq_data *bfqd, + if (bfqq) + bfq_put_queue(bfqq); + bfqq = bfq_get_queue(bfqd, bio, is_sync, bic); ++ BUG_ON(!hlist_unhashed(&bfqq->burst_list_node)); + + bic_set_bfqq(bic, bfqq, is_sync); + if (split && is_sync) { ++ bfq_log_bfqq(bfqd, bfqq, ++ "get_request: was_in_list %d " ++ "was_in_large_burst %d " ++ "large burst in progress %d", ++ bic->was_in_burst_list, ++ bic->saved_in_large_burst, ++ bfqd->large_burst); ++ + if ((bic->was_in_burst_list && bfqd->large_burst) || +- bic->saved_in_large_burst) ++ bic->saved_in_large_burst) { ++ bfq_log_bfqq(bfqd, bfqq, ++ "get_request: marking in " ++ "large burst"); + bfq_mark_bfqq_in_large_burst(bfqq); +- else { ++ } else { ++ bfq_log_bfqq(bfqd, bfqq, ++ "get_request: clearing in " ++ "large burst"); + bfq_clear_bfqq_in_large_burst(bfqq); + if (bic->was_in_burst_list) + hlist_add_head(&bfqq->burst_list_node, +@@ -4897,10 +4927,12 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + + bfqq = bfq_get_bfqq_handle_split(bfqd, bic, bio, false, is_sync, + &new_queue); ++ BUG_ON(bfqd->ioc_to_put); + + if (unlikely(!new_queue)) { + /* If the queue was seeky for too long, break it apart. */ + if (bfq_bfqq_coop(bfqq) && bfq_bfqq_split_coop(bfqq)) { ++ BUG_ON(!is_sync); + bfq_log_bfqq(bfqd, bfqq, "breaking apart bfqq"); + + /* Update bic before losing reference to bfqq */ +@@ -4923,6 +4955,9 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + NULL); + else + bfqq_already_existing = true; ++ ++ BUG_ON(!bfqq); ++ BUG_ON(bfqq == &bfqd->oom_bfqq); + } + } + +@@ -4976,6 +5011,8 @@ static void bfq_idle_slice_timer_body(struct bfq_queue *bfqq) + + BUG_ON(!bfqd); + spin_lock_irqsave(&bfqd->lock, flags); ++ BUG_ON(bfqd->ioc_to_put); ++ + bfq_log_bfqq(bfqd, bfqq, "handling slice_timer expiration"); + bfq_clear_bfqq_wait_request(bfqq); + +@@ -5083,6 +5120,7 @@ static void bfq_exit_queue(struct elevator_queue *e) + spin_lock_irq(&bfqd->lock); + list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) + bfq_deactivate_bfqq(bfqd, bfqq, false, false); ++ BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + + hrtimer_cancel(&bfqd->idle_slice_timer); +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index a5c8b4acd33c..85e59eeb3569 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -1906,6 +1906,7 @@ static void __bfq_bfqd_reset_in_service(struct bfq_data *bfqd) + + if (bfqd->in_service_bic) { + #ifdef BFQ_MQ ++ BUG_ON(bfqd->ioc_to_put); + /* + * Schedule the release of a reference to + * bfqd->in_service_bic->icq.ioc to right after the + +From 3d54cb804f1db2e08ce4a6cc335868538542f587 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 22 Feb 2017 11:30:01 +0100 +Subject: [PATCH 18/51] Fix unbalanced increment of rq_in_driver + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 52 +++++++++++++++++++++++++++++++++++++++++--------- + 1 file changed, 43 insertions(+), 9 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 5707d42b160d..9cbcb8d43d81 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -3936,9 +3936,45 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + rq = list_first_entry(&bfqd->dispatch, struct request, + queuelist); + list_del_init(&rq->queuelist); ++ + bfq_log(bfqd, + "dispatch requests: picked %p from dispatch list", rq); +- goto exit; ++ bfqq = RQ_BFQQ(rq); ++ ++ if (bfqq) { ++ /* ++ * Increment counters here, because this ++ * dispatch does not follow the standard ++ * dispatch flow (where counters are ++ * incremented) ++ */ ++ bfqq->dispatched++; ++ ++ goto inc_in_driver_start_rq; ++ } ++ ++ /* ++ * We exploit the put_rq_private hook to decrement ++ * rq_in_driver, but put_rq_private will not be ++ * invoked on this request. So, to avoid unbalance, ++ * just start this request, without incrementing ++ * rq_in_driver. As a negative consequence, ++ * rq_in_driver is deceptively lower than it should be ++ * while this request is in service. This may cause ++ * bfq_schedule_dispatch to be invoked uselessly. ++ * ++ * As for implementing an exact solution, the ++ * put_request hook, if defined, is probably invoked ++ * also on this request. So, by exploiting this hook, ++ * we could 1) increment rq_in_driver here, and 2) ++ * decrement it in put_request. Such a solution would ++ * let the value of the counter be always accurate, ++ * but it would entail using an extra interface ++ * function. This cost seems higher than the benefit, ++ * being the frequency of non-elevator-private ++ * requests very low. ++ */ ++ goto start_rq; + } + + bfq_log(bfqd, "dispatch requests: %d busy queues", bfqd->busy_queues); +@@ -3973,10 +4009,12 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + + BUG_ON(bfqq->next_rq == NULL && + bfqq->entity.budget < bfqq->entity.service); +-exit: ++ + if (rq) { +- rq->rq_flags |= RQF_STARTED; ++ inc_in_driver_start_rq: + bfqd->rq_in_driver++; ++ start_rq: ++ rq->rq_flags |= RQF_STARTED; + if (bfqq) + bfq_log_bfqq(bfqd, bfqq, + "dispatched %s request %p, rq_in_driver %d", +@@ -3992,6 +4030,7 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + "returned NULL request, rq_in_driver %d", + bfqd->rq_in_driver); + ++exit: + return rq; + } + +@@ -4591,15 +4630,10 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + + spin_lock_irq(&bfqd->lock); + if (at_head || blk_rq_is_passthrough(rq)) { +- struct bfq_queue *bfqq = RQ_BFQQ(rq); +- + if (at_head) + list_add(&rq->queuelist, &bfqd->dispatch); + else + list_add_tail(&rq->queuelist, &bfqd->dispatch); +- +- if (bfqq) +- bfqq->dispatched++; + } else { + __bfq_insert_request(bfqd, rq); + +@@ -4966,7 +5000,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + "get_request: new allocated %d", bfqq->allocated); + + bfqq->ref++; +- bfq_log_bfqq(bfqd, bfqq, "get_request: bfqq %p, %d", bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqd, bfqq, "get_request %p: bfqq %p, %d", rq, bfqq, bfqq->ref); + + rq->elv.priv[0] = bic; + rq->elv.priv[1] = bfqq; + +From 7ba977d696b239569b4cd233aebc99e136ecf487 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 3 Mar 2017 09:39:35 +0100 +Subject: [PATCH 19/51] Add checks and extra log messages - Part III + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 11 +++++++++++ + 1 file changed, 11 insertions(+) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 9cbcb8d43d81..24b529a2edc7 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4630,10 +4630,21 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + + spin_lock_irq(&bfqd->lock); + if (at_head || blk_rq_is_passthrough(rq)) { ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); ++ + if (at_head) + list_add(&rq->queuelist, &bfqd->dispatch); + else + list_add_tail(&rq->queuelist, &bfqd->dispatch); ++ ++ if (bfqq) ++ bfq_log_bfqq(bfqd, bfqq, ++ "insert_request %p in disp: at_head %d", ++ rq, at_head); ++ else ++ bfq_log(bfqd, ++ "insert_request %p in disp: at_head %d", ++ rq, at_head); + } else { + __bfq_insert_request(bfqd, rq); + + +From c94e47b2908600b8ba89f84b0ac7febddd313141 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 17 Feb 2017 14:28:02 +0100 +Subject: [PATCH 20/51] TESTING: Check wrong invocation of merge and + put_rq_priv functions + +Check that merge functions are not invoked on requests queued in the +dispatch queue, and that neither put_rq_private is invoked on these +requests if, in addition, they have not passed through get_rq_private. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 22 ++++++++++++++++++++++ + include/linux/blkdev.h | 2 ++ + 2 files changed, 24 insertions(+) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 24b529a2edc7..b4d40bb712d2 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -1746,6 +1746,8 @@ static int bfq_request_merge(struct request_queue *q, struct request **req, + static void bfq_request_merged(struct request_queue *q, struct request *req, + enum elv_merge type) + { ++ BUG_ON(req->rq_flags & RQF_DISP_LIST); ++ + if (type == ELEVATOR_FRONT_MERGE && + rb_prev(&req->rb_node) && + blk_rq_pos(req) < +@@ -1795,6 +1797,8 @@ static void bfq_requests_merged(struct request_queue *q, struct request *rq, + + BUG_ON(!RQ_BFQQ(rq)); + BUG_ON(!RQ_BFQQ(next)); ++ BUG_ON(rq->rq_flags & RQF_DISP_LIST); ++ BUG_ON(next->rq_flags & RQF_DISP_LIST); + + if (!RB_EMPTY_NODE(&rq->rb_node)) + goto end; +@@ -3936,6 +3940,7 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + rq = list_first_entry(&bfqd->dispatch, struct request, + queuelist); + list_del_init(&rq->queuelist); ++ rq->rq_flags &= ~RQF_DISP_LIST; + + bfq_log(bfqd, + "dispatch requests: picked %p from dispatch list", rq); +@@ -3950,6 +3955,17 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + */ + bfqq->dispatched++; + ++ /* ++ * TESTING: reset DISP_LIST flag, because: 1) ++ * this rq this request has passed through ++ * get_rq_private, 2) then it will have ++ * put_rq_private invoked on it, and 3) in ++ * put_rq_private we use this flag to check ++ * that put_rq_private is not invoked on ++ * requests for which get_rq_private has been ++ * invoked. ++ */ ++ rq->rq_flags &= ~RQF_DISP_LIST; + goto inc_in_driver_start_rq; + } + +@@ -4637,6 +4653,7 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + else + list_add_tail(&rq->queuelist, &bfqd->dispatch); + ++ rq->rq_flags |= RQF_DISP_LIST; + if (bfqq) + bfq_log_bfqq(bfqd, bfqq, + "insert_request %p in disp: at_head %d", +@@ -4824,6 +4841,10 @@ static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + bfqd = bfqq->bfqd; + BUG_ON(!bfqd); + ++ if (rq->rq_flags & RQF_DISP_LIST) { ++ pr_crit("putting disp rq %p for %d", rq, bfqq->pid); ++ BUG(); ++ } + BUG_ON(rq->rq_flags & RQF_QUEUED); + BUG_ON(!(rq->rq_flags & RQF_ELVPRIV)); + +@@ -5015,6 +5036,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + + rq->elv.priv[0] = bic; + rq->elv.priv[1] = bfqq; ++ rq->rq_flags &= ~RQF_DISP_LIST; + + /* + * If a bfq_queue has only one process reference, it is owned +diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h +index 10f892ca585d..0048e59e6d07 100644 +--- a/include/linux/blkdev.h ++++ b/include/linux/blkdev.h +@@ -121,6 +121,8 @@ typedef __u32 __bitwise req_flags_t; + /* Look at ->special_vec for the actual data payload instead of the + bio chain. */ + #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18)) ++/* DEBUG: rq in bfq-mq dispatch list */ ++#define RQF_DISP_LIST ((__force req_flags_t)(1 << 19)) + + /* flags that prevent us from merging requests: */ + #define RQF_NOMERGE_FLAGS \ + +From 49206f9052d13c96d49dbc36c612bed41b2d6552 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Sat, 25 Feb 2017 17:38:05 +0100 +Subject: [PATCH 21/51] Complete support for cgroups + +This commit completes cgroups support for bfq-mq. In particular, it deals with +a sort of circular dependency introduced in blk-mq: the function +blkcg_activate_policy, invoked during scheduler initialization, triggers the +invocation of the has_work scheduler hook (before the init function is +finished). To adress this issue, this commit moves the invocation of +blkcg_activate_policy after the initialization of all the fields that could be +initialized before invoking blkcg_activate_policy itself. This enables has_work +to correctly return false, and thus to prevent the blk-mq stack from invoking +further scheduler hooks before the init function is finished. + +Signed-off-by: Paolo Valente +--- + block/Kconfig.iosched | 9 +++++ + block/bfq-mq-iosched.c | 108 ++++++++++++++++++++++++++++--------------------- + block/bfq-mq.h | 2 +- + 3 files changed, 72 insertions(+), 47 deletions(-) + +diff --git a/block/Kconfig.iosched b/block/Kconfig.iosched +index 2d94af3d8b0a..299a6861fb90 100644 +--- a/block/Kconfig.iosched ++++ b/block/Kconfig.iosched +@@ -106,6 +106,15 @@ config MQ_IOSCHED_BFQ + guarantees a low latency to interactive and soft real-time + applications. Details in Documentation/block/bfq-iosched.txt + ++config MQ_BFQ_GROUP_IOSCHED ++ bool "BFQ-MQ hierarchical scheduling support" ++ depends on MQ_IOSCHED_BFQ && BLK_CGROUP ++ default n ++ ---help--- ++ ++ Enable hierarchical scheduling in BFQ-MQ, using the blkio ++ (cgroups-v1) or io (cgroups-v2) controller. ++ + config MQ_IOSCHED_DEADLINE + tristate "MQ deadline I/O scheduler" + default y +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index b4d40bb712d2..02a1e7fd0ea4 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -88,7 +88,6 @@ + #include "blk-mq.h" + #include "blk-mq-tag.h" + #include "blk-mq-sched.h" +-#undef CONFIG_BFQ_GROUP_IOSCHED /* cgroups support not yet functional */ + #include "bfq-mq.h" + + /* Expiration time of sync (0) and async (1) requests, in ns. */ +@@ -233,15 +232,6 @@ static struct bfq_io_cq *bfq_bic_lookup(struct bfq_data *bfqd, + return NULL; + } + +-#define BFQ_MQ +-#include "bfq-sched.c" +-#include "bfq-cgroup-included.c" +- +-#define bfq_class_idle(bfqq) ((bfqq)->ioprio_class == IOPRIO_CLASS_IDLE) +-#define bfq_class_rt(bfqq) ((bfqq)->ioprio_class == IOPRIO_CLASS_RT) +- +-#define bfq_sample_valid(samples) ((samples) > 80) +- + /* + * Scheduler run of queue, if there are requests pending and no one in the + * driver that will restart queueing. +@@ -255,6 +245,43 @@ static void bfq_schedule_dispatch(struct bfq_data *bfqd) + } + + /* ++ * Next two functions release bfqd->lock and put the io context ++ * pointed by bfqd->ioc_to_put. This delayed put is used to not risk ++ * to take an ioc->lock while the scheduler lock is being held. ++ */ ++static void bfq_unlock_put_ioc(struct bfq_data *bfqd) ++{ ++ struct io_context *ioc_to_put = bfqd->ioc_to_put; ++ ++ bfqd->ioc_to_put = NULL; ++ spin_unlock_irq(&bfqd->lock); ++ ++ if (ioc_to_put) ++ put_io_context(ioc_to_put); ++} ++ ++static void bfq_unlock_put_ioc_restore(struct bfq_data *bfqd, ++ unsigned long flags) ++{ ++ struct io_context *ioc_to_put = bfqd->ioc_to_put; ++ ++ bfqd->ioc_to_put = NULL; ++ spin_unlock_irqrestore(&bfqd->lock, flags); ++ ++ if (ioc_to_put) ++ put_io_context(ioc_to_put); ++} ++ ++#define BFQ_MQ ++#include "bfq-sched.c" ++#include "bfq-cgroup-included.c" ++ ++#define bfq_class_idle(bfqq) ((bfqq)->ioprio_class == IOPRIO_CLASS_IDLE) ++#define bfq_class_rt(bfqq) ((bfqq)->ioprio_class == IOPRIO_CLASS_RT) ++ ++#define bfq_sample_valid(samples) ((samples) > 80) ++ ++/* + * Lifted from AS - choose which of rq1 and rq2 that is best served now. + * We choose the request that is closesr to the head right now. Distance + * behind the head is penalized and only allowed to a certain extent. +@@ -4050,34 +4077,6 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + return rq; + } + +-/* +- * Next two functions release bfqd->lock and put the io context +- * pointed by bfqd->ioc_to_put. This delayed put is used to not risk +- * to take an ioc->lock while the scheduler lock is being held. +- */ +-static void bfq_unlock_put_ioc(struct bfq_data *bfqd) +-{ +- struct io_context *ioc_to_put = bfqd->ioc_to_put; +- +- bfqd->ioc_to_put = NULL; +- spin_unlock_irq(&bfqd->lock); +- +- if (ioc_to_put) +- put_io_context(ioc_to_put); +-} +- +-static void bfq_unlock_put_ioc_restore(struct bfq_data *bfqd, +- unsigned long flags) +-{ +- struct io_context *ioc_to_put = bfqd->ioc_to_put; +- +- bfqd->ioc_to_put = NULL; +- spin_unlock_irqrestore(&bfqd->lock, flags); +- +- if (ioc_to_put) +- put_io_context(ioc_to_put); +-} +- + static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + { + struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; +@@ -5239,6 +5238,10 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) + } + eq->elevator_data = bfqd; + ++ spin_lock_irq(q->queue_lock); ++ q->elevator = eq; ++ spin_unlock_irq(q->queue_lock); ++ + /* + * Our fallback bfqq if bfq_find_alloc_queue() runs into OOM issues. + * Grab a permanent reference to it, so that the normal code flow +@@ -5261,12 +5264,7 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) + bfqd->oom_bfqq.entity.prio_changed = 1; + + bfqd->queue = q; +- +- bfqd->root_group = bfq_create_group_hierarchy(bfqd, q->node); +- if (!bfqd->root_group) +- goto out_free; +- bfq_init_root_group(bfqd->root_group, bfqd); +- bfq_init_entity(&bfqd->oom_bfqq.entity, bfqd->root_group); ++ INIT_LIST_HEAD(&bfqd->dispatch); + + hrtimer_init(&bfqd->idle_slice_timer, CLOCK_MONOTONIC, + HRTIMER_MODE_REL); +@@ -5324,9 +5322,27 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) + bfqd->device_speed = BFQ_BFQD_FAST; + + spin_lock_init(&bfqd->lock); +- INIT_LIST_HEAD(&bfqd->dispatch); + +- q->elevator = eq; ++ /* ++ * The invocation of the next bfq_create_group_hierarchy ++ * function is the head of a chain of function calls ++ * (bfq_create_group_hierarchy->blkcg_activate_policy-> ++ * blk_mq_freeze_queue) that may lead to the invocation of the ++ * has_work hook function. For this reason, ++ * bfq_create_group_hierarchy is invoked only after all ++ * scheduler data has been initialized, apart from the fields ++ * that can be initialized only after invoking ++ * bfq_create_group_hierarchy. This, in particular, enables ++ * has_work to correctly return false. Of course, to avoid ++ * other inconsistencies, the blk-mq stack must then refrain ++ * from invoking further scheduler hooks before this init ++ * function is finished. ++ */ ++ bfqd->root_group = bfq_create_group_hierarchy(bfqd, q->node); ++ if (!bfqd->root_group) ++ goto out_free; ++ bfq_init_root_group(bfqd->root_group, bfqd); ++ bfq_init_entity(&bfqd->oom_bfqq.entity, bfqd->root_group); + + return 0; + +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index bd83f1c02573..2c81c02bccc4 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -20,7 +20,7 @@ + #include + + /* see comments on CONFIG_BFQ_GROUP_IOSCHED in bfq.h */ +-#ifdef CONFIG_BFQ_MQ_GROUP_IOSCHED ++#ifdef CONFIG_MQ_BFQ_GROUP_IOSCHED + #define BFQ_GROUP_IOSCHED_ENABLED + #endif + + +From 62d12db23ce14d2716b5cff7d2635fbc817b96d0 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 17 Mar 2017 06:15:18 +0100 +Subject: [PATCH 22/51] Remove all get and put of I/O contexts + +When a bfq queue is set in service and when it is merged, a reference +to the I/O context associated with the queue is taken. This reference +is then released when the queue is deselected from service or +split. More precisely, the release of the reference is postponed to +when the scheduler lock is released, to avoid nesting between the +scheduler and the I/O-context lock. In fact, such nesting would lead +to deadlocks, because of other code paths that take the same locks in +the opposite order. This postponing of I/O-context releases does +complicate code. + +This commit addresses this issue by modifying involved operations in +such a way to not need to get the above I/O-context references any +more. Then it also removes any get and release of these references. + +Signed-off-by: Paolo Valente +--- + block/bfq-cgroup-included.c | 2 +- + block/bfq-mq-iosched.c | 127 ++++++++------------------------------------ + block/bfq-mq.h | 11 ---- + block/bfq-sched.c | 17 ------ + 4 files changed, 22 insertions(+), 135 deletions(-) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index cf59eeb7f08e..dfacca799b5e 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -773,7 +773,7 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) + bfq_put_async_queues(bfqd, bfqg); + + #ifdef BFQ_MQ +- bfq_unlock_put_ioc_restore(bfqd, flags); ++ spin_unlock_irqrestore(&bfqd->lock, flags); + #endif + /* + * @blkg is going offline and will be ignored by +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 02a1e7fd0ea4..8e7589d3280f 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -244,34 +244,6 @@ static void bfq_schedule_dispatch(struct bfq_data *bfqd) + } + } + +-/* +- * Next two functions release bfqd->lock and put the io context +- * pointed by bfqd->ioc_to_put. This delayed put is used to not risk +- * to take an ioc->lock while the scheduler lock is being held. +- */ +-static void bfq_unlock_put_ioc(struct bfq_data *bfqd) +-{ +- struct io_context *ioc_to_put = bfqd->ioc_to_put; +- +- bfqd->ioc_to_put = NULL; +- spin_unlock_irq(&bfqd->lock); +- +- if (ioc_to_put) +- put_io_context(ioc_to_put); +-} +- +-static void bfq_unlock_put_ioc_restore(struct bfq_data *bfqd, +- unsigned long flags) +-{ +- struct io_context *ioc_to_put = bfqd->ioc_to_put; +- +- bfqd->ioc_to_put = NULL; +- spin_unlock_irqrestore(&bfqd->lock, flags); +- +- if (ioc_to_put) +- put_io_context(ioc_to_put); +-} +- + #define BFQ_MQ + #include "bfq-sched.c" + #include "bfq-cgroup-included.c" +@@ -1747,7 +1719,6 @@ static bool bfq_bio_merge(struct blk_mq_hw_ctx *hctx, struct bio *bio) + if (free) + blk_mq_free_request(free); + bfqd->bio_bfqq_set = false; +- BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + + return ret; +@@ -1812,7 +1783,6 @@ static void bfq_request_merged(struct request_queue *q, struct request *req, + bfq_updated_next_req(bfqd, bfqq); + bfq_pos_tree_add_move(bfqd, bfqq); + } +- BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + } + } +@@ -1858,7 +1828,6 @@ static void bfq_requests_merged(struct request_queue *q, struct request *rq, + + bfq_remove_request(q, next); + +- BUG_ON(bfqq->bfqd->ioc_to_put); + spin_unlock_irq(&bfqq->bfqd->lock); + end: + bfqg_stats_update_io_merged(bfqq_group(bfqq), next->cmd_flags); +@@ -2035,20 +2004,18 @@ bfq_setup_merge(struct bfq_queue *bfqq, struct bfq_queue *new_bfqq) + * first time that the requests of some process are redirected to + * it. + * +- * We redirect bfqq to new_bfqq and not the opposite, because we +- * are in the context of the process owning bfqq, hence we have +- * the io_cq of this process. So we can immediately configure this +- * io_cq to redirect the requests of the process to new_bfqq. ++ * We redirect bfqq to new_bfqq and not the opposite, because ++ * we are in the context of the process owning bfqq, thus we ++ * have the io_cq of this process. So we can immediately ++ * configure this io_cq to redirect the requests of the ++ * process to new_bfqq. In contrast, the io_cq of new_bfqq is ++ * not available any more (new_bfqq->bic == NULL). + * +- * NOTE, even if new_bfqq coincides with the in-service queue, the +- * io_cq of new_bfqq is not available, because, if the in-service +- * queue is shared, bfqd->in_service_bic may not point to the +- * io_cq of the in-service queue. +- * Redirecting the requests of the process owning bfqq to the +- * currently in-service queue is in any case the best option, as +- * we feed the in-service queue with new requests close to the +- * last request served and, by doing so, hopefully increase the +- * throughput. ++ * Anyway, even in case new_bfqq coincides with the in-service ++ * queue, redirecting requests the in-service queue is the ++ * best option, as we feed the in-service queue with new ++ * requests close to the last request served and, by doing so, ++ * are likely to increase the throughput. + */ + bfqq->new_bfqq = new_bfqq; + new_bfqq->ref += process_refs; +@@ -2147,13 +2114,13 @@ bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, + in_service_bfqq = bfqd->in_service_queue; + + if (in_service_bfqq && in_service_bfqq != bfqq && +- bfqd->in_service_bic && wr_from_too_long(in_service_bfqq) ++ wr_from_too_long(in_service_bfqq) + && likely(in_service_bfqq == &bfqd->oom_bfqq)) + bfq_log_bfqq(bfqd, bfqq, + "would have tried merge with in-service-queue, but wr"); + +- if (!in_service_bfqq || in_service_bfqq == bfqq || +- !bfqd->in_service_bic || wr_from_too_long(in_service_bfqq) || ++ if (!in_service_bfqq || in_service_bfqq == bfqq ++ || wr_from_too_long(in_service_bfqq) || + unlikely(in_service_bfqq == &bfqd->oom_bfqq)) + goto check_scheduled; + +@@ -2214,16 +2181,6 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq) + BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); + } + +-static void bfq_get_bic_reference(struct bfq_queue *bfqq) +-{ +- /* +- * If bfqq->bic has a non-NULL value, the bic to which it belongs +- * is about to begin using a shared bfq_queue. +- */ +- if (bfqq->bic) +- atomic_long_inc(&bfqq->bic->icq.ioc->refcount); +-} +- + static void + bfq_merge_bfqqs(struct bfq_data *bfqd, struct bfq_io_cq *bic, + struct bfq_queue *bfqq, struct bfq_queue *new_bfqq) +@@ -2280,12 +2237,6 @@ bfq_merge_bfqqs(struct bfq_data *bfqd, struct bfq_io_cq *bic, + bfqd->wr_busy_queues); + + /* +- * Grab a reference to the bic, to prevent it from being destroyed +- * before being possibly touched by a bfq_split_bfqq(). +- */ +- bfq_get_bic_reference(bfqq); +- bfq_get_bic_reference(new_bfqq); +- /* + * Merge queues (that is, let bic redirect its requests to new_bfqq) + */ + bic_set_bfqq(bic, new_bfqq, 1); +@@ -2472,16 +2423,10 @@ static struct bfq_queue *bfq_set_in_service_queue(struct bfq_data *bfqd) + static void bfq_arm_slice_timer(struct bfq_data *bfqd) + { + struct bfq_queue *bfqq = bfqd->in_service_queue; +- struct bfq_io_cq *bic; + u32 sl; + + BUG_ON(!RB_EMPTY_ROOT(&bfqq->sort_list)); + +- /* Processes have exited, don't wait. */ +- bic = bfqd->in_service_bic; +- if (!bic || atomic_read(&bic->icq.ioc->active_ref) == 0) +- return; +- + bfq_mark_bfqq_wait_request(bfqq); + + /* +@@ -3922,11 +3867,6 @@ static struct request *bfq_dispatch_rq_from_bfqq(struct bfq_data *bfqd, + bfq_bfqq_budget_left(bfqq), + bfqq->dispatched); + +- if (!bfqd->in_service_bic) { +- atomic_long_inc(&RQ_BIC(rq)->icq.ioc->refcount); +- bfqd->in_service_bic = RQ_BIC(rq); +- } +- + /* + * Expire bfqq, pretending that its budget expired, if bfqq + * belongs to CLASS_IDLE and other queues are waiting for +@@ -4085,7 +4025,7 @@ static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + spin_lock_irq(&bfqd->lock); + + rq = __bfq_dispatch_request(hctx); +- bfq_unlock_put_ioc(bfqd); ++ spin_unlock_irq(&bfqd->lock); + + return rq; + } +@@ -4184,21 +4124,10 @@ static void bfq_exit_icq_bfqq(struct bfq_io_cq *bic, bool is_sync) + unsigned long flags; + + spin_lock_irqsave(&bfqd->lock, flags); +- BUG_ON(bfqd->ioc_to_put); +- /* +- * If the bic is using a shared queue, put the +- * reference taken on the io_context when the bic +- * started using a shared bfq_queue. This put cannot +- * make ioc->ref_count reach 0, then no ioc->lock +- * risks to be taken (leading to possible deadlock +- * scenarios). +- */ +- if (is_sync && bfq_bfqq_coop(bfqq)) +- put_io_context(bic->icq.ioc); + + bfq_exit_bfqq(bfqd, bfqq); + bic_set_bfqq(bic, NULL, is_sync); +- bfq_unlock_put_ioc_restore(bfqd, flags); ++ spin_unlock_irqrestore(&bfqd->lock, flags); + } + } + +@@ -4633,12 +4562,10 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + + spin_lock_irq(&bfqd->lock); + if (blk_mq_sched_try_insert_merge(q, rq)) { +- BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + return; + } + +- BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + + blk_mq_sched_request_inserted(rq); +@@ -4671,7 +4598,7 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + } + } + +- bfq_unlock_put_ioc(bfqd); ++ spin_unlock_irq(&bfqd->lock); + } + + static void bfq_insert_requests(struct blk_mq_hw_ctx *hctx, +@@ -4864,12 +4791,11 @@ static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + unsigned long flags; + + spin_lock_irqsave(&bfqd->lock, flags); +- BUG_ON(bfqd->ioc_to_put); + + bfq_completed_request(bfqq, bfqd); + bfq_put_rq_priv_body(bfqq); + +- bfq_unlock_put_ioc_restore(bfqd, flags); ++ spin_unlock_irqrestore(&bfqd->lock, flags); + } else { + /* + * Request rq may be still/already in the scheduler, +@@ -4992,7 +4918,6 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + + bfqq = bfq_get_bfqq_handle_split(bfqd, bic, bio, false, is_sync, + &new_queue); +- BUG_ON(bfqd->ioc_to_put); + + if (unlikely(!new_queue)) { + /* If the queue was seeky for too long, break it apart. */ +@@ -5005,14 +4930,6 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + bic->saved_in_large_burst = true; + + bfqq = bfq_split_bfqq(bic, bfqq); +- /* +- * A reference to bic->icq.ioc needs to be +- * released after a queue split. Do not do it +- * immediately, to not risk to possibly take +- * an ioc->lock while holding the scheduler +- * lock. +- */ +- bfqd->ioc_to_put = bic->icq.ioc; + + if (!bfqq) + bfqq = bfq_get_bfqq_handle_split(bfqd, bic, bio, +@@ -5045,7 +4962,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + */ + if (likely(bfqq != &bfqd->oom_bfqq) && bfqq_process_refs(bfqq) == 1) { + bfqq->bic = bic; +- if (bfqd->ioc_to_put) { /* if true, then there has been a split */ ++ if (split) { + /* + * The queue has just been split from a shared + * queue: restore the idle window and the +@@ -5059,7 +4976,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + if (unlikely(bfq_bfqq_just_created(bfqq))) + bfq_handle_burst(bfqd, bfqq); + +- bfq_unlock_put_ioc(bfqd); ++ spin_unlock_irq(&bfqd->lock); + + return 0; + +@@ -5077,7 +4994,6 @@ static void bfq_idle_slice_timer_body(struct bfq_queue *bfqq) + + BUG_ON(!bfqd); + spin_lock_irqsave(&bfqd->lock, flags); +- BUG_ON(bfqd->ioc_to_put); + + bfq_log_bfqq(bfqd, bfqq, "handling slice_timer expiration"); + bfq_clear_bfqq_wait_request(bfqq); +@@ -5108,7 +5024,7 @@ static void bfq_idle_slice_timer_body(struct bfq_queue *bfqq) + bfq_bfqq_expire(bfqd, bfqq, true, reason); + + schedule_dispatch: +- bfq_unlock_put_ioc_restore(bfqd, flags); ++ spin_unlock_irqrestore(&bfqd->lock, flags); + bfq_schedule_dispatch(bfqd); + } + +@@ -5186,7 +5102,6 @@ static void bfq_exit_queue(struct elevator_queue *e) + spin_lock_irq(&bfqd->lock); + list_for_each_entry_safe(bfqq, n, &bfqd->idle_list, bfqq_list) + bfq_deactivate_bfqq(bfqd, bfqq, false, false); +- BUG_ON(bfqd->ioc_to_put); + spin_unlock_irq(&bfqd->lock); + + hrtimer_cancel(&bfqd->idle_slice_timer); +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 2c81c02bccc4..36ee24a87dda 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -458,8 +458,6 @@ struct bfq_data { + + /* bfq_queue in service */ + struct bfq_queue *in_service_queue; +- /* bfq_io_cq (bic) associated with the @in_service_queue */ +- struct bfq_io_cq *in_service_bic; + + /* on-disk position of the last served request */ + sector_t last_position; +@@ -621,15 +619,6 @@ struct bfq_data { + struct bfq_queue *bio_bfqq; + /* Extra flag used only for TESTING */ + bool bio_bfqq_set; +- +- /* +- * io context to put right after bfqd->lock is released. This +- * filed is used to perform put_io_context, when needed, to +- * after the scheduler lock has been released, and thus +- * prevent an ioc->lock from being possibly taken while the +- * scheduler lock is being held. +- */ +- struct io_context *ioc_to_put; + }; + + enum bfqq_state_flags { +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index 85e59eeb3569..9c4e6797d8c9 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -1904,23 +1904,6 @@ static void __bfq_bfqd_reset_in_service(struct bfq_data *bfqd) + struct bfq_entity *in_serv_entity = &in_serv_bfqq->entity; + struct bfq_entity *entity = in_serv_entity; + +- if (bfqd->in_service_bic) { +-#ifdef BFQ_MQ +- BUG_ON(bfqd->ioc_to_put); +- /* +- * Schedule the release of a reference to +- * bfqd->in_service_bic->icq.ioc to right after the +- * scheduler lock is released. This ioc is not +- * released immediately, to not risk to possibly take +- * an ioc->lock while holding the scheduler lock. +- */ +- bfqd->ioc_to_put = bfqd->in_service_bic->icq.ioc; +-#else +- put_io_context(bfqd->in_service_bic->icq.ioc); +-#endif +- bfqd->in_service_bic = NULL; +- } +- + bfq_clear_bfqq_wait_request(in_serv_bfqq); + hrtimer_try_to_cancel(&bfqd->idle_slice_timer); + bfqd->in_service_queue = NULL; + +From 1521ad11f8684cf0a1b7249249cd406fee50da6d Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 29 Mar 2017 18:41:46 +0200 +Subject: [PATCH 23/51] BUGFIX: Remove unneeded and deadlock-causing lock in + request_merged + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 2 -- + 1 file changed, 2 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 8e7589d3280f..bb046335ff4f 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -1761,7 +1761,6 @@ static void bfq_request_merged(struct request_queue *q, struct request *req, + BUG_ON(RQ_BFQQ(req) != bfqq); + elv_rb_add(&bfqq->sort_list, req); + +- spin_lock_irq(&bfqd->lock); + /* Choose next request to be served for bfqq */ + prev = bfqq->next_rq; + next_rq = bfq_choose_req(bfqd, bfqq->next_rq, req, +@@ -1783,7 +1782,6 @@ static void bfq_request_merged(struct request_queue *q, struct request *req, + bfq_updated_next_req(bfqd, bfqq); + bfq_pos_tree_add_move(bfqd, bfqq); + } +- spin_unlock_irq(&bfqd->lock); + } + } + + +From 9136b4c953918ea937254c57cfb787b55b5bc2c6 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 29 Mar 2017 18:55:30 +0200 +Subject: [PATCH 24/51] Fix wrong unlikely + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index bb046335ff4f..3ae9bd424b3f 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4917,7 +4917,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + bfqq = bfq_get_bfqq_handle_split(bfqd, bic, bio, false, is_sync, + &new_queue); + +- if (unlikely(!new_queue)) { ++ if (likely(!new_queue)) { + /* If the queue was seeky for too long, break it apart. */ + if (bfq_bfqq_coop(bfqq) && bfq_bfqq_split_coop(bfqq)) { + BUG_ON(!is_sync); + +From 8e05f722f19645f2278f6962368ca3b7c2a81c9c Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 12 May 2017 09:51:18 +0200 +Subject: [PATCH 25/51] Change cgroup params prefix to bfq-mq for bfq-mq + +Signed-off-by: Paolo Valente +--- + block/bfq-cgroup-included.c | 54 ++++++++++++++++++++++++++------------------- + 1 file changed, 31 insertions(+), 23 deletions(-) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index dfacca799b5e..9e9b0a09e26f 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -995,9 +995,15 @@ bfq_create_group_hierarchy(struct bfq_data *bfqd, int node) + return blkg_to_bfqg(bfqd->queue->root_blkg); + } + ++#ifdef BFQ_MQ ++#define BFQ_CGROUP_FNAME(param) "bfq-mq."#param ++#else ++#define BFQ_CGROUP_FNAME(param) "bfq."#param ++#endif ++ + static struct cftype bfq_blkcg_legacy_files[] = { + { +- .name = "bfq.weight", ++ .name = BFQ_CGROUP_FNAME(weight), + .flags = CFTYPE_NOT_ON_ROOT, + .seq_show = bfq_io_show_weight, + .write_u64 = bfq_io_set_weight_legacy, +@@ -1005,106 +1011,106 @@ static struct cftype bfq_blkcg_legacy_files[] = { + + /* statistics, covers only the tasks in the bfqg */ + { +- .name = "bfq.time", ++ .name = BFQ_CGROUP_FNAME(time), + .private = offsetof(struct bfq_group, stats.time), + .seq_show = bfqg_print_stat, + }, + { +- .name = "bfq.sectors", ++ .name = BFQ_CGROUP_FNAME(sectors), + .seq_show = bfqg_print_stat_sectors, + }, + { +- .name = "bfq.io_service_bytes", ++ .name = BFQ_CGROUP_FNAME(io_service_bytes), + .private = (unsigned long)&blkcg_policy_bfq, + .seq_show = blkg_print_stat_bytes, + }, + { +- .name = "bfq.io_serviced", ++ .name = BFQ_CGROUP_FNAME(io_serviced), + .private = (unsigned long)&blkcg_policy_bfq, + .seq_show = blkg_print_stat_ios, + }, + { +- .name = "bfq.io_service_time", ++ .name = BFQ_CGROUP_FNAME(io_service_time), + .private = offsetof(struct bfq_group, stats.service_time), + .seq_show = bfqg_print_rwstat, + }, + { +- .name = "bfq.io_wait_time", ++ .name = BFQ_CGROUP_FNAME(io_wait_time), + .private = offsetof(struct bfq_group, stats.wait_time), + .seq_show = bfqg_print_rwstat, + }, + { +- .name = "bfq.io_merged", ++ .name = BFQ_CGROUP_FNAME(io_merged), + .private = offsetof(struct bfq_group, stats.merged), + .seq_show = bfqg_print_rwstat, + }, + { +- .name = "bfq.io_queued", ++ .name = BFQ_CGROUP_FNAME(io_queued), + .private = offsetof(struct bfq_group, stats.queued), + .seq_show = bfqg_print_rwstat, + }, + + /* the same statictics which cover the bfqg and its descendants */ + { +- .name = "bfq.time_recursive", ++ .name = BFQ_CGROUP_FNAME(time_recursive), + .private = offsetof(struct bfq_group, stats.time), + .seq_show = bfqg_print_stat_recursive, + }, + { +- .name = "bfq.sectors_recursive", ++ .name = BFQ_CGROUP_FNAME(sectors_recursive), + .seq_show = bfqg_print_stat_sectors_recursive, + }, + { +- .name = "bfq.io_service_bytes_recursive", ++ .name = BFQ_CGROUP_FNAME(io_service_bytes_recursive), + .private = (unsigned long)&blkcg_policy_bfq, + .seq_show = blkg_print_stat_bytes_recursive, + }, + { +- .name = "bfq.io_serviced_recursive", ++ .name = BFQ_CGROUP_FNAME(io_serviced_recursive), + .private = (unsigned long)&blkcg_policy_bfq, + .seq_show = blkg_print_stat_ios_recursive, + }, + { +- .name = "bfq.io_service_time_recursive", ++ .name = BFQ_CGROUP_FNAME(io_service_time_recursive), + .private = offsetof(struct bfq_group, stats.service_time), + .seq_show = bfqg_print_rwstat_recursive, + }, + { +- .name = "bfq.io_wait_time_recursive", ++ .name = BFQ_CGROUP_FNAME(io_wait_time_recursive), + .private = offsetof(struct bfq_group, stats.wait_time), + .seq_show = bfqg_print_rwstat_recursive, + }, + { +- .name = "bfq.io_merged_recursive", ++ .name = BFQ_CGROUP_FNAME(io_merged_recursive), + .private = offsetof(struct bfq_group, stats.merged), + .seq_show = bfqg_print_rwstat_recursive, + }, + { +- .name = "bfq.io_queued_recursive", ++ .name = BFQ_CGROUP_FNAME(io_queued_recursive), + .private = offsetof(struct bfq_group, stats.queued), + .seq_show = bfqg_print_rwstat_recursive, + }, + { +- .name = "bfq.avg_queue_size", ++ .name = BFQ_CGROUP_FNAME(avg_queue_size), + .seq_show = bfqg_print_avg_queue_size, + }, + { +- .name = "bfq.group_wait_time", ++ .name = BFQ_CGROUP_FNAME(group_wait_time), + .private = offsetof(struct bfq_group, stats.group_wait_time), + .seq_show = bfqg_print_stat, + }, + { +- .name = "bfq.idle_time", ++ .name = BFQ_CGROUP_FNAME(idle_time), + .private = offsetof(struct bfq_group, stats.idle_time), + .seq_show = bfqg_print_stat, + }, + { +- .name = "bfq.empty_time", ++ .name = BFQ_CGROUP_FNAME(empty_time), + .private = offsetof(struct bfq_group, stats.empty_time), + .seq_show = bfqg_print_stat, + }, + { +- .name = "bfq.dequeue", ++ .name = BFQ_CGROUP_FNAME(dequeue), + .private = offsetof(struct bfq_group, stats.dequeue), + .seq_show = bfqg_print_stat, + }, +@@ -1113,7 +1119,7 @@ static struct cftype bfq_blkcg_legacy_files[] = { + + static struct cftype bfq_blkg_files[] = { + { +- .name = "bfq.weight", ++ .name = BFQ_CGROUP_FNAME(weight), + .flags = CFTYPE_NOT_ON_ROOT, + .seq_show = bfq_io_show_weight, + .write = bfq_io_set_weight, +@@ -1121,6 +1127,8 @@ static struct cftype bfq_blkg_files[] = { + {} /* terminate */ + }; + ++#undef BFQ_CGROUP_FNAME ++ + #else /* BFQ_GROUP_IOSCHED_ENABLED */ + + static inline void bfqg_stats_update_io_add(struct bfq_group *bfqg, + +From abdf7565dadbb00e78be5f4fb2cc9b157649840e Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 12 May 2017 11:56:13 +0200 +Subject: [PATCH 26/51] Add tentative extra tests on groups, reqs and queues + +Signed-off-by: Paolo Valente +--- + block/bfq-cgroup-included.c | 1 + + block/bfq-mq-iosched.c | 5 +++++ + include/linux/blkdev.h | 2 ++ + 3 files changed, 8 insertions(+) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index 9e9b0a09e26f..72107ad12220 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -412,6 +412,7 @@ static void bfq_pd_init(struct blkg_policy_data *pd) + BUG_ON(!blkg); + bfqg = blkg_to_bfqg(blkg); + bfqd = blkg->q->elevator->elevator_data; ++ BUG_ON(bfqg == bfqd->root_group); + entity = &bfqg->entity; + d = blkcg_to_bfqgd(blkg->blkcg); + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 3ae9bd424b3f..a9e3406fef06 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4494,6 +4494,7 @@ static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, + static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + { + struct bfq_queue *bfqq = RQ_BFQQ(rq), *new_bfqq; ++ BUG_ON(!bfqq); + + assert_spin_locked(&bfqd->lock); + +@@ -4587,6 +4588,9 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + "insert_request %p in disp: at_head %d", + rq, at_head); + } else { ++ BUG_ON(!(rq->rq_flags & RQF_GOT)); ++ rq->rq_flags &= ~RQF_GOT; ++ + __bfq_insert_request(bfqd, rq); + + if (rq_mergeable(rq)) { +@@ -4974,6 +4978,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + if (unlikely(bfq_bfqq_just_created(bfqq))) + bfq_handle_burst(bfqd, bfqq); + ++ rq->rq_flags |= RQF_GOT; + spin_unlock_irq(&bfqd->lock); + + return 0; +diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h +index 0048e59e6d07..9ae814743095 100644 +--- a/include/linux/blkdev.h ++++ b/include/linux/blkdev.h +@@ -123,6 +123,8 @@ typedef __u32 __bitwise req_flags_t; + #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18)) + /* DEBUG: rq in bfq-mq dispatch list */ + #define RQF_DISP_LIST ((__force req_flags_t)(1 << 19)) ++/* DEBUG: rq had get_rq_private executed on it */ ++#define RQF_GOT ((__force req_flags_t)(1 << 20)) + + /* flags that prevent us from merging requests: */ + #define RQF_NOMERGE_FLAGS \ + +From 9e1c1514bc947c4e04502331372b1cc58459d8d1 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Mon, 15 May 2017 22:25:03 +0200 +Subject: [PATCH 27/51] block, bfq-mq: access and cache blkg data only when + safe + +In blk-cgroup, operations on blkg objects are protected with the +request_queue lock. This is no more the lock that protects +I/O-scheduler operations in blk-mq. In fact, the latter are now +protected with a finer-grained per-scheduler-instance lock. As a +consequence, although blkg lookups are also rcu-protected, blk-mq I/O +schedulers may see inconsistent data when they access blkg and +blkg-related objects. BFQ does access these objects, and does incur +this problem, in the following case. + +The blkg_lookup performed in bfq_get_queue, being protected (only) +through rcu, may happen to return the address of a copy of the +original blkg. If this is the case, then the blkg_get performed in +bfq_get_queue, to pin down the blkg, is useless: it does not prevent +blk-cgroup code from destroying both the original blkg and all objects +directly or indirectly referred by the copy of the blkg. BFQ accesses +these objects, which typically causes a crash for NULL-pointer +dereference of memory-protection violation. + +Some additional protection mechanism should be added to blk-cgroup to +address this issue. In the meantime, this commit provides a quick +temporary fix for BFQ: cache (when safe) blkg data that might +disappear right after a blkg_lookup. + +In particular, this commit exploits the following facts to achieve its +goal without introducing further locks. Destroy operations on a blkg +invoke, as a first step, hooks of the scheduler associated with the +blkg. And these hooks are executed with bfqd->lock held for BFQ. As a +consequence, for any blkg associated with the request queue an +instance of BFQ is attached to, we are guaranteed that such a blkg is +not destroyed, and that all the pointers it contains are consistent, +while that instance is holding its bfqd->lock. A blkg_lookup performed +with bfqd->lock held then returns a fully consistent blkg, which +remains consistent until this lock is held. In more detail, this holds +even if the returned blkg is a copy of the original one. + +Finally, also the object describing a group inside BFQ needs to be +protected from destruction on the blkg_free of the original blkg +(which invokes bfq_pd_free). This commit adds private refcounting for +this object, to let it disappear only after no bfq_queue refers to it +any longer. + +This commit also removes or updates some stale comments on locking +issues related to blk-cgroup operations. + +Reported-by: Tomas Konir +Reported-by: Lee Tibbert +Reported-by: Marco Piazza +Signed-off-by: Paolo Valente +Tested-by: Tomas Konir +Tested-by: Lee Tibbert +Tested-by: Marco Piazza +--- + block/bfq-cgroup-included.c | 149 ++++++++++++++++++++++++++++++++++++++++---- + block/bfq-mq-iosched.c | 2 +- + block/bfq-mq.h | 26 +++----- + 3 files changed, 148 insertions(+), 29 deletions(-) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index 72107ad12220..d903393ee78a 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -43,7 +43,11 @@ BFQG_FLAG_FNS(idling) + BFQG_FLAG_FNS(empty) + #undef BFQG_FLAG_FNS + ++#ifdef BFQ_MQ ++/* This should be called with the scheduler lock held. */ ++#else + /* This should be called with the queue_lock held. */ ++#endif + static void bfqg_stats_update_group_wait_time(struct bfqg_stats *stats) + { + unsigned long long now; +@@ -58,7 +62,11 @@ static void bfqg_stats_update_group_wait_time(struct bfqg_stats *stats) + bfqg_stats_clear_waiting(stats); + } + ++#ifdef BFQ_MQ ++/* This should be called with the scheduler lock held. */ ++#else + /* This should be called with the queue_lock held. */ ++#endif + static void bfqg_stats_set_start_group_wait_time(struct bfq_group *bfqg, + struct bfq_group *curr_bfqg) + { +@@ -72,7 +80,11 @@ static void bfqg_stats_set_start_group_wait_time(struct bfq_group *bfqg, + bfqg_stats_mark_waiting(stats); + } + ++#ifdef BFQ_MQ ++/* This should be called with the scheduler lock held. */ ++#else + /* This should be called with the queue_lock held. */ ++#endif + static void bfqg_stats_end_empty_time(struct bfqg_stats *stats) + { + unsigned long long now; +@@ -198,14 +210,43 @@ static struct bfq_group *bfqq_group(struct bfq_queue *bfqq) + + static void bfqg_get(struct bfq_group *bfqg) + { +- return blkg_get(bfqg_to_blkg(bfqg)); ++#ifdef BFQ_MQ ++ bfqg->ref++; ++#else ++ blkg_get(bfqg_to_blkg(bfqg)); ++#endif + } + + static void bfqg_put(struct bfq_group *bfqg) + { +- return blkg_put(bfqg_to_blkg(bfqg)); ++#ifdef BFQ_MQ ++ bfqg->ref--; ++ ++ BUG_ON(bfqg->ref < 0); ++ if (bfqg->ref == 0) ++ kfree(bfqg); ++#else ++ blkg_put(bfqg_to_blkg(bfqg)); ++#endif ++} ++ ++#ifdef BFQ_MQ ++static void bfqg_and_blkg_get(struct bfq_group *bfqg) ++{ ++ /* see comments in bfq_bic_update_cgroup for why refcounting bfqg */ ++ bfqg_get(bfqg); ++ ++ blkg_get(bfqg_to_blkg(bfqg)); + } + ++static void bfqg_and_blkg_put(struct bfq_group *bfqg) ++{ ++ bfqg_put(bfqg); ++ ++ blkg_put(bfqg_to_blkg(bfqg)); ++} ++#endif ++ + static void bfqg_stats_update_io_add(struct bfq_group *bfqg, + struct bfq_queue *bfqq, + unsigned int op) +@@ -310,7 +351,15 @@ static void bfq_init_entity(struct bfq_entity *entity, + if (bfqq) { + bfqq->ioprio = bfqq->new_ioprio; + bfqq->ioprio_class = bfqq->new_ioprio_class; ++#ifdef BFQ_MQ ++ /* ++ * Make sure that bfqg and its associated blkg do not ++ * disappear before entity. ++ */ ++ bfqg_and_blkg_get(bfqg); ++#else + bfqg_get(bfqg); ++#endif + } + entity->parent = bfqg->my_entity; /* NULL for root group */ + entity->sched_data = &bfqg->sched_data; +@@ -397,6 +446,10 @@ static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, int node) + return NULL; + } + ++#ifdef BFQ_MQ ++ /* see comments in bfq_bic_update_cgroup for why refcounting */ ++ bfqg_get(bfqg); ++#endif + return &bfqg->pd; + } + +@@ -432,7 +485,11 @@ static void bfq_pd_free(struct blkg_policy_data *pd) + struct bfq_group *bfqg = pd_to_bfqg(pd); + + bfqg_stats_exit(&bfqg->stats); +- return kfree(bfqg); ++#ifdef BFQ_MQ ++ bfqg_put(bfqg); ++#else ++ kfree(bfqg); ++#endif + } + + static void bfq_pd_reset_stats(struct blkg_policy_data *pd) +@@ -516,9 +573,16 @@ static void bfq_bfqq_expire(struct bfq_data *bfqd, + * Move @bfqq to @bfqg, deactivating it from its old group and reactivating + * it on the new one. Avoid putting the entity on the old group idle tree. + * ++#ifdef BFQ_MQ ++ * Must be called under the scheduler lock, to make sure that the blkg ++ * owning @bfqg does not disappear (see comments in ++ * bfq_bic_update_cgroup on guaranteeing the consistency of blkg ++ * objects). ++#else + * Must be called under the queue lock; the cgroup owning @bfqg must + * not disappear (by now this just means that we are called under + * rcu_read_lock()). ++#endif + */ + static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, + struct bfq_group *bfqg) +@@ -555,16 +619,20 @@ static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, + entity->tree); + bfq_put_idle_entity(bfq_entity_service_tree(entity), entity); + } ++#ifdef BFQ_MQ ++ bfqg_and_blkg_put(bfqq_group(bfqq)); ++#else + bfqg_put(bfqq_group(bfqq)); ++#endif + +- /* +- * Here we use a reference to bfqg. We don't need a refcounter +- * as the cgroup reference will not be dropped, so that its +- * destroy() callback will not be invoked. +- */ + entity->parent = bfqg->my_entity; + entity->sched_data = &bfqg->sched_data; ++#ifdef BFQ_MQ ++ /* pin down bfqg and its associated blkg */ ++ bfqg_and_blkg_get(bfqg); ++#else + bfqg_get(bfqg); ++#endif + + BUG_ON(RB_EMPTY_ROOT(&bfqq->sort_list) && bfq_bfqq_busy(bfqq)); + if (bfq_bfqq_busy(bfqq)) { +@@ -585,8 +653,14 @@ static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, + * @bic: the bic to move. + * @blkcg: the blk-cgroup to move to. + * ++#ifdef BFQ_MQ ++ * Move bic to blkcg, assuming that bfqd->lock is held; which makes ++ * sure that the reference to cgroup is valid across the call (see ++ * comments in bfq_bic_update_cgroup on this issue) ++#else + * Move bic to blkcg, assuming that bfqd->queue is locked; the caller + * has to make sure that the reference to cgroup is valid across the call. ++#endif + * + * NOTE: an alternative approach might have been to store the current + * cgroup in bfqq and getting a reference to it, reducing the lookup +@@ -645,6 +719,59 @@ static void bfq_bic_update_cgroup(struct bfq_io_cq *bic, struct bio *bio) + goto out; + + bfqg = __bfq_bic_change_cgroup(bfqd, bic, bio_blkcg(bio)); ++#ifdef BFQ_MQ ++ /* ++ * Update blkg_path for bfq_log_* functions. We cache this ++ * path, and update it here, for the following ++ * reasons. Operations on blkg objects in blk-cgroup are ++ * protected with the request_queue lock, and not with the ++ * lock that protects the instances of this scheduler ++ * (bfqd->lock). This exposes BFQ to the following sort of ++ * race. ++ * ++ * The blkg_lookup performed in bfq_get_queue, protected ++ * through rcu, may happen to return the address of a copy of ++ * the original blkg. If this is the case, then the ++ * bfqg_and_blkg_get performed in bfq_get_queue, to pin down ++ * the blkg, is useless: it does not prevent blk-cgroup code ++ * from destroying both the original blkg and all objects ++ * directly or indirectly referred by the copy of the ++ * blkg. ++ * ++ * On the bright side, destroy operations on a blkg invoke, as ++ * a first step, hooks of the scheduler associated with the ++ * blkg. And these hooks are executed with bfqd->lock held for ++ * BFQ. As a consequence, for any blkg associated with the ++ * request queue this instance of the scheduler is attached ++ * to, we are guaranteed that such a blkg is not destroyed, and ++ * that all the pointers it contains are consistent, while we ++ * are holding bfqd->lock. A blkg_lookup performed with ++ * bfqd->lock held then returns a fully consistent blkg, which ++ * remains consistent until this lock is held. ++ * ++ * Thanks to the last fact, and to the fact that: (1) bfqg has ++ * been obtained through a blkg_lookup in the above ++ * assignment, and (2) bfqd->lock is being held, here we can ++ * safely use the policy data for the involved blkg (i.e., the ++ * field bfqg->pd) to get to the blkg associated with bfqg, ++ * and then we can safely use any field of blkg. After we ++ * release bfqd->lock, even just getting blkg through this ++ * bfqg may cause dangling references to be traversed, as ++ * bfqg->pd may not exist any more. ++ * ++ * In view of the above facts, here we cache, in the bfqg, any ++ * blkg data we may need for this bic, and for its associated ++ * bfq_queue. As of now, we need to cache only the path of the ++ * blkg, which is used in the bfq_log_* functions. ++ * ++ * Finally, note that bfqg itself needs to be protected from ++ * destruction on the blkg_free of the original blkg (which ++ * invokes bfq_pd_free). We use an additional private ++ * refcounter for bfqg, to let it disappear only after no ++ * bfq_queue refers to it any longer. ++ */ ++ blkg_path(bfqg_to_blkg(bfqg), bfqg->blkg_path, sizeof(bfqg->blkg_path)); ++#endif + bic->blkcg_serial_nr = serial_nr; + out: + rcu_read_unlock(); +@@ -682,8 +809,6 @@ static void bfq_reparent_leaf_entity(struct bfq_data *bfqd, + * @bfqd: the device data structure with the root group. + * @bfqg: the group to move from. + * @st: the service tree with the entities. +- * +- * Needs queue_lock to be taken and reference to be valid over the call. + */ + static void bfq_reparent_active_entities(struct bfq_data *bfqd, + struct bfq_group *bfqg, +@@ -736,6 +861,7 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) + #ifdef BFQ_MQ + spin_lock_irqsave(&bfqd->lock, flags); + #endif ++ + /* + * Empty all service_trees belonging to this group before + * deactivating the group itself. +@@ -746,8 +872,7 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) + /* + * The idle tree may still contain bfq_queues belonging + * to exited task because they never migrated to a different +- * cgroup from the one being destroyed now. No one else +- * can access them so it's safe to act without any lock. ++ * cgroup from the one being destroyed now. + */ + bfq_flush_idle_tree(st); + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index a9e3406fef06..4eb668eeacdc 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4073,7 +4073,7 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + + kmem_cache_free(bfq_pool, bfqq); + #ifdef BFQ_GROUP_IOSCHED_ENABLED +- bfqg_put(bfqg); ++ bfqg_and_blkg_put(bfqg); + #endif + } + +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 36ee24a87dda..77ab0f22ed22 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -695,23 +695,17 @@ static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); + static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ +- char __pbuf[128]; \ +- \ +- blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ + pr_crit("%s bfq%d%c %s " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ + (bfqq)->pid, \ + bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- __pbuf, ##args); \ ++ bfqq_group(bfqq)->blkg_path, ##args); \ + } while (0) + + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ +- char __pbuf[128]; \ +- \ +- blkg_path(bfqg_to_blkg(bfqg), __pbuf, sizeof(__pbuf)); \ + pr_crit("%s %s " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ +- __pbuf, ##args); \ ++ bfqg->blkg_path, ##args); \ + } while (0) + + #else /* BFQ_GROUP_IOSCHED_ENABLED */ +@@ -736,20 +730,14 @@ static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); + static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ +- char __pbuf[128]; \ +- \ +- blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ + blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s " fmt, \ + (bfqq)->pid, \ + bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- __pbuf, ##args); \ ++ bfqq_group(bfqq)->blkg_path, ##args); \ + } while (0) + + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ +- char __pbuf[128]; \ +- \ +- blkg_path(bfqg_to_blkg(bfqg), __pbuf, sizeof(__pbuf)); \ +- blk_add_trace_msg((bfqd)->queue, "%s " fmt, __pbuf, ##args); \ ++ blk_add_trace_msg((bfqd)->queue, "%s " fmt, bfqg->blkg_path, ##args);\ + } while (0) + + #else /* BFQ_GROUP_IOSCHED_ENABLED */ +@@ -860,6 +848,12 @@ struct bfq_group { + /* must be the first member */ + struct blkg_policy_data pd; + ++ /* cached path for this blkg (see comments in bfq_bic_update_cgroup) */ ++ char blkg_path[128]; ++ ++ /* reference counter (see comments in bfq_bic_update_cgroup) */ ++ int ref; ++ + struct bfq_entity entity; + struct bfq_sched_data sched_data; + + +From c9137b749aceef6c2dde88e99b2fc978d5952e76 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Sat, 17 Jun 2017 11:18:11 +0200 +Subject: [PATCH 28/51] bfq-mq: fix macro name in conditional invocation of + policy_unregister + +This commit fixes the name of the macro in the conditional group that +invokes blkcg_policy_unregister in bfq_exit for bfq-mq. Because of +this error, blkcg_policy_unregister was never invoked. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 4eb668eeacdc..bc1de3f70ea8 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -5669,7 +5669,7 @@ static int __init bfq_init(void) + static void __exit bfq_exit(void) + { + elv_unregister(&iosched_bfq_mq); +-#ifdef CONFIG_BFQ_GROUP_ENABLED ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_policy_unregister(&blkcg_policy_bfq); + #endif + bfq_slab_kill(); + +From c7ceb37496f63b2dba4d06946ab85ec97b87bfb5 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 5 Jul 2017 11:48:17 +0200 +Subject: [PATCH 29/51] Port of "blk-mq-sched: unify request finished methods" + +No need to have two different callouts of bfq vs kyber. + +Signed-off-by: Christoph Hellwig +Signed-off-by: Jens Axboe +--- + block/bfq-mq-iosched.c | 6 +++--- + 1 file changed, 3 insertions(+), 3 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index bc1de3f70ea8..2598602a0b10 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4753,7 +4753,7 @@ static void bfq_put_rq_priv_body(struct bfq_queue *bfqq) + bfq_put_queue(bfqq); + } + +-static void bfq_put_rq_private(struct request_queue *q, struct request *rq) ++static void bfq_finish_request(struct request *rq) + { + struct bfq_queue *bfqq; + struct bfq_data *bfqd; +@@ -4814,7 +4814,7 @@ static void bfq_put_rq_private(struct request_queue *q, struct request *rq) + + assert_spin_locked(&bfqd->lock); + if (!RB_EMPTY_NODE(&rq->rb_node)) +- bfq_remove_request(q, rq); ++ bfq_remove_request(rq->q, rq); + bfq_put_rq_priv_body(bfqq); + } + +@@ -5558,7 +5558,7 @@ static struct elv_fs_entry bfq_attrs[] = { + static struct elevator_type iosched_bfq_mq = { + .ops.mq = { + .get_rq_priv = bfq_get_rq_private, +- .put_rq_priv = bfq_put_rq_private, ++ .finish_request = bfq_finish_request, + .exit_icq = bfq_exit_icq, + .insert_requests = bfq_insert_requests, + .dispatch_request = bfq_dispatch_request, + +From 12bef026fe114ab5e2e284772ddc52a8be83fdbc Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 5 Jul 2017 11:54:57 +0200 +Subject: [PATCH 30/51] Port of "bfq-iosched: fix NULL ioc check in + bfq_get_rq_private" + +icq_to_bic is a container_of operation, so we need to check for NULL +before it. Also move the check outside the spinlock while we're at +it. + +Signed-off-by: Christoph Hellwig +Signed-off-by: Jens Axboe +--- + block/bfq-mq-iosched.c | 15 +++++---------- + 1 file changed, 5 insertions(+), 10 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 2598602a0b10..c57774a60911 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4903,16 +4903,17 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + struct bio *bio) + { + struct bfq_data *bfqd = q->elevator->elevator_data; +- struct bfq_io_cq *bic = icq_to_bic(rq->elv.icq); ++ struct bfq_io_cq *bic; + const int is_sync = rq_is_sync(rq); + struct bfq_queue *bfqq; + bool bfqq_already_existing = false, split = false; + bool new_queue = false; + +- spin_lock_irq(&bfqd->lock); ++ if (!rq->elv.icq) ++ return 1; ++ bic = icq_to_bic(rq->elv.icq); + +- if (!bic) +- goto queue_fail; ++ spin_lock_irq(&bfqd->lock); + + bfq_check_ioprio_change(bic, bio); + +@@ -4980,13 +4981,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + + rq->rq_flags |= RQF_GOT; + spin_unlock_irq(&bfqd->lock); +- + return 0; +- +-queue_fail: +- spin_unlock_irq(&bfqd->lock); +- +- return 1; + } + + static void bfq_idle_slice_timer_body(struct bfq_queue *bfqq) + +From 633e5711347df1bf4ca935fd0aa9118a0054f75d Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 5 Jul 2017 12:02:16 +0200 +Subject: [PATCH 31/51] Port of "blk-mq-sched: unify request prepare methods" + +This patch makes sure we always allocate requests in the core blk-mq +code and use a common prepare_request method to initialize them for +both mq I/O schedulers. For Kyber and additional limit_depth method +is added that is called before allocating the request. + +Also because none of the intializations can really fail the new method +does not return an error - instead the bfq finish method is hardened +to deal with the no-IOC case. + +Last but not least this removes the abuse of RQF_QUEUE by the blk-mq +scheduling code as RQF_ELFPRIV is all that is needed now. + +Signed-off-by: Christoph Hellwig +Signed-off-by: Jens Axboe +--- + block/bfq-mq-iosched.c | 13 ++++++++----- + 1 file changed, 8 insertions(+), 5 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index c57774a60911..49ffca1ad6e7 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4760,6 +4760,10 @@ static void bfq_finish_request(struct request *rq) + struct bfq_io_cq *bic; + + BUG_ON(!rq); ++ ++ if (!rq->elv.icq) ++ return; ++ + bfqq = RQ_BFQQ(rq); + BUG_ON(!bfqq); + +@@ -4899,9 +4903,9 @@ static struct bfq_queue *bfq_get_bfqq_handle_split(struct bfq_data *bfqd, + /* + * Allocate bfq data structures associated with this request. + */ +-static int bfq_get_rq_private(struct request_queue *q, struct request *rq, +- struct bio *bio) ++static void bfq_prepare_request(struct request *rq, struct bio *bio) + { ++ struct request_queue *q = rq->q; + struct bfq_data *bfqd = q->elevator->elevator_data; + struct bfq_io_cq *bic; + const int is_sync = rq_is_sync(rq); +@@ -4910,7 +4914,7 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + bool new_queue = false; + + if (!rq->elv.icq) +- return 1; ++ return; + bic = icq_to_bic(rq->elv.icq); + + spin_lock_irq(&bfqd->lock); +@@ -4981,7 +4985,6 @@ static int bfq_get_rq_private(struct request_queue *q, struct request *rq, + + rq->rq_flags |= RQF_GOT; + spin_unlock_irq(&bfqd->lock); +- return 0; + } + + static void bfq_idle_slice_timer_body(struct bfq_queue *bfqq) +@@ -5552,7 +5555,7 @@ static struct elv_fs_entry bfq_attrs[] = { + + static struct elevator_type iosched_bfq_mq = { + .ops.mq = { +- .get_rq_priv = bfq_get_rq_private, ++ .prepare_request = bfq_prepare_request, + .finish_request = bfq_finish_request, + .exit_icq = bfq_exit_icq, + .insert_requests = bfq_insert_requests, + +From 5a321acfce282c3e58ac63582faf6f928ad17f27 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 5 Jul 2017 12:43:22 +0200 +Subject: [PATCH 32/51] Add list of bfq instances to documentation + +Signed-off-by: Paolo Valente +--- + Documentation/block/bfq-iosched.txt | 11 ++++++++++- + 1 file changed, 10 insertions(+), 1 deletion(-) + +diff --git a/Documentation/block/bfq-iosched.txt b/Documentation/block/bfq-iosched.txt +index 3d6951d63489..8ce6b9a9bacd 100644 +--- a/Documentation/block/bfq-iosched.txt ++++ b/Documentation/block/bfq-iosched.txt +@@ -11,6 +11,15 @@ controllers), BFQ's main features are: + groups (switching back to time distribution when needed to keep + throughput high). + ++If bfq-mq patches have been applied, then the following three ++instances of BFQ are available (otherwise only the first instance): ++- bfq: mainline version of BFQ, for blk-mq ++- bfq-mq: development version of BFQ for blk-mq; this version contains ++ also all latest features not yet landed in mainline, plus many ++ safety checks ++- bfq: BFQ for legacy blk; also this version contains both latest ++ features and safety checks ++ + In its default configuration, BFQ privileges latency over + throughput. So, when needed for achieving a lower latency, BFQ builds + schedules that may lead to a lower throughput. If your main or only +@@ -27,7 +36,7 @@ sequential I/O (e.g., 8-12 GB/s if I/O requests are 256 KB large), and + to 120-200 MB/s with 4KB random I/O. BFQ is currently being tested on + multi-queue devices too. + +-The table of contents follow. Impatients can just jump to Section 3. ++The table of contents follows. Impatients can just jump to Section 3. + + CONTENTS + + +From 9f2e5b27227fd9254cc258572dc2d4531838c30b Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 5 Jul 2017 16:28:00 +0200 +Subject: [PATCH 33/51] bfq-sq: fix prefix of names of cgroups parameters + +Signed-off-by: Paolo Valente +--- + Documentation/block/bfq-iosched.txt | 12 +++++++----- + block/bfq-cgroup-included.c | 2 +- + 2 files changed, 8 insertions(+), 6 deletions(-) + +diff --git a/Documentation/block/bfq-iosched.txt b/Documentation/block/bfq-iosched.txt +index 8ce6b9a9bacd..965d82f94db9 100644 +--- a/Documentation/block/bfq-iosched.txt ++++ b/Documentation/block/bfq-iosched.txt +@@ -503,10 +503,12 @@ To get proportional sharing of bandwidth with BFQ for a given device, + BFQ must of course be the active scheduler for that device. + + Within each group directory, the names of the files associated with +-BFQ-specific cgroup parameters and stats begin with the "bfq." +-prefix. So, with cgroups-v1 or cgroups-v2, the full prefix for +-BFQ-specific files is "blkio.bfq." or "io.bfq." For example, the group +-parameter to set the weight of a group with BFQ is blkio.bfq.weight ++BFQ-specific cgroup parameters and stats begin with the "bfq.", ++"bfq-sq." or "bfq-mq." prefix, depending on which instance of bfq you ++want to use. So, with cgroups-v1 or cgroups-v2, the full prefix for ++BFQ-specific files is "blkio.bfqX." or "io.bfqX.", where X can be "" ++(i.e., null string), "-sq" or "-mq". For example, the group parameter ++to set the weight of a group with the mainline BFQ is blkio.bfq.weight + or io.bfq.weight. + + Parameters to set +@@ -514,7 +516,7 @@ Parameters to set + + For each group, there is only the following parameter to set. + +-weight (namely blkio.bfq.weight or io.bfq-weight): the weight of the ++weight (namely blkio.bfqX.weight or io.bfqX.weight): the weight of the + group inside its parent. Available values: 1..10000 (default 100). The + linear mapping between ioprio and weights, described at the beginning + of the tunable section, is still valid, but all weights higher than +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index d903393ee78a..631e53d9150d 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -1124,7 +1124,7 @@ bfq_create_group_hierarchy(struct bfq_data *bfqd, int node) + #ifdef BFQ_MQ + #define BFQ_CGROUP_FNAME(param) "bfq-mq."#param + #else +-#define BFQ_CGROUP_FNAME(param) "bfq."#param ++#define BFQ_CGROUP_FNAME(param) "bfq-sq."#param + #endif + + static struct cftype bfq_blkcg_legacy_files[] = { + +From 92b42df8166939ccf26aa450125b5b575cf6d505 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 5 Jul 2017 21:08:32 +0200 +Subject: [PATCH 34/51] Add to documentation that bfq-mq and bfq-sq contain + last fixes too + +Signed-off-by: Paolo Valente +--- + Documentation/block/bfq-iosched.txt | 6 +++--- + 1 file changed, 3 insertions(+), 3 deletions(-) + +diff --git a/Documentation/block/bfq-iosched.txt b/Documentation/block/bfq-iosched.txt +index 965d82f94db9..0e59f1c9d30e 100644 +--- a/Documentation/block/bfq-iosched.txt ++++ b/Documentation/block/bfq-iosched.txt +@@ -15,10 +15,10 @@ If bfq-mq patches have been applied, then the following three + instances of BFQ are available (otherwise only the first instance): + - bfq: mainline version of BFQ, for blk-mq + - bfq-mq: development version of BFQ for blk-mq; this version contains +- also all latest features not yet landed in mainline, plus many ++ also all latest features and fixes not yet landed in mainline, plus many + safety checks +-- bfq: BFQ for legacy blk; also this version contains both latest +- features and safety checks ++- bfq: BFQ for legacy blk; also this version contains latest features ++ and fixes, as well as safety checks + + In its default configuration, BFQ privileges latency over + throughput. So, when needed for achieving a lower latency, BFQ builds + +From 7f9bdd433b848d4f53c167258bf4d0b3f1ae1923 Mon Sep 17 00:00:00 2001 +From: Lee Tibbert +Date: Wed, 19 Jul 2017 10:28:32 -0400 +Subject: [PATCH 35/51] Improve most frequently used no-logging path + +This patch originated as a fix for compiler unused-variable warnings +issued when compiling bfq-mq with logging disabled (both +CONFIG_BLK_DEV_IO_TRACE and CONFIG_BFQ_REDIRECT_TO_CONSOLE +undefined). + +It turns out to also have benefits for the bfq-sq path as well. + +In most performance sensitive production builds blktrace_api logging +will probably be turned off, so it is worth making the no-logging path +compile without warnings. Any performance benefit is a bonus. + +Thank you to T. B. on the bfq-iosched@googlegroups.com list +for ((void) (bfqq)) simplification/suggestion/improvement. All bugs +and unclear descriptions are my own doing. + +The discussion below is based on the gcc compiler with optimization +level of at least 02. Lower optimization levels are unlikely to +remove no-op instruction equivalents. + +Provide three improvements in this likely case. + + 1) Fix multiple occurrences of an unused-variable warning + issued when compiling bfq-mq with no logging. The warning + occurred each time the bfq_log_bfqg macro was expanded inside + a code block such as the following snippet from + block/bfq-sched.c, line 139 and few following, lightly edited for + indentation in order to pass checkpatch.pl maximum line lengths. + +else { + struct bfq_group *bfqg = + container_of(next_in_service, + struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, + "update_next_in_service: chosen this entity"); + } + + Previously bfq-mq.h expanded bfq_log_bfqg to blk_add_trace_msg. + When both bfq console logging and blktrace_api logging are + disabled, include/linux/blktrace_api expands to + do { } while (0), leaving the code block local variable unused. + + bfq_log_bfqq() had similar behavior but is never called with + a potentially unused variable. This patch fixes that macro for + consistency. + + bfq-sq.h (single queue) with blktrace_api enabled, and the bfq + console logging macros have code paths which not trigger this + warning. + + kernel.org (4.12 & 4.13) bfq (bfq-iosched.h) could trigger + the warning but no code does so now. This patch fixes + bfq-iosched.h for consistency. + + The style above enables a software engineering approach where + complex expressions are moved to a local variable before the + bfq_log* call. This makes it easier to read the expression and + use breakpoints to verify it. bfq-mq uses this approach in + several places. + + New bfq_log* macros are provided for the no-logging case. + I touch only the second argument, because current code never + uses the local variable approach with the first or other + arguments. I tried to balance consistency with simplicity. + + 2) For bfq-sq, reduce to zero, the number of instructions executed + when no logging is configured. No sense marshaling arguments + which are never going to be used. + + On a trial V8R11 builds, this reduced the size of bfq-iosched.o + by 14.3 KiB. The size went from 70304 to 55664 bytes. + + bfq-mq and kernel.org bfq code size does not change because + existing macros already optimize to zero bytes when not logging. + The current changes maintains consistency with the bfq-sq path + and makes the bfq-mq & bfq no-logging paths resistant to future + logging path macro changes which might cause generated code. + + 3) Slightly reduce compile time of all bfq variants by including + blktrace_api.h only when it will be used. + +Signed-off-by: Lee Tibbert +--- + block/bfq-mq.h | 18 +++++++++++++++++- + block/bfq.h | 18 +++++++++++++++++- + 2 files changed, 34 insertions(+), 2 deletions(-) + +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 77ab0f22ed22..7ed2cc29be57 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -15,7 +15,6 @@ + #ifndef _BFQ_H + #define _BFQ_H + +-#include + #include + #include + +@@ -725,6 +724,21 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + ##args) + + #else /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ ++ ++#if !defined(CONFIG_BLK_DEV_IO_TRACE) ++ ++/* Avoid possible "unused-variable" warning. See commit message. */ ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) ((void) (bfqq)) ++ ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) ((void) (bfqg)) ++ ++#define bfq_log(bfqd, fmt, args...) do {} while (0) ++ ++#else /* CONFIG_BLK_DEV_IO_TRACE */ ++ ++#include ++ + #ifdef BFQ_GROUP_IOSCHED_ENABLED + static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); + static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); +@@ -752,6 +766,8 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + + #define bfq_log(bfqd, fmt, args...) \ + blk_add_trace_msg((bfqd)->queue, "bfq " fmt, ##args) ++ ++#endif /* CONFIG_BLK_DEV_IO_TRACE */ + #endif /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ + + /* Expiration reasons. */ +diff --git a/block/bfq.h b/block/bfq.h +index 53954d1b87f8..15d326f466b7 100644 +--- a/block/bfq.h ++++ b/block/bfq.h +@@ -15,7 +15,6 @@ + #ifndef _BFQ_H + #define _BFQ_H + +-#include + #include + #include + +@@ -725,6 +724,21 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + ##args) + + #else /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ ++ ++#if !defined(CONFIG_BLK_DEV_IO_TRACE) ++ ++/* Avoid possible "unused-variable" warning. See commit message. */ ++ ++#define bfq_log_bfqq(bfqd, bfqq, fmt, args...) ((void) (bfqq)) ++ ++#define bfq_log_bfqg(bfqd, bfqg, fmt, args...) ((void) (bfqg)) ++ ++#define bfq_log(bfqd, fmt, args...) do {} while (0) ++ ++#else /* CONFIG_BLK_DEV_IO_TRACE */ ++ ++#include ++ + #ifdef BFQ_GROUP_IOSCHED_ENABLED + static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); + static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); +@@ -759,6 +773,8 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + + #define bfq_log(bfqd, fmt, args...) \ + blk_add_trace_msg((bfqd)->queue, "bfq " fmt, ##args) ++ ++#endif /* CONFIG_BLK_DEV_IO_TRACE */ + #endif /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ + + /* Expiration reasons. */ + +From f11a0e751e741bf94c6a48234824d50b3c0100ad Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 9 Aug 2017 16:40:39 +0200 +Subject: [PATCH 36/51] bfq-sq: fix commit "Remove all get and put of I/O + contexts" in branch bfq-mq + +The commit "Remove all get and put of I/O contexts" erroneously removed +the reset of the field in_service_bic for bfq-sq. This commit re-adds +that missing reset. + +Signed-off-by: Paolo Valente +--- + block/bfq-sched.c | 7 +++++++ + block/bfq-sq-iosched.c | 1 + + 2 files changed, 8 insertions(+) + +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index 9c4e6797d8c9..7425824c26b8 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -1904,6 +1904,13 @@ static void __bfq_bfqd_reset_in_service(struct bfq_data *bfqd) + struct bfq_entity *in_serv_entity = &in_serv_bfqq->entity; + struct bfq_entity *entity = in_serv_entity; + ++#ifndef BFQ_MQ ++ if (bfqd->in_service_bic) { ++ put_io_context(bfqd->in_service_bic->icq.ioc); ++ bfqd->in_service_bic = NULL; ++ } ++#endif ++ + bfq_clear_bfqq_wait_request(in_serv_bfqq); + hrtimer_try_to_cancel(&bfqd->idle_slice_timer); + bfqd->in_service_queue = NULL; +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 25da0d1c0622..e1960bf149d8 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -3765,6 +3765,7 @@ static int bfq_dispatch_request(struct bfq_data *bfqd, + if (!bfqd->in_service_bic) { + atomic_long_inc(&RQ_BIC(rq)->icq.ioc->refcount); + bfqd->in_service_bic = RQ_BIC(rq); ++ BUG_ON(!bfqd->in_service_bic); + } + + if (bfqd->busy_queues > 1 && bfq_class_idle(bfqq)) + +From eceae5457530df8598557767d7be258ca9384de4 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 9 Aug 2017 22:29:01 +0200 +Subject: [PATCH 37/51] bfq-sq-mq: make lookup_next_entity push up vtime on + expirations + +To provide a very smooth service, bfq starts to serve a bfq_queue +only if the queue is 'eligible', i.e., if the same queue would +have started to be served in the ideal, perfectly fair system that +bfq simulates internally. This is obtained by associating each +queue with a virtual start time, and by computing a special system +virtual time quantity: a queue is eligible only if the system +virtual time has reached the virtual start time of the +queue. Finally, bfq guarantees that, when a new queue must be set +in service, there is always at least one eligible entity for each +active parent entity in the scheduler. To provide this guarantee, +the function __bfq_lookup_next_entity pushes up, for each parent +entity on which it is invoked, the system virtual time to the +minimum among the virtual start times of the entities in the +active tree for the parent entity (more precisely, the push up +occurs if the system virtual time happens to be lower than all +such virtual start times). + +There is however a circumstance in which __bfq_lookup_next_entity +cannot push up the system virtual time for a parent entity, even +if the system virtual time is lower than the virtual start times +of all the child entities in the active tree. It happens if one of +the child entities is in service. In fact, in such a case, there +is already an eligible entity, the in-service one, even if it may +not be not present in the active tree (because in-service entities +may be removed from the active tree). + +Unfortunately, in the last re-design of the +hierarchical-scheduling engine, the reset of the pointer to the +in-service entity for a given parent entity--reset to be done as a +consequence of the expiration of the in-service entity--always +happens after the function __bfq_lookup_next_entity has been +invoked. This causes the function to think that there is still an +entity in service for the parent entity, and then that the system +virtual time cannot be pushed up, even if actually such a +no-more-in-service entity has already been properly reinserted +into the active tree (or in some other tree if no more +active). Yet, the system virtual time *had* to be pushed up, to be +ready to correctly choose the next queue to serve. Because of the +lack of this push up, bfq may wrongly set in service a queue that +had been speculatively pre-computed as the possible +next-in-service queue, but that would no more be the one to serve +after the expiration and the reinsertion into the active trees of +the previously in-service entities. + +This commit addresses this issue by making +__bfq_lookup_next_entity properly push up the system virtual time +if an expiration is occurring. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 4 +-- + block/bfq-sched.c | 77 ++++++++++++++++++++++++++++++++------------------ + block/bfq-sq-iosched.c | 4 +-- + 3 files changed, 53 insertions(+), 32 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 49ffca1ad6e7..b5c848650375 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -682,7 +682,7 @@ static void bfq_updated_next_req(struct bfq_data *bfqd, + entity->budget = new_budget; + bfq_log_bfqq(bfqd, bfqq, "updated next rq: new budget %lu", + new_budget); +- bfq_requeue_bfqq(bfqd, bfqq); ++ bfq_requeue_bfqq(bfqd, bfqq, false); + } + } + +@@ -2822,7 +2822,7 @@ static void __bfq_bfqq_expire(struct bfq_data *bfqd, struct bfq_queue *bfqq) + + bfq_del_bfqq_busy(bfqd, bfqq, true); + } else { +- bfq_requeue_bfqq(bfqd, bfqq); ++ bfq_requeue_bfqq(bfqd, bfqq, true); + /* + * Resort priority tree of potential close cooperators. + */ +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index 7425824c26b8..f3001af37256 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -33,7 +33,8 @@ static struct bfq_entity *bfq_root_active_entity(struct rb_root *tree) + return rb_entry(node, struct bfq_entity, rb_node); + } + +-static struct bfq_entity *bfq_lookup_next_entity(struct bfq_sched_data *sd); ++static struct bfq_entity *bfq_lookup_next_entity(struct bfq_sched_data *sd, ++ bool expiration); + + static bool bfq_update_parent_budget(struct bfq_entity *next_in_service); + +@@ -43,6 +44,8 @@ static bool bfq_update_parent_budget(struct bfq_entity *next_in_service); + * @new_entity: if not NULL, pointer to the entity whose activation, + * requeueing or repositionig triggered the invocation of + * this function. ++ * @expiration: id true, this function is being invoked after the ++ * expiration of the in-service entity + * + * This function is called to update sd->next_in_service, which, in + * its turn, may change as a consequence of the insertion or +@@ -61,7 +64,8 @@ static bool bfq_update_parent_budget(struct bfq_entity *next_in_service); + * entity. + */ + static bool bfq_update_next_in_service(struct bfq_sched_data *sd, +- struct bfq_entity *new_entity) ++ struct bfq_entity *new_entity, ++ bool expiration) + { + struct bfq_entity *next_in_service = sd->next_in_service; + struct bfq_queue *bfqq; +@@ -120,7 +124,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + if (replace_next) + next_in_service = new_entity; + } else /* invoked because of a deactivation: lookup needed */ +- next_in_service = bfq_lookup_next_entity(sd); ++ next_in_service = bfq_lookup_next_entity(sd, expiration); + + if (next_in_service) { + parent_sched_may_change = !sd->next_in_service || +@@ -1291,10 +1295,12 @@ static void __bfq_activate_requeue_entity(struct bfq_entity *entity, + * @requeue: true if this is a requeue, which implies that bfqq is + * being expired; thus ALL its ancestors stop being served and must + * therefore be requeued ++ * @expiration: true if this function is being invoked in the expiration path ++ * of the in-service queue + */ + static void bfq_activate_requeue_entity(struct bfq_entity *entity, + bool non_blocking_wait_rq, +- bool requeue) ++ bool requeue, bool expiration) + { + struct bfq_sched_data *sd; + +@@ -1307,7 +1313,8 @@ static void bfq_activate_requeue_entity(struct bfq_entity *entity, + RB_EMPTY_ROOT(&(sd->service_tree+1)->active) && + RB_EMPTY_ROOT(&(sd->service_tree+2)->active)); + +- if (!bfq_update_next_in_service(sd, entity) && !requeue) { ++ if (!bfq_update_next_in_service(sd, entity, expiration) && ++ !requeue) { + BUG_ON(!sd->next_in_service); + break; + } +@@ -1373,6 +1380,8 @@ static bool __bfq_deactivate_entity(struct bfq_entity *entity, + * bfq_deactivate_entity - deactivate an entity representing a bfq_queue. + * @entity: the entity to deactivate. + * @ins_into_idle_tree: true if the entity can be put into the idle tree ++ * @expiration: true if this function is being invoked in the expiration path ++ * of the in-service queue + */ + static void bfq_deactivate_entity(struct bfq_entity *entity, + bool ins_into_idle_tree, +@@ -1417,7 +1426,7 @@ static void bfq_deactivate_entity(struct bfq_entity *entity, + * then, since entity has just been + * deactivated, a new one must be found. + */ +- bfq_update_next_in_service(sd, NULL); ++ bfq_update_next_in_service(sd, NULL, expiration); + + if (sd->next_in_service || sd->in_service_entity) { + /* +@@ -1495,7 +1504,7 @@ static void bfq_deactivate_entity(struct bfq_entity *entity, + "invoking udpdate_next for this entity"); + } + #endif +- if (!bfq_update_next_in_service(sd, entity) && ++ if (!bfq_update_next_in_service(sd, entity, expiration) && + !expiration) + /* + * next_in_service unchanged or not causing +@@ -1524,7 +1533,7 @@ static u64 bfq_calc_vtime_jump(struct bfq_service_tree *st) + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, + "calc_vtime_jump: new value %llu", +- root_entity->min_start); ++ ((root_entity->min_start>>10)*1000)>>12); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = +@@ -1533,7 +1542,7 @@ static u64 bfq_calc_vtime_jump(struct bfq_service_tree *st) + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, + "calc_vtime_jump: new value %llu", +- root_entity->min_start); ++ ((root_entity->min_start>>10)*1000)>>12); + } + #endif + return root_entity->min_start; +@@ -1615,17 +1624,9 @@ static struct bfq_entity *bfq_first_active_entity(struct bfq_service_tree *st, + * 3) is idle. + */ + static struct bfq_entity * +-__bfq_lookup_next_entity(struct bfq_service_tree *st, bool in_service +-#if 0 +- , bool force +-#endif +- ) ++__bfq_lookup_next_entity(struct bfq_service_tree *st, bool in_service) + { +- struct bfq_entity *entity +-#if 0 +- , *new_next_in_service = NULL +-#endif +- ; ++ struct bfq_entity *entity; + u64 new_vtime; + struct bfq_queue *bfqq; + +@@ -1667,8 +1668,9 @@ __bfq_lookup_next_entity(struct bfq_service_tree *st, bool in_service + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "__lookup_next: start %llu vtime %llu st %p", ++ "__lookup_next: start %llu vtime %llu (%llu) st %p", + ((entity->start>>10)*1000)>>12, ++ ((st->vtime>>10)*1000)>>12, + ((new_vtime>>10)*1000)>>12, st); + } + #endif +@@ -1681,12 +1683,14 @@ __bfq_lookup_next_entity(struct bfq_service_tree *st, bool in_service + /** + * bfq_lookup_next_entity - return the first eligible entity in @sd. + * @sd: the sched_data. ++ * @expiration: true if we are on the expiration path of the in-service queue + * + * This function is invoked when there has been a change in the trees +- * for sd, and we need know what is the new next entity after this +- * change. ++ * for sd, and we need to know what is the new next entity to serve ++ * after this change. + */ +-static struct bfq_entity *bfq_lookup_next_entity(struct bfq_sched_data *sd) ++static struct bfq_entity *bfq_lookup_next_entity(struct bfq_sched_data *sd, ++ bool expiration) + { + struct bfq_service_tree *st = sd->service_tree; + struct bfq_service_tree *idle_class_st = st + (BFQ_IOPRIO_CLASSES - 1); +@@ -1716,8 +1720,24 @@ static struct bfq_entity *bfq_lookup_next_entity(struct bfq_sched_data *sd) + * class, unless the idle class needs to be served. + */ + for (; class_idx < BFQ_IOPRIO_CLASSES; class_idx++) { ++ /* ++ * If expiration is true, then bfq_lookup_next_entity ++ * is being invoked as a part of the expiration path ++ * of the in-service queue. In this case, even if ++ * sd->in_service_entity is not NULL, ++ * sd->in_service_entiy at this point is actually not ++ * in service any more, and, if needed, has already ++ * been properly queued or requeued into the right ++ * tree. The reason why sd->in_service_entity is still ++ * not NULL here, even if expiration is true, is that ++ * sd->in_service_entiy is reset as a last step in the ++ * expiration path. So, if expiration is true, tell ++ * __bfq_lookup_next_entity that there is no ++ * sd->in_service_entity. ++ */ + entity = __bfq_lookup_next_entity(st + class_idx, +- sd->in_service_entity); ++ sd->in_service_entity && ++ !expiration); + + if (entity) + break; +@@ -1891,7 +1911,7 @@ static struct bfq_queue *bfq_get_next_queue(struct bfq_data *bfqd) + for_each_entity(entity) { + struct bfq_sched_data *sd = entity->sched_data; + +- if(!bfq_update_next_in_service(sd, NULL)) ++ if (!bfq_update_next_in_service(sd, NULL, false)) + break; + } + +@@ -1951,16 +1971,17 @@ static void bfq_activate_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) + entity->on_st); + + bfq_activate_requeue_entity(entity, bfq_bfqq_non_blocking_wait_rq(bfqq), +- false); ++ false, false); + bfq_clear_bfqq_non_blocking_wait_rq(bfqq); + } + +-static void bfq_requeue_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) ++static void bfq_requeue_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, ++ bool expiration) + { + struct bfq_entity *entity = &bfqq->entity; + + bfq_activate_requeue_entity(entity, false, +- bfqq == bfqd->in_service_queue); ++ bfqq == bfqd->in_service_queue, expiration); + } + + static void bfqg_stats_update_dequeue(struct bfq_group *bfqg); +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index e1960bf149d8..42393ab889a9 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -644,7 +644,7 @@ static void bfq_updated_next_req(struct bfq_data *bfqd, + entity->budget = new_budget; + bfq_log_bfqq(bfqd, bfqq, "updated next rq: new budget %lu", + new_budget); +- bfq_requeue_bfqq(bfqd, bfqq); ++ bfq_requeue_bfqq(bfqd, bfqq, false); + } + } + +@@ -2715,7 +2715,7 @@ static void __bfq_bfqq_expire(struct bfq_data *bfqd, struct bfq_queue *bfqq) + + bfq_del_bfqq_busy(bfqd, bfqq, true); + } else { +- bfq_requeue_bfqq(bfqd, bfqq); ++ bfq_requeue_bfqq(bfqd, bfqq, true); + /* + * Resort priority tree of potential close cooperators. + */ + +From ee9f95b24e1d88ffba4845981c2a4684aefd0245 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 9 Aug 2017 22:53:00 +0200 +Subject: [PATCH 38/51] bfq-sq-mq: remove direct switch to an entity in higher + class + +If the function bfq_update_next_in_service is invoked as a consequence +of the activation or requeueing of an entity, say E, and finds out +that E belongs to a higher-priority class than that of the current +next-in-service entity, then it sets next_in_service directly to +E. But this may lead to anomalous schedules, because E may happen not +be eligible for service, because its virtual start time is higher than +the system virtual time for its service tree. + +This commit addresses this issue by simply removing this direct +switch. + +Signed-off-by: Paolo Valente +--- + block/bfq-sched.c | 19 +++++-------------- + 1 file changed, 5 insertions(+), 14 deletions(-) + +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index f3001af37256..b1a59088db88 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -76,9 +76,8 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + * or repositiong of an entity that does not coincide with + * sd->next_in_service, then a full lookup in the active tree + * can be avoided. In fact, it is enough to check whether the +- * just-modified entity has a higher priority than +- * sd->next_in_service, or, even if it has the same priority +- * as sd->next_in_service, is eligible and has a lower virtual ++ * just-modified entity has the same priority as ++ * sd->next_in_service, is eligible and has a lower virtual + * finish time than sd->next_in_service. If this compound + * condition holds, then the new entity becomes the new + * next_in_service. Otherwise no change is needed. +@@ -94,9 +93,8 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + + /* + * If there is already a next_in_service candidate +- * entity, then compare class priorities or timestamps +- * to decide whether to replace sd->service_tree with +- * new_entity. ++ * entity, then compare timestamps to decide whether ++ * to replace sd->service_tree with new_entity. + */ + if (next_in_service) { + unsigned int new_entity_class_idx = +@@ -104,10 +102,6 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + struct bfq_service_tree *st = + sd->service_tree + new_entity_class_idx; + +- /* +- * For efficiency, evaluate the most likely +- * sub-condition first. +- */ + replace_next = + (new_entity_class_idx == + bfq_class_idx(next_in_service) +@@ -115,10 +109,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + !bfq_gt(new_entity->start, st->vtime) + && + bfq_gt(next_in_service->finish, +- new_entity->finish)) +- || +- new_entity_class_idx < +- bfq_class_idx(next_in_service); ++ new_entity->finish)); + } + + if (replace_next) + +From a3fdc5af40537355b68c1f0d3997c5a5fb54b9ce Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 10 Aug 2017 08:15:50 +0200 +Subject: [PATCH 39/51] bfq-sq-mq: guarantee update_next_in_service always + returns an eligible entity + +If the function bfq_update_next_in_service is invoked as a consequence +of the activation or requeueing of an entity, say E, then it doesn't +invoke bfq_lookup_next_entity to get the next-in-service entity. In +contrast, it follows a shorter path: if E happens to be eligible (see +commit "bfq-sq-mq: make lookup_next_entity push up vtime on +expirations" for details on eligibility) and to have a lower virtual +finish time than the current candidate as next-in-service entity, then +E directly becomes the next-in-service entity. Unfortunately, there is +a corner case for which this shorter path makes +bfq_update_next_in_service choose a non eligible entity: it occurs if +both E and the current next-in-service entity happen to be non +eligible when bfq_update_next_in_service is invoked. In this case, E +is not set as next-in-service, and, since bfq_lookup_next_entity is +not invoked, the state of the parent entity is not updated so as to +end up with an eligible entity as the proper next-in-service entity. + +In this respect, next-in-service is actually allowed to be non +eligible while some queue is in service: since no system-virtual-time +push-up can be performed in that case (see again commit "bfq-sq-mq: +make lookup_next_entity push up vtime on expirations" for details), +next-in-service is chosen, speculatively, as a function of the +possible value that the system virtual time may get after a push +up. But the correctness of the schedule breaks if next-in-service is +still a non eligible entity when it is time to set in service the next +entity. Unfortunately, this may happen in the above corner case. + +This commit fixes this problem by making bfq_update_next_in_service +invoke bfq_lookup_next_entity not only if the above shorter path +cannot be taken, but also if the shorter path is taken but fails to +yield an eligible next-in-service entity. + +Signed-off-by: Paolo Valente +--- + block/bfq-sched.c | 38 ++++++++++++++++++++++++++++---------- + 1 file changed, 28 insertions(+), 10 deletions(-) + +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index b1a59088db88..e4a2553a2d2c 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -70,6 +70,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + struct bfq_entity *next_in_service = sd->next_in_service; + struct bfq_queue *bfqq; + bool parent_sched_may_change = false; ++ bool change_without_lookup = false; + + /* + * If this update is triggered by the activation, requeueing +@@ -89,7 +90,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + * set to true, and left as true if + * sd->next_in_service is NULL. + */ +- bool replace_next = true; ++ change_without_lookup = true; + + /* + * If there is already a next_in_service candidate +@@ -102,7 +103,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + struct bfq_service_tree *st = + sd->service_tree + new_entity_class_idx; + +- replace_next = ++ change_without_lookup = + (new_entity_class_idx == + bfq_class_idx(next_in_service) + && +@@ -112,15 +113,32 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + new_entity->finish)); + } + +- if (replace_next) ++ if (change_without_lookup) { + next_in_service = new_entity; +- } else /* invoked because of a deactivation: lookup needed */ ++ bfqq = bfq_entity_to_bfqq(next_in_service); ++ ++ if (bfqq) ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "update_next_in_service: chose without lookup"); ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ else { ++ struct bfq_group *bfqg = ++ container_of(next_in_service, ++ struct bfq_group, entity); ++ ++ bfq_log_bfqg((struct bfq_data*)bfqg->bfqd, bfqg, ++ "update_next_in_service: chose without lookup"); ++ } ++#endif ++ } ++ } ++ ++ if (!change_without_lookup) /* lookup needed */ + next_in_service = bfq_lookup_next_entity(sd, expiration); + +- if (next_in_service) { ++ if (next_in_service) + parent_sched_may_change = !sd->next_in_service || + bfq_update_parent_budget(next_in_service); +- } + + sd->next_in_service = next_in_service; + +@@ -1053,7 +1071,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + + if (bfqq) { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "__activate_entity: new queue finish %llu", ++ "update_fin_time_enqueue: new queue finish %llu", + ((entity->finish>>10)*1000)>>12); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { +@@ -1061,7 +1079,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "__activate_entity: new group finish %llu", ++ "update_fin_time_enqueue: new group finish %llu", + ((entity->finish>>10)*1000)>>12); + #endif + } +@@ -1071,7 +1089,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + + if (bfqq) { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "__activate_entity: queue %seligible in st %p", ++ "update_fin_time_enqueue: queue %seligible in st %p", + entity->start <= st->vtime ? "" : "non ", st); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { +@@ -1079,7 +1097,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "__activate_entity: group %seligible in st %p", ++ "update_fin_time_enqueue: group %seligible in st %p", + entity->start <= st->vtime ? "" : "non ", st); + #endif + } + +From 6565e4d1aac029b6f0a5d86a4c6ef38608838eac Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 31 Aug 2017 19:24:26 +0200 +Subject: [PATCH 40/51] doc, block, bfq: fix some typos and stale sentences + +Signed-off-by: Paolo Valente +Reviewed-by: Jeremy Hickman +Reviewed-by: Laurentiu Nicola +--- + Documentation/block/bfq-iosched.txt | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/Documentation/block/bfq-iosched.txt b/Documentation/block/bfq-iosched.txt +index 0e59f1c9d30e..dcfe15523da3 100644 +--- a/Documentation/block/bfq-iosched.txt ++++ b/Documentation/block/bfq-iosched.txt +@@ -17,7 +17,7 @@ instances of BFQ are available (otherwise only the first instance): + - bfq-mq: development version of BFQ for blk-mq; this version contains + also all latest features and fixes not yet landed in mainline, plus many + safety checks +-- bfq: BFQ for legacy blk; also this version contains latest features ++- bfq-sq: BFQ for legacy blk; also this version contains latest features + and fixes, as well as safety checks + + In its default configuration, BFQ privileges latency over + +From 261ee8cc9f43e03d790a07184f0bcaa504ee6737 Mon Sep 17 00:00:00 2001 +From: Luca Miccio +Date: Wed, 13 Sep 2017 12:03:56 +0200 +Subject: [PATCH 41/51] bfq-mq, bfq-sq: Disable writeback throttling + +Similarly to CFQ, BFQ has its write-throttling heuristics, and it +is better not to combine them with further write-throttling +heuristics of a different nature. +So this commit disables write-back throttling for a device if BFQ +is used as I/O scheduler for that device. + +Signed-off-by: Luca Miccio +Signed-off-by: Paolo Valente +Tested-by: Oleksandr Natalenko +--- + block/bfq-mq-iosched.c | 2 ++ + block/bfq-sq-iosched.c | 7 +++++++ + 2 files changed, 9 insertions(+) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index b5c848650375..7d27d5b3befb 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -89,6 +89,7 @@ + #include "blk-mq-tag.h" + #include "blk-mq-sched.h" + #include "bfq-mq.h" ++#include "blk-wbt.h" + + /* Expiration time of sync (0) and async (1) requests, in ns. */ + static const u64 bfq_fifo_expire[2] = { NSEC_PER_SEC / 4, NSEC_PER_SEC / 8 }; +@@ -5260,6 +5261,7 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) + bfq_init_root_group(bfqd->root_group, bfqd); + bfq_init_entity(&bfqd->oom_bfqq.entity, bfqd->root_group); + ++ wbt_disable_default(q); + return 0; + + out_free: +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 42393ab889a9..6fdc3b1d5bb8 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -83,6 +83,7 @@ + #include + #include "blk.h" + #include "bfq.h" ++#include "blk-wbt.h" + + /* Expiration time of sync (0) and async (1) requests, in ns. */ + static const u64 bfq_fifo_expire[2] = { NSEC_PER_SEC / 4, NSEC_PER_SEC / 8 }; +@@ -4976,6 +4977,11 @@ static int bfq_init_queue(struct request_queue *q, struct elevator_type *e) + return -ENOMEM; + } + ++static void bfq_registered_queue(struct request_queue *q) ++{ ++ wbt_disable_default(q); ++} ++ + static void bfq_slab_kill(void) + { + kmem_cache_destroy(bfq_pool); +@@ -5285,6 +5291,7 @@ static struct elevator_type iosched_bfq = { + .elevator_may_queue_fn = bfq_may_queue, + .elevator_init_fn = bfq_init_queue, + .elevator_exit_fn = bfq_exit_queue, ++ .elevator_registered_fn = bfq_registered_queue, + }, + .icq_size = sizeof(struct bfq_io_cq), + .icq_align = __alignof__(struct bfq_io_cq), + +From 40ea0aed088791da27fcfa51f3b64d1f96b0d06e Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Tue, 12 Sep 2017 16:45:53 +0200 +Subject: [PATCH 42/51] bfq-mq, bfq-sq: fix wrong init of saved start time for + weight raising + +This commit fixes a bug that causes bfq to fail to guarantee a high +responsiveness on some drives, if there is heavy random read+write I/O +in the background. More precisely, such a failure allowed this bug to +be found [1], but the bug may well cause other yet unreported +anomalies. + +BFQ raises the weight of the bfq_queues associated with soft real-time +applications, to privilege the I/O, and thus reduce latency, for these +applications. This mechanism is named soft-real-time weight raising in +BFQ. A soft real-time period may happen to be nested into an +interactive weight raising period, i.e., it may happen that, when a +bfq_queue switches to a soft real-time weight-raised state, the +bfq_queue is already being weight-raised because deemed interactive +too. In this case, BFQ saves in a special variable +wr_start_at_switch_to_srt, the time instant when the interactive +weight-raising period started for the bfq_queue, i.e., the time +instant when BFQ started to deem the bfq_queue interactive. This value +is then used to check whether the interactive weight-raising period +would still be in progress when the soft real-time weight-raising +period ends. If so, interactive weight raising is restored for the +bfq_queue. This restore is useful, in particular, because it prevents +bfq_queues from losing their interactive weight raising prematurely, +as a consequence of spurious, short-lived soft real-time +weight-raising periods caused by wrong detections as soft real-time. + +If, instead, a bfq_queue switches to soft-real-time weight raising +while it *is not* already in an interactive weight-raising period, +then the variable wr_start_at_switch_to_srt has no meaning during the +following soft real-time weight-raising period. Unfortunately the +handling of this case is wrong in BFQ: not only the variable is not +flagged somehow as meaningless, but it is also set to the time when +the switch to soft real-time weight-raising occurs. This may cause an +interactive weight-raising period to be considered mistakenly as still +in progress, and thus a spurious interactive weight-raising period to +start for the bfq_queue, at the end of the soft-real-time +weight-raising period. In particular the spurious interactive +weight-raising period will be considered as still in progress, if the +soft-real-time weight-raising period does not last very long. The +bfq_queue will then be wrongly privileged and, if I/O bound, will +unjustly steal bandwidth to truly interactive or soft real-time +bfq_queues, harming responsiveness and low latency. + +This commit fixes this issue by just setting wr_start_at_switch_to_srt +to minus infinity (farthest past time instant according to jiffies +macros): when the soft-real-time weight-raising period ends, certainly +no interactive weight-raising period will be considered as still in +progress. + +[1] Background I/O Type: Random - Background I/O mix: Reads and writes +- Application to start: LibreOffice Writer in +http://www.phoronix.com/scan.php?page=news_item&px=Linux-4.13-IO-Laptop + +Signed-off-by: Paolo Valente +Signed-off-by: Angelo Ruocco +Tested-by: Oleksandr Natalenko +Tested-by: Lee Tibbert +Tested-by: Mirko Montanari +--- + block/bfq-mq-iosched.c | 50 +++++++++++++++++++++++++++++++------------------- + block/bfq-sq-iosched.c | 50 +++++++++++++++++++++++++++++++------------------- + 2 files changed, 62 insertions(+), 38 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 7d27d5b3befb..f378519b6d33 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -1204,6 +1204,24 @@ static bool bfq_bfqq_update_budg_for_activation(struct bfq_data *bfqd, + return wr_or_deserves_wr; + } + ++/* ++ * Return the farthest future time instant according to jiffies ++ * macros. ++ */ ++static unsigned long bfq_greatest_from_now(void) ++{ ++ return jiffies + MAX_JIFFY_OFFSET; ++} ++ ++/* ++ * Return the farthest past time instant according to jiffies ++ * macros. ++ */ ++static unsigned long bfq_smallest_from_now(void) ++{ ++ return jiffies - MAX_JIFFY_OFFSET; ++} ++ + static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd, + struct bfq_queue *bfqq, + unsigned int old_wr_coeff, +@@ -1218,7 +1236,19 @@ static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd, + bfqq->wr_coeff = bfqd->bfq_wr_coeff; + bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); + } else { +- bfqq->wr_start_at_switch_to_srt = jiffies; ++ /* ++ * No interactive weight raising in progress ++ * here: assign minus infinity to ++ * wr_start_at_switch_to_srt, to make sure ++ * that, at the end of the soft-real-time ++ * weight raising periods that is starting ++ * now, no interactive weight-raising period ++ * may be wrongly considered as still in ++ * progress (and thus actually started by ++ * mistake). ++ */ ++ bfqq->wr_start_at_switch_to_srt = ++ bfq_smallest_from_now(); + bfqq->wr_coeff = bfqd->bfq_wr_coeff * + BFQ_SOFTRT_WEIGHT_FACTOR; + bfqq->wr_cur_max_time = +@@ -3174,24 +3204,6 @@ static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, + jiffies + nsecs_to_jiffies(bfqq->bfqd->bfq_slice_idle) + 4); + } + +-/* +- * Return the farthest future time instant according to jiffies +- * macros. +- */ +-static unsigned long bfq_greatest_from_now(void) +-{ +- return jiffies + MAX_JIFFY_OFFSET; +-} +- +-/* +- * Return the farthest past time instant according to jiffies +- * macros. +- */ +-static unsigned long bfq_smallest_from_now(void) +-{ +- return jiffies - MAX_JIFFY_OFFSET; +-} +- + /** + * bfq_bfqq_expire - expire a queue. + * @bfqd: device owning the queue. +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 6fdc3b1d5bb8..f4654436cd55 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -1165,6 +1165,24 @@ static bool bfq_bfqq_update_budg_for_activation(struct bfq_data *bfqd, + return wr_or_deserves_wr; + } + ++/* ++ * Return the farthest future time instant according to jiffies ++ * macros. ++ */ ++static unsigned long bfq_greatest_from_now(void) ++{ ++ return jiffies + MAX_JIFFY_OFFSET; ++} ++ ++/* ++ * Return the farthest past time instant according to jiffies ++ * macros. ++ */ ++static unsigned long bfq_smallest_from_now(void) ++{ ++ return jiffies - MAX_JIFFY_OFFSET; ++} ++ + static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd, + struct bfq_queue *bfqq, + unsigned int old_wr_coeff, +@@ -1179,7 +1197,19 @@ static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd, + bfqq->wr_coeff = bfqd->bfq_wr_coeff; + bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); + } else { +- bfqq->wr_start_at_switch_to_srt = jiffies; ++ /* ++ * No interactive weight raising in progress ++ * here: assign minus infinity to ++ * wr_start_at_switch_to_srt, to make sure ++ * that, at the end of the soft-real-time ++ * weight raising periods that is starting ++ * now, no interactive weight-raising period ++ * may be wrongly considered as still in ++ * progress (and thus actually started by ++ * mistake). ++ */ ++ bfqq->wr_start_at_switch_to_srt = ++ bfq_smallest_from_now(); + bfqq->wr_coeff = bfqd->bfq_wr_coeff * + BFQ_SOFTRT_WEIGHT_FACTOR; + bfqq->wr_cur_max_time = +@@ -3067,24 +3097,6 @@ static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, + jiffies + nsecs_to_jiffies(bfqq->bfqd->bfq_slice_idle) + 4); + } + +-/* +- * Return the farthest future time instant according to jiffies +- * macros. +- */ +-static unsigned long bfq_greatest_from_now(void) +-{ +- return jiffies + MAX_JIFFY_OFFSET; +-} +- +-/* +- * Return the farthest past time instant according to jiffies +- * macros. +- */ +-static unsigned long bfq_smallest_from_now(void) +-{ +- return jiffies - MAX_JIFFY_OFFSET; +-} +- + /** + * bfq_bfqq_expire - expire a queue. + * @bfqd: device owning the queue. + +From 9dbea44b6f721baeff35b9fdf628ec55fe00e09d Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 14 Sep 2017 05:12:58 -0400 +Subject: [PATCH 43/51] Fix commit "Unnest request-queue and ioc locks from + scheduler locks" + +The commit "Unnest request-queue and ioc locks from scheduler locks" +mistakenly removed the setting of the split flag in function +bfq_prepare_request. This commit puts this missing instruction back in +its place. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 12 ++++++++++++ + 1 file changed, 12 insertions(+) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index f378519b6d33..288078e68a2a 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -744,6 +744,12 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + bfqq->wr_cur_max_time = bic->saved_wr_cur_max_time; + BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); + ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "[%s] bic %p wr_coeff %d start_finish %lu max_time %lu", ++ __func__, ++ bic, bfqq->wr_coeff, bfqq->last_wr_start_finish, ++ bfqq->wr_cur_max_time); ++ + if (bfqq->wr_coeff > 1 && (bfq_bfqq_in_large_burst(bfqq) || + time_is_before_jiffies(bfqq->last_wr_start_finish + + bfqq->wr_cur_max_time))) { +@@ -2208,6 +2214,11 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq) + bic->saved_last_wr_start_finish = bfqq->last_wr_start_finish; + bic->saved_wr_cur_max_time = bfqq->wr_cur_max_time; + BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "[%s] bic %p wr_coeff %d start_finish %lu max_time %lu", ++ __func__, ++ bic, bfqq->wr_coeff, bfqq->last_wr_start_finish, ++ bfqq->wr_cur_max_time); + } + + static void +@@ -4950,6 +4961,7 @@ static void bfq_prepare_request(struct request *rq, struct bio *bio) + bic->saved_in_large_burst = true; + + bfqq = bfq_split_bfqq(bic, bfqq); ++ split = true; + + if (!bfqq) + bfqq = bfq_get_bfqq_handle_split(bfqd, bic, bio, + +From d4ebb2a66a23dc183792088c521f2be2193b56db Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 15 Sep 2017 01:53:51 -0400 +Subject: [PATCH 44/51] bfq-sq, bfq-mq: check and switch back to interactive wr + also on queue split + +As already explained in the message of commit "bfq-mq, bfq-sq: fix +wrong init of saved start time for weight raising", if a soft +real-time weight-raising period happens to be nested in a larger +interactive weight-raising period, then BFQ restores the interactive +weight raising at the end of the soft real-time weight raising. In +particular, BFQ checks whether the latter has ended only on request +dispatches. + +Unfortunately, the above scheme fails to restore interactive weight +raising in the following corner case: if a bfq_queue, say Q, +1) Is merged with another bfq_queue while it is in a nested soft +real-time weight-raising period. The weight-raising state of Q is +then saved, and not considered any longer until a split occurs. +2) Is split from the other bfq_queue(s) at a time instant when its +soft real-time weight raising is already finished. +On the split, while resuming the previous, soft real-time +weight-raised state of the bfq_queue Q, BFQ checks whether the +current soft real-time weight-raising period is actually over. If so, +BFQ switches weight raising off for Q, *without* checking whether the +soft real-time period was actually nested in a non-yet-finished +interactive weight-raising period. + +This commit addresses this issue by adding the above missing check in +bfq_queue splits, and restoring interactive weight raising if needed. + +Signed-off-by: Paolo Valente +Tested-by: Angelo Ruocco +Tested-by: Mirko Montanari +--- + block/bfq-mq-iosched.c | 29 +++++++++++++++++++++-------- + block/bfq-sq-iosched.c | 35 +++++++++++++++++++++++++++-------- + 2 files changed, 48 insertions(+), 16 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 288078e68a2a..6130a95c6497 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -716,6 +716,15 @@ static unsigned int bfq_wr_duration(struct bfq_data *bfqd) + return dur; + } + ++/* switch back from soft real-time to interactive weight raising */ ++static void switch_back_to_interactive_wr(struct bfq_queue *bfqq, ++ struct bfq_data *bfqd) ++{ ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ bfqq->last_wr_start_finish = bfqq->wr_start_at_switch_to_srt; ++} ++ + static void + bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + struct bfq_io_cq *bic, bool bfq_already_existing) +@@ -753,12 +762,20 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + if (bfqq->wr_coeff > 1 && (bfq_bfqq_in_large_burst(bfqq) || + time_is_before_jiffies(bfqq->last_wr_start_finish + + bfqq->wr_cur_max_time))) { +- bfq_log_bfqq(bfqq->bfqd, bfqq, ++ if (bfqq->wr_cur_max_time == bfqd->bfq_wr_rt_max_time && ++ !bfq_bfqq_in_large_burst(bfqq) && ++ time_is_after_eq_jiffies(bfqq->wr_start_at_switch_to_srt + ++ bfq_wr_duration(bfqd))) { ++ switch_back_to_interactive_wr(bfqq, bfqd); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "resume state: switching back to interactive"); ++ } else { ++ bfqq->wr_coeff = 1; ++ bfq_log_bfqq(bfqq->bfqd, bfqq, + "resume state: switching off wr (%lu + %lu < %lu)", + bfqq->last_wr_start_finish, bfqq->wr_cur_max_time, + jiffies); +- +- bfqq->wr_coeff = 1; ++ } + } + + /* make sure weight will be updated, however we got here */ +@@ -3820,11 +3837,7 @@ static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfq_wr_duration(bfqd))) + bfq_bfqq_end_wr(bfqq); + else { +- /* switch back to interactive wr */ +- bfqq->wr_coeff = bfqd->bfq_wr_coeff; +- bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); +- bfqq->last_wr_start_finish = +- bfqq->wr_start_at_switch_to_srt; ++ switch_back_to_interactive_wr(bfqq, bfqd); + BUG_ON(time_is_after_jiffies( + bfqq->last_wr_start_finish)); + bfqq->entity.prio_changed = 1; +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index f4654436cd55..e07d5d1c0d40 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -678,6 +678,15 @@ static unsigned int bfq_wr_duration(struct bfq_data *bfqd) + return dur; + } + ++/* switch back from soft real-time to interactive weight raising */ ++static void switch_back_to_interactive_wr(struct bfq_queue *bfqq, ++ struct bfq_data *bfqd) ++{ ++ bfqq->wr_coeff = bfqd->bfq_wr_coeff; ++ bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); ++ bfqq->last_wr_start_finish = bfqq->wr_start_at_switch_to_srt; ++} ++ + static void + bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + struct bfq_io_cq *bic, bool bfq_already_existing) +@@ -705,15 +714,29 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + bfqq->wr_cur_max_time = bic->saved_wr_cur_max_time; + BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); + ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "[%s] bic %p wr_coeff %d start_finish %lu max_time %lu", ++ __func__, ++ bic, bfqq->wr_coeff, bfqq->last_wr_start_finish, ++ bfqq->wr_cur_max_time); ++ + if (bfqq->wr_coeff > 1 && (bfq_bfqq_in_large_burst(bfqq) || + time_is_before_jiffies(bfqq->last_wr_start_finish + + bfqq->wr_cur_max_time))) { +- bfq_log_bfqq(bfqq->bfqd, bfqq, ++ if (bfqq->wr_cur_max_time == bfqd->bfq_wr_rt_max_time && ++ !bfq_bfqq_in_large_burst(bfqq) && ++ time_is_after_eq_jiffies(bfqq->wr_start_at_switch_to_srt + ++ bfq_wr_duration(bfqd))) { ++ switch_back_to_interactive_wr(bfqq, bfqd); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "resume state: switching back to interactive"); ++ } else { ++ bfqq->wr_coeff = 1; ++ bfq_log_bfqq(bfqq->bfqd, bfqq, + "resume state: switching off wr (%lu + %lu < %lu)", + bfqq->last_wr_start_finish, bfqq->wr_cur_max_time, + jiffies); +- +- bfqq->wr_coeff = 1; ++ } + } + + /* make sure weight will be updated, however we got here */ +@@ -3703,11 +3726,7 @@ static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfq_wr_duration(bfqd))) + bfq_bfqq_end_wr(bfqq); + else { +- /* switch back to interactive wr */ +- bfqq->wr_coeff = bfqd->bfq_wr_coeff; +- bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); +- bfqq->last_wr_start_finish = +- bfqq->wr_start_at_switch_to_srt; ++ switch_back_to_interactive_wr(bfqq, bfqd); + BUG_ON(time_is_after_jiffies( + bfqq->last_wr_start_finish)); + bfqq->entity.prio_changed = 1; + +From 9eaec0c3a2d675763b09da81c9117a9c43bce942 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 15 Sep 2017 04:58:33 -0400 +Subject: [PATCH 45/51] bfq-sq, bfq-mq: let early-merged queues be + weight-raised on split too + +A just-created bfq_queue, say Q, may happen to be merged with another +bfq_queue on the very first invocation of the function +__bfq_insert_request. In such a case, even if Q would clearly deserve +interactive weight raising (as it has just been created), the function +bfq_add_request does not make it to be invoked for Q, and thus to +activate weight raising for Q. As a consequence, when the state of Q +is saved for a possible future restore, after a split of Q from the +other bfq_queue(s), such a state happens to be (unjustly) +non-weight-raised. Then the bfq_queue will not enjoy any weight +raising on the split, even if should still be in an interactive +weight-raising period when the split occurs. + +This commit solves this problem as follows, for a just-created +bfq_queue that is being early-merged: it stores directly, in the saved +state of the bfq_queue, the weight-raising state that would have been +assigned to the bfq_queue if not early-merged. + +Signed-off-by: Paolo Valente +Tested-by: Angelo Ruocco +Tested-by: Mirko Montanari +--- + block/bfq-mq-iosched.c | 28 +++++++++++++++++++++++----- + block/bfq-sq-iosched.c | 28 +++++++++++++++++++++++----- + 2 files changed, 46 insertions(+), 10 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 6130a95c6497..af84e506e897 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -2226,10 +2226,27 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq) + bic->saved_IO_bound = bfq_bfqq_IO_bound(bfqq); + bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq); + bic->was_in_burst_list = !hlist_unhashed(&bfqq->burst_list_node); +- bic->saved_wr_coeff = bfqq->wr_coeff; +- bic->saved_wr_start_at_switch_to_srt = bfqq->wr_start_at_switch_to_srt; +- bic->saved_last_wr_start_finish = bfqq->last_wr_start_finish; +- bic->saved_wr_cur_max_time = bfqq->wr_cur_max_time; ++ if (unlikely(bfq_bfqq_just_created(bfqq) && ++ !bfq_bfqq_in_large_burst(bfqq))) { ++ /* ++ * bfqq being merged ritgh after being created: bfqq ++ * would have deserved interactive weight raising, but ++ * did not make it to be set in a weight-raised state, ++ * because of this early merge. Store directly the ++ * weight-raising state that would have been assigned ++ * to bfqq, so that to avoid that bfqq unjustly fails ++ * to enjoy weight raising if split soon. ++ */ ++ bic->saved_wr_coeff = bfqq->bfqd->bfq_wr_coeff; ++ bic->saved_wr_cur_max_time = bfq_wr_duration(bfqq->bfqd); ++ bic->saved_last_wr_start_finish = jiffies; ++ } else { ++ bic->saved_wr_coeff = bfqq->wr_coeff; ++ bic->saved_wr_start_at_switch_to_srt = ++ bfqq->wr_start_at_switch_to_srt; ++ bic->saved_last_wr_start_finish = bfqq->last_wr_start_finish; ++ bic->saved_wr_cur_max_time = bfqq->wr_cur_max_time; ++ } + BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); + bfq_log_bfqq(bfqq->bfqd, bfqq, + "[%s] bic %p wr_coeff %d start_finish %lu max_time %lu", +@@ -4560,7 +4577,6 @@ static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + bfqq->allocated); + + new_bfqq->ref++; +- bfq_clear_bfqq_just_created(bfqq); + /* + * If the bic associated with the process + * issuing this request still points to bfqq +@@ -4572,6 +4588,8 @@ static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + if (bic_to_bfqq(RQ_BIC(rq), 1) == bfqq) + bfq_merge_bfqqs(bfqd, RQ_BIC(rq), + bfqq, new_bfqq); ++ ++ bfq_clear_bfqq_just_created(bfqq); + /* + * rq is about to be enqueued into new_bfqq, + * release rq reference on bfqq +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index e07d5d1c0d40..0c48f527fe3f 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -2105,10 +2105,27 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq) + bic->saved_IO_bound = bfq_bfqq_IO_bound(bfqq); + bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq); + bic->was_in_burst_list = !hlist_unhashed(&bfqq->burst_list_node); +- bic->saved_wr_coeff = bfqq->wr_coeff; +- bic->saved_wr_start_at_switch_to_srt = bfqq->wr_start_at_switch_to_srt; +- bic->saved_last_wr_start_finish = bfqq->last_wr_start_finish; +- bic->saved_wr_cur_max_time = bfqq->wr_cur_max_time; ++ if (unlikely(bfq_bfqq_just_created(bfqq) && ++ !bfq_bfqq_in_large_burst(bfqq))) { ++ /* ++ * bfqq being merged ritgh after being created: bfqq ++ * would have deserved interactive weight raising, but ++ * did not make it to be set in a weight-raised state, ++ * because of this early merge. Store directly the ++ * weight-raising state that would have been assigned ++ * to bfqq, so that to avoid that bfqq unjustly fails ++ * to enjoy weight raising if split soon. ++ */ ++ bic->saved_wr_coeff = bfqq->bfqd->bfq_wr_coeff; ++ bic->saved_wr_cur_max_time = bfq_wr_duration(bfqq->bfqd); ++ bic->saved_last_wr_start_finish = jiffies; ++ } else { ++ bic->saved_wr_coeff = bfqq->wr_coeff; ++ bic->saved_wr_start_at_switch_to_srt = ++ bfqq->wr_start_at_switch_to_srt; ++ bic->saved_last_wr_start_finish = bfqq->last_wr_start_finish; ++ bic->saved_wr_cur_max_time = bfqq->wr_cur_max_time; ++ } + BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); + } + +@@ -4383,10 +4400,11 @@ static void bfq_insert_request(struct request_queue *q, struct request *rq) + new_bfqq->allocated[rq_data_dir(rq)]++; + bfqq->allocated[rq_data_dir(rq)]--; + new_bfqq->ref++; +- bfq_clear_bfqq_just_created(bfqq); + if (bic_to_bfqq(RQ_BIC(rq), 1) == bfqq) + bfq_merge_bfqqs(bfqd, RQ_BIC(rq), + bfqq, new_bfqq); ++ ++ bfq_clear_bfqq_just_created(bfqq); + /* + * rq is about to be enqueued into new_bfqq, + * release rq reference on bfqq + +From cb05150675095cb97ab22e4955eb82e4fe2e9dbe Mon Sep 17 00:00:00 2001 +From: omcira +Date: Mon, 18 Sep 2017 10:49:48 +0200 +Subject: [PATCH 46/51] bfq-sq, bfq-mq: decrease burst size when queues in + burst exit + +If many queues belonging to the same group happen to be created +shortly after each other, then the concurrent processes associated +with these queues have typically a common goal, and they get it done +as soon as possible if not hampered by device idling. Examples are +processes spawned by git grep, or by systemd during boot. As for +device idling, this mechanism is currently necessary for weight +raising to succeed in its goal: privileging I/O. In view of these +facts, BFQ does not provide the above queues with either weight +raising or device idling. + +On the other hand, a burst of queue creations may be caused also by +the start-up of a complex application. In this case, these queues need +usually to be served one after the other, and as quickly as possible, +to maximise responsiveness. Therefore, in this case the best strategy +is to weight-raise all the queues created during the burst, i.e., the +exact opposite of the strategy for the above case. + +To distinguish between the two cases, BFQ uses an empirical burst-size +threshold, found through extensive tests and monitoring of daily +usage. Only large bursts, i.e., burst with a size above this +threshold, are considered as generated by a high number of parallel +processes. In this respect, upstart-based boot proved to be rather +hard to detect as generating a large burst of queue creations, because +with upstart most of the queues created in a burst exit *before* the +next queues in the same burst are created. To address this issue, I +changed the burst-detection mechanism so as to not decrease the size +of the current burst even if one of the queues in the burst is +eliminated. + +Unfortunately, this missing decrease causes false positives on very +fast systems: on the start-up of a complex application, such as +libreoffice writer, so many queues are created, served and exited +shortly after each other, that a large burst of queue creations is +wrongly detected as occurring. These false positives just disappear if +the size of a burst is decreased when one of the queues in the burst +exits. This commit restores the missing burst-size decrease, relying +of the fact that upstart is apparently unlikely to be used on systems +running this and future versions of the kernel. + +Signed-off-by: Paolo Valente +Signed-off-by: Mauro Andreolini +Signed-off-by: Angelo Ruocco +Tested-by: Mirko Montanari +--- + block/bfq-mq-iosched.c | 12 +++--------- + block/bfq-sq-iosched.c | 12 +++--------- + 2 files changed, 6 insertions(+), 18 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index af84e506e897..6e413d7236ce 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4111,16 +4111,10 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + BUG_ON(bfqq->entity.tree); + BUG_ON(bfq_bfqq_busy(bfqq)); + +- if (bfq_bfqq_sync(bfqq)) +- /* +- * The fact that this queue is being destroyed does not +- * invalidate the fact that this queue may have been +- * activated during the current burst. As a consequence, +- * although the queue does not exist anymore, and hence +- * needs to be removed from the burst list if there, +- * the burst size has not to be decremented. +- */ ++ if (bfq_bfqq_sync(bfqq) && !hlist_unhashed(&bfqq->burst_list_node)) { + hlist_del_init(&bfqq->burst_list_node); ++ bfqq->bfqd->burst_size--; ++ } + + if (bfqq->bfqd) + bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 0c48f527fe3f..93034dd7b801 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -3945,16 +3945,10 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + BUG_ON(bfqq->entity.tree); + BUG_ON(bfq_bfqq_busy(bfqq)); + +- if (bfq_bfqq_sync(bfqq)) +- /* +- * The fact that this queue is being destroyed does not +- * invalidate the fact that this queue may have been +- * activated during the current burst. As a consequence, +- * although the queue does not exist anymore, and hence +- * needs to be removed from the burst list if there, +- * the burst size has not to be decremented. +- */ ++ if (bfq_bfqq_sync(bfqq) && !hlist_unhashed(&bfqq->burst_list_node)) { + hlist_del_init(&bfqq->burst_list_node); ++ bfqq->bfqd->burst_size--; ++ } + + bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); + + +From 60de7307d5e3ed7f272f12c900f631bdfe114db2 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 6 Oct 2017 19:35:38 +0200 +Subject: [PATCH 47/51] bfq-sq, bfq-mq: fix unbalanced decrements of burst size +MIME-Version: 1.0 +Content-Type: text/plain; charset=UTF-8 +Content-Transfer-Encoding: 8bit + +The commit "bfq-sq, bfq-mq: decrease burst size when queues in burst +exit" introduced the decrement of burst_size on the removal of a +bfq_queue from the burst list. Unfortunately, this decrement can +happen to be performed even when burst size is already equal to 0, +because of unbalanced decrements. A description follows of the cause +of these unbalanced decrements, namely a wrong assumption, and of the +way how this wrong assumption leads to unbalanced decrements. + +The wrong assumption is that a bfq_queue can exit only if the process +associated with the bfq_queue has exited. This is false, because a +bfq_queue, say Q, may exit also as a consequence of a merge with +another bfq_queue. In this case, Q exits because the I/O of its +associated process has been redirected to another bfq_queue. + +The decrement unbalance occurs because Q may then be re-created after +a split, and added back to the current burst list, *without* +incrementing burst_size. burst_size is not incremented because Q is +not a new bfq_queue added to the burst list, but a bfq_queue only +temporarily removed from the list, and, before the commit "bfq-sq, +bfq-mq: decrease burst size when queues in burst exit", burst_size was +not decremented when Q was removed. + +This commit addresses this issue by just checking whether the exiting +bfq_queue is a merged bfq_queue, and, in that case, not decrementing +burst_size. Unfortunately, this still leaves room for unbalanced +decrements, in the following rarer case: on a split, the bfq_queue +happens to be inserted into a different burst list than that it was +removed from when merged. If this happens, the number of elements in +the new burst list becomes higher than burst_size (by one). When the +bfq_queue then exits, it is of course not in a merged state any +longer, thus burst_size is decremented, which results in an unbalanced +decrement. To handle this sporadic, unlucky case in a simple way, +this commit also checks that burst_size is larger than 0 before +decrementing it. + +Finally, this commit removes an useless, extra check: the check that +the bfq_queue is sync, performed before checking whether the bfq_queue +is in the burst list. This extra check is redundant, because only sync +bfq_queues can be inserted into the burst list. + +Reported-by: Philip Müller +Signed-off-by: Paolo Valente +Signed-off-by: Angelo Ruocco +Tested-by: Philip Müller +Tested-by: Oleksandr Natalenko +Tested-by: Lee Tibbert +--- + block/bfq-mq-iosched.c | 59 ++++++++++++++++++++++++++++++++++++++++++++++++-- + block/bfq-sq-iosched.c | 59 ++++++++++++++++++++++++++++++++++++++++++++++++-- + 2 files changed, 114 insertions(+), 4 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 6e413d7236ce..816bac6cdd3d 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4111,9 +4111,36 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + BUG_ON(bfqq->entity.tree); + BUG_ON(bfq_bfqq_busy(bfqq)); + +- if (bfq_bfqq_sync(bfqq) && !hlist_unhashed(&bfqq->burst_list_node)) { ++ if (!hlist_unhashed(&bfqq->burst_list_node)) { + hlist_del_init(&bfqq->burst_list_node); +- bfqq->bfqd->burst_size--; ++ /* ++ * Decrement also burst size after the removal, if the ++ * process associated with bfqq is exiting, and thus ++ * does not contribute to the burst any longer. This ++ * decrement helps filter out false positives of large ++ * bursts, when some short-lived process (often due to ++ * the execution of commands by some service) happens ++ * to start and exit while a complex application is ++ * starting, and thus spawning several processes that ++ * do I/O (and that *must not* be treated as a large ++ * burst, see comments on bfq_handle_burst). ++ * ++ * In particular, the decrement is performed only if: ++ * 1) bfqq is not a merged queue, because, if it is, ++ * then this free of bfqq is not triggered by the exit ++ * of the process bfqq is associated with, but exactly ++ * by the fact that bfqq has just been merged. ++ * 2) burst_size is greater than 0, to handle ++ * unbalanced decrements. Unbalanced decrements may ++ * happen in te following case: bfqq is inserted into ++ * the current burst list--without incrementing ++ * bust_size--because of a split, but the current ++ * burst list is not the burst list bfqq belonged to ++ * (see comments on the case of a split in ++ * bfq_set_request). ++ */ ++ if (bfqq->bic && bfqq->bfqd->burst_size > 0) ++ bfqq->bfqd->burst_size--; + } + + if (bfqq->bfqd) +@@ -4940,6 +4967,34 @@ static struct bfq_queue *bfq_get_bfqq_handle_split(struct bfq_data *bfqd, + "large burst"); + bfq_clear_bfqq_in_large_burst(bfqq); + if (bic->was_in_burst_list) ++ /* ++ * If bfqq was in the current ++ * burst list before being ++ * merged, then we have to add ++ * it back. And we do not need ++ * to increase burst_size, as ++ * we did not decrement ++ * burst_size when we removed ++ * bfqq from the burst list as ++ * a consequence of a merge ++ * (see comments in ++ * bfq_put_queue). In this ++ * respect, it would be rather ++ * costly to know whether the ++ * current burst list is still ++ * the same burst list from ++ * which bfqq was removed on ++ * the merge. To avoid this ++ * cost, if bfqq was in a ++ * burst list, then we add ++ * bfqq to the current burst ++ * list without any further ++ * check. This can cause ++ * inappropriate insertions, ++ * but rarely enough to not ++ * harm the detection of large ++ * bursts significantly. ++ */ + hlist_add_head(&bfqq->burst_list_node, + &bfqd->burst_list); + } +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 93034dd7b801..4bbd7f4c0154 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -3945,9 +3945,36 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + BUG_ON(bfqq->entity.tree); + BUG_ON(bfq_bfqq_busy(bfqq)); + +- if (bfq_bfqq_sync(bfqq) && !hlist_unhashed(&bfqq->burst_list_node)) { ++ if (!hlist_unhashed(&bfqq->burst_list_node)) { + hlist_del_init(&bfqq->burst_list_node); +- bfqq->bfqd->burst_size--; ++ /* ++ * Decrement also burst size after the removal, if the ++ * process associated with bfqq is exiting, and thus ++ * does not contribute to the burst any longer. This ++ * decrement helps filter out false positives of large ++ * bursts, when some short-lived process (often due to ++ * the execution of commands by some service) happens ++ * to start and exit while a complex application is ++ * starting, and thus spawning several processes that ++ * do I/O (and that *must not* be treated as a large ++ * burst, see comments on bfq_handle_burst). ++ * ++ * In particular, the decrement is performed only if: ++ * 1) bfqq is not a merged queue, because, if it is, ++ * then this free of bfqq is not triggered by the exit ++ * of the process bfqq is associated with, but exactly ++ * by the fact that bfqq has just been merged. ++ * 2) burst_size is greater than 0, to handle ++ * unbalanced decrements. Unbalanced decrements may ++ * happen in te following case: bfqq is inserted into ++ * the current burst list--without incrementing ++ * bust_size--because of a split, but the current ++ * burst list is not the burst list bfqq belonged to ++ * (see comments on the case of a split in ++ * bfq_set_request). ++ */ ++ if (bfqq->bic && bfqq->bfqd->burst_size > 0) ++ bfqq->bfqd->burst_size--; + } + + bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); +@@ -4691,6 +4718,34 @@ static int bfq_set_request(struct request_queue *q, struct request *rq, + "large burst"); + bfq_clear_bfqq_in_large_burst(bfqq); + if (bic->was_in_burst_list) ++ /* ++ * If bfqq was in the current ++ * burst list before being ++ * merged, then we have to add ++ * it back. And we do not need ++ * to increase burst_size, as ++ * we did not decrement ++ * burst_size when we removed ++ * bfqq from the burst list as ++ * a consequence of a merge ++ * (see comments in ++ * bfq_put_queue). In this ++ * respect, it would be rather ++ * costly to know whether the ++ * current burst list is still ++ * the same burst list from ++ * which bfqq was removed on ++ * the merge. To avoid this ++ * cost, if bfqq was in a ++ * burst list, then we add ++ * bfqq to the current burst ++ * list without any further ++ * check. This can cause ++ * inappropriate insertions, ++ * but rarely enough to not ++ * harm the detection of large ++ * bursts significantly. ++ */ + hlist_add_head(&bfqq->burst_list_node, + &bfqd->burst_list); + } + +From 09adbd0f46f4ba395964b35bf611b7cc3dd84b4d Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Mon, 30 Oct 2017 16:50:50 +0100 +Subject: [PATCH 48/51] doc, block, bfq-mq: update max IOPS sustainable with + BFQ + +We have investigated more deeply the performance of BFQ, in terms of +number of IOPS that can be processed by the CPU when BFQ is used as +I/O scheduler. In more detail, using the script [1], we have measured +the number of IOPS reached on top of a null block device configured +with zero latency, as a function of the workload (sequential read, +sequential write, random read, random write) and of the system (we +considered desktops, laptops and embedded systems). + +Basing on the resulting figures, with this commit we update the +current, conservative IOPS range reported in BFQ documentation. In +particular, the documentation now reports, for each of three different +systems, the lowest number of IOPS obtained for that system with the +above test (namely, the value obtained with the workload leading to +the lowest IOPS). + +[1] https://github.com/Algodev-github/IOSpeed + +Signed-off-by: Paolo Valente +Signed-off-by: Luca Miccio +--- + Documentation/block/bfq-iosched.txt | 19 +++++++++++++------ + 1 file changed, 13 insertions(+), 6 deletions(-) + +diff --git a/Documentation/block/bfq-iosched.txt b/Documentation/block/bfq-iosched.txt +index dcfe15523da3..595ff7a5ff34 100644 +--- a/Documentation/block/bfq-iosched.txt ++++ b/Documentation/block/bfq-iosched.txt +@@ -29,12 +29,19 @@ for that device, by setting low_latency to 0. See Section 3 for + details on how to configure BFQ for the desired tradeoff between + latency and throughput, or on how to maximize throughput. + +-On average CPUs, the current version of BFQ can handle devices +-performing at most ~30K IOPS; at most ~50 KIOPS on faster CPUs. As a +-reference, 30-50 KIOPS correspond to very high bandwidths with +-sequential I/O (e.g., 8-12 GB/s if I/O requests are 256 KB large), and +-to 120-200 MB/s with 4KB random I/O. BFQ is currently being tested on +-multi-queue devices too. ++BFQ has a non-null overhead, which limits the maximum IOPS that the ++CPU can process for a device scheduled with BFQ. To give an idea of ++the limits on slow or average CPUs, here are BFQ limits for three ++different CPUs, on, respectively, an average laptop, an old desktop, ++and a cheap embedded system, in case full hierarchical support is ++enabled (i.e., CONFIG_BFQ_SQ_GROUP_IOSCHED is set for bfq-sq, or ++CONFIG_MQ_BFQ_GROUP_IOSCHED is set for bfq-mq, or, finally, ++CONFIG_BFQ_GROUP_IOSCHED is set for bfq): ++- Intel i7-4850HQ: 250 KIOPS ++- AMD A8-3850: 170 KIOPS ++- ARM CortexTM-A53 Octa-core: 45 KIOPS ++ ++BFQ works for multi-queue devices too (bfq and bfq-mq instances). + + The table of contents follows. Impatients can just jump to Section 3. + + +From be94f97b577dc587593185224a7718aa59ac43f7 Mon Sep 17 00:00:00 2001 +From: Luca Miccio +Date: Tue, 31 Oct 2017 09:50:11 +0100 +Subject: [PATCH 49/51] block, bfq-mq: add missing invocations of + bfqg_stats_update_io_add/remove + +bfqg_stats_update_io_add and bfqg_stats_update_io_remove are to be +invoked, respectively, when an I/O request enters and when an I/O +request exits the scheduler. Unfortunately, bfq-mq does not fully comply +with this scheme, because it does not invoke these functions for +requests that are inserted into or extracted from its priority +dispatch list. This commit fixes this mistake. + +Signed-off-by: Paolo Valente +Signed-off-by: Luca Miccio +--- + block/bfq-mq-iosched.c | 24 +++++++++++++++++++----- + 1 file changed, 19 insertions(+), 5 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 816bac6cdd3d..fbf28804c220 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -1394,7 +1394,6 @@ static void bfq_bfqq_handle_idle_busy_switch(struct bfq_data *bfqd, + BUG_ON(bfqq->entity.budget < bfqq->entity.service); + + BUG_ON(bfqq == bfqd->in_service_queue); +- bfqg_stats_update_io_add(bfqq_group(RQ_BFQQ(rq)), bfqq, rq->cmd_flags); + + /* + * bfqq deserves to be weight-raised if: +@@ -1734,7 +1733,6 @@ static void bfq_remove_request(struct request_queue *q, + BUG_ON(bfqq->meta_pending == 0); + bfqq->meta_pending--; + } +- bfqg_stats_update_io_remove(bfqq_group(bfqq), rq->cmd_flags); + } + + static bool bfq_bio_merge(struct blk_mq_hw_ctx *hctx, struct bio *bio) +@@ -1879,6 +1877,7 @@ static void bfq_requests_merged(struct request_queue *q, struct request *rq, + bfqq->next_rq = rq; + + bfq_remove_request(q, next); ++ bfqg_stats_update_io_remove(bfqq_group(bfqq), next->cmd_flags); + + spin_unlock_irq(&bfqq->bfqd->lock); + end: +@@ -4077,6 +4076,10 @@ static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + spin_lock_irq(&bfqd->lock); + + rq = __bfq_dispatch_request(hctx); ++ if (rq && RQ_BFQQ(rq)) ++ bfqg_stats_update_io_remove(bfqq_group(RQ_BFQQ(rq)), ++ rq->cmd_flags); ++ + spin_unlock_irq(&bfqd->lock); + + return rq; +@@ -4634,6 +4637,7 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + { + struct request_queue *q = hctx->queue; + struct bfq_data *bfqd = q->elevator->elevator_data; ++ struct bfq_queue *bfqq = RQ_BFQQ(rq); + + spin_lock_irq(&bfqd->lock); + if (blk_mq_sched_try_insert_merge(q, rq)) { +@@ -4647,8 +4651,6 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + + spin_lock_irq(&bfqd->lock); + if (at_head || blk_rq_is_passthrough(rq)) { +- struct bfq_queue *bfqq = RQ_BFQQ(rq); +- + if (at_head) + list_add(&rq->queuelist, &bfqd->dispatch); + else +@@ -4668,6 +4670,12 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + rq->rq_flags &= ~RQF_GOT; + + __bfq_insert_request(bfqd, rq); ++ /* ++ * Update bfqq, because, if a queue merge has occurred ++ * in __bfq_insert_request, then rq has been ++ * redirected into a new queue. ++ */ ++ bfqq = RQ_BFQQ(rq); + + if (rq_mergeable(rq)) { + elv_rqhash_add(q, rq); +@@ -4676,6 +4684,9 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + } + } + ++ if (bfqq) ++ bfqg_stats_update_io_add(bfqq_group(bfqq), bfqq, rq->cmd_flags); ++ + spin_unlock_irq(&bfqd->lock); + } + +@@ -4893,8 +4904,11 @@ static void bfq_finish_request(struct request *rq) + BUG_ON(in_interrupt()); + + assert_spin_locked(&bfqd->lock); +- if (!RB_EMPTY_NODE(&rq->rb_node)) ++ if (!RB_EMPTY_NODE(&rq->rb_node)) { + bfq_remove_request(rq->q, rq); ++ bfqg_stats_update_io_remove(bfqq_group(bfqq), ++ rq->cmd_flags); ++ } + bfq_put_rq_priv_body(bfqq); + } + + +From 8659a1549d2bf241129a0f7c90429bddd9c2bc53 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 8 Nov 2017 19:07:40 +0100 +Subject: [PATCH 50/51] block, bfq-mq: update blkio stats outside the scheduler + lock + +bfq-mq invokes various blkg_*stats_* functions to update the statistics +contained in the special files blkio.bfq-mq.* in the blkio controller +groups, i.e., the I/O accounting related to the proportional-share +policy provided by bfq-mq. The execution of these functions takes a +considerable percentage, about 40%, of the total per-request execution +time of bfq-mq (i.e., of the sum of the execution time of all the bfq-mq +functions that have to be executed to process an I/O request from its +creation to its destruction). This reduces the request-processing +rate sustainable by bfq-mq noticeably, even on a multicore CPU. In fact, +the bfq-mq functions that invoke blkg_*stats_* functions cannot be +executed in parallel with the rest of the code of bfq-mq, because +both are executed under the same same per-device scheduler lock. + +To reduce this slowdown, this commit moves, wherever possible, the +invocation of these functions (more precisely, of the bfq-mq functions +that invoke blkg_*stats_* functions) outside the critical sections +protected by the scheduler lock. + +With this change, and with all blkio.bfq-mq.* statistics enabled, the +throughput grows, e.g., from 250 to 310 KIOPS (+25%) on an Intel +i7-4850HQ, in case of 8 threads doing random I/O in parallel on +null_blk, with the latter configured with 0 latency. We obtained the +same or higher throughput boosts, up to +30%, with other processors +(some figures are reported in the documentation). For our tests, we +used the script [1], with which our results can be easily reproduced. + +NOTE. This commit still protects the invocation of blkg_*stats_* +functions with the request_queue lock, because the group these +functions are invoked on may otherwise disappear before or while these +functions are executed. Fortunately, tests without even this lock +show, by difference, that the serialization caused by this lock has a +little impact (at most ~5% of throughput reduction). + +[1] https://github.com/Algodev-github/IOSpeed + +Signed-off-by: Paolo Valente +Signed-off-by: Luca Miccio +--- + Documentation/block/bfq-iosched.txt | 18 ++++-- + block/bfq-mq-iosched.c | 112 +++++++++++++++++++++++++++++++----- + block/bfq-sched.c | 2 + + 3 files changed, 112 insertions(+), 20 deletions(-) + +diff --git a/Documentation/block/bfq-iosched.txt b/Documentation/block/bfq-iosched.txt +index 595ff7a5ff34..c816c595082d 100644 +--- a/Documentation/block/bfq-iosched.txt ++++ b/Documentation/block/bfq-iosched.txt +@@ -31,16 +31,22 @@ latency and throughput, or on how to maximize throughput. + + BFQ has a non-null overhead, which limits the maximum IOPS that the + CPU can process for a device scheduled with BFQ. To give an idea of +-the limits on slow or average CPUs, here are BFQ limits for three +-different CPUs, on, respectively, an average laptop, an old desktop, +-and a cheap embedded system, in case full hierarchical support is +-enabled (i.e., CONFIG_BFQ_SQ_GROUP_IOSCHED is set for bfq-sq, or +-CONFIG_MQ_BFQ_GROUP_IOSCHED is set for bfq-mq, or, finally, +-CONFIG_BFQ_GROUP_IOSCHED is set for bfq): ++the limits on slow or average CPUs, here are, first, the limits of ++bfq-sq for three different CPUs, on, respectively, an average laptop, ++an old desktop, and a cheap embedded system, in case full hierarchical ++support is enabled (i.e., CONFIG_BFQ_SQ_GROUP_IOSCHED is set): + - Intel i7-4850HQ: 250 KIOPS + - AMD A8-3850: 170 KIOPS + - ARM CortexTM-A53 Octa-core: 45 KIOPS + ++bfq-mq and bfq instances reach, instead, a higher sustainable ++throughput. Their limits, on the same systems as above, are, with full ++hierarchical support enabled (i.e., CONFIG_MQ_BFQ_GROUP_IOSCHED set ++for bfq-mq, or CONFIG_BFQ_GROUP_IOSCHED set for bfq): ++- Intel i7-4850HQ: 310 KIOPS ++- AMD A8-3850: 200 KIOPS ++- ARM CortexTM-A53 Octa-core: 56 KIOPS ++ + BFQ works for multi-queue devices too (bfq and bfq-mq instances). + + The table of contents follows. Impatients can just jump to Section 3. +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index fbf28804c220..ab3b83d612c2 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -1822,7 +1822,7 @@ static void bfq_request_merged(struct request_queue *q, struct request *req, + bfqq->next_rq = next_rq; + + bfq_log_bfqq(bfqd, bfqq, +- "requests_merged: req %p prev %p next_rq %p bfqq %p", ++ "request_merged: req %p prev %p next_rq %p bfqq %p", + req, prev, next_rq, bfqq); + + /* +@@ -2415,7 +2415,6 @@ static void __bfq_set_in_service_queue(struct bfq_data *bfqd, + struct bfq_queue *bfqq) + { + if (bfqq) { +- bfqg_stats_update_avg_queue_size(bfqq_group(bfqq)); + bfq_clear_bfqq_fifo_expire(bfqq); + + bfqd->budgets_assigned = (bfqd->budgets_assigned*7 + 256) / 8; +@@ -3784,7 +3783,6 @@ static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) + */ + bfq_clear_bfqq_wait_request(bfqq); + hrtimer_try_to_cancel(&bfqd->idle_slice_timer); +- bfqg_stats_update_idle_time(bfqq_group(bfqq)); + } + goto keep_queue; + } +@@ -4072,16 +4070,67 @@ static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + { + struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; + struct request *rq; ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ struct bfq_queue *in_serv_queue, *bfqq; ++ bool waiting_rq, idle_timer_disabled; ++#endif + + spin_lock_irq(&bfqd->lock); + ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ in_serv_queue = bfqd->in_service_queue; ++ waiting_rq = in_serv_queue && bfq_bfqq_wait_request(in_serv_queue); ++ + rq = __bfq_dispatch_request(hctx); +- if (rq && RQ_BFQQ(rq)) +- bfqg_stats_update_io_remove(bfqq_group(RQ_BFQQ(rq)), +- rq->cmd_flags); + ++ idle_timer_disabled = ++ waiting_rq && !bfq_bfqq_wait_request(in_serv_queue); ++ ++#else ++ rq = __bfq_dispatch_request(hctx); ++#endif + spin_unlock_irq(&bfqd->lock); + ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ bfqq = rq ? RQ_BFQQ(rq) : NULL; ++ if (!idle_timer_disabled && !bfqq) ++ return rq; ++ ++ /* ++ * rq and bfqq are guaranteed to exist until this function ++ * ends, for the following reasons. First, rq can be ++ * dispatched to the device, and then can be completed and ++ * freed, only after this function ends. Second, rq cannot be ++ * merged (and thus freed because of a merge) any longer, ++ * because it has already started. Thus rq cannot be freed ++ * before this function ends, and, since rq has a reference to ++ * bfqq, the same guarantee holds for bfqq too. ++ * ++ * In addition, the following queue lock guarantees that ++ * bfqq_group(bfqq) exists as well. ++ */ ++ spin_lock_irq(hctx->queue->queue_lock); ++ if (idle_timer_disabled) ++ /* ++ * Since the idle timer has been disabled, ++ * in_serv_queue contained some request when ++ * __bfq_dispatch_request was invoked above, which ++ * implies that rq was picked exactly from ++ * in_serv_queue. Thus in_serv_queue == bfqq, and is ++ * therefore guaranteed to exist because of the above ++ * arguments. ++ */ ++ bfqg_stats_update_idle_time(bfqq_group(in_serv_queue)); ++ if (bfqq) { ++ struct bfq_group *bfqg = bfqq_group(bfqq); ++ ++ bfqg_stats_update_avg_queue_size(bfqg); ++ bfqg_stats_set_start_empty_time(bfqg); ++ bfqg_stats_update_io_remove(bfqg, rq->cmd_flags); ++ } ++ spin_unlock_irq(hctx->queue->queue_lock); ++#endif ++ + return rq; + } + +@@ -4200,7 +4249,6 @@ static void bfq_exit_icq_bfqq(struct bfq_io_cq *bic, bool is_sync) + unsigned long flags; + + spin_lock_irqsave(&bfqd->lock, flags); +- + bfq_exit_bfqq(bfqd, bfqq); + bic_set_bfqq(bic, NULL, is_sync); + spin_unlock_irqrestore(&bfqd->lock, flags); +@@ -4554,7 +4602,6 @@ static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, + */ + bfq_clear_bfqq_wait_request(bfqq); + hrtimer_try_to_cancel(&bfqd->idle_slice_timer); +- bfqg_stats_update_idle_time(bfqq_group(bfqq)); + + /* + * The queue is not empty, because a new request just +@@ -4569,9 +4616,11 @@ static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, + } + } + +-static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) ++/* returns true if it causes the idle timer to be disabled */ ++static bool __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + { + struct bfq_queue *bfqq = RQ_BFQQ(rq), *new_bfqq; ++ bool waiting, idle_timer_disabled = false; + BUG_ON(!bfqq); + + assert_spin_locked(&bfqd->lock); +@@ -4624,12 +4673,16 @@ static void __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + } + } + ++ waiting = bfqq && bfq_bfqq_wait_request(bfqq); + bfq_add_request(rq); ++ idle_timer_disabled = waiting && !bfq_bfqq_wait_request(bfqq); + + rq->fifo_time = ktime_get_ns() + bfqd->bfq_fifo_expire[rq_is_sync(rq)]; + list_add_tail(&rq->queuelist, &bfqq->fifo); + + bfq_rq_enqueued(bfqd, bfqq, rq); ++ ++ return idle_timer_disabled; + } + + static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, +@@ -4638,6 +4691,10 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + struct request_queue *q = hctx->queue; + struct bfq_data *bfqd = q->elevator->elevator_data; + struct bfq_queue *bfqq = RQ_BFQQ(rq); ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ bool idle_timer_disabled = false; ++ unsigned int cmd_flags; ++#endif + + spin_lock_irq(&bfqd->lock); + if (blk_mq_sched_try_insert_merge(q, rq)) { +@@ -4669,13 +4726,17 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + BUG_ON(!(rq->rq_flags & RQF_GOT)); + rq->rq_flags &= ~RQF_GOT; + +- __bfq_insert_request(bfqd, rq); ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ idle_timer_disabled = __bfq_insert_request(bfqd, rq); + /* + * Update bfqq, because, if a queue merge has occurred + * in __bfq_insert_request, then rq has been + * redirected into a new queue. + */ + bfqq = RQ_BFQQ(rq); ++#else ++ __bfq_insert_request(bfqd, rq); ++#endif + + if (rq_mergeable(rq)) { + elv_rqhash_add(q, rq); +@@ -4683,11 +4744,34 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + q->last_merge = rq; + } + } +- +- if (bfqq) +- bfqg_stats_update_io_add(bfqq_group(bfqq), bfqq, rq->cmd_flags); +- ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ /* ++ * Cache cmd_flags before releasing scheduler lock, because rq ++ * may disappear afterwards (for example, because of a request ++ * merge). ++ */ ++ cmd_flags = rq->cmd_flags; ++#endif + spin_unlock_irq(&bfqd->lock); ++#ifdef BFQ_GROUP_IOSCHED_ENABLED ++ if (!bfqq) ++ return; ++ /* ++ * bfqq still exists, because it can disappear only after ++ * either it is merged with another queue, or the process it ++ * is associated with exits. But both actions must be taken by ++ * the same process currently executing this flow of ++ * instruction. ++ * ++ * In addition, the following queue lock guarantees that ++ * bfqq_group(bfqq) exists as well. ++ */ ++ spin_lock_irq(q->queue_lock); ++ bfqg_stats_update_io_add(bfqq_group(bfqq), bfqq, cmd_flags); ++ if (idle_timer_disabled) ++ bfqg_stats_update_idle_time(bfqq_group(bfqq)); ++ spin_unlock_irq(q->queue_lock); ++#endif + } + + static void bfq_insert_requests(struct blk_mq_hw_ctx *hctx, +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index e4a2553a2d2c..616c0692335a 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -949,9 +949,11 @@ static void bfq_bfqq_served(struct bfq_queue *bfqq, int served) + st->vtime += bfq_delta(served, st->wsum); + bfq_forget_idle(st); + } ++#ifndef BFQ_MQ + #ifdef BFQ_GROUP_IOSCHED_ENABLED + bfqg_stats_set_start_empty_time(bfqq_group(bfqq)); + #endif ++#endif + st = bfq_entity_service_tree(&bfqq->entity); + bfq_log_bfqq(bfqq->bfqd, bfqq, "bfqq_served %d secs, vtime %llu on %p", + served, ((st->vtime>>10)*1000)>>12, st); + +From abdfb33a3325df55ec0261fd824ca61ddac13575 Mon Sep 17 00:00:00 2001 +From: Luca Miccio +Date: Wed, 8 Nov 2017 19:07:41 +0100 +Subject: [PATCH 51/51] block, bfq-sq, bfq-mq: move debug blkio stats behind + CONFIG_DEBUG_BLK_CGROUP + +BFQ (both bfq-mq and bfq-sq) currently creates, and updates, its own +instance of the whole set of blkio statistics that cfq creates. Yet, +from the comments of Tejun Heo in [1], it turned out that most of +these statistics are meant/useful only for debugging. This commit +makes BFQ create the latter, debugging statistics only if the option +CONFIG_DEBUG_BLK_CGROUP is set. + +By doing so, this commit also enables BFQ to enjoy a high perfomance +boost. The reason is that, if CONFIG_DEBUG_BLK_CGROUP is not set, then +BFQ has to update far fewer statistics, and, in particular, not the +heaviest to update. To give an idea of the benefits, if +CONFIG_DEBUG_BLK_CGROUP is not set, then, on an Intel i7-4850HQ, and +with 8 threads doing random I/O in parallel on null_blk (configured +with 0 latency), the throughput of bfq-mq grows from 310 to 400 KIOPS +(+30%). We have measured similar or even much higher boosts with other +CPUs: e.g., +45% with an ARM CortexTM-A53 Octa-core. Our results have +been obtained and can be reproduced very easily with the script in [1]. + +[1] https://www.spinics.net/lists/linux-block/msg18943.html + +Reported-by: Tejun Heo +Signed-off-by: Luca Miccio +Signed-off-by: Paolo Valente +--- + Documentation/block/bfq-iosched.txt | 59 ++++++++++--- + block/bfq-cgroup-included.c | 163 ++++++++++++++++++++---------------- + block/bfq-mq-iosched.c | 14 ++-- + block/bfq-mq.h | 4 +- + block/bfq.h | 4 +- + 5 files changed, 147 insertions(+), 97 deletions(-) + +diff --git a/Documentation/block/bfq-iosched.txt b/Documentation/block/bfq-iosched.txt +index c816c595082d..30ef2dba85ad 100644 +--- a/Documentation/block/bfq-iosched.txt ++++ b/Documentation/block/bfq-iosched.txt +@@ -29,24 +29,41 @@ for that device, by setting low_latency to 0. See Section 3 for + details on how to configure BFQ for the desired tradeoff between + latency and throughput, or on how to maximize throughput. + +-BFQ has a non-null overhead, which limits the maximum IOPS that the +-CPU can process for a device scheduled with BFQ. To give an idea of +-the limits on slow or average CPUs, here are, first, the limits of +-bfq-sq for three different CPUs, on, respectively, an average laptop, ++BFQ has a non-null overhead, which limits the maximum IOPS that a CPU ++can process for a device scheduled with BFQ. To give an idea of the ++limits on slow or average CPUs, here are, first, the limits of bfq-mq ++and bfq for three different CPUs, on, respectively, an average laptop, + an old desktop, and a cheap embedded system, in case full hierarchical +-support is enabled (i.e., CONFIG_BFQ_SQ_GROUP_IOSCHED is set): +-- Intel i7-4850HQ: 250 KIOPS +-- AMD A8-3850: 170 KIOPS +-- ARM CortexTM-A53 Octa-core: 45 KIOPS +- +-bfq-mq and bfq instances reach, instead, a higher sustainable +-throughput. Their limits, on the same systems as above, are, with full +-hierarchical support enabled (i.e., CONFIG_MQ_BFQ_GROUP_IOSCHED set +-for bfq-mq, or CONFIG_BFQ_GROUP_IOSCHED set for bfq): ++support is enabled (i.e., CONFIG_MQ_BFQ_GROUP_IOSCHED is set for ++bfq-mq, or CONFIG_BFQ_GROUP_IOSCHED is set for bfq), but ++CONFIG_DEBUG_BLK_CGROUP is not set (Section 4-2): ++- Intel i7-4850HQ: 400 KIOPS ++- AMD A8-3850: 250 KIOPS ++- ARM CortexTM-A53 Octa-core: 80 KIOPS ++ ++As for bfq-sq, it cannot reach the above IOPS, because of the ++inherent, lower parallelism of legacy blk and of the components within ++it (including bfq-sq itself). In particular, results with ++CONFIG_DEBUG_BLK_CGROUP unset are rather fluctuating. The limits ++reported below for the case CONFIG_DEBUG_BLK_CGROUP set will however ++provide a lower bound to the limits of bfq-sq. ++ ++Turning back to bfq-mq and bfq, If CONFIG_DEBUG_BLK_CGROUP is set (and ++of course full hierarchical support is enabled), then the sustainable ++throughput with bfq-mq and bfq decreases, because all blkio.bfq* ++statistics are created and updated (Section 4-2). For bfq-mq and bfq, ++this leads to the following maximum sustainable throughputs, on the ++same systems as above: + - Intel i7-4850HQ: 310 KIOPS + - AMD A8-3850: 200 KIOPS + - ARM CortexTM-A53 Octa-core: 56 KIOPS + ++Finally, if CONFIG_DEBUG_BLK_CGROUP is set (and full hierarchical ++support is enabled), then bfq-sq exhibits the following limits: ++- Intel i7-4850HQ: 250 KIOPS ++- AMD A8-3850: 170 KIOPS ++- ARM CortexTM-A53 Octa-core: 45 KIOPS ++ + BFQ works for multi-queue devices too (bfq and bfq-mq instances). + + The table of contents follows. Impatients can just jump to Section 3. +@@ -524,6 +541,22 @@ BFQ-specific files is "blkio.bfqX." or "io.bfqX.", where X can be "" + to set the weight of a group with the mainline BFQ is blkio.bfq.weight + or io.bfq.weight. + ++As for cgroups-v1 (blkio controller), the exact set of stat files ++created, and kept up-to-date by bfq*, depends on whether ++CONFIG_DEBUG_BLK_CGROUP is set. If it is set, then bfq* creates all ++the stat files documented in ++Documentation/cgroup-v1/blkio-controller.txt. If, instead, ++CONFIG_DEBUG_BLK_CGROUP is not set, then bfq* creates only the files ++blkio.bfq*.io_service_bytes ++blkio.bfq*.io_service_bytes_recursive ++blkio.bfq*.io_serviced ++blkio.bfq*.io_serviced_recursive ++ ++The value of CONFIG_DEBUG_BLK_CGROUP greatly influences the maximum ++throughput sustainable with bfq*, because updating the blkio.bfq* ++stats is rather costly, especially for some of the stats enabled by ++CONFIG_DEBUG_BLK_CGROUP. ++ + Parameters to set + ----------------- + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index 631e53d9150d..562b0ce581a7 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -15,7 +15,7 @@ + * file. + */ + +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + + /* bfqg stats flags */ + enum bfqg_stats_flags { +@@ -155,6 +155,63 @@ static void bfqg_stats_update_avg_queue_size(struct bfq_group *bfqg) + bfqg_stats_update_group_wait_time(stats); + } + ++static void bfqg_stats_update_io_add(struct bfq_group *bfqg, ++ struct bfq_queue *bfqq, unsigned int op) ++{ ++ blkg_rwstat_add(&bfqg->stats.queued, op, 1); ++ bfqg_stats_end_empty_time(&bfqg->stats); ++ if (!(bfqq == ((struct bfq_data *)bfqg->bfqd)->in_service_queue)) ++ bfqg_stats_set_start_group_wait_time(bfqg, bfqq_group(bfqq)); ++} ++ ++static void bfqg_stats_update_io_remove(struct bfq_group *bfqg, unsigned int op) ++{ ++ blkg_rwstat_add(&bfqg->stats.queued, op, -1); ++} ++ ++static void bfqg_stats_update_io_merged(struct bfq_group *bfqg, unsigned int op) ++{ ++ blkg_rwstat_add(&bfqg->stats.merged, op, 1); ++} ++ ++static void bfqg_stats_update_completion(struct bfq_group *bfqg, ++ uint64_t start_time, uint64_t io_start_time, unsigned int op) ++{ ++ struct bfqg_stats *stats = &bfqg->stats; ++ unsigned long long now = sched_clock(); ++ ++ if (time_after64(now, io_start_time)) ++ blkg_rwstat_add(&stats->service_time, op, ++ now - io_start_time); ++ if (time_after64(io_start_time, start_time)) ++ blkg_rwstat_add(&stats->wait_time, op, ++ io_start_time - start_time); ++} ++ ++#else /* BFQ_GROUP_IOSCHED_ENABLED && CONFIG_DEBUG_BLK_CGROUP */ ++ ++static inline void bfqg_stats_update_io_add(struct bfq_group *bfqg, ++ struct bfq_queue *bfqq, unsigned int op) { } ++static inline void ++bfqg_stats_update_io_remove(struct bfq_group *bfqg, unsigned int op) { } ++static inline void ++bfqg_stats_update_io_merged(struct bfq_group *bfqg, unsigned int op) { } ++static inline void bfqg_stats_update_completion(struct bfq_group *bfqg, ++ uint64_t start_time, uint64_t io_start_time, ++ unsigned int op) { } ++static inline void ++bfqg_stats_set_start_group_wait_time(struct bfq_group *bfqg, ++ struct bfq_group *curr_bfqg) { } ++static inline void bfqg_stats_end_empty_time(struct bfqg_stats *stats) { } ++static inline void bfqg_stats_update_dequeue(struct bfq_group *bfqg) { } ++static inline void bfqg_stats_set_start_empty_time(struct bfq_group *bfqg) { } ++static inline void bfqg_stats_update_idle_time(struct bfq_group *bfqg) { } ++static inline void bfqg_stats_set_start_idle_time(struct bfq_group *bfqg) { } ++static inline void bfqg_stats_update_avg_queue_size(struct bfq_group *bfqg) { } ++ ++#endif /* BFQ_GROUP_IOSCHED_ENABLED && CONFIG_DEBUG_BLK_CGROUP */ ++ ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + static struct blkcg_policy blkcg_policy_bfq; + + /* +@@ -247,44 +304,10 @@ static void bfqg_and_blkg_put(struct bfq_group *bfqg) + } + #endif + +-static void bfqg_stats_update_io_add(struct bfq_group *bfqg, +- struct bfq_queue *bfqq, +- unsigned int op) +-{ +- blkg_rwstat_add(&bfqg->stats.queued, op, 1); +- bfqg_stats_end_empty_time(&bfqg->stats); +- if (!(bfqq == ((struct bfq_data *)bfqg->bfqd)->in_service_queue)) +- bfqg_stats_set_start_group_wait_time(bfqg, bfqq_group(bfqq)); +-} +- +-static void bfqg_stats_update_io_remove(struct bfq_group *bfqg, unsigned int op) +-{ +- blkg_rwstat_add(&bfqg->stats.queued, op, -1); +-} +- +-static void bfqg_stats_update_io_merged(struct bfq_group *bfqg, unsigned int op) +-{ +- blkg_rwstat_add(&bfqg->stats.merged, op, 1); +-} +- +-static void bfqg_stats_update_completion(struct bfq_group *bfqg, +- uint64_t start_time, uint64_t io_start_time, +- unsigned int op) +-{ +- struct bfqg_stats *stats = &bfqg->stats; +- unsigned long long now = sched_clock(); +- +- if (time_after64(now, io_start_time)) +- blkg_rwstat_add(&stats->service_time, op, +- now - io_start_time); +- if (time_after64(io_start_time, start_time)) +- blkg_rwstat_add(&stats->wait_time, op, +- io_start_time - start_time); +-} +- + /* @stats = 0 */ + static void bfqg_stats_reset(struct bfqg_stats *stats) + { ++#ifdef CONFIG_DEBUG_BLK_CGROUP + /* queued stats shouldn't be cleared */ + blkg_rwstat_reset(&stats->merged); + blkg_rwstat_reset(&stats->service_time); +@@ -296,6 +319,7 @@ static void bfqg_stats_reset(struct bfqg_stats *stats) + blkg_stat_reset(&stats->group_wait_time); + blkg_stat_reset(&stats->idle_time); + blkg_stat_reset(&stats->empty_time); ++#endif + } + + /* @to += @from */ +@@ -304,6 +328,7 @@ static void bfqg_stats_add_aux(struct bfqg_stats *to, struct bfqg_stats *from) + if (!to || !from) + return; + ++#ifdef CONFIG_DEBUG_BLK_CGROUP + /* queued stats shouldn't be cleared */ + blkg_rwstat_add_aux(&to->merged, &from->merged); + blkg_rwstat_add_aux(&to->service_time, &from->service_time); +@@ -316,6 +341,7 @@ static void bfqg_stats_add_aux(struct bfqg_stats *to, struct bfqg_stats *from) + blkg_stat_add_aux(&to->group_wait_time, &from->group_wait_time); + blkg_stat_add_aux(&to->idle_time, &from->idle_time); + blkg_stat_add_aux(&to->empty_time, &from->empty_time); ++#endif + } + + /* +@@ -367,6 +393,7 @@ static void bfq_init_entity(struct bfq_entity *entity, + + static void bfqg_stats_exit(struct bfqg_stats *stats) + { ++#ifdef CONFIG_DEBUG_BLK_CGROUP + blkg_rwstat_exit(&stats->merged); + blkg_rwstat_exit(&stats->service_time); + blkg_rwstat_exit(&stats->wait_time); +@@ -378,10 +405,12 @@ static void bfqg_stats_exit(struct bfqg_stats *stats) + blkg_stat_exit(&stats->group_wait_time); + blkg_stat_exit(&stats->idle_time); + blkg_stat_exit(&stats->empty_time); ++#endif + } + + static int bfqg_stats_init(struct bfqg_stats *stats, gfp_t gfp) + { ++#ifdef CONFIG_DEBUG_BLK_CGROUP + if (blkg_rwstat_init(&stats->merged, gfp) || + blkg_rwstat_init(&stats->service_time, gfp) || + blkg_rwstat_init(&stats->wait_time, gfp) || +@@ -396,6 +425,7 @@ static int bfqg_stats_init(struct bfqg_stats *stats, gfp_t gfp) + bfqg_stats_exit(stats); + return -ENOMEM; + } ++#endif + + return 0; + } +@@ -1003,6 +1033,7 @@ static ssize_t bfq_io_set_weight(struct kernfs_open_file *of, + return bfq_io_set_weight_legacy(of_css(of), NULL, weight); + } + ++#ifdef CONFIG_DEBUG_BLK_CGROUP + static int bfqg_print_stat(struct seq_file *sf, void *v) + { + blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), blkg_prfill_stat, +@@ -1108,6 +1139,7 @@ static int bfqg_print_avg_queue_size(struct seq_file *sf, void *v) + 0, false); + return 0; + } ++#endif /* CONFIG_DEBUG_BLK_CGROUP */ + + static struct bfq_group * + bfq_create_group_hierarchy(struct bfq_data *bfqd, int node) +@@ -1137,15 +1169,6 @@ static struct cftype bfq_blkcg_legacy_files[] = { + + /* statistics, covers only the tasks in the bfqg */ + { +- .name = BFQ_CGROUP_FNAME(time), +- .private = offsetof(struct bfq_group, stats.time), +- .seq_show = bfqg_print_stat, +- }, +- { +- .name = BFQ_CGROUP_FNAME(sectors), +- .seq_show = bfqg_print_stat_sectors, +- }, +- { + .name = BFQ_CGROUP_FNAME(io_service_bytes), + .private = (unsigned long)&blkcg_policy_bfq, + .seq_show = blkg_print_stat_bytes, +@@ -1155,6 +1178,16 @@ static struct cftype bfq_blkcg_legacy_files[] = { + .private = (unsigned long)&blkcg_policy_bfq, + .seq_show = blkg_print_stat_ios, + }, ++#ifdef CONFIG_DEBUG_BLK_CGROUP ++ { ++ .name = BFQ_CGROUP_FNAME(time), ++ .private = offsetof(struct bfq_group, stats.time), ++ .seq_show = bfqg_print_stat, ++ }, ++ { ++ .name = BFQ_CGROUP_FNAME(sectors), ++ .seq_show = bfqg_print_stat_sectors, ++ }, + { + .name = BFQ_CGROUP_FNAME(io_service_time), + .private = offsetof(struct bfq_group, stats.service_time), +@@ -1175,18 +1208,10 @@ static struct cftype bfq_blkcg_legacy_files[] = { + .private = offsetof(struct bfq_group, stats.queued), + .seq_show = bfqg_print_rwstat, + }, ++#endif /* CONFIG_DEBUG_BLK_CGROUP */ + + /* the same statictics which cover the bfqg and its descendants */ + { +- .name = BFQ_CGROUP_FNAME(time_recursive), +- .private = offsetof(struct bfq_group, stats.time), +- .seq_show = bfqg_print_stat_recursive, +- }, +- { +- .name = BFQ_CGROUP_FNAME(sectors_recursive), +- .seq_show = bfqg_print_stat_sectors_recursive, +- }, +- { + .name = BFQ_CGROUP_FNAME(io_service_bytes_recursive), + .private = (unsigned long)&blkcg_policy_bfq, + .seq_show = blkg_print_stat_bytes_recursive, +@@ -1196,6 +1221,16 @@ static struct cftype bfq_blkcg_legacy_files[] = { + .private = (unsigned long)&blkcg_policy_bfq, + .seq_show = blkg_print_stat_ios_recursive, + }, ++#ifdef CONFIG_DEBUG_BLK_CGROUP ++ { ++ .name = BFQ_CGROUP_FNAME(time_recursive), ++ .private = offsetof(struct bfq_group, stats.time), ++ .seq_show = bfqg_print_stat_recursive, ++ }, ++ { ++ .name = BFQ_CGROUP_FNAME(sectors_recursive), ++ .seq_show = bfqg_print_stat_sectors_recursive, ++ }, + { + .name = BFQ_CGROUP_FNAME(io_service_time_recursive), + .private = offsetof(struct bfq_group, stats.service_time), +@@ -1240,6 +1275,7 @@ static struct cftype bfq_blkcg_legacy_files[] = { + .private = offsetof(struct bfq_group, stats.dequeue), + .seq_show = bfqg_print_stat, + }, ++#endif /* CONFIG_DEBUG_BLK_CGROUP */ + { } /* terminate */ + }; + +@@ -1257,25 +1293,6 @@ static struct cftype bfq_blkg_files[] = { + + #else /* BFQ_GROUP_IOSCHED_ENABLED */ + +-static inline void bfqg_stats_update_io_add(struct bfq_group *bfqg, +- struct bfq_queue *bfqq, unsigned int op) { } +-static inline void +-bfqg_stats_update_io_remove(struct bfq_group *bfqg, unsigned int op) { } +-static inline void +-bfqg_stats_update_io_merged(struct bfq_group *bfqg, unsigned int op) { } +-static inline void bfqg_stats_update_completion(struct bfq_group *bfqg, +- uint64_t start_time, uint64_t io_start_time, +- unsigned int op) { } +-static inline void +-bfqg_stats_set_start_group_wait_time(struct bfq_group *bfqg, +- struct bfq_group *curr_bfqg) { } +-static inline void bfqg_stats_end_empty_time(struct bfqg_stats *stats) { } +-static inline void bfqg_stats_update_dequeue(struct bfq_group *bfqg) { } +-static inline void bfqg_stats_set_start_empty_time(struct bfq_group *bfqg) { } +-static inline void bfqg_stats_update_idle_time(struct bfq_group *bfqg) { } +-static inline void bfqg_stats_set_start_idle_time(struct bfq_group *bfqg) { } +-static inline void bfqg_stats_update_avg_queue_size(struct bfq_group *bfqg) { } +- + static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, + struct bfq_group *bfqg) {} + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index ab3b83d612c2..0c09609a6099 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4070,14 +4070,14 @@ static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + { + struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; + struct request *rq; +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + struct bfq_queue *in_serv_queue, *bfqq; + bool waiting_rq, idle_timer_disabled; + #endif + + spin_lock_irq(&bfqd->lock); + +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + in_serv_queue = bfqd->in_service_queue; + waiting_rq = in_serv_queue && bfq_bfqq_wait_request(in_serv_queue); + +@@ -4091,7 +4091,7 @@ static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + #endif + spin_unlock_irq(&bfqd->lock); + +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + bfqq = rq ? RQ_BFQQ(rq) : NULL; + if (!idle_timer_disabled && !bfqq) + return rq; +@@ -4691,7 +4691,7 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + struct request_queue *q = hctx->queue; + struct bfq_data *bfqd = q->elevator->elevator_data; + struct bfq_queue *bfqq = RQ_BFQQ(rq); +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + bool idle_timer_disabled = false; + unsigned int cmd_flags; + #endif +@@ -4726,7 +4726,7 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + BUG_ON(!(rq->rq_flags & RQF_GOT)); + rq->rq_flags &= ~RQF_GOT; + +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + idle_timer_disabled = __bfq_insert_request(bfqd, rq); + /* + * Update bfqq, because, if a queue merge has occurred +@@ -4744,7 +4744,7 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + q->last_merge = rq; + } + } +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + /* + * Cache cmd_flags before releasing scheduler lock, because rq + * may disappear afterwards (for example, because of a request +@@ -4753,7 +4753,7 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + cmd_flags = rq->cmd_flags; + #endif + spin_unlock_irq(&bfqd->lock); +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + if (!bfqq) + return; + /* +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 7ed2cc29be57..1cb05bb853d2 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -784,7 +784,7 @@ enum bfqq_expiration { + + + struct bfqg_stats { +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + /* number of ios merged */ + struct blkg_rwstat merged; + /* total time spent on device in ns, may not be accurate w/ queueing */ +@@ -812,7 +812,7 @@ struct bfqg_stats { + uint64_t start_idle_time; + uint64_t start_empty_time; + uint16_t flags; +-#endif ++#endif /* BFQ_GROUP_IOSCHED_ENABLED && CONFIG_DEBUG_BLK_CGROUP */ + }; + + #ifdef BFQ_GROUP_IOSCHED_ENABLED +diff --git a/block/bfq.h b/block/bfq.h +index 15d326f466b7..47cd4d5a8c32 100644 +--- a/block/bfq.h ++++ b/block/bfq.h +@@ -791,7 +791,7 @@ enum bfqq_expiration { + + + struct bfqg_stats { +-#ifdef BFQ_GROUP_IOSCHED_ENABLED ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + /* number of ios merged */ + struct blkg_rwstat merged; + /* total time spent on device in ns, may not be accurate w/ queueing */ +@@ -819,7 +819,7 @@ struct bfqg_stats { + uint64_t start_idle_time; + uint64_t start_empty_time; + uint16_t flags; +-#endif ++#endif /* BFQ_GROUP_IOSCHED_ENABLED && CONFIG_DEBUG_BLK_CGROUP */ + }; + + #ifdef BFQ_GROUP_IOSCHED_ENABLED diff --git a/sys-kernel/linux-sources-redcore-lts/files/0001-MuQSS-version-0.162-CPU-scheduler-linux-hardened.patch b/sys-kernel/linux-sources-redcore-lts/files/0001-MuQSS-version-0.162-CPU-scheduler-linux-hardened.patch new file mode 100644 index 00000000..8f2c8783 --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0001-MuQSS-version-0.162-CPU-scheduler-linux-hardened.patch @@ -0,0 +1,9571 @@ +diff -Nur a/arch/powerpc/platforms/cell/spufs/sched.c b/arch/powerpc/platforms/cell/spufs/sched.c +--- a/arch/powerpc/platforms/cell/spufs/sched.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/arch/powerpc/platforms/cell/spufs/sched.c 2018-11-03 16:06:32.704528679 +0000 +@@ -65,11 +65,6 @@ + static struct timer_list spuloadavg_timer; + + /* +- * Priority of a normal, non-rt, non-niced'd process (aka nice level 0). +- */ +-#define NORMAL_PRIO 120 +- +-/* + * Frequency of the spu scheduler tick. By default we do one SPU scheduler + * tick for every 10 CPU scheduler ticks. + */ +diff -Nur a/arch/x86/Kconfig b/arch/x86/Kconfig +--- a/arch/x86/Kconfig 2018-11-03 16:00:51.897619785 +0000 ++++ b/arch/x86/Kconfig 2018-11-03 16:06:32.705528711 +0000 +@@ -963,10 +963,26 @@ + depends on SMP + ---help--- + SMT scheduler support improves the CPU scheduler's decision making +- when dealing with Intel Pentium 4 chips with HyperThreading at a ++ when dealing with Intel P4/Core 2 chips with HyperThreading at a + cost of slightly increased overhead in some places. If unsure say + N here. + ++config SMT_NICE ++ bool "SMT (Hyperthreading) aware nice priority and policy support" ++ depends on SCHED_MUQSS && SCHED_SMT ++ default y ++ ---help--- ++ Enabling Hyperthreading on Intel CPUs decreases the effectiveness ++ of the use of 'nice' levels and different scheduling policies ++ (e.g. realtime) due to sharing of CPU power between hyperthreads. ++ SMT nice support makes each logical CPU aware of what is running on ++ its hyperthread siblings, maintaining appropriate distribution of ++ CPU according to nice levels and scheduling policies at the expense ++ of slightly increased overhead. ++ ++ If unsure say Y here. ++ ++ + config SCHED_MC + def_bool y + prompt "Multi-core scheduler support" +diff -Nur a/Documentation/scheduler/sched-BFS.txt b/Documentation/scheduler/sched-BFS.txt +--- a/Documentation/scheduler/sched-BFS.txt 1970-01-01 01:00:00.000000000 +0100 ++++ b/Documentation/scheduler/sched-BFS.txt 2018-11-03 16:06:32.702528615 +0000 +@@ -0,0 +1,351 @@ ++BFS - The Brain Fuck Scheduler by Con Kolivas. ++ ++Goals. ++ ++The goal of the Brain Fuck Scheduler, referred to as BFS from here on, is to ++completely do away with the complex designs of the past for the cpu process ++scheduler and instead implement one that is very simple in basic design. ++The main focus of BFS is to achieve excellent desktop interactivity and ++responsiveness without heuristics and tuning knobs that are difficult to ++understand, impossible to model and predict the effect of, and when tuned to ++one workload cause massive detriment to another. ++ ++ ++Design summary. ++ ++BFS is best described as a single runqueue, O(n) lookup, earliest effective ++virtual deadline first design, loosely based on EEVDF (earliest eligible virtual ++deadline first) and my previous Staircase Deadline scheduler. Each component ++shall be described in order to understand the significance of, and reasoning for ++it. The codebase when the first stable version was released was approximately ++9000 lines less code than the existing mainline linux kernel scheduler (in ++2.6.31). This does not even take into account the removal of documentation and ++the cgroups code that is not used. ++ ++Design reasoning. ++ ++The single runqueue refers to the queued but not running processes for the ++entire system, regardless of the number of CPUs. The reason for going back to ++a single runqueue design is that once multiple runqueues are introduced, ++per-CPU or otherwise, there will be complex interactions as each runqueue will ++be responsible for the scheduling latency and fairness of the tasks only on its ++own runqueue, and to achieve fairness and low latency across multiple CPUs, any ++advantage in throughput of having CPU local tasks causes other disadvantages. ++This is due to requiring a very complex balancing system to at best achieve some ++semblance of fairness across CPUs and can only maintain relatively low latency ++for tasks bound to the same CPUs, not across them. To increase said fairness ++and latency across CPUs, the advantage of local runqueue locking, which makes ++for better scalability, is lost due to having to grab multiple locks. ++ ++A significant feature of BFS is that all accounting is done purely based on CPU ++used and nowhere is sleep time used in any way to determine entitlement or ++interactivity. Interactivity "estimators" that use some kind of sleep/run ++algorithm are doomed to fail to detect all interactive tasks, and to falsely tag ++tasks that aren't interactive as being so. The reason for this is that it is ++close to impossible to determine that when a task is sleeping, whether it is ++doing it voluntarily, as in a userspace application waiting for input in the ++form of a mouse click or otherwise, or involuntarily, because it is waiting for ++another thread, process, I/O, kernel activity or whatever. Thus, such an ++estimator will introduce corner cases, and more heuristics will be required to ++cope with those corner cases, introducing more corner cases and failed ++interactivity detection and so on. Interactivity in BFS is built into the design ++by virtue of the fact that tasks that are waking up have not used up their quota ++of CPU time, and have earlier effective deadlines, thereby making it very likely ++they will preempt any CPU bound task of equivalent nice level. See below for ++more information on the virtual deadline mechanism. Even if they do not preempt ++a running task, because the rr interval is guaranteed to have a bound upper ++limit on how long a task will wait for, it will be scheduled within a timeframe ++that will not cause visible interface jitter. ++ ++ ++Design details. ++ ++Task insertion. ++ ++BFS inserts tasks into each relevant queue as an O(1) insertion into a double ++linked list. On insertion, *every* running queue is checked to see if the newly ++queued task can run on any idle queue, or preempt the lowest running task on the ++system. This is how the cross-CPU scheduling of BFS achieves significantly lower ++latency per extra CPU the system has. In this case the lookup is, in the worst ++case scenario, O(n) where n is the number of CPUs on the system. ++ ++Data protection. ++ ++BFS has one single lock protecting the process local data of every task in the ++global queue. Thus every insertion, removal and modification of task data in the ++global runqueue needs to grab the global lock. However, once a task is taken by ++a CPU, the CPU has its own local data copy of the running process' accounting ++information which only that CPU accesses and modifies (such as during a ++timer tick) thus allowing the accounting data to be updated lockless. Once a ++CPU has taken a task to run, it removes it from the global queue. Thus the ++global queue only ever has, at most, ++ ++ (number of tasks requesting cpu time) - (number of logical CPUs) + 1 ++ ++tasks in the global queue. This value is relevant for the time taken to look up ++tasks during scheduling. This will increase if many tasks with CPU affinity set ++in their policy to limit which CPUs they're allowed to run on if they outnumber ++the number of CPUs. The +1 is because when rescheduling a task, the CPU's ++currently running task is put back on the queue. Lookup will be described after ++the virtual deadline mechanism is explained. ++ ++Virtual deadline. ++ ++The key to achieving low latency, scheduling fairness, and "nice level" ++distribution in BFS is entirely in the virtual deadline mechanism. The one ++tunable in BFS is the rr_interval, or "round robin interval". This is the ++maximum time two SCHED_OTHER (or SCHED_NORMAL, the common scheduling policy) ++tasks of the same nice level will be running for, or looking at it the other ++way around, the longest duration two tasks of the same nice level will be ++delayed for. When a task requests cpu time, it is given a quota (time_slice) ++equal to the rr_interval and a virtual deadline. The virtual deadline is ++offset from the current time in jiffies by this equation: ++ ++ jiffies + (prio_ratio * rr_interval) ++ ++The prio_ratio is determined as a ratio compared to the baseline of nice -20 ++and increases by 10% per nice level. The deadline is a virtual one only in that ++no guarantee is placed that a task will actually be scheduled by this time, but ++it is used to compare which task should go next. There are three components to ++how a task is next chosen. First is time_slice expiration. If a task runs out ++of its time_slice, it is descheduled, the time_slice is refilled, and the ++deadline reset to that formula above. Second is sleep, where a task no longer ++is requesting CPU for whatever reason. The time_slice and deadline are _not_ ++adjusted in this case and are just carried over for when the task is next ++scheduled. Third is preemption, and that is when a newly waking task is deemed ++higher priority than a currently running task on any cpu by virtue of the fact ++that it has an earlier virtual deadline than the currently running task. The ++earlier deadline is the key to which task is next chosen for the first and ++second cases. Once a task is descheduled, it is put back on the queue, and an ++O(n) lookup of all queued-but-not-running tasks is done to determine which has ++the earliest deadline and that task is chosen to receive CPU next. ++ ++The CPU proportion of different nice tasks works out to be approximately the ++ ++ (prio_ratio difference)^2 ++ ++The reason it is squared is that a task's deadline does not change while it is ++running unless it runs out of time_slice. Thus, even if the time actually ++passes the deadline of another task that is queued, it will not get CPU time ++unless the current running task deschedules, and the time "base" (jiffies) is ++constantly moving. ++ ++Task lookup. ++ ++BFS has 103 priority queues. 100 of these are dedicated to the static priority ++of realtime tasks, and the remaining 3 are, in order of best to worst priority, ++SCHED_ISO (isochronous), SCHED_NORMAL, and SCHED_IDLEPRIO (idle priority ++scheduling). When a task of these priorities is queued, a bitmap of running ++priorities is set showing which of these priorities has tasks waiting for CPU ++time. When a CPU is made to reschedule, the lookup for the next task to get ++CPU time is performed in the following way: ++ ++First the bitmap is checked to see what static priority tasks are queued. If ++any realtime priorities are found, the corresponding queue is checked and the ++first task listed there is taken (provided CPU affinity is suitable) and lookup ++is complete. If the priority corresponds to a SCHED_ISO task, they are also ++taken in FIFO order (as they behave like SCHED_RR). If the priority corresponds ++to either SCHED_NORMAL or SCHED_IDLEPRIO, then the lookup becomes O(n). At this ++stage, every task in the runlist that corresponds to that priority is checked ++to see which has the earliest set deadline, and (provided it has suitable CPU ++affinity) it is taken off the runqueue and given the CPU. If a task has an ++expired deadline, it is taken and the rest of the lookup aborted (as they are ++chosen in FIFO order). ++ ++Thus, the lookup is O(n) in the worst case only, where n is as described ++earlier, as tasks may be chosen before the whole task list is looked over. ++ ++ ++Scalability. ++ ++The major limitations of BFS will be that of scalability, as the separate ++runqueue designs will have less lock contention as the number of CPUs rises. ++However they do not scale linearly even with separate runqueues as multiple ++runqueues will need to be locked concurrently on such designs to be able to ++achieve fair CPU balancing, to try and achieve some sort of nice-level fairness ++across CPUs, and to achieve low enough latency for tasks on a busy CPU when ++other CPUs would be more suited. BFS has the advantage that it requires no ++balancing algorithm whatsoever, as balancing occurs by proxy simply because ++all CPUs draw off the global runqueue, in priority and deadline order. Despite ++the fact that scalability is _not_ the prime concern of BFS, it both shows very ++good scalability to smaller numbers of CPUs and is likely a more scalable design ++at these numbers of CPUs. ++ ++It also has some very low overhead scalability features built into the design ++when it has been deemed their overhead is so marginal that they're worth adding. ++The first is the local copy of the running process' data to the CPU it's running ++on to allow that data to be updated lockless where possible. Then there is ++deference paid to the last CPU a task was running on, by trying that CPU first ++when looking for an idle CPU to use the next time it's scheduled. Finally there ++is the notion of cache locality beyond the last running CPU. The sched_domains ++information is used to determine the relative virtual "cache distance" that ++other CPUs have from the last CPU a task was running on. CPUs with shared ++caches, such as SMT siblings, or multicore CPUs with shared caches, are treated ++as cache local. CPUs without shared caches are treated as not cache local, and ++CPUs on different NUMA nodes are treated as very distant. This "relative cache ++distance" is used by modifying the virtual deadline value when doing lookups. ++Effectively, the deadline is unaltered between "cache local" CPUs, doubled for ++"cache distant" CPUs, and quadrupled for "very distant" CPUs. The reasoning ++behind the doubling of deadlines is as follows. The real cost of migrating a ++task from one CPU to another is entirely dependant on the cache footprint of ++the task, how cache intensive the task is, how long it's been running on that ++CPU to take up the bulk of its cache, how big the CPU cache is, how fast and ++how layered the CPU cache is, how fast a context switch is... and so on. In ++other words, it's close to random in the real world where we do more than just ++one sole workload. The only thing we can be sure of is that it's not free. So ++BFS uses the principle that an idle CPU is a wasted CPU and utilising idle CPUs ++is more important than cache locality, and cache locality only plays a part ++after that. Doubling the effective deadline is based on the premise that the ++"cache local" CPUs will tend to work on the same tasks up to double the number ++of cache local CPUs, and once the workload is beyond that amount, it is likely ++that none of the tasks are cache warm anywhere anyway. The quadrupling for NUMA ++is a value I pulled out of my arse. ++ ++When choosing an idle CPU for a waking task, the cache locality is determined ++according to where the task last ran and then idle CPUs are ranked from best ++to worst to choose the most suitable idle CPU based on cache locality, NUMA ++node locality and hyperthread sibling business. They are chosen in the ++following preference (if idle): ++ ++* Same core, idle or busy cache, idle threads ++* Other core, same cache, idle or busy cache, idle threads. ++* Same node, other CPU, idle cache, idle threads. ++* Same node, other CPU, busy cache, idle threads. ++* Same core, busy threads. ++* Other core, same cache, busy threads. ++* Same node, other CPU, busy threads. ++* Other node, other CPU, idle cache, idle threads. ++* Other node, other CPU, busy cache, idle threads. ++* Other node, other CPU, busy threads. ++ ++This shows the SMT or "hyperthread" awareness in the design as well which will ++choose a real idle core first before a logical SMT sibling which already has ++tasks on the physical CPU. ++ ++Early benchmarking of BFS suggested scalability dropped off at the 16 CPU mark. ++However this benchmarking was performed on an earlier design that was far less ++scalable than the current one so it's hard to know how scalable it is in terms ++of both CPUs (due to the global runqueue) and heavily loaded machines (due to ++O(n) lookup) at this stage. Note that in terms of scalability, the number of ++_logical_ CPUs matters, not the number of _physical_ CPUs. Thus, a dual (2x) ++quad core (4X) hyperthreaded (2X) machine is effectively a 16X. Newer benchmark ++results are very promising indeed, without needing to tweak any knobs, features ++or options. Benchmark contributions are most welcome. ++ ++ ++Features ++ ++As the initial prime target audience for BFS was the average desktop user, it ++was designed to not need tweaking, tuning or have features set to obtain benefit ++from it. Thus the number of knobs and features has been kept to an absolute ++minimum and should not require extra user input for the vast majority of cases. ++There are precisely 2 tunables, and 2 extra scheduling policies. The rr_interval ++and iso_cpu tunables, and the SCHED_ISO and SCHED_IDLEPRIO policies. In addition ++to this, BFS also uses sub-tick accounting. What BFS does _not_ now feature is ++support for CGROUPS. The average user should neither need to know what these ++are, nor should they need to be using them to have good desktop behaviour. ++ ++rr_interval ++ ++There is only one "scheduler" tunable, the round robin interval. This can be ++accessed in ++ ++ /proc/sys/kernel/rr_interval ++ ++The value is in milliseconds, and the default value is set to 6 on a ++uniprocessor machine, and automatically set to a progressively higher value on ++multiprocessor machines. The reasoning behind increasing the value on more CPUs ++is that the effective latency is decreased by virtue of there being more CPUs on ++BFS (for reasons explained above), and increasing the value allows for less ++cache contention and more throughput. Valid values are from 1 to 1000 ++Decreasing the value will decrease latencies at the cost of decreasing ++throughput, while increasing it will improve throughput, but at the cost of ++worsening latencies. The accuracy of the rr interval is limited by HZ resolution ++of the kernel configuration. Thus, the worst case latencies are usually slightly ++higher than this actual value. The default value of 6 is not an arbitrary one. ++It is based on the fact that humans can detect jitter at approximately 7ms, so ++aiming for much lower latencies is pointless under most circumstances. It is ++worth noting this fact when comparing the latency performance of BFS to other ++schedulers. Worst case latencies being higher than 7ms are far worse than ++average latencies not being in the microsecond range. ++ ++Isochronous scheduling. ++ ++Isochronous scheduling is a unique scheduling policy designed to provide ++near-real-time performance to unprivileged (ie non-root) users without the ++ability to starve the machine indefinitely. Isochronous tasks (which means ++"same time") are set using, for example, the schedtool application like so: ++ ++ schedtool -I -e amarok ++ ++This will start the audio application "amarok" as SCHED_ISO. How SCHED_ISO works ++is that it has a priority level between true realtime tasks and SCHED_NORMAL ++which would allow them to preempt all normal tasks, in a SCHED_RR fashion (ie, ++if multiple SCHED_ISO tasks are running, they purely round robin at rr_interval ++rate). However if ISO tasks run for more than a tunable finite amount of time, ++they are then demoted back to SCHED_NORMAL scheduling. This finite amount of ++time is the percentage of _total CPU_ available across the machine, configurable ++as a percentage in the following "resource handling" tunable (as opposed to a ++scheduler tunable): ++ ++ /proc/sys/kernel/iso_cpu ++ ++and is set to 70% by default. It is calculated over a rolling 5 second average ++Because it is the total CPU available, it means that on a multi CPU machine, it ++is possible to have an ISO task running as realtime scheduling indefinitely on ++just one CPU, as the other CPUs will be available. Setting this to 100 is the ++equivalent of giving all users SCHED_RR access and setting it to 0 removes the ++ability to run any pseudo-realtime tasks. ++ ++A feature of BFS is that it detects when an application tries to obtain a ++realtime policy (SCHED_RR or SCHED_FIFO) and the caller does not have the ++appropriate privileges to use those policies. When it detects this, it will ++give the task SCHED_ISO policy instead. Thus it is transparent to the user. ++Because some applications constantly set their policy as well as their nice ++level, there is potential for them to undo the override specified by the user ++on the command line of setting the policy to SCHED_ISO. To counter this, once ++a task has been set to SCHED_ISO policy, it needs superuser privileges to set ++it back to SCHED_NORMAL. This will ensure the task remains ISO and all child ++processes and threads will also inherit the ISO policy. ++ ++Idleprio scheduling. ++ ++Idleprio scheduling is a scheduling policy designed to give out CPU to a task ++_only_ when the CPU would be otherwise idle. The idea behind this is to allow ++ultra low priority tasks to be run in the background that have virtually no ++effect on the foreground tasks. This is ideally suited to distributed computing ++clients (like setiathome, folding, mprime etc) but can also be used to start ++a video encode or so on without any slowdown of other tasks. To avoid this ++policy from grabbing shared resources and holding them indefinitely, if it ++detects a state where the task is waiting on I/O, the machine is about to ++suspend to ram and so on, it will transiently schedule them as SCHED_NORMAL. As ++per the Isochronous task management, once a task has been scheduled as IDLEPRIO, ++it cannot be put back to SCHED_NORMAL without superuser privileges. Tasks can ++be set to start as SCHED_IDLEPRIO with the schedtool command like so: ++ ++ schedtool -D -e ./mprime ++ ++Subtick accounting. ++ ++It is surprisingly difficult to get accurate CPU accounting, and in many cases, ++the accounting is done by simply determining what is happening at the precise ++moment a timer tick fires off. This becomes increasingly inaccurate as the ++timer tick frequency (HZ) is lowered. It is possible to create an application ++which uses almost 100% CPU, yet by being descheduled at the right time, records ++zero CPU usage. While the main problem with this is that there are possible ++security implications, it is also difficult to determine how much CPU a task ++really does use. BFS tries to use the sub-tick accounting from the TSC clock, ++where possible, to determine real CPU usage. This is not entirely reliable, but ++is far more likely to produce accurate CPU usage data than the existing designs ++and will not show tasks as consuming no CPU usage when they actually are. Thus, ++the amount of CPU reported as being used by BFS will more accurately represent ++how much CPU the task itself is using (as is shown for example by the 'time' ++application), so the reported values may be quite different to other schedulers. ++Values reported as the 'load' are more prone to problems with this design, but ++per process values are closer to real usage. When comparing throughput of BFS ++to other designs, it is important to compare the actual completed work in terms ++of total wall clock time taken and total work done, rather than the reported ++"cpu usage". ++ ++ ++Con Kolivas Fri Aug 27 2010 +diff -Nur a/Documentation/scheduler/sched-MuQSS.txt b/Documentation/scheduler/sched-MuQSS.txt +--- a/Documentation/scheduler/sched-MuQSS.txt 1970-01-01 01:00:00.000000000 +0100 ++++ b/Documentation/scheduler/sched-MuQSS.txt 2018-11-03 16:06:32.702528615 +0000 +@@ -0,0 +1,347 @@ ++MuQSS - The Multiple Queue Skiplist Scheduler by Con Kolivas. ++ ++MuQSS is a per-cpu runqueue variant of the original BFS scheduler with ++one 8 level skiplist per runqueue, and fine grained locking for much more ++scalability. ++ ++ ++Goals. ++ ++The goal of the Multiple Queue Skiplist Scheduler, referred to as MuQSS from ++here on (pronounced mux) is to completely do away with the complex designs of ++the past for the cpu process scheduler and instead implement one that is very ++simple in basic design. The main focus of MuQSS is to achieve excellent desktop ++interactivity and responsiveness without heuristics and tuning knobs that are ++difficult to understand, impossible to model and predict the effect of, and when ++tuned to one workload cause massive detriment to another, while still being ++scalable to many CPUs and processes. ++ ++ ++Design summary. ++ ++MuQSS is best described as per-cpu multiple runqueue, O(log n) insertion, O(1) ++lookup, earliest effective virtual deadline first tickless design, loosely based ++on EEVDF (earliest eligible virtual deadline first) and my previous Staircase ++Deadline scheduler, and evolved from the single runqueue O(n) BFS scheduler. ++Each component shall be described in order to understand the significance of, ++and reasoning for it. ++ ++ ++Design reasoning. ++ ++In BFS, the use of a single runqueue across all CPUs meant that each CPU would ++need to scan the entire runqueue looking for the process with the earliest ++deadline and schedule that next, regardless of which CPU it originally came ++from. This made BFS deterministic with respect to latency and provided ++guaranteed latencies dependent on number of processes and CPUs. The single ++runqueue, however, meant that all CPUs would compete for the single lock ++protecting it, which would lead to increasing lock contention as the number of ++CPUs rose and appeared to limit scalability of common workloads beyond 16 ++logical CPUs. Additionally, the O(n) lookup of the runqueue list obviously ++increased overhead proportionate to the number of queued proecesses and led to ++cache thrashing while iterating over the linked list. ++ ++MuQSS is an evolution of BFS, designed to maintain the same scheduling ++decision mechanism and be virtually deterministic without relying on the ++constrained design of the single runqueue by splitting out the single runqueue ++to be per-CPU and use skiplists instead of linked lists. ++ ++The original reason for going back to a single runqueue design for BFS was that ++once multiple runqueues are introduced, per-CPU or otherwise, there will be ++complex interactions as each runqueue will be responsible for the scheduling ++latency and fairness of the tasks only on its own runqueue, and to achieve ++fairness and low latency across multiple CPUs, any advantage in throughput of ++having CPU local tasks causes other disadvantages. This is due to requiring a ++very complex balancing system to at best achieve some semblance of fairness ++across CPUs and can only maintain relatively low latency for tasks bound to the ++same CPUs, not across them. To increase said fairness and latency across CPUs, ++the advantage of local runqueue locking, which makes for better scalability, is ++lost due to having to grab multiple locks. ++ ++MuQSS works around the problems inherent in multiple runqueue designs by ++making its skip lists priority ordered and through novel use of lockless ++examination of each other runqueue it can decide if it should take the earliest ++deadline task from another runqueue for latency reasons, or for CPU balancing ++reasons. It still does not have a balancing system, choosing to allow the ++next task scheduling decision and task wakeup CPU choice to allow balancing to ++happen by virtue of its choices. ++ ++ ++Design details. ++ ++Custom skip list implementation: ++ ++To avoid the overhead of building up and tearing down skip list structures, ++the variant used by MuQSS has a number of optimisations making it specific for ++its use case in the scheduler. It uses static arrays of 8 'levels' instead of ++building up and tearing down structures dynamically. This makes each runqueue ++only scale O(log N) up to 64k tasks. However as there is one runqueue per CPU ++it means that it scales O(log N) up to 64k x number of logical CPUs which is ++far beyond the realistic task limits each CPU could handle. By being 8 levels ++it also makes the array exactly one cacheline in size. Additionally, each ++skip list node is bidirectional making insertion and removal amortised O(1), ++being O(k) where k is 1-8. Uniquely, we are only ever interested in the very ++first entry in each list at all times with MuQSS, so there is never a need to ++do a search and thus look up is always O(1). In interactive mode, the queues ++will be searched beyond their first entry if the first task is not suitable ++for affinity or SMT nice reasons. ++ ++Task insertion: ++ ++MuQSS inserts tasks into a per CPU runqueue as an O(log N) insertion into ++a custom skip list as described above (based on the original design by William ++Pugh). Insertion is ordered in such a way that there is never a need to do a ++search by ordering tasks according to static priority primarily, and then ++virtual deadline at the time of insertion. ++ ++Niffies: ++ ++Niffies are a monotonic forward moving timer not unlike the "jiffies" but are ++of nanosecond resolution. Niffies are calculated per-runqueue from the high ++resolution TSC timers, and in order to maintain fairness are synchronised ++between CPUs whenever both runqueues are locked concurrently. ++ ++Virtual deadline: ++ ++The key to achieving low latency, scheduling fairness, and "nice level" ++distribution in MuQSS is entirely in the virtual deadline mechanism. The one ++tunable in MuQSS is the rr_interval, or "round robin interval". This is the ++maximum time two SCHED_OTHER (or SCHED_NORMAL, the common scheduling policy) ++tasks of the same nice level will be running for, or looking at it the other ++way around, the longest duration two tasks of the same nice level will be ++delayed for. When a task requests cpu time, it is given a quota (time_slice) ++equal to the rr_interval and a virtual deadline. The virtual deadline is ++offset from the current time in niffies by this equation: ++ ++ niffies + (prio_ratio * rr_interval) ++ ++The prio_ratio is determined as a ratio compared to the baseline of nice -20 ++and increases by 10% per nice level. The deadline is a virtual one only in that ++no guarantee is placed that a task will actually be scheduled by this time, but ++it is used to compare which task should go next. There are three components to ++how a task is next chosen. First is time_slice expiration. If a task runs out ++of its time_slice, it is descheduled, the time_slice is refilled, and the ++deadline reset to that formula above. Second is sleep, where a task no longer ++is requesting CPU for whatever reason. The time_slice and deadline are _not_ ++adjusted in this case and are just carried over for when the task is next ++scheduled. Third is preemption, and that is when a newly waking task is deemed ++higher priority than a currently running task on any cpu by virtue of the fact ++that it has an earlier virtual deadline than the currently running task. The ++earlier deadline is the key to which task is next chosen for the first and ++second cases. ++ ++The CPU proportion of different nice tasks works out to be approximately the ++ ++ (prio_ratio difference)^2 ++ ++The reason it is squared is that a task's deadline does not change while it is ++running unless it runs out of time_slice. Thus, even if the time actually ++passes the deadline of another task that is queued, it will not get CPU time ++unless the current running task deschedules, and the time "base" (niffies) is ++constantly moving. ++ ++Task lookup: ++ ++As tasks are already pre-ordered according to anticipated scheduling order in ++the skip lists, lookup for the next suitable task per-runqueue is always a ++matter of simply selecting the first task in the 0th level skip list entry. ++In order to maintain optimal latency and fairness across CPUs, MuQSS does a ++novel examination of every other runqueue in cache locality order, choosing the ++best task across all runqueues. This provides near-determinism of how long any ++task across the entire system may wait before receiving CPU time. The other ++runqueues are first examine lockless and then trylocked to minimise the ++potential lock contention if they are likely to have a suitable better task. ++Each other runqueue lock is only held for as long as it takes to examine the ++entry for suitability. In "interactive" mode, the default setting, MuQSS will ++look for the best deadline task across all CPUs, while in !interactive mode, ++it will only select a better deadline task from another CPU if it is more ++heavily laden than the current one. ++ ++Lookup is therefore O(k) where k is number of CPUs. ++ ++ ++Latency. ++ ++Through the use of virtual deadlines to govern the scheduling order of normal ++tasks, queue-to-activation latency per runqueue is guaranteed to be bound by ++the rr_interval tunable which is set to 6ms by default. This means that the ++longest a CPU bound task will wait for more CPU is proportional to the number ++of running tasks and in the common case of 0-2 running tasks per CPU, will be ++under the 7ms threshold for human perception of jitter. Additionally, as newly ++woken tasks will have an early deadline from their previous runtime, the very ++tasks that are usually latency sensitive will have the shortest interval for ++activation, usually preempting any existing CPU bound tasks. ++ ++Tickless expiry: ++ ++A feature of MuQSS is that it is not tied to the resolution of the chosen tick ++rate in Hz, instead depending entirely on the high resolution timers where ++possible for sub-millisecond accuracy on timeouts regarless of the underlying ++tick rate. This allows MuQSS to be run with the low overhead of low Hz rates ++such as 100 by default, benefiting from the improved throughput and lower ++power usage it provides. Another advantage of this approach is that in ++combination with the Full No HZ option, which disables ticks on running task ++CPUs instead of just idle CPUs, the tick can be disabled at all times ++regardless of how many tasks are running instead of being limited to just one ++running task. Note that this option is NOT recommended for regular desktop ++users. ++ ++ ++Scalability and balancing. ++ ++Unlike traditional approaches where balancing is a combination of CPU selection ++at task wakeup and intermittent balancing based on a vast array of rules set ++according to architecture, busyness calculations and special case management, ++MuQSS indirectly balances on the fly at task wakeup and next task selection. ++During initialisation, MuQSS creates a cache coherency ordered list of CPUs for ++each logical CPU and uses this to aid task/CPU selection when CPUs are busy. ++Additionally it selects any idle CPUs, if they are available, at any time over ++busy CPUs according to the following preference: ++ ++ * Same thread, idle or busy cache, idle or busy threads ++ * Other core, same cache, idle or busy cache, idle threads. ++ * Same node, other CPU, idle cache, idle threads. ++ * Same node, other CPU, busy cache, idle threads. ++ * Other core, same cache, busy threads. ++ * Same node, other CPU, busy threads. ++ * Other node, other CPU, idle cache, idle threads. ++ * Other node, other CPU, busy cache, idle threads. ++ * Other node, other CPU, busy threads. ++ ++Mux is therefore SMT, MC and Numa aware without the need for extra ++intermittent balancing to maintain CPUs busy and make the most of cache ++coherency. ++ ++ ++Features ++ ++As the initial prime target audience for MuQSS was the average desktop user, it ++was designed to not need tweaking, tuning or have features set to obtain benefit ++from it. Thus the number of knobs and features has been kept to an absolute ++minimum and should not require extra user input for the vast majority of cases. ++There are 3 optional tunables, and 2 extra scheduling policies. The rr_interval, ++interactive, and iso_cpu tunables, and the SCHED_ISO and SCHED_IDLEPRIO ++policies. In addition to this, MuQSS also uses sub-tick accounting. What MuQSS ++does _not_ now feature is support for CGROUPS. The average user should neither ++need to know what these are, nor should they need to be using them to have good ++desktop behaviour. However since some applications refuse to work without ++cgroups, one can enable them with MuQSS as a stub and the filesystem will be ++created which will allow the applications to work. ++ ++rr_interval: ++ ++ /proc/sys/kernel/rr_interval ++ ++The value is in milliseconds, and the default value is set to 6. Valid values ++are from 1 to 1000 Decreasing the value will decrease latencies at the cost of ++decreasing throughput, while increasing it will improve throughput, but at the ++cost of worsening latencies. It is based on the fact that humans can detect ++jitter at approximately 7ms, so aiming for much lower latencies is pointless ++under most circumstances. It is worth noting this fact when comparing the ++latency performance of MuQSS to other schedulers. Worst case latencies being ++higher than 7ms are far worse than average latencies not being in the ++microsecond range. ++ ++interactive: ++ ++ /proc/sys/kernel/interactive ++ ++The value is a simple boolean of 1 for on and 0 for off and is set to on by ++default. Disabling this will disable the near-determinism of MuQSS when ++selecting the next task by not examining all CPUs for the earliest deadline ++task, or which CPU to wake to, instead prioritising CPU balancing for improved ++throughput. Latency will still be bound by rr_interval, but on a per-CPU basis ++instead of across the whole system. ++ ++Isochronous scheduling: ++ ++Isochronous scheduling is a unique scheduling policy designed to provide ++near-real-time performance to unprivileged (ie non-root) users without the ++ability to starve the machine indefinitely. Isochronous tasks (which means ++"same time") are set using, for example, the schedtool application like so: ++ ++ schedtool -I -e amarok ++ ++This will start the audio application "amarok" as SCHED_ISO. How SCHED_ISO works ++is that it has a priority level between true realtime tasks and SCHED_NORMAL ++which would allow them to preempt all normal tasks, in a SCHED_RR fashion (ie, ++if multiple SCHED_ISO tasks are running, they purely round robin at rr_interval ++rate). However if ISO tasks run for more than a tunable finite amount of time, ++they are then demoted back to SCHED_NORMAL scheduling. This finite amount of ++time is the percentage of CPU available per CPU, configurable as a percentage in ++the following "resource handling" tunable (as opposed to a scheduler tunable): ++ ++iso_cpu: ++ ++ /proc/sys/kernel/iso_cpu ++ ++and is set to 70% by default. It is calculated over a rolling 5 second average ++Because it is the total CPU available, it means that on a multi CPU machine, it ++is possible to have an ISO task running as realtime scheduling indefinitely on ++just one CPU, as the other CPUs will be available. Setting this to 100 is the ++equivalent of giving all users SCHED_RR access and setting it to 0 removes the ++ability to run any pseudo-realtime tasks. ++ ++A feature of MuQSS is that it detects when an application tries to obtain a ++realtime policy (SCHED_RR or SCHED_FIFO) and the caller does not have the ++appropriate privileges to use those policies. When it detects this, it will ++give the task SCHED_ISO policy instead. Thus it is transparent to the user. ++ ++ ++Idleprio scheduling: ++ ++Idleprio scheduling is a scheduling policy designed to give out CPU to a task ++_only_ when the CPU would be otherwise idle. The idea behind this is to allow ++ultra low priority tasks to be run in the background that have virtually no ++effect on the foreground tasks. This is ideally suited to distributed computing ++clients (like setiathome, folding, mprime etc) but can also be used to start a ++video encode or so on without any slowdown of other tasks. To avoid this policy ++from grabbing shared resources and holding them indefinitely, if it detects a ++state where the task is waiting on I/O, the machine is about to suspend to ram ++and so on, it will transiently schedule them as SCHED_NORMAL. Once a task has ++been scheduled as IDLEPRIO, it cannot be put back to SCHED_NORMAL without ++superuser privileges since it is effectively a lower scheduling policy. Tasks ++can be set to start as SCHED_IDLEPRIO with the schedtool command like so: ++ ++schedtool -D -e ./mprime ++ ++Subtick accounting: ++ ++It is surprisingly difficult to get accurate CPU accounting, and in many cases, ++the accounting is done by simply determining what is happening at the precise ++moment a timer tick fires off. This becomes increasingly inaccurate as the timer ++tick frequency (HZ) is lowered. It is possible to create an application which ++uses almost 100% CPU, yet by being descheduled at the right time, records zero ++CPU usage. While the main problem with this is that there are possible security ++implications, it is also difficult to determine how much CPU a task really does ++use. Mux uses sub-tick accounting from the TSC clock to determine real CPU ++usage. Thus, the amount of CPU reported as being used by MuQSS will more ++accurately represent how much CPU the task itself is using (as is shown for ++example by the 'time' application), so the reported values may be quite ++different to other schedulers. When comparing throughput of MuQSS to other ++designs, it is important to compare the actual completed work in terms of total ++wall clock time taken and total work done, rather than the reported "cpu usage". ++ ++Symmetric MultiThreading (SMT) aware nice: ++ ++SMT, a.k.a. hyperthreading, is a very common feature on modern CPUs. While the ++logical CPU count rises by adding thread units to each CPU core, allowing more ++than one task to be run simultaneously on the same core, the disadvantage of it ++is that the CPU power is shared between the tasks, not summating to the power ++of two CPUs. The practical upshot of this is that two tasks running on ++separate threads of the same core run significantly slower than if they had one ++core each to run on. While smart CPU selection allows each task to have a core ++to itself whenever available (as is done on MuQSS), it cannot offset the ++slowdown that occurs when the cores are all loaded and only a thread is left. ++Most of the time this is harmless as the CPU is effectively overloaded at this ++point and the extra thread is of benefit. However when running a niced task in ++the presence of an un-niced task (say nice 19 v nice 0), the nice task gets ++precisely the same amount of CPU power as the unniced one. MuQSS has an ++optional configuration feature known as SMT-NICE which selectively idles the ++secondary niced thread for a period proportional to the nice difference, ++allowing CPU distribution according to nice level to be maintained, at the ++expense of a small amount of extra overhead. If this is configured in on a ++machine without SMT threads, the overhead is minimal. ++ ++ ++Con Kolivas Sat, 29th October 2016 +diff -Nur a/Documentation/sysctl/kernel.txt b/Documentation/sysctl/kernel.txt +--- a/Documentation/sysctl/kernel.txt 2018-11-03 16:00:51.893619657 +0000 ++++ b/Documentation/sysctl/kernel.txt 2018-11-03 16:06:32.703528647 +0000 +@@ -39,6 +39,7 @@ + - hung_task_timeout_secs + - hung_task_warnings + - kexec_load_disabled ++- iso_cpu + - kptr_restrict + - l2cr [ PPC only ] + - modprobe ==> Documentation/debugging-modules.txt +@@ -73,6 +74,7 @@ + - randomize_va_space + - real-root-dev ==> Documentation/admin-guide/initrd.rst + - reboot-cmd [ SPARC only ] ++- rr_interval + - rtsig-max + - rtsig-nr + - seccomp/ ==> Documentation/userspace-api/seccomp_filter.rst +@@ -95,6 +97,7 @@ + - unknown_nmi_panic + - watchdog + - watchdog_thresh ++- yield_type + - version + + ============================================================== +@@ -397,6 +400,16 @@ + + ============================================================== + ++iso_cpu: (MuQSS CPU scheduler only). ++ ++This sets the percentage cpu that the unprivileged SCHED_ISO tasks can ++run effectively at realtime priority, averaged over a rolling five ++seconds over the -whole- system, meaning all cpus. ++ ++Set to 70 (percent) by default. ++ ++============================================================== ++ + l2cr: (PPC only) + + This flag controls the L2 cache of G3 processor boards. If +@@ -823,6 +836,20 @@ + + ============================================================== + ++rr_interval: (MuQSS CPU scheduler only) ++ ++This is the smallest duration that any cpu process scheduling unit ++will run for. Increasing this value can increase throughput of cpu ++bound tasks substantially but at the expense of increased latencies ++overall. Conversely decreasing it will decrease average and maximum ++latencies but at the expense of throughput. This value is in ++milliseconds and the default value chosen depends on the number of ++cpus available at scheduler initialisation with a minimum of 6. ++ ++Valid values are from 1-1000. ++ ++============================================================== ++ + rtsig-max & rtsig-nr: + + The file rtsig-max can be used to tune the maximum number +@@ -1081,3 +1108,13 @@ + tunable to zero will disable lockup detection altogether. + + ============================================================== ++ ++yield_type: (MuQSS CPU scheduler only) ++ ++This determines what type of yield calls to sched_yield will perform. ++ ++ 0: No yield. ++ 1: Yield only to better priority/deadline tasks. (default) ++ 2: Expire timeslice and recalculate deadline. ++ ++============================================================== +diff -Nur a/fs/proc/base.c b/fs/proc/base.c +--- a/fs/proc/base.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/fs/proc/base.c 2018-11-03 16:06:32.706528743 +0000 +@@ -481,7 +481,7 @@ + seq_printf(m, "0 0 0\n"); + else + seq_printf(m, "%llu %llu %lu\n", +- (unsigned long long)task->se.sum_exec_runtime, ++ (unsigned long long)tsk_seruntime(task), + (unsigned long long)task->sched_info.run_delay, + task->sched_info.pcount); + +diff -Nur a/include/linux/init_task.h b/include/linux/init_task.h +--- a/include/linux/init_task.h 2018-10-10 07:54:28.000000000 +0100 ++++ b/include/linux/init_task.h 2018-11-03 16:06:32.706528743 +0000 +@@ -172,8 +172,6 @@ + # define INIT_VTIME(tsk) + #endif + +-#define INIT_TASK_COMM "swapper" +- + #ifdef CONFIG_RT_MUTEXES + # define INIT_RT_MUTEXES(tsk) \ + .pi_waiters = RB_ROOT_CACHED, \ +@@ -223,6 +221,80 @@ + * INIT_TASK is used to set up the first task table, touch at + * your own risk!. Base=0, limit=0x1fffff (=2MB) + */ ++#ifdef CONFIG_SCHED_MUQSS ++#define INIT_TASK_COMM "MuQSS" ++#define INIT_TASK(tsk) \ ++{ \ ++ INIT_TASK_TI(tsk) \ ++ .state = 0, \ ++ .stack = init_stack, \ ++ .usage = ATOMIC_INIT(2), \ ++ .flags = PF_KTHREAD, \ ++ .prio = NORMAL_PRIO, \ ++ .static_prio = MAX_PRIO-20, \ ++ .normal_prio = NORMAL_PRIO, \ ++ .deadline = 0, \ ++ .policy = SCHED_NORMAL, \ ++ .cpus_allowed = CPU_MASK_ALL, \ ++ .mm = NULL, \ ++ .active_mm = &init_mm, \ ++ .restart_block = { \ ++ .fn = do_no_restart_syscall, \ ++ }, \ ++ .time_slice = 1000000, \ ++ .tasks = LIST_HEAD_INIT(tsk.tasks), \ ++ INIT_PUSHABLE_TASKS(tsk) \ ++ .ptraced = LIST_HEAD_INIT(tsk.ptraced), \ ++ .ptrace_entry = LIST_HEAD_INIT(tsk.ptrace_entry), \ ++ .real_parent = &tsk, \ ++ .parent = &tsk, \ ++ .children = LIST_HEAD_INIT(tsk.children), \ ++ .sibling = LIST_HEAD_INIT(tsk.sibling), \ ++ .group_leader = &tsk, \ ++ RCU_POINTER_INITIALIZER(real_cred, &init_cred), \ ++ RCU_POINTER_INITIALIZER(cred, &init_cred), \ ++ .comm = INIT_TASK_COMM, \ ++ .thread = INIT_THREAD, \ ++ .fs = &init_fs, \ ++ .files = &init_files, \ ++ .signal = &init_signals, \ ++ .sighand = &init_sighand, \ ++ .nsproxy = &init_nsproxy, \ ++ .pending = { \ ++ .list = LIST_HEAD_INIT(tsk.pending.list), \ ++ .signal = {{0}}}, \ ++ .blocked = {{0}}, \ ++ .alloc_lock = __SPIN_LOCK_UNLOCKED(tsk.alloc_lock), \ ++ .journal_info = NULL, \ ++ INIT_CPU_TIMERS(tsk) \ ++ .pi_lock = __RAW_SPIN_LOCK_UNLOCKED(tsk.pi_lock), \ ++ .timer_slack_ns = 50000, /* 50 usec default slack */ \ ++ .pids = { \ ++ [PIDTYPE_PID] = INIT_PID_LINK(PIDTYPE_PID), \ ++ [PIDTYPE_PGID] = INIT_PID_LINK(PIDTYPE_PGID), \ ++ [PIDTYPE_SID] = INIT_PID_LINK(PIDTYPE_SID), \ ++ }, \ ++ .thread_group = LIST_HEAD_INIT(tsk.thread_group), \ ++ .thread_node = LIST_HEAD_INIT(init_signals.thread_head), \ ++ INIT_IDS \ ++ INIT_PERF_EVENTS(tsk) \ ++ INIT_TRACE_IRQFLAGS \ ++ INIT_LOCKDEP \ ++ INIT_FTRACE_GRAPH \ ++ INIT_TRACE_RECURSION \ ++ INIT_TASK_RCU_PREEMPT(tsk) \ ++ INIT_TASK_RCU_TASKS(tsk) \ ++ INIT_CPUSET_SEQ(tsk) \ ++ INIT_RT_MUTEXES(tsk) \ ++ INIT_PREV_CPUTIME(tsk) \ ++ INIT_VTIME(tsk) \ ++ INIT_NUMA_BALANCING(tsk) \ ++ INIT_KASAN(tsk) \ ++ INIT_LIVEPATCH(tsk) \ ++ INIT_TASK_SECURITY \ ++} ++#else /* CONFIG_SCHED_MUQSS */ ++#define INIT_TASK_COMM "swapper" + #define INIT_TASK(tsk) \ + { \ + INIT_TASK_TI(tsk) \ +@@ -300,7 +372,7 @@ + INIT_LIVEPATCH(tsk) \ + INIT_TASK_SECURITY \ + } +- ++#endif /* CONFIG_SCHED_MUQSS */ + + /* Attach to the init_task data structure for proper alignment */ + #define __init_task_data __attribute__((__section__(".data..init_task"))) +diff -Nur a/include/linux/ioprio.h b/include/linux/ioprio.h +--- a/include/linux/ioprio.h 2018-10-10 07:54:28.000000000 +0100 ++++ b/include/linux/ioprio.h 2018-11-03 16:06:32.706528743 +0000 +@@ -52,6 +52,8 @@ + */ + static inline int task_nice_ioprio(struct task_struct *task) + { ++ if (iso_task(task)) ++ return 0; + return (task_nice(task) + 20) / 5; + } + +diff -Nur a/include/linux/sched/nohz.h b/include/linux/sched/nohz.h +--- a/include/linux/sched/nohz.h 2018-10-10 07:54:28.000000000 +0100 ++++ b/include/linux/sched/nohz.h 2018-11-03 16:06:32.707528775 +0000 +@@ -6,7 +6,7 @@ + * This is the interface between the scheduler and nohz/dynticks: + */ + +-#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) ++#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) && !defined(CONFIG_SCHED_MUQSS) + extern void cpu_load_update_nohz_start(void); + extern void cpu_load_update_nohz_stop(void); + #else +@@ -23,7 +23,7 @@ + static inline void set_cpu_sd_state_idle(void) { } + #endif + +-#ifdef CONFIG_NO_HZ_COMMON ++#if defined(CONFIG_NO_HZ_COMMON) && !defined(CONFIG_SCHED_MUQSS) + void calc_load_nohz_start(void); + void calc_load_nohz_stop(void); + #else +diff -Nur a/include/linux/sched/prio.h b/include/linux/sched/prio.h +--- a/include/linux/sched/prio.h 2018-10-10 07:54:28.000000000 +0100 ++++ b/include/linux/sched/prio.h 2018-11-03 16:06:32.707528775 +0000 +@@ -20,8 +20,20 @@ + */ + + #define MAX_USER_RT_PRIO 100 ++ ++#ifdef CONFIG_SCHED_MUQSS ++/* Note different MAX_RT_PRIO */ ++#define MAX_RT_PRIO (MAX_USER_RT_PRIO + 1) ++ ++#define ISO_PRIO (MAX_RT_PRIO) ++#define NORMAL_PRIO (MAX_RT_PRIO + 1) ++#define IDLE_PRIO (MAX_RT_PRIO + 2) ++#define PRIO_LIMIT ((IDLE_PRIO) + 1) ++#else /* CONFIG_SCHED_MUQSS */ + #define MAX_RT_PRIO MAX_USER_RT_PRIO + ++#endif /* CONFIG_SCHED_MUQSS */ ++ + #define MAX_PRIO (MAX_RT_PRIO + NICE_WIDTH) + #define DEFAULT_PRIO (MAX_RT_PRIO + NICE_WIDTH / 2) + +diff -Nur a/include/linux/sched/task.h b/include/linux/sched/task.h +--- a/include/linux/sched/task.h 2018-10-10 07:54:28.000000000 +0100 ++++ b/include/linux/sched/task.h 2018-11-03 16:06:32.707528775 +0000 +@@ -80,7 +80,7 @@ + extern void free_task(struct task_struct *tsk); + + /* sched_exec is called by processes performing an exec */ +-#ifdef CONFIG_SMP ++#if defined(CONFIG_SMP) && !defined(CONFIG_SCHED_MUQSS) + extern void sched_exec(void); + #else + #define sched_exec() {} +diff -Nur a/include/linux/sched.h b/include/linux/sched.h +--- a/include/linux/sched.h 2018-10-10 07:54:28.000000000 +0100 ++++ b/include/linux/sched.h 2018-11-03 16:06:32.707528775 +0000 +@@ -27,6 +27,9 @@ + #include + #include + #include ++#ifdef CONFIG_SCHED_MUQSS ++#include ++#endif + + /* task_struct member predeclarations (sorted alphabetically): */ + struct audit_context; +@@ -579,9 +582,11 @@ + unsigned int flags; + unsigned int ptrace; + ++#if defined(CONFIG_SMP) || defined(CONFIG_SCHED_MUQSS) ++ int on_cpu; ++#endif + #ifdef CONFIG_SMP + struct llist_node wake_entry; +- int on_cpu; + #ifdef CONFIG_THREAD_INFO_IN_TASK + /* Current CPU: */ + unsigned int cpu; +@@ -598,10 +603,25 @@ + int static_prio; + int normal_prio; + unsigned int rt_priority; ++#ifdef CONFIG_SCHED_MUQSS ++ int time_slice; ++ u64 deadline; ++ skiplist_node node; /* Skip list node */ ++ u64 last_ran; ++ u64 sched_time; /* sched_clock time spent running */ ++#ifdef CONFIG_SMT_NICE ++ int smt_bias; /* Policy/nice level bias across smt siblings */ ++#endif ++#ifdef CONFIG_HOTPLUG_CPU ++ bool zerobound; /* Bound to CPU0 for hotplug */ ++#endif ++ unsigned long rt_timeout; ++#else /* CONFIG_SCHED_MUQSS */ + + const struct sched_class *sched_class; + struct sched_entity se; + struct sched_rt_entity rt; ++#endif + #ifdef CONFIG_CGROUP_SCHED + struct task_group *sched_task_group; + #endif +@@ -751,6 +771,10 @@ + u64 utimescaled; + u64 stimescaled; + #endif ++#ifdef CONFIG_SCHED_MUQSS ++ /* Unbanked cpu time */ ++ unsigned long utime_ns, stime_ns; ++#endif + u64 gtime; + struct prev_cputime prev_cputime; + #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN +@@ -1155,6 +1179,40 @@ + */ + }; + ++#ifdef CONFIG_SCHED_MUQSS ++#define tsk_seruntime(t) ((t)->sched_time) ++#define tsk_rttimeout(t) ((t)->rt_timeout) ++ ++static inline void tsk_cpus_current(struct task_struct *p) ++{ ++} ++ ++void print_scheduler_version(void); ++ ++static inline bool iso_task(struct task_struct *p) ++{ ++ return (p->policy == SCHED_ISO); ++} ++#else /* CFS */ ++#define tsk_seruntime(t) ((t)->se.sum_exec_runtime) ++#define tsk_rttimeout(t) ((t)->rt.timeout) ++ ++static inline void tsk_cpus_current(struct task_struct *p) ++{ ++ p->nr_cpus_allowed = current->nr_cpus_allowed; ++} ++ ++static inline void print_scheduler_version(void) ++{ ++ printk(KERN_INFO "CFS CPU scheduler.\n"); ++} ++ ++static inline bool iso_task(struct task_struct *p) ++{ ++ return false; ++} ++#endif /* CONFIG_SCHED_MUQSS */ ++ + static inline struct pid *task_pid(struct task_struct *task) + { + return task->pids[PIDTYPE_PID].pid; +diff -Nur a/include/linux/skip_list.h b/include/linux/skip_list.h +--- a/include/linux/skip_list.h 1970-01-01 01:00:00.000000000 +0100 ++++ b/include/linux/skip_list.h 2018-11-03 16:06:32.708528807 +0000 +@@ -0,0 +1,33 @@ ++#ifndef _LINUX_SKIP_LISTS_H ++#define _LINUX_SKIP_LISTS_H ++typedef u64 keyType; ++typedef void *valueType; ++ ++typedef struct nodeStructure skiplist_node; ++ ++struct nodeStructure { ++ int level; /* Levels in this structure */ ++ keyType key; ++ valueType value; ++ skiplist_node *next[8]; ++ skiplist_node *prev[8]; ++}; ++ ++typedef struct listStructure { ++ int entries; ++ int level; /* Maximum level of the list ++ (1 more than the number of levels in the list) */ ++ skiplist_node *header; /* pointer to header */ ++} skiplist; ++ ++void skiplist_init(skiplist_node *slnode); ++skiplist *new_skiplist(skiplist_node *slnode); ++void free_skiplist(skiplist *l); ++void skiplist_node_init(skiplist_node *node); ++void skiplist_insert(skiplist *l, skiplist_node *node, keyType key, valueType value, unsigned int randseed); ++void skiplist_delete(skiplist *l, skiplist_node *node); ++ ++static inline bool skiplist_node_empty(skiplist_node *node) { ++ return (!node->next[0]); ++} ++#endif /* _LINUX_SKIP_LISTS_H */ +diff -Nur a/include/uapi/linux/sched.h b/include/uapi/linux/sched.h +--- a/include/uapi/linux/sched.h 2018-10-10 07:54:28.000000000 +0100 ++++ b/include/uapi/linux/sched.h 2018-11-03 16:06:32.708528807 +0000 +@@ -37,9 +37,16 @@ + #define SCHED_FIFO 1 + #define SCHED_RR 2 + #define SCHED_BATCH 3 +-/* SCHED_ISO: reserved but not implemented yet */ ++/* SCHED_ISO: Implemented on MuQSS only */ + #define SCHED_IDLE 5 ++#ifdef CONFIG_SCHED_MUQSS ++#define SCHED_ISO 4 ++#define SCHED_IDLEPRIO SCHED_IDLE ++#define SCHED_MAX (SCHED_IDLEPRIO) ++#define SCHED_RANGE(policy) ((policy) <= SCHED_MAX) ++#else /* CONFIG_SCHED_MUQSS */ + #define SCHED_DEADLINE 6 ++#endif /* CONFIG_SCHED_MUQSS */ + + /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */ + #define SCHED_RESET_ON_FORK 0x40000000 +diff -Nur a/init/Kconfig b/init/Kconfig +--- a/init/Kconfig 2018-11-03 16:00:51.921620552 +0000 ++++ b/init/Kconfig 2018-11-03 16:06:32.709528839 +0000 +@@ -38,6 +38,18 @@ + + menu "General setup" + ++config SCHED_MUQSS ++ bool "MuQSS cpu scheduler" ++ select HIGH_RES_TIMERS ++ ---help--- ++ The Multiple Queue Skiplist Scheduler for excellent interactivity and ++ responsiveness on the desktop and highly scalable deterministic ++ low latency on any hardware. ++ ++ Say Y here. ++ default y ++ ++ + config BROKEN + bool + +@@ -621,6 +633,7 @@ + depends on ARCH_SUPPORTS_NUMA_BALANCING + depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY + depends on SMP && NUMA && MIGRATION ++ depends on !SCHED_MUQSS + help + This option adds support for automatic NUMA aware memory/task placement. + The mechanism is quite primitive and is based on migrating memory when +@@ -723,9 +736,13 @@ + help + This feature lets CPU scheduler recognize task groups and control CPU + bandwidth allocation to such task groups. It uses cgroups to group +- tasks. ++ tasks. In combination with MuQSS this is purely a STUB to create the ++ files associated with the CPU controller cgroup but most of the ++ controls do nothing. This is useful for working in environments and ++ with applications that will only work if this control group is ++ present. + +-if CGROUP_SCHED ++if CGROUP_SCHED && !SCHED_MUQSS + config FAIR_GROUP_SCHED + bool "Group scheduling for SCHED_OTHER" + depends on CGROUP_SCHED +@@ -832,6 +849,7 @@ + + config CGROUP_CPUACCT + bool "Simple CPU accounting controller" ++ depends on !SCHED_MUQSS + help + Provides a simple controller for monitoring the + total CPU consumed by the tasks in a cgroup. +@@ -950,6 +968,7 @@ + + config SCHED_AUTOGROUP + bool "Automatic process group scheduling" ++ depends on !SCHED_MUQSS + select CGROUPS + select CGROUP_SCHED + select FAIR_GROUP_SCHED +diff -Nur a/init/main.c b/init/main.c +--- a/init/main.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/init/main.c 2018-11-03 16:06:32.709528839 +0000 +@@ -841,7 +841,6 @@ + return ret; + } + +- + extern initcall_t __initcall_start[]; + extern initcall_t __initcall0_start[]; + extern initcall_t __initcall1_start[]; +@@ -1008,6 +1007,8 @@ + + rcu_end_inkernel_boot(); + ++ print_scheduler_version(); ++ + if (ramdisk_execute_command) { + ret = run_init_process(ramdisk_execute_command); + if (!ret) +diff -Nur a/kernel/delayacct.c b/kernel/delayacct.c +--- a/kernel/delayacct.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/delayacct.c 2018-11-03 16:06:32.710528871 +0000 +@@ -115,7 +115,7 @@ + */ + t1 = tsk->sched_info.pcount; + t2 = tsk->sched_info.run_delay; +- t3 = tsk->se.sum_exec_runtime; ++ t3 = tsk_seruntime(tsk); + + d->cpu_count += t1; + +diff -Nur a/kernel/exit.c b/kernel/exit.c +--- a/kernel/exit.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/exit.c 2018-11-03 16:06:32.710528871 +0000 +@@ -129,7 +129,7 @@ + sig->curr_target = next_thread(tsk); + } + +- add_device_randomness((const void*) &tsk->se.sum_exec_runtime, ++ add_device_randomness((const void*) &tsk_seruntime(tsk), + sizeof(unsigned long long)); + + /* +@@ -150,7 +150,7 @@ + sig->inblock += task_io_get_inblock(tsk); + sig->oublock += task_io_get_oublock(tsk); + task_io_accounting_add(&sig->ioac, &tsk->ioac); +- sig->sum_sched_runtime += tsk->se.sum_exec_runtime; ++ sig->sum_sched_runtime += tsk_seruntime(tsk); + sig->nr_threads--; + __unhash_process(tsk, group_dead); + write_sequnlock(&sig->stats_lock); +diff -Nur a/kernel/kthread.c b/kernel/kthread.c +--- a/kernel/kthread.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/kthread.c 2018-11-03 16:06:32.711528903 +0000 +@@ -410,6 +410,34 @@ + } + EXPORT_SYMBOL(kthread_bind); + ++#if defined(CONFIG_SCHED_MUQSS) && defined(CONFIG_SMP) ++extern void __do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask); ++ ++/* ++ * new_kthread_bind is a special variant of __kthread_bind_mask. ++ * For new threads to work on muqss we want to call do_set_cpus_allowed ++ * without the task_cpu being set and the task rescheduled until they're ++ * rescheduled on their own so we call __do_set_cpus_allowed directly which ++ * only changes the cpumask. This is particularly important for smpboot threads ++ * to work. ++ */ ++static void new_kthread_bind(struct task_struct *p, unsigned int cpu) ++{ ++ unsigned long flags; ++ ++ if (WARN_ON(!wait_task_inactive(p, TASK_UNINTERRUPTIBLE))) ++ return; ++ ++ /* It's safe because the task is inactive. */ ++ raw_spin_lock_irqsave(&p->pi_lock, flags); ++ __do_set_cpus_allowed(p, cpumask_of(cpu)); ++ p->flags |= PF_NO_SETAFFINITY; ++ raw_spin_unlock_irqrestore(&p->pi_lock, flags); ++} ++#else ++#define new_kthread_bind(p, cpu) kthread_bind(p, cpu) ++#endif ++ + /** + * kthread_create_on_cpu - Create a cpu bound kthread + * @threadfn: the function to run until signal_pending(current). +@@ -431,7 +459,7 @@ + cpu); + if (IS_ERR(p)) + return p; +- kthread_bind(p, cpu); ++ new_kthread_bind(p, cpu); + /* CPU hotplug need to bind once again when unparking the thread. */ + set_bit(KTHREAD_IS_PER_CPU, &to_kthread(p)->flags); + to_kthread(p)->cpu = cpu; +diff -Nur a/kernel/livepatch/transition.c b/kernel/livepatch/transition.c +--- a/kernel/livepatch/transition.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/livepatch/transition.c 2018-11-03 16:06:32.711528903 +0000 +@@ -277,6 +277,12 @@ + return 0; + } + ++#ifdef CONFIG_SCHED_MUQSS ++typedef unsigned long rq_flags_t; ++#else ++typedef struct rq_flags rq_flag_t; ++#endif ++ + /* + * Try to safely switch a task to the target patch state. If it's currently + * running, or it's sleeping on a to-be-patched or to-be-unpatched function, or +@@ -285,7 +291,7 @@ + static bool klp_try_switch_task(struct task_struct *task) + { + struct rq *rq; +- struct rq_flags flags; ++ rq_flags_t flags; + int ret; + bool success = false; + char err_buf[STACK_ERR_BUF_SIZE]; +diff -Nur a/kernel/Makefile b/kernel/Makefile +--- a/kernel/Makefile 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/Makefile 2018-11-03 16:06:32.709528839 +0000 +@@ -10,7 +10,7 @@ + extable.o params.o \ + kthread.o sys_ni.o nsproxy.o \ + notifier.o ksysfs.o cred.o reboot.o \ +- async.o range.o smpboot.o ucount.o ++ async.o range.o smpboot.o ucount.o skip_list.o + + obj-$(CONFIG_MODULES) += kmod.o + obj-$(CONFIG_MULTIUSER) += groups.o +diff -Nur a/kernel/rcu/Kconfig b/kernel/rcu/Kconfig +--- a/kernel/rcu/Kconfig 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/rcu/Kconfig 2018-11-03 16:06:32.711528903 +0000 +@@ -93,7 +93,7 @@ + config CONTEXT_TRACKING_FORCE + bool "Force context tracking" + depends on CONTEXT_TRACKING +- default y if !NO_HZ_FULL ++ default y if !NO_HZ_FULL && !SCHED_MUQSS + help + The major pre-requirement for full dynticks to work is to + support the context tracking subsystem. But there are also +diff -Nur a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c +--- a/kernel/sched/cpufreq_schedutil.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/sched/cpufreq_schedutil.c 2018-11-03 16:06:32.716529064 +0000 +@@ -176,6 +176,17 @@ + return cpufreq_driver_resolve_freq(policy, freq); + } + ++#ifdef CONFIG_SCHED_MUQSS ++static void sugov_get_util(unsigned long *util, unsigned long *max, int cpu) ++{ ++ struct rq *rq = cpu_rq(cpu); ++ ++ *util = rq->load_avg; ++ if (*util > SCHED_CAPACITY_SCALE) ++ *util = SCHED_CAPACITY_SCALE; ++ *max = SCHED_CAPACITY_SCALE; ++} ++#else /* CONFIG_SCHED_MUQSS */ + static void sugov_get_util(unsigned long *util, unsigned long *max, int cpu) + { + struct rq *rq = cpu_rq(cpu); +@@ -186,6 +197,7 @@ + *util = min(rq->cfs.avg.util_avg, cfs_max); + *max = cfs_max; + } ++#endif /* CONFIG_SCHED_MUQSS */ + + static void sugov_set_iowait_boost(struct sugov_cpu *sg_cpu, u64 time, + unsigned int flags) +diff -Nur a/kernel/sched/cputime.c b/kernel/sched/cputime.c +--- a/kernel/sched/cputime.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/sched/cputime.c 2018-11-03 16:06:32.716529064 +0000 +@@ -270,26 +270,6 @@ + return accounted; + } + +-#ifdef CONFIG_64BIT +-static inline u64 read_sum_exec_runtime(struct task_struct *t) +-{ +- return t->se.sum_exec_runtime; +-} +-#else +-static u64 read_sum_exec_runtime(struct task_struct *t) +-{ +- u64 ns; +- struct rq_flags rf; +- struct rq *rq; +- +- rq = task_rq_lock(t, &rf); +- ns = t->se.sum_exec_runtime; +- task_rq_unlock(rq, t, &rf); +- +- return ns; +-} +-#endif +- + /* + * Accumulate raw cputime values of dead tasks (sig->[us]time) and live + * tasks (sum on group iteration) belonging to @tsk's group. +@@ -661,7 +641,7 @@ + void task_cputime_adjusted(struct task_struct *p, u64 *ut, u64 *st) + { + struct task_cputime cputime = { +- .sum_exec_runtime = p->se.sum_exec_runtime, ++ .sum_exec_runtime = tsk_seruntime(p), + }; + + task_cputime(p, &cputime.utime, &cputime.stime); +diff -Nur a/kernel/sched/idle.c b/kernel/sched/idle.c +--- a/kernel/sched/idle.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/sched/idle.c 2018-11-03 16:06:32.716529064 +0000 +@@ -209,6 +209,9 @@ + */ + static void do_idle(void) + { ++ int cpu = smp_processor_id(); ++ bool pending = false; ++ + /* + * If the arch has a polling bit, we maintain an invariant: + * +@@ -220,13 +223,16 @@ + + __current_set_polling(); + quiet_vmstat(); +- tick_nohz_idle_enter(); ++ if (unlikely(softirq_pending(cpu))) ++ pending = true; ++ else ++ tick_nohz_idle_enter(); + + while (!need_resched()) { + check_pgt_cache(); + rmb(); + +- if (cpu_is_offline(smp_processor_id())) { ++ if (cpu_is_offline(cpu)) { + cpuhp_report_idle_dead(); + arch_cpu_idle_dead(); + } +@@ -255,7 +261,8 @@ + * an IPI to fold the state for us. + */ + preempt_set_need_resched(); +- tick_nohz_idle_exit(); ++ if (!pending) ++ tick_nohz_idle_exit(); + __current_clr_polling(); + + /* +diff -Nur a/kernel/sched/Makefile b/kernel/sched/Makefile +--- a/kernel/sched/Makefile 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/sched/Makefile 2018-11-03 16:06:32.711528903 +0000 +@@ -16,14 +16,20 @@ + CFLAGS_core.o := $(PROFILING) -fno-omit-frame-pointer + endif + +-obj-y += core.o loadavg.o clock.o cputime.o ++ifdef CONFIG_SCHED_MUQSS ++obj-y += MuQSS.o clock.o ++else ++obj-y += core.o loadavg.o clock.o + obj-y += idle_task.o fair.o rt.o deadline.o +-obj-y += wait.o wait_bit.o swait.o completion.o idle.o +-obj-$(CONFIG_SMP) += cpupri.o cpudeadline.o topology.o stop_task.o ++obj-$(CONFIG_SMP) += cpudeadline.o stop_task.o + obj-$(CONFIG_SCHED_AUTOGROUP) += autogroup.o +-obj-$(CONFIG_SCHEDSTATS) += stats.o + obj-$(CONFIG_SCHED_DEBUG) += debug.o + obj-$(CONFIG_CGROUP_CPUACCT) += cpuacct.o ++endif ++obj-y += cputime.o ++obj-y += wait.o wait_bit.o swait.o completion.o idle.o ++obj-$(CONFIG_SMP) += cpupri.o topology.o ++obj-$(CONFIG_SCHEDSTATS) += stats.o + obj-$(CONFIG_CPU_FREQ) += cpufreq.o + obj-$(CONFIG_CPU_FREQ_GOV_SCHEDUTIL) += cpufreq_schedutil.o + obj-$(CONFIG_MEMBARRIER) += membarrier.o +diff -Nur a/kernel/sched/MuQSS.c b/kernel/sched/MuQSS.c +--- a/kernel/sched/MuQSS.c 1970-01-01 01:00:00.000000000 +0100 ++++ b/kernel/sched/MuQSS.c 2018-11-03 16:06:32.715529032 +0000 +@@ -0,0 +1,6923 @@ ++// SPDX-License-Identifier: GPL-2.0 ++/* ++ * kernel/sched/MuQSS.c, was kernel/sched.c ++ * ++ * Kernel scheduler and related syscalls ++ * ++ * Copyright (C) 1991-2002 Linus Torvalds ++ * ++ * 1996-12-23 Modified by Dave Grothe to fix bugs in semaphores and ++ * make semaphores SMP safe ++ * 1998-11-19 Implemented schedule_timeout() and related stuff ++ * by Andrea Arcangeli ++ * 2002-01-04 New ultra-scalable O(1) scheduler by Ingo Molnar: ++ * hybrid priority-list and round-robin design with ++ * an array-switch method of distributing timeslices ++ * and per-CPU runqueues. Cleanups and useful suggestions ++ * by Davide Libenzi, preemptible kernel bits by Robert Love. ++ * 2003-09-03 Interactivity tuning by Con Kolivas. ++ * 2004-04-02 Scheduler domains code by Nick Piggin ++ * 2007-04-15 Work begun on replacing all interactivity tuning with a ++ * fair scheduling design by Con Kolivas. ++ * 2007-05-05 Load balancing (smp-nice) and other improvements ++ * by Peter Williams ++ * 2007-05-06 Interactivity improvements to CFS by Mike Galbraith ++ * 2007-07-01 Group scheduling enhancements by Srivatsa Vaddagiri ++ * 2007-11-29 RT balancing improvements by Steven Rostedt, Gregory Haskins, ++ * Thomas Gleixner, Mike Kravetz ++ * 2009-08-13 Brainfuck deadline scheduling policy by Con Kolivas deletes ++ * a whole lot of those previous things. ++ * 2016-10-01 Multiple Queue Skiplist Scheduler scalable evolution of BFS ++ * scheduler by Con Kolivas. ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++#include ++#ifdef CONFIG_PARAVIRT ++#include ++#endif ++ ++#include "../workqueue_internal.h" ++#include "../smpboot.h" ++ ++#define CREATE_TRACE_POINTS ++#include ++ ++#include "MuQSS.h" ++ ++#define rt_prio(prio) unlikely((prio) < MAX_RT_PRIO) ++#define rt_task(p) rt_prio((p)->prio) ++#define batch_task(p) (unlikely((p)->policy == SCHED_BATCH)) ++#define is_rt_policy(policy) ((policy) == SCHED_FIFO || \ ++ (policy) == SCHED_RR) ++#define has_rt_policy(p) unlikely(is_rt_policy((p)->policy)) ++ ++#define is_idle_policy(policy) ((policy) == SCHED_IDLEPRIO) ++#define idleprio_task(p) unlikely(is_idle_policy((p)->policy)) ++#define task_running_idle(p) unlikely((p)->prio == IDLE_PRIO) ++ ++#define is_iso_policy(policy) ((policy) == SCHED_ISO) ++#define iso_task(p) unlikely(is_iso_policy((p)->policy)) ++#define task_running_iso(p) unlikely((p)->prio == ISO_PRIO) ++ ++#define rq_idle(rq) ((rq)->rq_prio == PRIO_LIMIT) ++ ++#define ISO_PERIOD (5 * HZ) ++ ++#define STOP_PRIO (MAX_RT_PRIO - 1) ++ ++/* ++ * Some helpers for converting to/from various scales. Use shifts to get ++ * approximate multiples of ten for less overhead. ++ */ ++#define JIFFIES_TO_NS(TIME) ((TIME) * (1073741824 / HZ)) ++#define JIFFY_NS (1073741824 / HZ) ++#define JIFFY_US (1048576 / HZ) ++#define NS_TO_JIFFIES(TIME) ((TIME) / JIFFY_NS) ++#define HALF_JIFFY_NS (1073741824 / HZ / 2) ++#define HALF_JIFFY_US (1048576 / HZ / 2) ++#define MS_TO_NS(TIME) ((TIME) << 20) ++#define MS_TO_US(TIME) ((TIME) << 10) ++#define NS_TO_MS(TIME) ((TIME) >> 20) ++#define NS_TO_US(TIME) ((TIME) >> 10) ++#define US_TO_NS(TIME) ((TIME) << 10) ++ ++#define RESCHED_US (100) /* Reschedule if less than this many μs left */ ++ ++void print_scheduler_version(void) ++{ ++ printk(KERN_INFO "MuQSS CPU scheduler v0.162 by Con Kolivas.\n"); ++} ++ ++/* ++ * This is the time all tasks within the same priority round robin. ++ * Value is in ms and set to a minimum of 6ms. ++ * Tunable via /proc interface. ++ */ ++int rr_interval __read_mostly = 6; ++ ++/* ++ * Tunable to choose whether to prioritise latency or throughput, simple ++ * binary yes or no ++ */ ++int sched_interactive __read_mostly = 1; ++ ++/* ++ * sched_iso_cpu - sysctl which determines the cpu percentage SCHED_ISO tasks ++ * are allowed to run five seconds as real time tasks. This is the total over ++ * all online cpus. ++ */ ++int sched_iso_cpu __read_mostly = 70; ++ ++/* ++ * sched_yield_type - Choose what sort of yield sched_yield will perform. ++ * 0: No yield. ++ * 1: Yield only to better priority/deadline tasks. (default) ++ * 2: Expire timeslice and recalculate deadline. ++ */ ++int sched_yield_type __read_mostly = 1; ++ ++/* ++ * The relative length of deadline for each priority(nice) level. ++ */ ++static int prio_ratios[NICE_WIDTH] __read_mostly; ++ ++/* ++ * The quota handed out to tasks of all priority levels when refilling their ++ * time_slice. ++ */ ++static inline int timeslice(void) ++{ ++ return MS_TO_US(rr_interval); ++} ++ ++#ifdef CONFIG_SMP ++static cpumask_t cpu_idle_map ____cacheline_aligned_in_smp; ++#endif ++ ++/* CPUs with isolated domains */ ++cpumask_var_t cpu_isolated_map; ++ ++DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues); ++#ifdef CONFIG_SMP ++struct rq *cpu_rq(int cpu) ++{ ++ return &per_cpu(runqueues, (cpu)); ++} ++#define cpu_curr(cpu) (cpu_rq(cpu)->curr) ++ ++/* ++ * For asym packing, by default the lower numbered cpu has higher priority. ++ */ ++int __weak arch_asym_cpu_priority(int cpu) ++{ ++ return -cpu; ++} ++ ++int __weak arch_sd_sibling_asym_packing(void) ++{ ++ return 0*SD_ASYM_PACKING; ++} ++#else ++struct rq *uprq; ++#endif /* CONFIG_SMP */ ++ ++#include "stats.h" ++ ++#ifndef prepare_arch_switch ++# define prepare_arch_switch(next) do { } while (0) ++#endif ++#ifndef finish_arch_switch ++# define finish_arch_switch(prev) do { } while (0) ++#endif ++#ifndef finish_arch_post_lock_switch ++# define finish_arch_post_lock_switch() do { } while (0) ++#endif ++ ++/* ++ * All common locking functions performed on rq->lock. rq->clock is local to ++ * the CPU accessing it so it can be modified just with interrupts disabled ++ * when we're not updating niffies. ++ * Looking up task_rq must be done under rq->lock to be safe. ++ */ ++ ++/* ++ * RQ-clock updating methods: ++ */ ++ ++static void update_rq_clock_task(struct rq *rq, s64 delta) ++{ ++/* ++ * In theory, the compile should just see 0 here, and optimize out the call ++ * to sched_rt_avg_update. But I don't trust it... ++ */ ++#ifdef CONFIG_IRQ_TIME_ACCOUNTING ++ s64 irq_delta = irq_time_read(cpu_of(rq)) - rq->prev_irq_time; ++ ++ /* ++ * Since irq_time is only updated on {soft,}irq_exit, we might run into ++ * this case when a previous update_rq_clock() happened inside a ++ * {soft,}irq region. ++ * ++ * When this happens, we stop ->clock_task and only update the ++ * prev_irq_time stamp to account for the part that fit, so that a next ++ * update will consume the rest. This ensures ->clock_task is ++ * monotonic. ++ * ++ * It does however cause some slight miss-attribution of {soft,}irq ++ * time, a more accurate solution would be to update the irq_time using ++ * the current rq->clock timestamp, except that would require using ++ * atomic ops. ++ */ ++ if (irq_delta > delta) ++ irq_delta = delta; ++ ++ rq->prev_irq_time += irq_delta; ++ delta -= irq_delta; ++#endif ++#ifdef CONFIG_PARAVIRT_TIME_ACCOUNTING ++ if (static_key_false((¶virt_steal_rq_enabled))) { ++ s64 steal = paravirt_steal_clock(cpu_of(rq)); ++ ++ steal -= rq->prev_steal_time_rq; ++ ++ if (unlikely(steal > delta)) ++ steal = delta; ++ ++ rq->prev_steal_time_rq += steal; ++ ++ delta -= steal; ++ } ++#endif ++ rq->clock_task += delta; ++} ++ ++static inline void update_rq_clock(struct rq *rq) ++{ ++ s64 delta = sched_clock_cpu(cpu_of(rq)) - rq->clock; ++ ++ if (unlikely(delta < 0)) ++ return; ++ rq->clock += delta; ++ update_rq_clock_task(rq, delta); ++} ++ ++/* ++ * Niffies are a globally increasing nanosecond counter. They're only used by ++ * update_load_avg and time_slice_expired, however deadlines are based on them ++ * across CPUs. Update them whenever we will call one of those functions, and ++ * synchronise them across CPUs whenever we hold both runqueue locks. ++ */ ++static inline void update_clocks(struct rq *rq) ++{ ++ s64 ndiff, minndiff; ++ long jdiff; ++ ++ update_rq_clock(rq); ++ ndiff = rq->clock - rq->old_clock; ++ rq->old_clock = rq->clock; ++ jdiff = jiffies - rq->last_jiffy; ++ ++ /* Subtract any niffies added by balancing with other rqs */ ++ ndiff -= rq->niffies - rq->last_niffy; ++ minndiff = JIFFIES_TO_NS(jdiff) - rq->niffies + rq->last_jiffy_niffies; ++ if (minndiff < 0) ++ minndiff = 0; ++ ndiff = max(ndiff, minndiff); ++ rq->niffies += ndiff; ++ rq->last_niffy = rq->niffies; ++ if (jdiff) { ++ rq->last_jiffy += jdiff; ++ rq->last_jiffy_niffies = rq->niffies; ++ } ++} ++ ++static inline int task_on_rq_queued(struct task_struct *p) ++{ ++ return p->on_rq == TASK_ON_RQ_QUEUED; ++} ++ ++static inline int task_on_rq_migrating(struct task_struct *p) ++{ ++ return p->on_rq == TASK_ON_RQ_MIGRATING; ++} ++ ++static inline int rq_trylock(struct rq *rq) ++ __acquires(rq->lock) ++{ ++ return raw_spin_trylock(&rq->lock); ++} ++ ++/* ++ * Any time we have two runqueues locked we use that as an opportunity to ++ * synchronise niffies to the highest value as idle ticks may have artificially ++ * kept niffies low on one CPU and the truth can only be later. ++ */ ++static inline void synchronise_niffies(struct rq *rq1, struct rq *rq2) ++{ ++ if (rq1->niffies > rq2->niffies) ++ rq2->niffies = rq1->niffies; ++ else ++ rq1->niffies = rq2->niffies; ++} ++ ++/* ++ * double_rq_lock - safely lock two runqueues ++ * ++ * Note this does not disable interrupts like task_rq_lock, ++ * you need to do so manually before calling. ++ */ ++ ++/* For when we know rq1 != rq2 */ ++static inline void __double_rq_lock(struct rq *rq1, struct rq *rq2) ++ __acquires(rq1->lock) ++ __acquires(rq2->lock) ++{ ++ if (rq1 < rq2) { ++ raw_spin_lock(&rq1->lock); ++ raw_spin_lock_nested(&rq2->lock, SINGLE_DEPTH_NESTING); ++ } else { ++ raw_spin_lock(&rq2->lock); ++ raw_spin_lock_nested(&rq1->lock, SINGLE_DEPTH_NESTING); ++ } ++} ++ ++static inline void double_rq_lock(struct rq *rq1, struct rq *rq2) ++ __acquires(rq1->lock) ++ __acquires(rq2->lock) ++{ ++ BUG_ON(!irqs_disabled()); ++ if (rq1 == rq2) { ++ raw_spin_lock(&rq1->lock); ++ __acquire(rq2->lock); /* Fake it out ;) */ ++ } else ++ __double_rq_lock(rq1, rq2); ++ synchronise_niffies(rq1, rq2); ++} ++ ++/* ++ * double_rq_unlock - safely unlock two runqueues ++ * ++ * Note this does not restore interrupts like task_rq_unlock, ++ * you need to do so manually after calling. ++ */ ++static inline void double_rq_unlock(struct rq *rq1, struct rq *rq2) ++ __releases(rq1->lock) ++ __releases(rq2->lock) ++{ ++ raw_spin_unlock(&rq1->lock); ++ if (rq1 != rq2) ++ raw_spin_unlock(&rq2->lock); ++ else ++ __release(rq2->lock); ++} ++ ++static inline void lock_all_rqs(void) ++{ ++ int cpu; ++ ++ preempt_disable(); ++ for_each_possible_cpu(cpu) { ++ struct rq *rq = cpu_rq(cpu); ++ ++ do_raw_spin_lock(&rq->lock); ++ } ++} ++ ++static inline void unlock_all_rqs(void) ++{ ++ int cpu; ++ ++ for_each_possible_cpu(cpu) { ++ struct rq *rq = cpu_rq(cpu); ++ ++ do_raw_spin_unlock(&rq->lock); ++ } ++ preempt_enable(); ++} ++ ++/* Specially nest trylock an rq */ ++static inline bool trylock_rq(struct rq *this_rq, struct rq *rq) ++{ ++ if (unlikely(!do_raw_spin_trylock(&rq->lock))) ++ return false; ++ spin_acquire(&rq->lock.dep_map, SINGLE_DEPTH_NESTING, 1, _RET_IP_); ++ synchronise_niffies(this_rq, rq); ++ return true; ++} ++ ++/* Unlock a specially nested trylocked rq */ ++static inline void unlock_rq(struct rq *rq) ++{ ++ spin_release(&rq->lock.dep_map, 1, _RET_IP_); ++ do_raw_spin_unlock(&rq->lock); ++} ++ ++/* ++ * cmpxchg based fetch_or, macro so it works for different integer types ++ */ ++#define fetch_or(ptr, mask) \ ++ ({ \ ++ typeof(ptr) _ptr = (ptr); \ ++ typeof(mask) _mask = (mask); \ ++ typeof(*_ptr) _old, _val = *_ptr; \ ++ \ ++ for (;;) { \ ++ _old = cmpxchg(_ptr, _val, _val | _mask); \ ++ if (_old == _val) \ ++ break; \ ++ _val = _old; \ ++ } \ ++ _old; \ ++}) ++ ++#if defined(CONFIG_SMP) && defined(TIF_POLLING_NRFLAG) ++/* ++ * Atomically set TIF_NEED_RESCHED and test for TIF_POLLING_NRFLAG, ++ * this avoids any races wrt polling state changes and thereby avoids ++ * spurious IPIs. ++ */ ++static bool set_nr_and_not_polling(struct task_struct *p) ++{ ++ struct thread_info *ti = task_thread_info(p); ++ return !(fetch_or(&ti->flags, _TIF_NEED_RESCHED) & _TIF_POLLING_NRFLAG); ++} ++ ++/* ++ * Atomically set TIF_NEED_RESCHED if TIF_POLLING_NRFLAG is set. ++ * ++ * If this returns true, then the idle task promises to call ++ * sched_ttwu_pending() and reschedule soon. ++ */ ++static bool set_nr_if_polling(struct task_struct *p) ++{ ++ struct thread_info *ti = task_thread_info(p); ++ typeof(ti->flags) old, val = READ_ONCE(ti->flags); ++ ++ for (;;) { ++ if (!(val & _TIF_POLLING_NRFLAG)) ++ return false; ++ if (val & _TIF_NEED_RESCHED) ++ return true; ++ old = cmpxchg(&ti->flags, val, val | _TIF_NEED_RESCHED); ++ if (old == val) ++ break; ++ val = old; ++ } ++ return true; ++} ++ ++#else ++static bool set_nr_and_not_polling(struct task_struct *p) ++{ ++ set_tsk_need_resched(p); ++ return true; ++} ++ ++#ifdef CONFIG_SMP ++static bool set_nr_if_polling(struct task_struct *p) ++{ ++ return false; ++} ++#endif ++#endif ++ ++void wake_q_add(struct wake_q_head *head, struct task_struct *task) ++{ ++ struct wake_q_node *node = &task->wake_q; ++ ++ /* ++ * Atomically grab the task, if ->wake_q is !nil already it means ++ * its already queued (either by us or someone else) and will get the ++ * wakeup due to that. ++ * ++ * This cmpxchg() implies a full barrier, which pairs with the write ++ * barrier implied by the wakeup in wake_up_q(). ++ */ ++ if (cmpxchg(&node->next, NULL, WAKE_Q_TAIL)) ++ return; ++ ++ get_task_struct(task); ++ ++ /* ++ * The head is context local, there can be no concurrency. ++ */ ++ *head->lastp = node; ++ head->lastp = &node->next; ++} ++ ++void wake_up_q(struct wake_q_head *head) ++{ ++ struct wake_q_node *node = head->first; ++ ++ while (node != WAKE_Q_TAIL) { ++ struct task_struct *task; ++ ++ task = container_of(node, struct task_struct, wake_q); ++ BUG_ON(!task); ++ /* Task can safely be re-inserted now */ ++ node = node->next; ++ task->wake_q.next = NULL; ++ ++ /* ++ * wake_up_process() implies a wmb() to pair with the queueing ++ * in wake_q_add() so as not to miss wakeups. ++ */ ++ wake_up_process(task); ++ put_task_struct(task); ++ } ++} ++ ++static inline void prepare_lock_switch(struct rq *rq, struct task_struct *next) ++{ ++ next->on_cpu = 1; ++} ++ ++static inline void smp_sched_reschedule(int cpu) ++{ ++ if (likely(cpu_online(cpu))) ++ smp_send_reschedule(cpu); ++} ++ ++/* ++ * resched_task - mark a task 'to be rescheduled now'. ++ * ++ * On UP this means the setting of the need_resched flag, on SMP it ++ * might also involve a cross-CPU call to trigger the scheduler on ++ * the target CPU. ++ */ ++void resched_task(struct task_struct *p) ++{ ++ int cpu; ++#ifdef CONFIG_LOCKDEP ++ /* Kernel threads call this when creating workqueues while still ++ * inactive from __kthread_bind_mask, holding only the pi_lock */ ++ if (!(p->flags & PF_KTHREAD)) { ++ struct rq *rq = task_rq(p); ++ ++ lockdep_assert_held(&rq->lock); ++ } ++#endif ++ if (test_tsk_need_resched(p)) ++ return; ++ ++ cpu = task_cpu(p); ++ if (cpu == smp_processor_id()) { ++ set_tsk_need_resched(p); ++ set_preempt_need_resched(); ++ return; ++ } ++ ++ if (set_nr_and_not_polling(p)) ++ smp_sched_reschedule(cpu); ++ else ++ trace_sched_wake_idle_without_ipi(cpu); ++} ++ ++/* ++ * A task that is not running or queued will not have a node set. ++ * A task that is queued but not running will have a node set. ++ * A task that is currently running will have ->on_cpu set but no node set. ++ */ ++static inline bool task_queued(struct task_struct *p) ++{ ++ return !skiplist_node_empty(&p->node); ++} ++ ++static void enqueue_task(struct rq *rq, struct task_struct *p, int flags); ++static inline void resched_if_idle(struct rq *rq); ++ ++/* Dodgy workaround till we figure out where the softirqs are going */ ++static inline void do_pending_softirq(struct rq *rq, struct task_struct *next) ++{ ++ if (unlikely(next == rq->idle && local_softirq_pending() && !in_interrupt())) ++ do_softirq_own_stack(); ++} ++ ++static inline void finish_lock_switch(struct rq *rq, struct task_struct *prev) ++{ ++#ifdef CONFIG_SMP ++ /* ++ * After ->on_cpu is cleared, the task can be moved to a different CPU. ++ * We must ensure this doesn't happen until the switch is completely ++ * finished. ++ * ++ * In particular, the load of prev->state in finish_task_switch() must ++ * happen before this. ++ * ++ * Pairs with the smp_cond_load_acquire() in try_to_wake_up(). ++ */ ++ smp_store_release(&prev->on_cpu, 0); ++#endif ++#ifdef CONFIG_DEBUG_SPINLOCK ++ /* this is a valid case when another task releases the spinlock */ ++ rq->lock.owner = current; ++#endif ++ /* ++ * If we are tracking spinlock dependencies then we have to ++ * fix up the runqueue lock - which gets 'carried over' from ++ * prev into current: ++ */ ++ spin_acquire(&rq->lock.dep_map, 0, 0, _THIS_IP_); ++ ++#ifdef CONFIG_SMP ++ /* ++ * If prev was marked as migrating to another CPU in return_task, drop ++ * the local runqueue lock but leave interrupts disabled and grab the ++ * remote lock we're migrating it to before enabling them. ++ */ ++ if (unlikely(task_on_rq_migrating(prev))) { ++ sched_info_dequeued(rq, prev); ++ /* ++ * We move the ownership of prev to the new cpu now. ttwu can't ++ * activate prev to the wrong cpu since it has to grab this ++ * runqueue in ttwu_remote. ++ */ ++#ifdef CONFIG_THREAD_INFO_IN_TASK ++ prev->cpu = prev->wake_cpu; ++#else ++ task_thread_info(prev)->cpu = prev->wake_cpu; ++#endif ++ raw_spin_unlock(&rq->lock); ++ ++ raw_spin_lock(&prev->pi_lock); ++ rq = __task_rq_lock(prev); ++ /* Check that someone else hasn't already queued prev */ ++ if (likely(!task_queued(prev))) { ++ enqueue_task(rq, prev, 0); ++ prev->on_rq = TASK_ON_RQ_QUEUED; ++ /* Wake up the CPU if it's not already running */ ++ resched_if_idle(rq); ++ } ++ raw_spin_unlock(&prev->pi_lock); ++ } ++#endif ++ /* Accurately set nr_running here for load average calculations */ ++ rq->nr_running = rq->sl->entries + !rq_idle(rq); ++ rq_unlock(rq); ++ ++ do_pending_softirq(rq, current); ++ ++ local_irq_enable(); ++} ++ ++static inline bool deadline_before(u64 deadline, u64 time) ++{ ++ return (deadline < time); ++} ++ ++/* ++ * Deadline is "now" in niffies + (offset by priority). Setting the deadline ++ * is the key to everything. It distributes cpu fairly amongst tasks of the ++ * same nice value, it proportions cpu according to nice level, it means the ++ * task that last woke up the longest ago has the earliest deadline, thus ++ * ensuring that interactive tasks get low latency on wake up. The CPU ++ * proportion works out to the square of the virtual deadline difference, so ++ * this equation will give nice 19 3% CPU compared to nice 0. ++ */ ++static inline u64 prio_deadline_diff(int user_prio) ++{ ++ return (prio_ratios[user_prio] * rr_interval * (MS_TO_NS(1) / 128)); ++} ++ ++static inline u64 task_deadline_diff(struct task_struct *p) ++{ ++ return prio_deadline_diff(TASK_USER_PRIO(p)); ++} ++ ++static inline u64 static_deadline_diff(int static_prio) ++{ ++ return prio_deadline_diff(USER_PRIO(static_prio)); ++} ++ ++static inline int longest_deadline_diff(void) ++{ ++ return prio_deadline_diff(39); ++} ++ ++static inline int ms_longest_deadline_diff(void) ++{ ++ return NS_TO_MS(longest_deadline_diff()); ++} ++ ++static inline bool rq_local(struct rq *rq); ++ ++#ifndef SCHED_CAPACITY_SCALE ++#define SCHED_CAPACITY_SCALE 1024 ++#endif ++ ++static inline int rq_load(struct rq *rq) ++{ ++ return rq->nr_running; ++} ++ ++/* ++ * Update the load average for feeding into cpu frequency governors. Use a ++ * rough estimate of a rolling average with ~ time constant of 32ms. ++ * 80/128 ~ 0.63. * 80 / 32768 / 128 == * 5 / 262144 ++ * Make sure a call to update_clocks has been made before calling this to get ++ * an updated rq->niffies. ++ */ ++static void update_load_avg(struct rq *rq, unsigned int flags) ++{ ++ unsigned long us_interval, curload; ++ long load; ++ ++ if (unlikely(rq->niffies <= rq->load_update)) ++ return; ++ ++ us_interval = NS_TO_US(rq->niffies - rq->load_update); ++ curload = rq_load(rq); ++ load = rq->load_avg - (rq->load_avg * us_interval * 5 / 262144); ++ if (unlikely(load < 0)) ++ load = 0; ++ load += curload * curload * SCHED_CAPACITY_SCALE * us_interval * 5 / 262144; ++ rq->load_avg = load; ++ ++ rq->load_update = rq->niffies; ++ if (likely(rq_local(rq))) ++ cpufreq_trigger(rq, flags); ++} ++ ++/* ++ * Removing from the runqueue. Enter with rq locked. Deleting a task ++ * from the skip list is done via the stored node reference in the task struct ++ * and does not require a full look up. Thus it occurs in O(k) time where k ++ * is the "level" of the list the task was stored at - usually < 4, max 8. ++ */ ++static void dequeue_task(struct rq *rq, struct task_struct *p, int flags) ++{ ++ skiplist_delete(rq->sl, &p->node); ++ rq->best_key = rq->node.next[0]->key; ++ update_clocks(rq); ++ ++ if (!(flags & DEQUEUE_SAVE)) ++ sched_info_dequeued(task_rq(p), p); ++ update_load_avg(rq, flags); ++} ++ ++#ifdef CONFIG_PREEMPT_RCU ++static bool rcu_read_critical(struct task_struct *p) ++{ ++ return p->rcu_read_unlock_special.b.blocked; ++} ++#else /* CONFIG_PREEMPT_RCU */ ++#define rcu_read_critical(p) (false) ++#endif /* CONFIG_PREEMPT_RCU */ ++ ++/* ++ * To determine if it's safe for a task of SCHED_IDLEPRIO to actually run as ++ * an idle task, we ensure none of the following conditions are met. ++ */ ++static bool idleprio_suitable(struct task_struct *p) ++{ ++ return (!(task_contributes_to_load(p)) && !(p->flags & (PF_EXITING)) && ++ !signal_pending(p) && !rcu_read_critical(p) && !freezing(p)); ++} ++ ++/* ++ * To determine if a task of SCHED_ISO can run in pseudo-realtime, we check ++ * that the iso_refractory flag is not set. ++ */ ++static inline bool isoprio_suitable(struct rq *rq) ++{ ++ return !rq->iso_refractory; ++} ++ ++/* ++ * Adding to the runqueue. Enter with rq locked. ++ */ ++static void enqueue_task(struct rq *rq, struct task_struct *p, int flags) ++{ ++ unsigned int randseed, cflags = 0; ++ u64 sl_id; ++ ++ if (!rt_task(p)) { ++ /* Check it hasn't gotten rt from PI */ ++ if ((idleprio_task(p) && idleprio_suitable(p)) || ++ (iso_task(p) && isoprio_suitable(rq))) ++ p->prio = p->normal_prio; ++ else ++ p->prio = NORMAL_PRIO; ++ } ++ /* ++ * The sl_id key passed to the skiplist generates a sorted list. ++ * Realtime and sched iso tasks run FIFO so they only need be sorted ++ * according to priority. The skiplist will put tasks of the same ++ * key inserted later in FIFO order. Tasks of sched normal, batch ++ * and idleprio are sorted according to their deadlines. Idleprio ++ * tasks are offset by an impossibly large deadline value ensuring ++ * they get sorted into last positions, but still according to their ++ * own deadlines. This creates a "landscape" of skiplists running ++ * from priority 0 realtime in first place to the lowest priority ++ * idleprio tasks last. Skiplist insertion is an O(log n) process. ++ */ ++ if (p->prio <= ISO_PRIO) { ++ sl_id = p->prio; ++ cflags = SCHED_CPUFREQ_RT; ++ } else { ++ sl_id = p->deadline; ++ if (idleprio_task(p)) { ++ if (p->prio == IDLE_PRIO) ++ sl_id |= 0xF000000000000000; ++ else ++ sl_id += longest_deadline_diff(); ++ } ++ } ++ /* ++ * Some architectures don't have better than microsecond resolution ++ * so mask out ~microseconds as the random seed for skiplist insertion. ++ */ ++ update_clocks(rq); ++ if (!(flags & ENQUEUE_RESTORE)) ++ sched_info_queued(rq, p); ++ randseed = (rq->niffies >> 10) & 0xFFFFFFFF; ++ skiplist_insert(rq->sl, &p->node, sl_id, p, randseed); ++ rq->best_key = rq->node.next[0]->key; ++ if (p->in_iowait) ++ cflags |= SCHED_CPUFREQ_IOWAIT; ++ update_load_avg(rq, cflags); ++} ++ ++/* ++ * Returns the relative length of deadline all compared to the shortest ++ * deadline which is that of nice -20. ++ */ ++static inline int task_prio_ratio(struct task_struct *p) ++{ ++ return prio_ratios[TASK_USER_PRIO(p)]; ++} ++ ++/* ++ * task_timeslice - all tasks of all priorities get the exact same timeslice ++ * length. CPU distribution is handled by giving different deadlines to ++ * tasks of different priorities. Use 128 as the base value for fast shifts. ++ */ ++static inline int task_timeslice(struct task_struct *p) ++{ ++ return (rr_interval * task_prio_ratio(p) / 128); ++} ++ ++#ifdef CONFIG_SMP ++/* Entered with rq locked */ ++static inline void resched_if_idle(struct rq *rq) ++{ ++ if (rq_idle(rq)) ++ resched_task(rq->curr); ++} ++ ++static inline bool rq_local(struct rq *rq) ++{ ++ return (rq->cpu == smp_processor_id()); ++} ++#ifdef CONFIG_SMT_NICE ++static const cpumask_t *thread_cpumask(int cpu); ++ ++/* Find the best real time priority running on any SMT siblings of cpu and if ++ * none are running, the static priority of the best deadline task running. ++ * The lookups to the other runqueues is done lockless as the occasional wrong ++ * value would be harmless. */ ++static int best_smt_bias(struct rq *this_rq) ++{ ++ int other_cpu, best_bias = 0; ++ ++ for_each_cpu(other_cpu, &this_rq->thread_mask) { ++ struct rq *rq = cpu_rq(other_cpu); ++ ++ if (rq_idle(rq)) ++ continue; ++ if (unlikely(!rq->online)) ++ continue; ++ if (!rq->rq_mm) ++ continue; ++ if (likely(rq->rq_smt_bias > best_bias)) ++ best_bias = rq->rq_smt_bias; ++ } ++ return best_bias; ++} ++ ++static int task_prio_bias(struct task_struct *p) ++{ ++ if (rt_task(p)) ++ return 1 << 30; ++ else if (task_running_iso(p)) ++ return 1 << 29; ++ else if (task_running_idle(p)) ++ return 0; ++ return MAX_PRIO - p->static_prio; ++} ++ ++static bool smt_always_schedule(struct task_struct __maybe_unused *p, struct rq __maybe_unused *this_rq) ++{ ++ return true; ++} ++ ++static bool (*smt_schedule)(struct task_struct *p, struct rq *this_rq) = &smt_always_schedule; ++ ++/* We've already decided p can run on CPU, now test if it shouldn't for SMT ++ * nice reasons. */ ++static bool smt_should_schedule(struct task_struct *p, struct rq *this_rq) ++{ ++ int best_bias, task_bias; ++ ++ /* Kernel threads always run */ ++ if (unlikely(!p->mm)) ++ return true; ++ if (rt_task(p)) ++ return true; ++ if (!idleprio_suitable(p)) ++ return true; ++ best_bias = best_smt_bias(this_rq); ++ /* The smt siblings are all idle or running IDLEPRIO */ ++ if (best_bias < 1) ++ return true; ++ task_bias = task_prio_bias(p); ++ if (task_bias < 1) ++ return false; ++ if (task_bias >= best_bias) ++ return true; ++ /* Dither 25% cpu of normal tasks regardless of nice difference */ ++ if (best_bias % 4 == 1) ++ return true; ++ /* Sorry, you lose */ ++ return false; ++} ++#else /* CONFIG_SMT_NICE */ ++#define smt_schedule(p, this_rq) (true) ++#endif /* CONFIG_SMT_NICE */ ++ ++static inline void atomic_set_cpu(int cpu, cpumask_t *cpumask) ++{ ++ set_bit(cpu, (volatile unsigned long *)cpumask); ++} ++ ++/* ++ * The cpu_idle_map stores a bitmap of all the CPUs currently idle to ++ * allow easy lookup of whether any suitable idle CPUs are available. ++ * It's cheaper to maintain a binary yes/no if there are any idle CPUs on the ++ * idle_cpus variable than to do a full bitmask check when we are busy. The ++ * bits are set atomically but read locklessly as occasional false positive / ++ * negative is harmless. ++ */ ++static inline void set_cpuidle_map(int cpu) ++{ ++ if (likely(cpu_online(cpu))) ++ atomic_set_cpu(cpu, &cpu_idle_map); ++} ++ ++static inline void atomic_clear_cpu(int cpu, cpumask_t *cpumask) ++{ ++ clear_bit(cpu, (volatile unsigned long *)cpumask); ++} ++ ++static inline void clear_cpuidle_map(int cpu) ++{ ++ atomic_clear_cpu(cpu, &cpu_idle_map); ++} ++ ++static bool suitable_idle_cpus(struct task_struct *p) ++{ ++ return (cpumask_intersects(&p->cpus_allowed, &cpu_idle_map)); ++} ++ ++/* ++ * Resched current on rq. We don't know if rq is local to this CPU nor if it ++ * is locked so we do not use an intermediate variable for the task to avoid ++ * having it dereferenced. ++ */ ++static void resched_curr(struct rq *rq) ++{ ++ int cpu; ++ ++ if (test_tsk_need_resched(rq->curr)) ++ return; ++ ++ rq->preempt = rq->curr; ++ cpu = rq->cpu; ++ ++ /* We're doing this without holding the rq lock if it's not task_rq */ ++ ++ if (cpu == smp_processor_id()) { ++ set_tsk_need_resched(rq->curr); ++ set_preempt_need_resched(); ++ return; ++ } ++ ++ if (set_nr_and_not_polling(rq->curr)) ++ smp_sched_reschedule(cpu); ++ else ++ trace_sched_wake_idle_without_ipi(cpu); ++} ++ ++#define CPUIDLE_DIFF_THREAD (1) ++#define CPUIDLE_DIFF_CORE (2) ++#define CPUIDLE_CACHE_BUSY (4) ++#define CPUIDLE_DIFF_CPU (8) ++#define CPUIDLE_THREAD_BUSY (16) ++#define CPUIDLE_DIFF_NODE (32) ++ ++/* ++ * The best idle CPU is chosen according to the CPUIDLE ranking above where the ++ * lowest value would give the most suitable CPU to schedule p onto next. The ++ * order works out to be the following: ++ * ++ * Same thread, idle or busy cache, idle or busy threads ++ * Other core, same cache, idle or busy cache, idle threads. ++ * Same node, other CPU, idle cache, idle threads. ++ * Same node, other CPU, busy cache, idle threads. ++ * Other core, same cache, busy threads. ++ * Same node, other CPU, busy threads. ++ * Other node, other CPU, idle cache, idle threads. ++ * Other node, other CPU, busy cache, idle threads. ++ * Other node, other CPU, busy threads. ++ */ ++static int best_mask_cpu(int best_cpu, struct rq *rq, cpumask_t *tmpmask) ++{ ++ int best_ranking = CPUIDLE_DIFF_NODE | CPUIDLE_THREAD_BUSY | ++ CPUIDLE_DIFF_CPU | CPUIDLE_CACHE_BUSY | CPUIDLE_DIFF_CORE | ++ CPUIDLE_DIFF_THREAD; ++ int cpu_tmp; ++ ++ if (cpumask_test_cpu(best_cpu, tmpmask)) ++ goto out; ++ ++ for_each_cpu(cpu_tmp, tmpmask) { ++ int ranking, locality; ++ struct rq *tmp_rq; ++ ++ ranking = 0; ++ tmp_rq = cpu_rq(cpu_tmp); ++ ++ locality = rq->cpu_locality[cpu_tmp]; ++#ifdef CONFIG_NUMA ++ if (locality > 3) ++ ranking |= CPUIDLE_DIFF_NODE; ++ else ++#endif ++ if (locality > 2) ++ ranking |= CPUIDLE_DIFF_CPU; ++#ifdef CONFIG_SCHED_MC ++ else if (locality == 2) ++ ranking |= CPUIDLE_DIFF_CORE; ++ else if (!(tmp_rq->cache_idle(tmp_rq))) ++ ranking |= CPUIDLE_CACHE_BUSY; ++#endif ++#ifdef CONFIG_SCHED_SMT ++ if (locality == 1) ++ ranking |= CPUIDLE_DIFF_THREAD; ++ if (!(tmp_rq->siblings_idle(tmp_rq))) ++ ranking |= CPUIDLE_THREAD_BUSY; ++#endif ++ if (ranking < best_ranking) { ++ best_cpu = cpu_tmp; ++ best_ranking = ranking; ++ } ++ } ++out: ++ return best_cpu; ++} ++ ++bool cpus_share_cache(int this_cpu, int that_cpu) ++{ ++ struct rq *this_rq = cpu_rq(this_cpu); ++ ++ return (this_rq->cpu_locality[that_cpu] < 3); ++} ++ ++/* As per resched_curr but only will resched idle task */ ++static inline void resched_idle(struct rq *rq) ++{ ++ if (test_tsk_need_resched(rq->idle)) ++ return; ++ ++ rq->preempt = rq->idle; ++ ++ set_tsk_need_resched(rq->idle); ++ ++ if (rq_local(rq)) { ++ set_preempt_need_resched(); ++ return; ++ } ++ ++ smp_sched_reschedule(rq->cpu); ++} ++ ++static struct rq *resched_best_idle(struct task_struct *p, int cpu) ++{ ++ cpumask_t tmpmask; ++ struct rq *rq; ++ int best_cpu; ++ ++ cpumask_and(&tmpmask, &p->cpus_allowed, &cpu_idle_map); ++ best_cpu = best_mask_cpu(cpu, task_rq(p), &tmpmask); ++ rq = cpu_rq(best_cpu); ++ if (!smt_schedule(p, rq)) ++ return NULL; ++ rq->preempt = p; ++ resched_idle(rq); ++ return rq; ++} ++ ++static inline void resched_suitable_idle(struct task_struct *p) ++{ ++ if (suitable_idle_cpus(p)) ++ resched_best_idle(p, task_cpu(p)); ++} ++ ++static inline struct rq *rq_order(struct rq *rq, int cpu) ++{ ++ return rq->rq_order[cpu]; ++} ++#else /* CONFIG_SMP */ ++static inline void set_cpuidle_map(int cpu) ++{ ++} ++ ++static inline void clear_cpuidle_map(int cpu) ++{ ++} ++ ++static inline bool suitable_idle_cpus(struct task_struct *p) ++{ ++ return uprq->curr == uprq->idle; ++} ++ ++static inline void resched_suitable_idle(struct task_struct *p) ++{ ++} ++ ++static inline void resched_curr(struct rq *rq) ++{ ++ resched_task(rq->curr); ++} ++ ++static inline void resched_if_idle(struct rq *rq) ++{ ++} ++ ++static inline bool rq_local(struct rq *rq) ++{ ++ return true; ++} ++ ++static inline struct rq *rq_order(struct rq *rq, int cpu) ++{ ++ return rq; ++} ++ ++static inline bool smt_schedule(struct task_struct *p, struct rq *rq) ++{ ++ return true; ++} ++#endif /* CONFIG_SMP */ ++ ++static inline int normal_prio(struct task_struct *p) ++{ ++ if (has_rt_policy(p)) ++ return MAX_RT_PRIO - 1 - p->rt_priority; ++ if (idleprio_task(p)) ++ return IDLE_PRIO; ++ if (iso_task(p)) ++ return ISO_PRIO; ++ return NORMAL_PRIO; ++} ++ ++/* ++ * Calculate the current priority, i.e. the priority ++ * taken into account by the scheduler. This value might ++ * be boosted by RT tasks as it will be RT if the task got ++ * RT-boosted. If not then it returns p->normal_prio. ++ */ ++static int effective_prio(struct task_struct *p) ++{ ++ p->normal_prio = normal_prio(p); ++ /* ++ * If we are RT tasks or we were boosted to RT priority, ++ * keep the priority unchanged. Otherwise, update priority ++ * to the normal priority: ++ */ ++ if (!rt_prio(p->prio)) ++ return p->normal_prio; ++ return p->prio; ++} ++ ++/* ++ * activate_task - move a task to the runqueue. Enter with rq locked. ++ */ ++static void activate_task(struct task_struct *p, struct rq *rq) ++{ ++ resched_if_idle(rq); ++ ++ /* ++ * Sleep time is in units of nanosecs, so shift by 20 to get a ++ * milliseconds-range estimation of the amount of time that the task ++ * spent sleeping: ++ */ ++ if (unlikely(prof_on == SLEEP_PROFILING)) { ++ if (p->state == TASK_UNINTERRUPTIBLE) ++ profile_hits(SLEEP_PROFILING, (void *)get_wchan(p), ++ (rq->niffies - p->last_ran) >> 20); ++ } ++ ++ p->prio = effective_prio(p); ++ if (task_contributes_to_load(p)) ++ rq->nr_uninterruptible--; ++ ++ enqueue_task(rq, p, 0); ++ p->on_rq = TASK_ON_RQ_QUEUED; ++} ++ ++/* ++ * deactivate_task - If it's running, it's not on the runqueue and we can just ++ * decrement the nr_running. Enter with rq locked. ++ */ ++static inline void deactivate_task(struct task_struct *p, struct rq *rq) ++{ ++ if (task_contributes_to_load(p)) ++ rq->nr_uninterruptible++; ++ ++ p->on_rq = 0; ++ sched_info_dequeued(rq, p); ++} ++ ++#ifdef CONFIG_SMP ++void set_task_cpu(struct task_struct *p, unsigned int new_cpu) ++{ ++ struct rq *rq; ++ ++ if (task_cpu(p) == new_cpu) ++ return; ++ ++ /* Do NOT call set_task_cpu on a currently queued task as we will not ++ * be reliably holding the rq lock after changing CPU. */ ++ BUG_ON(task_queued(p)); ++ rq = task_rq(p); ++ ++#ifdef CONFIG_LOCKDEP ++ /* ++ * The caller should hold either p->pi_lock or rq->lock, when changing ++ * a task's CPU. ->pi_lock for waking tasks, rq->lock for runnable tasks. ++ * ++ * Furthermore, all task_rq users should acquire both locks, see ++ * task_rq_lock(). ++ */ ++ WARN_ON_ONCE(debug_locks && !(lockdep_is_held(&p->pi_lock) || ++ lockdep_is_held(&rq->lock))); ++#endif ++ ++ trace_sched_migrate_task(p, new_cpu); ++ perf_event_task_migrate(p); ++ ++ /* ++ * After ->cpu is set up to a new value, task_rq_lock(p, ...) can be ++ * successfully executed on another CPU. We must ensure that updates of ++ * per-task data have been completed by this moment. ++ */ ++ smp_wmb(); ++ ++ p->wake_cpu = new_cpu; ++ ++ if (task_running(rq, p)) { ++ /* ++ * We should only be calling this on a running task if we're ++ * holding rq lock. ++ */ ++ lockdep_assert_held(&rq->lock); ++ ++ /* ++ * We can't change the task_thread_info CPU on a running task ++ * as p will still be protected by the rq lock of the CPU it ++ * is still running on so we only set the wake_cpu for it to be ++ * lazily updated once off the CPU. ++ */ ++ return; ++ } ++ ++#ifdef CONFIG_THREAD_INFO_IN_TASK ++ p->cpu = new_cpu; ++#else ++ task_thread_info(p)->cpu = new_cpu; ++#endif ++ /* We're no longer protecting p after this point since we're holding ++ * the wrong runqueue lock. */ ++} ++#endif /* CONFIG_SMP */ ++ ++/* ++ * Move a task off the runqueue and take it to a cpu for it will ++ * become the running task. ++ */ ++static inline void take_task(struct rq *rq, int cpu, struct task_struct *p) ++{ ++ struct rq *p_rq = task_rq(p); ++ ++ dequeue_task(p_rq, p, DEQUEUE_SAVE); ++ if (p_rq != rq) { ++ sched_info_dequeued(p_rq, p); ++ sched_info_queued(rq, p); ++ } ++ set_task_cpu(p, cpu); ++} ++ ++/* ++ * Returns a descheduling task to the runqueue unless it is being ++ * deactivated. ++ */ ++static inline void return_task(struct task_struct *p, struct rq *rq, ++ int cpu, bool deactivate) ++{ ++ if (deactivate) ++ deactivate_task(p, rq); ++ else { ++#ifdef CONFIG_SMP ++ /* ++ * set_task_cpu was called on the running task that doesn't ++ * want to deactivate so it has to be enqueued to a different ++ * CPU and we need its lock. Tag it to be moved with as the ++ * lock is dropped in finish_lock_switch. ++ */ ++ if (unlikely(p->wake_cpu != cpu)) ++ p->on_rq = TASK_ON_RQ_MIGRATING; ++ else ++#endif ++ enqueue_task(rq, p, ENQUEUE_RESTORE); ++ } ++} ++ ++/* Enter with rq lock held. We know p is on the local cpu */ ++static inline void __set_tsk_resched(struct task_struct *p) ++{ ++ set_tsk_need_resched(p); ++ set_preempt_need_resched(); ++} ++ ++/** ++ * task_curr - is this task currently executing on a CPU? ++ * @p: the task in question. ++ * ++ * Return: 1 if the task is currently executing. 0 otherwise. ++ */ ++inline int task_curr(const struct task_struct *p) ++{ ++ return cpu_curr(task_cpu(p)) == p; ++} ++ ++#ifdef CONFIG_SMP ++/* ++ * wait_task_inactive - wait for a thread to unschedule. ++ * ++ * If @match_state is nonzero, it's the @p->state value just checked and ++ * not expected to change. If it changes, i.e. @p might have woken up, ++ * then return zero. When we succeed in waiting for @p to be off its CPU, ++ * we return a positive number (its total switch count). If a second call ++ * a short while later returns the same number, the caller can be sure that ++ * @p has remained unscheduled the whole time. ++ * ++ * The caller must ensure that the task *will* unschedule sometime soon, ++ * else this function might spin for a *long* time. This function can't ++ * be called with interrupts off, or it may introduce deadlock with ++ * smp_call_function() if an IPI is sent by the same process we are ++ * waiting to become inactive. ++ */ ++unsigned long wait_task_inactive(struct task_struct *p, long match_state) ++{ ++ int running, queued; ++ unsigned long flags; ++ unsigned long ncsw; ++ struct rq *rq; ++ ++ for (;;) { ++ rq = task_rq(p); ++ ++ /* ++ * If the task is actively running on another CPU ++ * still, just relax and busy-wait without holding ++ * any locks. ++ * ++ * NOTE! Since we don't hold any locks, it's not ++ * even sure that "rq" stays as the right runqueue! ++ * But we don't care, since this will return false ++ * if the runqueue has changed and p is actually now ++ * running somewhere else! ++ */ ++ while (task_running(rq, p)) { ++ if (match_state && unlikely(p->state != match_state)) ++ return 0; ++ cpu_relax(); ++ } ++ ++ /* ++ * Ok, time to look more closely! We need the rq ++ * lock now, to be *sure*. If we're wrong, we'll ++ * just go back and repeat. ++ */ ++ rq = task_rq_lock(p, &flags); ++ trace_sched_wait_task(p); ++ running = task_running(rq, p); ++ queued = task_on_rq_queued(p); ++ ncsw = 0; ++ if (!match_state || p->state == match_state) ++ ncsw = p->nvcsw | LONG_MIN; /* sets MSB */ ++ task_rq_unlock(rq, p, &flags); ++ ++ /* ++ * If it changed from the expected state, bail out now. ++ */ ++ if (unlikely(!ncsw)) ++ break; ++ ++ /* ++ * Was it really running after all now that we ++ * checked with the proper locks actually held? ++ * ++ * Oops. Go back and try again.. ++ */ ++ if (unlikely(running)) { ++ cpu_relax(); ++ continue; ++ } ++ ++ /* ++ * It's not enough that it's not actively running, ++ * it must be off the runqueue _entirely_, and not ++ * preempted! ++ * ++ * So if it was still runnable (but just not actively ++ * running right now), it's preempted, and we should ++ * yield - it could be a while. ++ */ ++ if (unlikely(queued)) { ++ ktime_t to = NSEC_PER_SEC / HZ; ++ ++ set_current_state(TASK_UNINTERRUPTIBLE); ++ schedule_hrtimeout(&to, HRTIMER_MODE_REL); ++ continue; ++ } ++ ++ /* ++ * Ahh, all good. It wasn't running, and it wasn't ++ * runnable, which means that it will never become ++ * running in the future either. We're all done! ++ */ ++ break; ++ } ++ ++ return ncsw; ++} ++ ++/*** ++ * kick_process - kick a running thread to enter/exit the kernel ++ * @p: the to-be-kicked thread ++ * ++ * Cause a process which is running on another CPU to enter ++ * kernel-mode, without any delay. (to get signals handled.) ++ * ++ * NOTE: this function doesn't have to take the runqueue lock, ++ * because all it wants to ensure is that the remote task enters ++ * the kernel. If the IPI races and the task has been migrated ++ * to another CPU then no harm is done and the purpose has been ++ * achieved as well. ++ */ ++void kick_process(struct task_struct *p) ++{ ++ int cpu; ++ ++ preempt_disable(); ++ cpu = task_cpu(p); ++ if ((cpu != smp_processor_id()) && task_curr(p)) ++ smp_sched_reschedule(cpu); ++ preempt_enable(); ++} ++EXPORT_SYMBOL_GPL(kick_process); ++#endif ++ ++/* ++ * RT tasks preempt purely on priority. SCHED_NORMAL tasks preempt on the ++ * basis of earlier deadlines. SCHED_IDLEPRIO don't preempt anything else or ++ * between themselves, they cooperatively multitask. An idle rq scores as ++ * prio PRIO_LIMIT so it is always preempted. ++ */ ++static inline bool ++can_preempt(struct task_struct *p, int prio, u64 deadline) ++{ ++ /* Better static priority RT task or better policy preemption */ ++ if (p->prio < prio) ++ return true; ++ if (p->prio > prio) ++ return false; ++ if (p->policy == SCHED_BATCH) ++ return false; ++ /* SCHED_NORMAL and ISO will preempt based on deadline */ ++ if (!deadline_before(p->deadline, deadline)) ++ return false; ++ return true; ++} ++ ++#ifdef CONFIG_SMP ++/* ++ * Check to see if p can run on cpu, and if not, whether there are any online ++ * CPUs it can run on instead. ++ */ ++static inline bool needs_other_cpu(struct task_struct *p, int cpu) ++{ ++ if (unlikely(!cpumask_test_cpu(cpu, &p->cpus_allowed))) ++ return true; ++ return false; ++} ++#define cpu_online_map (*(cpumask_t *)cpu_online_mask) ++ ++static void try_preempt(struct task_struct *p, struct rq *this_rq) ++{ ++ int i, this_entries = rq_load(this_rq); ++ cpumask_t tmp; ++ ++ if (suitable_idle_cpus(p) && resched_best_idle(p, task_cpu(p))) ++ return; ++ ++ /* IDLEPRIO tasks never preempt anything but idle */ ++ if (p->policy == SCHED_IDLEPRIO) ++ return; ++ ++ cpumask_and(&tmp, &cpu_online_map, &p->cpus_allowed); ++ ++ for (i = 0; i < num_possible_cpus(); i++) { ++ struct rq *rq = this_rq->rq_order[i]; ++ ++ if (!cpumask_test_cpu(rq->cpu, &tmp)) ++ continue; ++ ++ if (!sched_interactive && rq != this_rq && rq_load(rq) <= this_entries) ++ continue; ++ if (smt_schedule(p, rq) && can_preempt(p, rq->rq_prio, rq->rq_deadline)) { ++ /* We set rq->preempting lockless, it's a hint only */ ++ rq->preempting = p; ++ resched_curr(rq); ++ return; ++ } ++ } ++} ++ ++static int __set_cpus_allowed_ptr(struct task_struct *p, ++ const struct cpumask *new_mask, bool check); ++#else /* CONFIG_SMP */ ++static inline bool needs_other_cpu(struct task_struct *p, int cpu) ++{ ++ return false; ++} ++ ++static void try_preempt(struct task_struct *p, struct rq *this_rq) ++{ ++ if (p->policy == SCHED_IDLEPRIO) ++ return; ++ if (can_preempt(p, uprq->rq_prio, uprq->rq_deadline)) ++ resched_curr(uprq); ++} ++ ++static inline int __set_cpus_allowed_ptr(struct task_struct *p, ++ const struct cpumask *new_mask, bool check) ++{ ++ return set_cpus_allowed_ptr(p, new_mask); ++} ++#endif /* CONFIG_SMP */ ++ ++/* ++ * wake flags ++ */ ++#define WF_SYNC 0x01 /* waker goes to sleep after wakeup */ ++#define WF_FORK 0x02 /* child wakeup after fork */ ++#define WF_MIGRATED 0x04 /* internal use, task got migrated */ ++ ++static void ++ttwu_stat(struct task_struct *p, int cpu, int wake_flags) ++{ ++ struct rq *rq; ++ ++ if (!schedstat_enabled()) ++ return; ++ ++ rq = this_rq(); ++ ++#ifdef CONFIG_SMP ++ if (cpu == rq->cpu) ++ schedstat_inc(rq->ttwu_local); ++ else { ++ struct sched_domain *sd; ++ ++ rcu_read_lock(); ++ for_each_domain(rq->cpu, sd) { ++ if (cpumask_test_cpu(cpu, sched_domain_span(sd))) { ++ schedstat_inc(sd->ttwu_wake_remote); ++ break; ++ } ++ } ++ rcu_read_unlock(); ++ } ++ ++#endif /* CONFIG_SMP */ ++ ++ schedstat_inc(rq->ttwu_count); ++} ++ ++static inline void ttwu_activate(struct rq *rq, struct task_struct *p) ++{ ++ activate_task(p, rq); ++ ++ /* if a worker is waking up, notify the workqueue */ ++ if (p->flags & PF_WQ_WORKER) ++ wq_worker_waking_up(p, cpu_of(rq)); ++} ++ ++/* ++ * Mark the task runnable and perform wakeup-preemption. ++ */ ++static void ttwu_do_wakeup(struct rq *rq, struct task_struct *p, int wake_flags) ++{ ++ /* ++ * Sync wakeups (i.e. those types of wakeups where the waker ++ * has indicated that it will leave the CPU in short order) ++ * don't trigger a preemption if there are no idle cpus, ++ * instead waiting for current to deschedule. ++ */ ++ if (wake_flags & WF_SYNC) ++ resched_suitable_idle(p); ++ else ++ try_preempt(p, rq); ++ p->state = TASK_RUNNING; ++ trace_sched_wakeup(p); ++} ++ ++static void ++ttwu_do_activate(struct rq *rq, struct task_struct *p, int wake_flags) ++{ ++ lockdep_assert_held(&rq->lock); ++ ++#ifdef CONFIG_SMP ++ if (p->sched_contributes_to_load) ++ rq->nr_uninterruptible--; ++#endif ++ ++ ttwu_activate(rq, p); ++ ttwu_do_wakeup(rq, p, wake_flags); ++} ++ ++/* ++ * Called in case the task @p isn't fully descheduled from its runqueue, ++ * in this case we must do a remote wakeup. Its a 'light' wakeup though, ++ * since all we need to do is flip p->state to TASK_RUNNING, since ++ * the task is still ->on_rq. ++ */ ++static int ttwu_remote(struct task_struct *p, int wake_flags) ++{ ++ struct rq *rq; ++ int ret = 0; ++ ++ rq = __task_rq_lock(p); ++ if (likely(task_on_rq_queued(p))) { ++ ttwu_do_wakeup(rq, p, wake_flags); ++ ret = 1; ++ } ++ __task_rq_unlock(rq); ++ ++ return ret; ++} ++ ++#ifdef CONFIG_SMP ++void sched_ttwu_pending(void) ++{ ++ struct rq *rq = this_rq(); ++ struct llist_node *llist = llist_del_all(&rq->wake_list); ++ struct task_struct *p, *t; ++ unsigned long flags; ++ ++ if (!llist) ++ return; ++ ++ rq_lock_irqsave(rq, &flags); ++ ++ llist_for_each_entry_safe(p, t, llist, wake_entry) ++ ttwu_do_activate(rq, p, 0); ++ ++ rq_unlock_irqrestore(rq, &flags); ++} ++ ++void scheduler_ipi(void) ++{ ++ /* ++ * Fold TIF_NEED_RESCHED into the preempt_count; anybody setting ++ * TIF_NEED_RESCHED remotely (for the first time) will also send ++ * this IPI. ++ */ ++ preempt_fold_need_resched(); ++ ++ if (llist_empty(&this_rq()->wake_list) && (!idle_cpu(smp_processor_id()) || need_resched())) ++ return; ++ ++ /* ++ * Not all reschedule IPI handlers call irq_enter/irq_exit, since ++ * traditionally all their work was done from the interrupt return ++ * path. Now that we actually do some work, we need to make sure ++ * we do call them. ++ * ++ * Some archs already do call them, luckily irq_enter/exit nest ++ * properly. ++ * ++ * Arguably we should visit all archs and update all handlers, ++ * however a fair share of IPIs are still resched only so this would ++ * somewhat pessimize the simple resched case. ++ */ ++ irq_enter(); ++ sched_ttwu_pending(); ++ irq_exit(); ++} ++ ++static void ttwu_queue_remote(struct task_struct *p, int cpu, int wake_flags) ++{ ++ struct rq *rq = cpu_rq(cpu); ++ ++ if (llist_add(&p->wake_entry, &cpu_rq(cpu)->wake_list)) { ++ if (!set_nr_if_polling(rq->idle)) ++ smp_sched_reschedule(cpu); ++ else ++ trace_sched_wake_idle_without_ipi(cpu); ++ } ++} ++ ++void wake_up_if_idle(int cpu) ++{ ++ struct rq *rq = cpu_rq(cpu); ++ unsigned long flags; ++ ++ rcu_read_lock(); ++ ++ if (!is_idle_task(rcu_dereference(rq->curr))) ++ goto out; ++ ++ if (set_nr_if_polling(rq->idle)) { ++ trace_sched_wake_idle_without_ipi(cpu); ++ } else { ++ rq_lock_irqsave(rq, &flags); ++ if (likely(is_idle_task(rq->curr))) ++ smp_sched_reschedule(cpu); ++ /* Else cpu is not in idle, do nothing here */ ++ rq_unlock_irqrestore(rq, &flags); ++ } ++ ++out: ++ rcu_read_unlock(); ++} ++ ++static int valid_task_cpu(struct task_struct *p) ++{ ++ cpumask_t valid_mask; ++ ++ if (p->flags & PF_KTHREAD) ++ cpumask_and(&valid_mask, &p->cpus_allowed, cpu_online_mask); ++ else ++ cpumask_and(&valid_mask, &p->cpus_allowed, cpu_active_mask); ++ ++ if (unlikely(!cpumask_weight(&valid_mask))) { ++ /* Hotplug boot threads do this before the CPU is up */ ++ printk(KERN_INFO "SCHED: No cpumask for %s/%d weight %d\n", p->comm, p->pid, cpumask_weight(&p->cpus_allowed)); ++ return cpumask_any(&p->cpus_allowed); ++ } ++ return cpumask_any(&valid_mask); ++} ++ ++/* ++ * For a task that's just being woken up we have a valuable balancing ++ * opportunity so choose the nearest cache most lightly loaded runqueue. ++ * Entered with rq locked and returns with the chosen runqueue locked. ++ */ ++static inline int select_best_cpu(struct task_struct *p) ++{ ++ unsigned int idlest = ~0U; ++ struct rq *rq = NULL; ++ int i; ++ ++ if (suitable_idle_cpus(p)) { ++ int cpu = task_cpu(p); ++ ++ if (unlikely(needs_other_cpu(p, cpu))) ++ cpu = valid_task_cpu(p); ++ rq = resched_best_idle(p, cpu); ++ if (likely(rq)) ++ return rq->cpu; ++ } ++ ++ for (i = 0; i < num_possible_cpus(); i++) { ++ struct rq *other_rq = task_rq(p)->rq_order[i]; ++ int entries; ++ ++ if (!other_rq->online) ++ continue; ++ if (needs_other_cpu(p, other_rq->cpu)) ++ continue; ++ entries = rq_load(other_rq); ++ if (entries >= idlest) ++ continue; ++ idlest = entries; ++ rq = other_rq; ++ } ++ if (unlikely(!rq)) ++ return task_cpu(p); ++ return rq->cpu; ++} ++#else /* CONFIG_SMP */ ++static int valid_task_cpu(struct task_struct *p) ++{ ++ return 0; ++} ++ ++static inline int select_best_cpu(struct task_struct *p) ++{ ++ return 0; ++} ++ ++static struct rq *resched_best_idle(struct task_struct *p, int cpu) ++{ ++ return NULL; ++} ++#endif /* CONFIG_SMP */ ++ ++static void ttwu_queue(struct task_struct *p, int cpu, int wake_flags) ++{ ++ struct rq *rq = cpu_rq(cpu); ++ ++#if defined(CONFIG_SMP) ++ if (!cpus_share_cache(smp_processor_id(), cpu)) { ++ sched_clock_cpu(cpu); /* Sync clocks across CPUs */ ++ ttwu_queue_remote(p, cpu, wake_flags); ++ return; ++ } ++#endif ++ rq_lock(rq); ++ ttwu_do_activate(rq, p, wake_flags); ++ rq_unlock(rq); ++} ++ ++/*** ++ * try_to_wake_up - wake up a thread ++ * @p: the thread to be awakened ++ * @state: the mask of task states that can be woken ++ * @wake_flags: wake modifier flags (WF_*) ++ * ++ * Put it on the run-queue if it's not already there. The "current" ++ * thread is always on the run-queue (except when the actual ++ * re-schedule is in progress), and as such you're allowed to do ++ * the simpler "current->state = TASK_RUNNING" to mark yourself ++ * runnable without the overhead of this. ++ * ++ * Return: %true if @p was woken up, %false if it was already running. ++ * or @state didn't match @p's state. ++ */ ++static int ++try_to_wake_up(struct task_struct *p, unsigned int state, int wake_flags) ++{ ++ unsigned long flags; ++ int cpu, success = 0; ++ ++ /* ++ * If we are going to wake up a thread waiting for CONDITION we ++ * need to ensure that CONDITION=1 done by the caller can not be ++ * reordered with p->state check below. This pairs with mb() in ++ * set_current_state() the waiting thread does. ++ */ ++ raw_spin_lock_irqsave(&p->pi_lock, flags); ++ smp_mb__after_spinlock(); ++ /* state is a volatile long, どうして、分からない */ ++ if (!((unsigned int)p->state & state)) ++ goto out; ++ ++ trace_sched_waking(p); ++ ++ /* We're going to change ->state: */ ++ success = 1; ++ cpu = task_cpu(p); ++ ++ /* ++ * Ensure we load p->on_rq _after_ p->state, otherwise it would ++ * be possible to, falsely, observe p->on_rq == 0 and get stuck ++ * in smp_cond_load_acquire() below. ++ * ++ * sched_ttwu_pending() try_to_wake_up() ++ * [S] p->on_rq = 1; [L] P->state ++ * UNLOCK rq->lock -----. ++ * \ ++ * +--- RMB ++ * schedule() / ++ * LOCK rq->lock -----' ++ * UNLOCK rq->lock ++ * ++ * [task p] ++ * [S] p->state = UNINTERRUPTIBLE [L] p->on_rq ++ * ++ * Pairs with the UNLOCK+LOCK on rq->lock from the ++ * last wakeup of our task and the schedule that got our task ++ * current. ++ */ ++ smp_rmb(); ++ if (p->on_rq && ttwu_remote(p, wake_flags)) ++ goto stat; ++ ++#ifdef CONFIG_SMP ++ /* ++ * Ensure we load p->on_cpu _after_ p->on_rq, otherwise it would be ++ * possible to, falsely, observe p->on_cpu == 0. ++ * ++ * One must be running (->on_cpu == 1) in order to remove oneself ++ * from the runqueue. ++ * ++ * [S] ->on_cpu = 1; [L] ->on_rq ++ * UNLOCK rq->lock ++ * RMB ++ * LOCK rq->lock ++ * [S] ->on_rq = 0; [L] ->on_cpu ++ * ++ * Pairs with the full barrier implied in the UNLOCK+LOCK on rq->lock ++ * from the consecutive calls to schedule(); the first switching to our ++ * task, the second putting it to sleep. ++ */ ++ smp_rmb(); ++ ++ /* ++ * If the owning (remote) CPU is still in the middle of schedule() with ++ * this task as prev, wait until its done referencing the task. ++ * ++ * Pairs with the smp_store_release() in finish_lock_switch(). ++ * ++ * This ensures that tasks getting woken will be fully ordered against ++ * their previous state and preserve Program Order. ++ */ ++ smp_cond_load_acquire(&p->on_cpu, !VAL); ++ ++ p->sched_contributes_to_load = !!task_contributes_to_load(p); ++ p->state = TASK_WAKING; ++ ++ if (p->in_iowait) { ++ delayacct_blkio_end(); ++ atomic_dec(&task_rq(p)->nr_iowait); ++ } ++ ++ cpu = select_best_cpu(p); ++ if (task_cpu(p) != cpu) ++ set_task_cpu(p, cpu); ++ ++#else /* CONFIG_SMP */ ++ ++ if (p->in_iowait) { ++ delayacct_blkio_end(); ++ atomic_dec(&task_rq(p)->nr_iowait); ++ } ++ ++#endif /* CONFIG_SMP */ ++ ++ ttwu_queue(p, cpu, wake_flags); ++stat: ++ ttwu_stat(p, cpu, wake_flags); ++out: ++ raw_spin_unlock_irqrestore(&p->pi_lock, flags); ++ ++ return success; ++} ++ ++/** ++ * try_to_wake_up_local - try to wake up a local task with rq lock held ++ * @p: the thread to be awakened ++ * ++ * Put @p on the run-queue if it's not already there. The caller must ++ * ensure that rq is locked and, @p is not the current task. ++ * rq stays locked over invocation. ++ */ ++static void try_to_wake_up_local(struct task_struct *p) ++{ ++ struct rq *rq = task_rq(p); ++ ++ if (WARN_ON_ONCE(rq != this_rq()) || ++ WARN_ON_ONCE(p == current)) ++ return; ++ ++ lockdep_assert_held(&rq->lock); ++ ++ if (!raw_spin_trylock(&p->pi_lock)) { ++ /* ++ * This is OK, because current is on_cpu, which avoids it being ++ * picked for load-balance and preemption/IRQs are still ++ * disabled avoiding further scheduler activity on it and we've ++ * not yet picked a replacement task. ++ */ ++ rq_unlock(rq); ++ raw_spin_lock(&p->pi_lock); ++ rq_lock(rq); ++ } ++ ++ if (!(p->state & TASK_NORMAL)) ++ goto out; ++ ++ trace_sched_waking(p); ++ ++ if (!task_on_rq_queued(p)) { ++ if (p->in_iowait) { ++ delayacct_blkio_end(); ++ atomic_dec(&rq->nr_iowait); ++ } ++ ttwu_activate(rq, p); ++ } ++ ++ ttwu_do_wakeup(rq, p, 0); ++ ttwu_stat(p, smp_processor_id(), 0); ++out: ++ raw_spin_unlock(&p->pi_lock); ++} ++ ++/** ++ * wake_up_process - Wake up a specific process ++ * @p: The process to be woken up. ++ * ++ * Attempt to wake up the nominated process and move it to the set of runnable ++ * processes. ++ * ++ * Return: 1 if the process was woken up, 0 if it was already running. ++ * ++ * It may be assumed that this function implies a write memory barrier before ++ * changing the task state if and only if any tasks are woken up. ++ */ ++int wake_up_process(struct task_struct *p) ++{ ++ return try_to_wake_up(p, TASK_NORMAL, 0); ++} ++EXPORT_SYMBOL(wake_up_process); ++ ++int wake_up_state(struct task_struct *p, unsigned int state) ++{ ++ return try_to_wake_up(p, state, 0); ++} ++ ++static void time_slice_expired(struct task_struct *p, struct rq *rq); ++ ++/* ++ * Perform scheduler related setup for a newly forked process p. ++ * p is forked by current. ++ */ ++int sched_fork(unsigned long __maybe_unused clone_flags, struct task_struct *p) ++{ ++ unsigned long flags; ++ int cpu = get_cpu(); ++ ++#ifdef CONFIG_PREEMPT_NOTIFIERS ++ INIT_HLIST_HEAD(&p->preempt_notifiers); ++#endif ++ /* ++ * We mark the process as NEW here. This guarantees that ++ * nobody will actually run it, and a signal or other external ++ * event cannot wake it up and insert it on the runqueue either. ++ */ ++ p->state = TASK_NEW; ++ ++ /* ++ * The process state is set to the same value of the process executing ++ * do_fork() code. That is running. This guarantees that nobody will ++ * actually run it, and a signal or other external event cannot wake ++ * it up and insert it on the runqueue either. ++ */ ++ ++ /* Should be reset in fork.c but done here for ease of MuQSS patching */ ++ p->on_cpu = ++ p->on_rq = ++ p->utime = ++ p->stime = ++ p->sched_time = ++ p->stime_ns = ++ p->utime_ns = 0; ++ skiplist_node_init(&p->node); ++ ++ /* ++ * Revert to default priority/policy on fork if requested. ++ */ ++ if (unlikely(p->sched_reset_on_fork)) { ++ if (p->policy == SCHED_FIFO || p->policy == SCHED_RR) { ++ p->policy = SCHED_NORMAL; ++ p->normal_prio = normal_prio(p); ++ } ++ ++ if (PRIO_TO_NICE(p->static_prio) < 0) { ++ p->static_prio = NICE_TO_PRIO(0); ++ p->normal_prio = p->static_prio; ++ } ++ ++ /* ++ * We don't need the reset flag anymore after the fork. It has ++ * fulfilled its duty: ++ */ ++ p->sched_reset_on_fork = 0; ++ } ++ ++ /* ++ * Silence PROVE_RCU. ++ */ ++ raw_spin_lock_irqsave(&p->pi_lock, flags); ++ set_task_cpu(p, cpu); ++ raw_spin_unlock_irqrestore(&p->pi_lock, flags); ++ ++#ifdef CONFIG_SCHED_INFO ++ if (unlikely(sched_info_on())) ++ memset(&p->sched_info, 0, sizeof(p->sched_info)); ++#endif ++ init_task_preempt_count(p); ++ ++ put_cpu(); ++ return 0; ++} ++ ++#ifdef CONFIG_SCHEDSTATS ++ ++DEFINE_STATIC_KEY_FALSE(sched_schedstats); ++static bool __initdata __sched_schedstats = false; ++ ++static void set_schedstats(bool enabled) ++{ ++ if (enabled) ++ static_branch_enable(&sched_schedstats); ++ else ++ static_branch_disable(&sched_schedstats); ++} ++ ++void force_schedstat_enabled(void) ++{ ++ if (!schedstat_enabled()) { ++ pr_info("kernel profiling enabled schedstats, disable via kernel.sched_schedstats.\n"); ++ static_branch_enable(&sched_schedstats); ++ } ++} ++ ++static int __init setup_schedstats(char *str) ++{ ++ int ret = 0; ++ if (!str) ++ goto out; ++ ++ /* ++ * This code is called before jump labels have been set up, so we can't ++ * change the static branch directly just yet. Instead set a temporary ++ * variable so init_schedstats() can do it later. ++ */ ++ if (!strcmp(str, "enable")) { ++ __sched_schedstats = true; ++ ret = 1; ++ } else if (!strcmp(str, "disable")) { ++ __sched_schedstats = false; ++ ret = 1; ++ } ++out: ++ if (!ret) ++ pr_warn("Unable to parse schedstats=\n"); ++ ++ return ret; ++} ++__setup("schedstats=", setup_schedstats); ++ ++static void __init init_schedstats(void) ++{ ++ set_schedstats(__sched_schedstats); ++} ++ ++#ifdef CONFIG_PROC_SYSCTL ++int sysctl_schedstats(struct ctl_table *table, int write, ++ void __user *buffer, size_t *lenp, loff_t *ppos) ++{ ++ struct ctl_table t; ++ int err; ++ int state = static_branch_likely(&sched_schedstats); ++ ++ if (write && !capable(CAP_SYS_ADMIN)) ++ return -EPERM; ++ ++ t = *table; ++ t.data = &state; ++ err = proc_dointvec_minmax(&t, write, buffer, lenp, ppos); ++ if (err < 0) ++ return err; ++ if (write) ++ set_schedstats(state); ++ return err; ++} ++#endif /* CONFIG_PROC_SYSCTL */ ++#else /* !CONFIG_SCHEDSTATS */ ++static inline void init_schedstats(void) {} ++#endif /* CONFIG_SCHEDSTATS */ ++ ++static void update_cpu_clock_switch(struct rq *rq, struct task_struct *p); ++ ++static void account_task_cpu(struct rq *rq, struct task_struct *p) ++{ ++ update_clocks(rq); ++ /* This isn't really a context switch but accounting is the same */ ++ update_cpu_clock_switch(rq, p); ++ p->last_ran = rq->niffies; ++} ++ ++bool sched_smp_initialized __read_mostly; ++ ++static inline int hrexpiry_enabled(struct rq *rq) ++{ ++ if (unlikely(!cpu_active(cpu_of(rq)) || !sched_smp_initialized)) ++ return 0; ++ return hrtimer_is_hres_active(&rq->hrexpiry_timer); ++} ++ ++/* ++ * Use HR-timers to deliver accurate preemption points. ++ */ ++static inline void hrexpiry_clear(struct rq *rq) ++{ ++ if (!hrexpiry_enabled(rq)) ++ return; ++ if (hrtimer_active(&rq->hrexpiry_timer)) ++ hrtimer_cancel(&rq->hrexpiry_timer); ++} ++ ++/* ++ * High-resolution time_slice expiry. ++ * Runs from hardirq context with interrupts disabled. ++ */ ++static enum hrtimer_restart hrexpiry(struct hrtimer *timer) ++{ ++ struct rq *rq = container_of(timer, struct rq, hrexpiry_timer); ++ struct task_struct *p; ++ ++ /* This can happen during CPU hotplug / resume */ ++ if (unlikely(cpu_of(rq) != smp_processor_id())) ++ goto out; ++ ++ /* ++ * We're doing this without the runqueue lock but this should always ++ * be run on the local CPU. Time slice should run out in __schedule ++ * but we set it to zero here in case niffies is slightly less. ++ */ ++ p = rq->curr; ++ p->time_slice = 0; ++ __set_tsk_resched(p); ++out: ++ return HRTIMER_NORESTART; ++} ++ ++/* ++ * Called to set the hrexpiry timer state. ++ * ++ * called with irqs disabled from the local CPU only ++ */ ++static void hrexpiry_start(struct rq *rq, u64 delay) ++{ ++ if (!hrexpiry_enabled(rq)) ++ return; ++ ++ hrtimer_start(&rq->hrexpiry_timer, ns_to_ktime(delay), ++ HRTIMER_MODE_REL_PINNED); ++} ++ ++static void init_rq_hrexpiry(struct rq *rq) ++{ ++ hrtimer_init(&rq->hrexpiry_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); ++ rq->hrexpiry_timer.function = hrexpiry; ++} ++ ++static inline int rq_dither(struct rq *rq) ++{ ++ if (!hrexpiry_enabled(rq)) ++ return HALF_JIFFY_US; ++ return 0; ++} ++ ++/* ++ * wake_up_new_task - wake up a newly created task for the first time. ++ * ++ * This function will do some initial scheduler statistics housekeeping ++ * that must be done for every newly created context, then puts the task ++ * on the runqueue and wakes it. ++ */ ++void wake_up_new_task(struct task_struct *p) ++{ ++ struct task_struct *parent, *rq_curr; ++ struct rq *rq, *new_rq; ++ unsigned long flags; ++ ++ parent = p->parent; ++ ++ raw_spin_lock_irqsave(&p->pi_lock, flags); ++ p->state = TASK_RUNNING; ++ /* Task_rq can't change yet on a new task */ ++ new_rq = rq = task_rq(p); ++ if (unlikely(needs_other_cpu(p, task_cpu(p)))) { ++ set_task_cpu(p, valid_task_cpu(p)); ++ new_rq = task_rq(p); ++ } ++ ++ double_rq_lock(rq, new_rq); ++ rq_curr = rq->curr; ++ ++ /* ++ * Make sure we do not leak PI boosting priority to the child. ++ */ ++ p->prio = rq_curr->normal_prio; ++ ++ trace_sched_wakeup_new(p); ++ ++ /* ++ * Share the timeslice between parent and child, thus the ++ * total amount of pending timeslices in the system doesn't change, ++ * resulting in more scheduling fairness. If it's negative, it won't ++ * matter since that's the same as being 0. rq->rq_deadline is only ++ * modified within schedule() so it is always equal to ++ * current->deadline. ++ */ ++ account_task_cpu(rq, rq_curr); ++ p->last_ran = rq_curr->last_ran; ++ if (likely(rq_curr->policy != SCHED_FIFO)) { ++ rq_curr->time_slice /= 2; ++ if (rq_curr->time_slice < RESCHED_US) { ++ /* ++ * Forking task has run out of timeslice. Reschedule it and ++ * start its child with a new time slice and deadline. The ++ * child will end up running first because its deadline will ++ * be slightly earlier. ++ */ ++ __set_tsk_resched(rq_curr); ++ time_slice_expired(p, new_rq); ++ if (suitable_idle_cpus(p)) ++ resched_best_idle(p, task_cpu(p)); ++ else if (unlikely(rq != new_rq)) ++ try_preempt(p, new_rq); ++ } else { ++ p->time_slice = rq_curr->time_slice; ++ if (rq_curr == parent && rq == new_rq && !suitable_idle_cpus(p)) { ++ /* ++ * The VM isn't cloned, so we're in a good position to ++ * do child-runs-first in anticipation of an exec. This ++ * usually avoids a lot of COW overhead. ++ */ ++ __set_tsk_resched(rq_curr); ++ } else { ++ /* ++ * Adjust the hrexpiry since rq_curr will keep ++ * running and its timeslice has been shortened. ++ */ ++ hrexpiry_start(rq, US_TO_NS(rq_curr->time_slice)); ++ try_preempt(p, new_rq); ++ } ++ } ++ } else { ++ time_slice_expired(p, new_rq); ++ try_preempt(p, new_rq); ++ } ++ activate_task(p, new_rq); ++ double_rq_unlock(rq, new_rq); ++ raw_spin_unlock_irqrestore(&p->pi_lock, flags); ++} ++ ++#ifdef CONFIG_PREEMPT_NOTIFIERS ++ ++static struct static_key preempt_notifier_key = STATIC_KEY_INIT_FALSE; ++ ++void preempt_notifier_inc(void) ++{ ++ static_key_slow_inc(&preempt_notifier_key); ++} ++EXPORT_SYMBOL_GPL(preempt_notifier_inc); ++ ++void preempt_notifier_dec(void) ++{ ++ static_key_slow_dec(&preempt_notifier_key); ++} ++EXPORT_SYMBOL_GPL(preempt_notifier_dec); ++ ++/** ++ * preempt_notifier_register - tell me when current is being preempted & rescheduled ++ * @notifier: notifier struct to register ++ */ ++void preempt_notifier_register(struct preempt_notifier *notifier) ++{ ++ if (!static_key_false(&preempt_notifier_key)) ++ WARN(1, "registering preempt_notifier while notifiers disabled\n"); ++ ++ hlist_add_head(¬ifier->link, ¤t->preempt_notifiers); ++} ++EXPORT_SYMBOL_GPL(preempt_notifier_register); ++ ++/** ++ * preempt_notifier_unregister - no longer interested in preemption notifications ++ * @notifier: notifier struct to unregister ++ * ++ * This is *not* safe to call from within a preemption notifier. ++ */ ++void preempt_notifier_unregister(struct preempt_notifier *notifier) ++{ ++ hlist_del(¬ifier->link); ++} ++EXPORT_SYMBOL_GPL(preempt_notifier_unregister); ++ ++static void __fire_sched_in_preempt_notifiers(struct task_struct *curr) ++{ ++ struct preempt_notifier *notifier; ++ ++ hlist_for_each_entry(notifier, &curr->preempt_notifiers, link) ++ notifier->ops->sched_in(notifier, raw_smp_processor_id()); ++} ++ ++static __always_inline void fire_sched_in_preempt_notifiers(struct task_struct *curr) ++{ ++ if (static_key_false(&preempt_notifier_key)) ++ __fire_sched_in_preempt_notifiers(curr); ++} ++ ++static void ++__fire_sched_out_preempt_notifiers(struct task_struct *curr, ++ struct task_struct *next) ++{ ++ struct preempt_notifier *notifier; ++ ++ hlist_for_each_entry(notifier, &curr->preempt_notifiers, link) ++ notifier->ops->sched_out(notifier, next); ++} ++ ++static __always_inline void ++fire_sched_out_preempt_notifiers(struct task_struct *curr, ++ struct task_struct *next) ++{ ++ if (static_key_false(&preempt_notifier_key)) ++ __fire_sched_out_preempt_notifiers(curr, next); ++} ++ ++#else /* !CONFIG_PREEMPT_NOTIFIERS */ ++ ++static inline void fire_sched_in_preempt_notifiers(struct task_struct *curr) ++{ ++} ++ ++static inline void ++fire_sched_out_preempt_notifiers(struct task_struct *curr, ++ struct task_struct *next) ++{ ++} ++ ++#endif /* CONFIG_PREEMPT_NOTIFIERS */ ++ ++/** ++ * prepare_task_switch - prepare to switch tasks ++ * @rq: the runqueue preparing to switch ++ * @next: the task we are going to switch to. ++ * ++ * This is called with the rq lock held and interrupts off. It must ++ * be paired with a subsequent finish_task_switch after the context ++ * switch. ++ * ++ * prepare_task_switch sets up locking and calls architecture specific ++ * hooks. ++ */ ++static inline void ++prepare_task_switch(struct rq *rq, struct task_struct *prev, ++ struct task_struct *next) ++{ ++ sched_info_switch(rq, prev, next); ++ perf_event_task_sched_out(prev, next); ++ fire_sched_out_preempt_notifiers(prev, next); ++ prepare_lock_switch(rq, next); ++ prepare_arch_switch(next); ++} ++ ++/** ++ * finish_task_switch - clean up after a task-switch ++ * @rq: runqueue associated with task-switch ++ * @prev: the thread we just switched away from. ++ * ++ * finish_task_switch must be called after the context switch, paired ++ * with a prepare_task_switch call before the context switch. ++ * finish_task_switch will reconcile locking set up by prepare_task_switch, ++ * and do any other architecture-specific cleanup actions. ++ * ++ * Note that we may have delayed dropping an mm in context_switch(). If ++ * so, we finish that here outside of the runqueue lock. (Doing it ++ * with the lock held can cause deadlocks; see schedule() for ++ * details.) ++ * ++ * The context switch have flipped the stack from under us and restored the ++ * local variables which were saved when this task called schedule() in the ++ * past. prev == current is still correct but we need to recalculate this_rq ++ * because prev may have moved to another CPU. ++ */ ++static void finish_task_switch(struct task_struct *prev) ++ __releases(rq->lock) ++{ ++ struct rq *rq = this_rq(); ++ struct mm_struct *mm = rq->prev_mm; ++ long prev_state; ++ ++ /* ++ * The previous task will have left us with a preempt_count of 2 ++ * because it left us after: ++ * ++ * schedule() ++ * preempt_disable(); // 1 ++ * __schedule() ++ * raw_spin_lock_irq(&rq->lock) // 2 ++ * ++ * Also, see FORK_PREEMPT_COUNT. ++ */ ++ if (WARN_ONCE(preempt_count() != 2*PREEMPT_DISABLE_OFFSET, ++ "corrupted preempt_count: %s/%d/0x%x\n", ++ current->comm, current->pid, preempt_count())) ++ preempt_count_set(FORK_PREEMPT_COUNT); ++ ++ rq->prev_mm = NULL; ++ ++ /* ++ * A task struct has one reference for the use as "current". ++ * If a task dies, then it sets TASK_DEAD in tsk->state and calls ++ * schedule one last time. The schedule call will never return, and ++ * the scheduled task must drop that reference. ++ * ++ * We must observe prev->state before clearing prev->on_cpu (in ++ * finish_lock_switch), otherwise a concurrent wakeup can get prev ++ * running on another CPU and we could rave with its RUNNING -> DEAD ++ * transition, resulting in a double drop. ++ */ ++ prev_state = prev->state; ++ vtime_task_switch(prev); ++ perf_event_task_sched_in(prev, current); ++ /* ++ * The membarrier system call requires a full memory barrier ++ * after storing to rq->curr, before going back to user-space. ++ * ++ * TODO: This smp_mb__after_unlock_lock can go away if PPC end ++ * up adding a full barrier to switch_mm(), or we should figure ++ * out if a smp_mb__after_unlock_lock is really the proper API ++ * to use. ++ */ ++ smp_mb__after_unlock_lock(); ++ finish_lock_switch(rq, prev); ++ finish_arch_post_lock_switch(); ++ ++ fire_sched_in_preempt_notifiers(current); ++ if (mm) ++ mmdrop(mm); ++ if (unlikely(prev_state == TASK_DEAD)) { ++ /* ++ * Remove function-return probe instances associated with this ++ * task and put them back on the free list. ++ */ ++ kprobe_flush_task(prev); ++ ++ /* Task is done with its stack. */ ++ put_task_stack(prev); ++ ++ put_task_struct(prev); ++ } ++} ++ ++/** ++ * schedule_tail - first thing a freshly forked thread must call. ++ * @prev: the thread we just switched away from. ++ */ ++asmlinkage __visible void schedule_tail(struct task_struct *prev) ++{ ++ /* ++ * New tasks start with FORK_PREEMPT_COUNT, see there and ++ * finish_task_switch() for details. ++ * ++ * finish_task_switch() will drop rq->lock() and lower preempt_count ++ * and the preempt_enable() will end up enabling preemption (on ++ * PREEMPT_COUNT kernels). ++ */ ++ ++ finish_task_switch(prev); ++ preempt_enable(); ++ ++ if (current->set_child_tid) ++ put_user(task_pid_vnr(current), current->set_child_tid); ++} ++ ++/* ++ * context_switch - switch to the new MM and the new thread's register state. ++ */ ++static __always_inline void ++context_switch(struct rq *rq, struct task_struct *prev, ++ struct task_struct *next) ++{ ++ struct mm_struct *mm, *oldmm; ++ ++ prepare_task_switch(rq, prev, next); ++ ++ mm = next->mm; ++ oldmm = prev->active_mm; ++ /* ++ * For paravirt, this is coupled with an exit in switch_to to ++ * combine the page table reload and the switch backend into ++ * one hypercall. ++ */ ++ arch_start_context_switch(prev); ++ ++ if (!mm) { ++ next->active_mm = oldmm; ++ mmgrab(oldmm); ++ enter_lazy_tlb(oldmm, next); ++ } else ++ switch_mm_irqs_off(oldmm, mm, next); ++ ++ if (!prev->mm) { ++ prev->active_mm = NULL; ++ rq->prev_mm = oldmm; ++ } ++ /* ++ * Since the runqueue lock will be released by the next ++ * task (which is an invalid locking op but in the case ++ * of the scheduler it's an obvious special-case), so we ++ * do an early lockdep release here: ++ */ ++ spin_release(&rq->lock.dep_map, 1, _THIS_IP_); ++ ++ /* Here we just switch the register state and the stack. */ ++ switch_to(prev, next, prev); ++ barrier(); ++ ++ finish_task_switch(prev); ++} ++ ++/* ++ * nr_running, nr_uninterruptible and nr_context_switches: ++ * ++ * externally visible scheduler statistics: current number of runnable ++ * threads, total number of context switches performed since bootup. ++ */ ++unsigned long nr_running(void) ++{ ++ unsigned long i, sum = 0; ++ ++ for_each_online_cpu(i) ++ sum += cpu_rq(i)->nr_running; ++ ++ return sum; ++} ++ ++static unsigned long nr_uninterruptible(void) ++{ ++ unsigned long i, sum = 0; ++ ++ for_each_online_cpu(i) ++ sum += cpu_rq(i)->nr_uninterruptible; ++ ++ return sum; ++} ++ ++/* ++ * Check if only the current task is running on the CPU. ++ * ++ * Caution: this function does not check that the caller has disabled ++ * preemption, thus the result might have a time-of-check-to-time-of-use ++ * race. The caller is responsible to use it correctly, for example: ++ * ++ * - from a non-preemptable section (of course) ++ * ++ * - from a thread that is bound to a single CPU ++ * ++ * - in a loop with very short iterations (e.g. a polling loop) ++ */ ++bool single_task_running(void) ++{ ++ struct rq *rq = cpu_rq(smp_processor_id()); ++ ++ if (rq_load(rq) == 1) ++ return true; ++ else ++ return false; ++} ++EXPORT_SYMBOL(single_task_running); ++ ++unsigned long long nr_context_switches(void) ++{ ++ int i; ++ unsigned long long sum = 0; ++ ++ for_each_possible_cpu(i) ++ sum += cpu_rq(i)->nr_switches; ++ ++ return sum; ++} ++ ++/* ++ * IO-wait accounting, and how its mostly bollocks (on SMP). ++ * ++ * The idea behind IO-wait account is to account the idle time that we could ++ * have spend running if it were not for IO. That is, if we were to improve the ++ * storage performance, we'd have a proportional reduction in IO-wait time. ++ * ++ * This all works nicely on UP, where, when a task blocks on IO, we account ++ * idle time as IO-wait, because if the storage were faster, it could've been ++ * running and we'd not be idle. ++ * ++ * This has been extended to SMP, by doing the same for each CPU. This however ++ * is broken. ++ * ++ * Imagine for instance the case where two tasks block on one CPU, only the one ++ * CPU will have IO-wait accounted, while the other has regular idle. Even ++ * though, if the storage were faster, both could've ran at the same time, ++ * utilising both CPUs. ++ * ++ * This means, that when looking globally, the current IO-wait accounting on ++ * SMP is a lower bound, by reason of under accounting. ++ * ++ * Worse, since the numbers are provided per CPU, they are sometimes ++ * interpreted per CPU, and that is nonsensical. A blocked task isn't strictly ++ * associated with any one particular CPU, it can wake to another CPU than it ++ * blocked on. This means the per CPU IO-wait number is meaningless. ++ * ++ * Task CPU affinities can make all that even more 'interesting'. ++ */ ++ ++unsigned long nr_iowait(void) ++{ ++ unsigned long i, sum = 0; ++ ++ for_each_possible_cpu(i) ++ sum += atomic_read(&cpu_rq(i)->nr_iowait); ++ ++ return sum; ++} ++ ++/* ++ * Consumers of these two interfaces, like for example the cpufreq menu ++ * governor are using nonsensical data. Boosting frequency for a CPU that has ++ * IO-wait which might not even end up running the task when it does become ++ * runnable. ++ */ ++ ++unsigned long nr_iowait_cpu(int cpu) ++{ ++ struct rq *this = cpu_rq(cpu); ++ return atomic_read(&this->nr_iowait); ++} ++ ++unsigned long nr_active(void) ++{ ++ return nr_running() + nr_uninterruptible(); ++} ++ ++/* ++ * I/O wait is the number of running or queued tasks with their ->rq pointer ++ * set to this cpu as being the CPU they're more likely to run on. ++ */ ++void get_iowait_load(unsigned long *nr_waiters, unsigned long *load) ++{ ++ struct rq *rq = this_rq(); ++ ++ *nr_waiters = atomic_read(&rq->nr_iowait); ++ *load = rq_load(rq); ++} ++ ++/* Variables and functions for calc_load */ ++static unsigned long calc_load_update; ++unsigned long avenrun[3]; ++EXPORT_SYMBOL(avenrun); ++ ++/** ++ * get_avenrun - get the load average array ++ * @loads: pointer to dest load array ++ * @offset: offset to add ++ * @shift: shift count to shift the result left ++ * ++ * These values are estimates at best, so no need for locking. ++ */ ++void get_avenrun(unsigned long *loads, unsigned long offset, int shift) ++{ ++ loads[0] = (avenrun[0] + offset) << shift; ++ loads[1] = (avenrun[1] + offset) << shift; ++ loads[2] = (avenrun[2] + offset) << shift; ++} ++ ++static unsigned long ++calc_load(unsigned long load, unsigned long exp, unsigned long active) ++{ ++ unsigned long newload; ++ ++ newload = load * exp + active * (FIXED_1 - exp); ++ if (active >= load) ++ newload += FIXED_1-1; ++ ++ return newload / FIXED_1; ++} ++ ++/* ++ * calc_load - update the avenrun load estimates every LOAD_FREQ seconds. ++ */ ++void calc_global_load(unsigned long ticks) ++{ ++ long active; ++ ++ if (time_before(jiffies, READ_ONCE(calc_load_update))) ++ return; ++ active = nr_active() * FIXED_1; ++ ++ avenrun[0] = calc_load(avenrun[0], EXP_1, active); ++ avenrun[1] = calc_load(avenrun[1], EXP_5, active); ++ avenrun[2] = calc_load(avenrun[2], EXP_15, active); ++ ++ calc_load_update = jiffies + LOAD_FREQ; ++} ++ ++DEFINE_PER_CPU(struct kernel_stat, kstat); ++DEFINE_PER_CPU(struct kernel_cpustat, kernel_cpustat); ++ ++EXPORT_PER_CPU_SYMBOL(kstat); ++EXPORT_PER_CPU_SYMBOL(kernel_cpustat); ++ ++#ifdef CONFIG_PARAVIRT ++static inline u64 steal_ticks(u64 steal) ++{ ++ if (unlikely(steal > NSEC_PER_SEC)) ++ return div_u64(steal, TICK_NSEC); ++ ++ return __iter_div_u64_rem(steal, TICK_NSEC, &steal); ++} ++#endif ++ ++#ifndef nsecs_to_cputime ++# define nsecs_to_cputime(__nsecs) nsecs_to_jiffies(__nsecs) ++#endif ++ ++/* ++ * On each tick, add the number of nanoseconds to the unbanked variables and ++ * once one tick's worth has accumulated, account it allowing for accurate ++ * sub-tick accounting and totals. ++ */ ++static void pc_idle_time(struct rq *rq, struct task_struct *idle, unsigned long ns) ++{ ++ u64 *cpustat = kcpustat_this_cpu->cpustat; ++ unsigned long ticks; ++ ++ if (atomic_read(&rq->nr_iowait) > 0) { ++ rq->iowait_ns += ns; ++ if (rq->iowait_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(rq->iowait_ns); ++ cpustat[CPUTIME_IOWAIT] += (__force u64)TICK_NSEC * ticks; ++ rq->iowait_ns %= JIFFY_NS; ++ } ++ } else { ++ rq->idle_ns += ns; ++ if (rq->idle_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(rq->idle_ns); ++ cpustat[CPUTIME_IDLE] += (__force u64)TICK_NSEC * ticks; ++ rq->idle_ns %= JIFFY_NS; ++ } ++ } ++ acct_update_integrals(idle); ++} ++ ++static void pc_system_time(struct rq *rq, struct task_struct *p, ++ int hardirq_offset, unsigned long ns) ++{ ++ u64 *cpustat = kcpustat_this_cpu->cpustat; ++ unsigned long ticks; ++ ++ p->stime_ns += ns; ++ if (p->stime_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(p->stime_ns); ++ p->stime_ns %= JIFFY_NS; ++ p->stime += (__force u64)TICK_NSEC * ticks; ++ account_group_system_time(p, TICK_NSEC * ticks); ++ } ++ p->sched_time += ns; ++ account_group_exec_runtime(p, ns); ++ ++ if (hardirq_count() - hardirq_offset) { ++ rq->irq_ns += ns; ++ if (rq->irq_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(rq->irq_ns); ++ cpustat[CPUTIME_IRQ] += (__force u64)TICK_NSEC * ticks; ++ rq->irq_ns %= JIFFY_NS; ++ } ++ } else if (in_serving_softirq()) { ++ rq->softirq_ns += ns; ++ if (rq->softirq_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(rq->softirq_ns); ++ cpustat[CPUTIME_SOFTIRQ] += (__force u64)TICK_NSEC * ticks; ++ rq->softirq_ns %= JIFFY_NS; ++ } ++ } else { ++ rq->system_ns += ns; ++ if (rq->system_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(rq->system_ns); ++ cpustat[CPUTIME_SYSTEM] += (__force u64)TICK_NSEC * ticks; ++ rq->system_ns %= JIFFY_NS; ++ } ++ } ++ acct_update_integrals(p); ++} ++ ++static void pc_user_time(struct rq *rq, struct task_struct *p, unsigned long ns) ++{ ++ u64 *cpustat = kcpustat_this_cpu->cpustat; ++ unsigned long ticks; ++ ++ p->utime_ns += ns; ++ if (p->utime_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(p->utime_ns); ++ p->utime_ns %= JIFFY_NS; ++ p->utime += (__force u64)TICK_NSEC * ticks; ++ account_group_user_time(p, TICK_NSEC * ticks); ++ } ++ p->sched_time += ns; ++ account_group_exec_runtime(p, ns); ++ ++ if (this_cpu_ksoftirqd() == p) { ++ /* ++ * ksoftirqd time do not get accounted in cpu_softirq_time. ++ * So, we have to handle it separately here. ++ */ ++ rq->softirq_ns += ns; ++ if (rq->softirq_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(rq->softirq_ns); ++ cpustat[CPUTIME_SOFTIRQ] += (__force u64)TICK_NSEC * ticks; ++ rq->softirq_ns %= JIFFY_NS; ++ } ++ } ++ ++ if (task_nice(p) > 0 || idleprio_task(p)) { ++ rq->nice_ns += ns; ++ if (rq->nice_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(rq->nice_ns); ++ cpustat[CPUTIME_NICE] += (__force u64)TICK_NSEC * ticks; ++ rq->nice_ns %= JIFFY_NS; ++ } ++ } else { ++ rq->user_ns += ns; ++ if (rq->user_ns >= JIFFY_NS) { ++ ticks = NS_TO_JIFFIES(rq->user_ns); ++ cpustat[CPUTIME_USER] += (__force u64)TICK_NSEC * ticks; ++ rq->user_ns %= JIFFY_NS; ++ } ++ } ++ acct_update_integrals(p); ++} ++ ++/* ++ * This is called on clock ticks. ++ * Bank in p->sched_time the ns elapsed since the last tick or switch. ++ * CPU scheduler quota accounting is also performed here in microseconds. ++ */ ++static void update_cpu_clock_tick(struct rq *rq, struct task_struct *p) ++{ ++ s64 account_ns = rq->niffies - p->last_ran; ++ struct task_struct *idle = rq->idle; ++ ++ /* Accurate tick timekeeping */ ++ if (user_mode(get_irq_regs())) ++ pc_user_time(rq, p, account_ns); ++ else if (p != idle || (irq_count() != HARDIRQ_OFFSET)) { ++ pc_system_time(rq, p, HARDIRQ_OFFSET, account_ns); ++ } else ++ pc_idle_time(rq, idle, account_ns); ++ ++ /* time_slice accounting is done in usecs to avoid overflow on 32bit */ ++ if (p->policy != SCHED_FIFO && p != idle) ++ p->time_slice -= NS_TO_US(account_ns); ++ ++ p->last_ran = rq->niffies; ++} ++ ++/* ++ * This is called on context switches. ++ * Bank in p->sched_time the ns elapsed since the last tick or switch. ++ * CPU scheduler quota accounting is also performed here in microseconds. ++ */ ++static void update_cpu_clock_switch(struct rq *rq, struct task_struct *p) ++{ ++ s64 account_ns = rq->niffies - p->last_ran; ++ struct task_struct *idle = rq->idle; ++ ++ /* Accurate subtick timekeeping */ ++ if (p != idle) ++ pc_user_time(rq, p, account_ns); ++ else ++ pc_idle_time(rq, idle, account_ns); ++ ++ /* time_slice accounting is done in usecs to avoid overflow on 32bit */ ++ if (p->policy != SCHED_FIFO && p != idle) ++ p->time_slice -= NS_TO_US(account_ns); ++} ++ ++/* ++ * Return any ns on the sched_clock that have not yet been accounted in ++ * @p in case that task is currently running. ++ * ++ * Called with task_rq_lock(p) held. ++ */ ++static inline u64 do_task_delta_exec(struct task_struct *p, struct rq *rq) ++{ ++ u64 ns = 0; ++ ++ /* ++ * Must be ->curr _and_ ->on_rq. If dequeued, we would ++ * project cycles that may never be accounted to this ++ * thread, breaking clock_gettime(). ++ */ ++ if (p == rq->curr && task_on_rq_queued(p)) { ++ update_clocks(rq); ++ ns = rq->niffies - p->last_ran; ++ } ++ ++ return ns; ++} ++ ++/* ++ * Return accounted runtime for the task. ++ * Return separately the current's pending runtime that have not been ++ * accounted yet. ++ * ++ */ ++unsigned long long task_sched_runtime(struct task_struct *p) ++{ ++ unsigned long flags; ++ struct rq *rq; ++ u64 ns; ++ ++#if defined(CONFIG_64BIT) && defined(CONFIG_SMP) ++ /* ++ * 64-bit doesn't need locks to atomically read a 64bit value. ++ * So we have a optimization chance when the task's delta_exec is 0. ++ * Reading ->on_cpu is racy, but this is ok. ++ * ++ * If we race with it leaving CPU, we'll take a lock. So we're correct. ++ * If we race with it entering CPU, unaccounted time is 0. This is ++ * indistinguishable from the read occurring a few cycles earlier. ++ * If we see ->on_cpu without ->on_rq, the task is leaving, and has ++ * been accounted, so we're correct here as well. ++ */ ++ if (!p->on_cpu || !task_on_rq_queued(p)) ++ return tsk_seruntime(p); ++#endif ++ ++ rq = task_rq_lock(p, &flags); ++ ns = p->sched_time + do_task_delta_exec(p, rq); ++ task_rq_unlock(rq, p, &flags); ++ ++ return ns; ++} ++ ++/* ++ * Functions to test for when SCHED_ISO tasks have used their allocated ++ * quota as real time scheduling and convert them back to SCHED_NORMAL. All ++ * data is modified only by the local runqueue during scheduler_tick with ++ * interrupts disabled. ++ */ ++ ++/* ++ * Test if SCHED_ISO tasks have run longer than their alloted period as RT ++ * tasks and set the refractory flag if necessary. There is 10% hysteresis ++ * for unsetting the flag. 115/128 is ~90/100 as a fast shift instead of a ++ * slow division. ++ */ ++static inline void iso_tick(struct rq *rq) ++{ ++ rq->iso_ticks = rq->iso_ticks * (ISO_PERIOD - 1) / ISO_PERIOD; ++ rq->iso_ticks += 100; ++ if (rq->iso_ticks > ISO_PERIOD * sched_iso_cpu) { ++ rq->iso_refractory = true; ++ if (unlikely(rq->iso_ticks > ISO_PERIOD * 100)) ++ rq->iso_ticks = ISO_PERIOD * 100; ++ } ++} ++ ++/* No SCHED_ISO task was running so decrease rq->iso_ticks */ ++static inline void no_iso_tick(struct rq *rq, int ticks) ++{ ++ if (rq->iso_ticks > 0 || rq->iso_refractory) { ++ rq->iso_ticks = rq->iso_ticks * (ISO_PERIOD - ticks) / ISO_PERIOD; ++ if (rq->iso_ticks < ISO_PERIOD * (sched_iso_cpu * 115 / 128)) { ++ rq->iso_refractory = false; ++ if (unlikely(rq->iso_ticks < 0)) ++ rq->iso_ticks = 0; ++ } ++ } ++} ++ ++/* This manages tasks that have run out of timeslice during a scheduler_tick */ ++static void task_running_tick(struct rq *rq) ++{ ++ struct task_struct *p = rq->curr; ++ ++ /* ++ * If a SCHED_ISO task is running we increment the iso_ticks. In ++ * order to prevent SCHED_ISO tasks from causing starvation in the ++ * presence of true RT tasks we account those as iso_ticks as well. ++ */ ++ if (rt_task(p) || task_running_iso(p)) ++ iso_tick(rq); ++ else ++ no_iso_tick(rq, 1); ++ ++ /* SCHED_FIFO tasks never run out of timeslice. */ ++ if (p->policy == SCHED_FIFO) ++ return; ++ ++ if (iso_task(p)) { ++ if (task_running_iso(p)) { ++ if (rq->iso_refractory) { ++ /* ++ * SCHED_ISO task is running as RT and limit ++ * has been hit. Force it to reschedule as ++ * SCHED_NORMAL by zeroing its time_slice ++ */ ++ p->time_slice = 0; ++ } ++ } else if (!rq->iso_refractory) { ++ /* Can now run again ISO. Reschedule to pick up prio */ ++ goto out_resched; ++ } ++ } ++ ++ /* ++ * Tasks that were scheduled in the first half of a tick are not ++ * allowed to run into the 2nd half of the next tick if they will ++ * run out of time slice in the interim. Otherwise, if they have ++ * less than RESCHED_US μs of time slice left they will be rescheduled. ++ * Dither is used as a backup for when hrexpiry is disabled or high res ++ * timers not configured in. ++ */ ++ if (p->time_slice - rq->dither >= RESCHED_US) ++ return; ++out_resched: ++ rq_lock(rq); ++ __set_tsk_resched(p); ++ rq_unlock(rq); ++} ++ ++#ifdef CONFIG_NO_HZ_FULL ++/* ++ * We can stop the timer tick any time highres timers are active since ++ * we rely entirely on highres timeouts for task expiry rescheduling. ++ */ ++static void sched_stop_tick(struct rq *rq, int cpu) ++{ ++ if (!hrexpiry_enabled(rq)) ++ return; ++ if (!tick_nohz_full_enabled()) ++ return; ++ if (!tick_nohz_full_cpu(cpu)) ++ return; ++ tick_nohz_dep_clear_cpu(cpu, TICK_DEP_BIT_SCHED); ++} ++ ++static inline void sched_start_tick(struct rq *rq, int cpu) ++{ ++ tick_nohz_dep_set_cpu(cpu, TICK_DEP_BIT_SCHED); ++} ++ ++/** ++ * scheduler_tick_max_deferment ++ * ++ * Keep at least one tick per second when a single ++ * active task is running. ++ * ++ * This makes sure that uptime continues to move forward, even ++ * with a very low granularity. ++ * ++ * Return: Maximum deferment in nanoseconds. ++ */ ++u64 scheduler_tick_max_deferment(void) ++{ ++ struct rq *rq = this_rq(); ++ unsigned long next, now = READ_ONCE(jiffies); ++ ++ next = rq->last_jiffy + HZ; ++ ++ if (time_before_eq(next, now)) ++ return 0; ++ ++ return jiffies_to_nsecs(next - now); ++} ++#else ++static inline void sched_stop_tick(struct rq *rq, int cpu) ++{ ++} ++ ++static inline void sched_start_tick(struct rq *rq, int cpu) ++{ ++} ++#endif ++ ++/* ++ * This function gets called by the timer code, with HZ frequency. ++ * We call it with interrupts disabled. ++ */ ++void scheduler_tick(void) ++{ ++ int cpu __maybe_unused = smp_processor_id(); ++ struct rq *rq = cpu_rq(cpu); ++ ++ sched_clock_tick(); ++ update_clocks(rq); ++ update_load_avg(rq, 0); ++ update_cpu_clock_tick(rq, rq->curr); ++ if (!rq_idle(rq)) ++ task_running_tick(rq); ++ else if (rq->last_jiffy > rq->last_scheduler_tick) ++ no_iso_tick(rq, rq->last_jiffy - rq->last_scheduler_tick); ++ rq->last_scheduler_tick = rq->last_jiffy; ++ rq->last_tick = rq->clock; ++ perf_event_task_tick(); ++ sched_stop_tick(rq, cpu); ++} ++ ++#if defined(CONFIG_PREEMPT) && (defined(CONFIG_DEBUG_PREEMPT) || \ ++ defined(CONFIG_PREEMPT_TRACER)) ++/* ++ * If the value passed in is equal to the current preempt count ++ * then we just disabled preemption. Start timing the latency. ++ */ ++static inline void preempt_latency_start(int val) ++{ ++ if (preempt_count() == val) { ++ unsigned long ip = get_lock_parent_ip(); ++#ifdef CONFIG_DEBUG_PREEMPT ++ current->preempt_disable_ip = ip; ++#endif ++ trace_preempt_off(CALLER_ADDR0, ip); ++ } ++} ++ ++void preempt_count_add(int val) ++{ ++#ifdef CONFIG_DEBUG_PREEMPT ++ /* ++ * Underflow? ++ */ ++ if (DEBUG_LOCKS_WARN_ON((preempt_count() < 0))) ++ return; ++#endif ++ __preempt_count_add(val); ++#ifdef CONFIG_DEBUG_PREEMPT ++ /* ++ * Spinlock count overflowing soon? ++ */ ++ DEBUG_LOCKS_WARN_ON((preempt_count() & PREEMPT_MASK) >= ++ PREEMPT_MASK - 10); ++#endif ++ preempt_latency_start(val); ++} ++EXPORT_SYMBOL(preempt_count_add); ++NOKPROBE_SYMBOL(preempt_count_add); ++ ++/* ++ * If the value passed in equals to the current preempt count ++ * then we just enabled preemption. Stop timing the latency. ++ */ ++static inline void preempt_latency_stop(int val) ++{ ++ if (preempt_count() == val) ++ trace_preempt_on(CALLER_ADDR0, get_lock_parent_ip()); ++} ++ ++void preempt_count_sub(int val) ++{ ++#ifdef CONFIG_DEBUG_PREEMPT ++ /* ++ * Underflow? ++ */ ++ if (DEBUG_LOCKS_WARN_ON(val > preempt_count())) ++ return; ++ /* ++ * Is the spinlock portion underflowing? ++ */ ++ if (DEBUG_LOCKS_WARN_ON((val < PREEMPT_MASK) && ++ !(preempt_count() & PREEMPT_MASK))) ++ return; ++#endif ++ ++ preempt_latency_stop(val); ++ __preempt_count_sub(val); ++} ++EXPORT_SYMBOL(preempt_count_sub); ++NOKPROBE_SYMBOL(preempt_count_sub); ++ ++#else ++static inline void preempt_latency_start(int val) { } ++static inline void preempt_latency_stop(int val) { } ++#endif ++ ++static inline unsigned long get_preempt_disable_ip(struct task_struct *p) ++{ ++#ifdef CONFIG_DEBUG_PREEMPT ++ return p->preempt_disable_ip; ++#else ++ return 0; ++#endif ++} ++ ++/* ++ * The time_slice is only refilled when it is empty and that is when we set a ++ * new deadline. Make sure update_clocks has been called recently to update ++ * rq->niffies. ++ */ ++static void time_slice_expired(struct task_struct *p, struct rq *rq) ++{ ++ p->time_slice = timeslice(); ++ p->deadline = rq->niffies + task_deadline_diff(p); ++#ifdef CONFIG_SMT_NICE ++ if (!p->mm) ++ p->smt_bias = 0; ++ else if (rt_task(p)) ++ p->smt_bias = 1 << 30; ++ else if (task_running_iso(p)) ++ p->smt_bias = 1 << 29; ++ else if (idleprio_task(p)) { ++ if (task_running_idle(p)) ++ p->smt_bias = 0; ++ else ++ p->smt_bias = 1; ++ } else if (--p->smt_bias < 1) ++ p->smt_bias = MAX_PRIO - p->static_prio; ++#endif ++} ++ ++/* ++ * Timeslices below RESCHED_US are considered as good as expired as there's no ++ * point rescheduling when there's so little time left. SCHED_BATCH tasks ++ * have been flagged be not latency sensitive and likely to be fully CPU ++ * bound so every time they're rescheduled they have their time_slice ++ * refilled, but get a new later deadline to have little effect on ++ * SCHED_NORMAL tasks. ++ ++ */ ++static inline void check_deadline(struct task_struct *p, struct rq *rq) ++{ ++ if (p->time_slice < RESCHED_US || batch_task(p)) ++ time_slice_expired(p, rq); ++} ++ ++/* ++ * Task selection with skiplists is a simple matter of picking off the first ++ * task in the sorted list, an O(1) operation. The lookup is amortised O(1) ++ * being bound to the number of processors. ++ * ++ * Runqueues are selectively locked based on their unlocked data and then ++ * unlocked if not needed. At most 3 locks will be held at any time and are ++ * released as soon as they're no longer needed. All balancing between CPUs ++ * is thus done here in an extremely simple first come best fit manner. ++ * ++ * This iterates over runqueues in cache locality order. In interactive mode ++ * it iterates over all CPUs and finds the task with the best key/deadline. ++ * In non-interactive mode it will only take a task if it's from the current ++ * runqueue or a runqueue with more tasks than the current one with a better ++ * key/deadline. ++ */ ++#ifdef CONFIG_SMP ++static inline struct task_struct ++*earliest_deadline_task(struct rq *rq, int cpu, struct task_struct *idle) ++{ ++ struct rq *locked = NULL, *chosen = NULL; ++ struct task_struct *edt = idle; ++ int i, best_entries = 0; ++ u64 best_key = ~0ULL; ++ ++ for (i = 0; i < num_possible_cpus(); i++) { ++ struct rq *other_rq = rq_order(rq, i); ++ int entries = other_rq->sl->entries; ++ skiplist_node *next; ++ ++ /* ++ * Check for queued entres lockless first. The local runqueue ++ * is locked so entries will always be accurate. ++ */ ++ if (!sched_interactive) { ++ /* ++ * Don't reschedule balance across nodes unless the CPU ++ * is idle. ++ */ ++ if (edt != idle && rq->cpu_locality[other_rq->cpu] > 3) ++ break; ++ if (entries <= best_entries) ++ continue; ++ } else if (!entries) ++ continue; ++ ++ /* if (i) implies other_rq != rq */ ++ if (i) { ++ /* Check for best id queued lockless first */ ++ if (other_rq->best_key >= best_key) ++ continue; ++ ++ if (unlikely(!trylock_rq(rq, other_rq))) ++ continue; ++ ++ /* Need to reevaluate entries after locking */ ++ entries = other_rq->sl->entries; ++ if (unlikely(!entries)) { ++ unlock_rq(other_rq); ++ continue; ++ } ++ } ++ ++ next = &other_rq->node; ++ /* ++ * In interactive mode we check beyond the best entry on other ++ * runqueues if we can't get the best for smt or affinity ++ * reasons. ++ */ ++ while ((next = next->next[0]) != &other_rq->node) { ++ struct task_struct *p; ++ u64 key = next->key; ++ ++ /* Reevaluate key after locking */ ++ if (key >= best_key) ++ break; ++ ++ p = next->value; ++ if (!smt_schedule(p, rq)) { ++ if (i && !sched_interactive) ++ break; ++ continue; ++ } ++ ++ /* Make sure affinity is ok */ ++ if (i) { ++ if (needs_other_cpu(p, cpu)) { ++ if (sched_interactive) ++ continue; ++ break; ++ } ++ /* From this point on p is the best so far */ ++ if (locked) ++ unlock_rq(locked); ++ chosen = locked = other_rq; ++ } ++ best_entries = entries; ++ best_key = key; ++ edt = p; ++ break; ++ } ++ /* rq->preempting is a hint only as the state may have changed ++ * since it was set with the resched call but if we have met ++ * the condition we can break out here. */ ++ if (edt == rq->preempting) ++ break; ++ if (i && other_rq != chosen) ++ unlock_rq(other_rq); ++ } ++ ++ if (likely(edt != idle)) ++ take_task(rq, cpu, edt); ++ ++ if (locked) ++ unlock_rq(locked); ++ ++ rq->preempting = NULL; ++ ++ return edt; ++} ++#else /* CONFIG_SMP */ ++static inline struct task_struct ++*earliest_deadline_task(struct rq *rq, int cpu, struct task_struct *idle) ++{ ++ struct task_struct *edt; ++ ++ if (unlikely(!rq->sl->entries)) ++ return idle; ++ edt = rq->node.next[0]->value; ++ take_task(rq, cpu, edt); ++ return edt; ++} ++#endif /* CONFIG_SMP */ ++ ++/* ++ * Print scheduling while atomic bug: ++ */ ++static noinline void __schedule_bug(struct task_struct *prev) ++{ ++ /* Save this before calling printk(), since that will clobber it */ ++ unsigned long preempt_disable_ip = get_preempt_disable_ip(current); ++ ++ if (oops_in_progress) ++ return; ++ ++ printk(KERN_ERR "BUG: scheduling while atomic: %s/%d/0x%08x\n", ++ prev->comm, prev->pid, preempt_count()); ++ ++ debug_show_held_locks(prev); ++ print_modules(); ++ if (irqs_disabled()) ++ print_irqtrace_events(prev); ++ if (IS_ENABLED(CONFIG_DEBUG_PREEMPT) ++ && in_atomic_preempt_off()) { ++ pr_err("Preemption disabled at:"); ++ print_ip_sym(preempt_disable_ip); ++ pr_cont("\n"); ++ } ++ dump_stack(); ++ add_taint(TAINT_WARN, LOCKDEP_STILL_OK); ++} ++ ++/* ++ * Various schedule()-time debugging checks and statistics: ++ */ ++static inline void schedule_debug(struct task_struct *prev) ++{ ++#ifdef CONFIG_SCHED_STACK_END_CHECK ++ if (task_stack_end_corrupted(prev)) ++ panic("corrupted stack end detected inside scheduler\n"); ++#endif ++ ++ if (unlikely(in_atomic_preempt_off())) { ++ __schedule_bug(prev); ++ preempt_count_set(PREEMPT_DISABLED); ++ } ++ rcu_sleep_check(); ++ ++ profile_hit(SCHED_PROFILING, __builtin_return_address(0)); ++ ++ schedstat_inc(this_rq()->sched_count); ++} ++ ++/* ++ * The currently running task's information is all stored in rq local data ++ * which is only modified by the local CPU. ++ */ ++static inline void set_rq_task(struct rq *rq, struct task_struct *p) ++{ ++ if (p == rq->idle || p->policy == SCHED_FIFO) ++ hrexpiry_clear(rq); ++ else ++ hrexpiry_start(rq, US_TO_NS(p->time_slice)); ++ if (rq->clock - rq->last_tick > HALF_JIFFY_NS) ++ rq->dither = 0; ++ else ++ rq->dither = rq_dither(rq); ++ ++ rq->rq_deadline = p->deadline; ++ rq->rq_prio = p->prio; ++#ifdef CONFIG_SMT_NICE ++ rq->rq_mm = p->mm; ++ rq->rq_smt_bias = p->smt_bias; ++#endif ++} ++ ++#ifdef CONFIG_SMT_NICE ++static void check_no_siblings(struct rq __maybe_unused *this_rq) {} ++static void wake_no_siblings(struct rq __maybe_unused *this_rq) {} ++static void (*check_siblings)(struct rq *this_rq) = &check_no_siblings; ++static void (*wake_siblings)(struct rq *this_rq) = &wake_no_siblings; ++ ++/* Iterate over smt siblings when we've scheduled a process on cpu and decide ++ * whether they should continue running or be descheduled. */ ++static void check_smt_siblings(struct rq *this_rq) ++{ ++ int other_cpu; ++ ++ for_each_cpu(other_cpu, &this_rq->thread_mask) { ++ struct task_struct *p; ++ struct rq *rq; ++ ++ rq = cpu_rq(other_cpu); ++ if (rq_idle(rq)) ++ continue; ++ p = rq->curr; ++ if (!smt_schedule(p, this_rq)) ++ resched_curr(rq); ++ } ++} ++ ++static void wake_smt_siblings(struct rq *this_rq) ++{ ++ int other_cpu; ++ ++ for_each_cpu(other_cpu, &this_rq->thread_mask) { ++ struct rq *rq; ++ ++ rq = cpu_rq(other_cpu); ++ if (rq_idle(rq)) ++ resched_idle(rq); ++ } ++} ++#else ++static void check_siblings(struct rq __maybe_unused *this_rq) {} ++static void wake_siblings(struct rq __maybe_unused *this_rq) {} ++#endif ++ ++/* ++ * schedule() is the main scheduler function. ++ * ++ * The main means of driving the scheduler and thus entering this function are: ++ * ++ * 1. Explicit blocking: mutex, semaphore, waitqueue, etc. ++ * ++ * 2. TIF_NEED_RESCHED flag is checked on interrupt and userspace return ++ * paths. For example, see arch/x86/entry_64.S. ++ * ++ * To drive preemption between tasks, the scheduler sets the flag in timer ++ * interrupt handler scheduler_tick(). ++ * ++ * 3. Wakeups don't really cause entry into schedule(). They add a ++ * task to the run-queue and that's it. ++ * ++ * Now, if the new task added to the run-queue preempts the current ++ * task, then the wakeup sets TIF_NEED_RESCHED and schedule() gets ++ * called on the nearest possible occasion: ++ * ++ * - If the kernel is preemptible (CONFIG_PREEMPT=y): ++ * ++ * - in syscall or exception context, at the next outmost ++ * preempt_enable(). (this might be as soon as the wake_up()'s ++ * spin_unlock()!) ++ * ++ * - in IRQ context, return from interrupt-handler to ++ * preemptible context ++ * ++ * - If the kernel is not preemptible (CONFIG_PREEMPT is not set) ++ * then at the next: ++ * ++ * - cond_resched() call ++ * - explicit schedule() call ++ * - return from syscall or exception to user-space ++ * - return from interrupt-handler to user-space ++ * ++ * WARNING: must be called with preemption disabled! ++ */ ++static void __sched notrace __schedule(bool preempt) ++{ ++ struct task_struct *prev, *next, *idle; ++ unsigned long *switch_count; ++ bool deactivate = false; ++ struct rq *rq; ++ u64 niffies; ++ int cpu; ++ ++ cpu = smp_processor_id(); ++ rq = cpu_rq(cpu); ++ prev = rq->curr; ++ idle = rq->idle; ++ ++ schedule_debug(prev); ++ ++ local_irq_disable(); ++ rcu_note_context_switch(preempt); ++ ++ /* ++ * Make sure that signal_pending_state()->signal_pending() below ++ * can't be reordered with __set_current_state(TASK_INTERRUPTIBLE) ++ * done by the caller to avoid the race with signal_wake_up(). ++ */ ++ rq_lock(rq); ++ smp_mb__after_spinlock(); ++#ifdef CONFIG_SMP ++ if (rq->preempt) { ++ /* ++ * Make sure resched_curr hasn't triggered a preemption ++ * locklessly on a task that has since scheduled away. Spurious ++ * wakeup of idle is okay though. ++ */ ++ if (unlikely(preempt && prev != idle && !test_tsk_need_resched(prev))) { ++ rq->preempt = NULL; ++ clear_preempt_need_resched(); ++ rq_unlock_irq(rq); ++ return; ++ } ++ rq->preempt = NULL; ++ } ++#endif ++ ++ switch_count = &prev->nivcsw; ++ if (!preempt && prev->state) { ++ if (unlikely(signal_pending_state(prev->state, prev))) { ++ prev->state = TASK_RUNNING; ++ } else { ++ deactivate = true; ++ prev->on_rq = 0; ++ ++ if (prev->in_iowait) { ++ atomic_inc(&rq->nr_iowait); ++ delayacct_blkio_start(); ++ } ++ ++ /* ++ * If a worker is going to sleep, notify and ++ * ask workqueue whether it wants to wake up a ++ * task to maintain concurrency. If so, wake ++ * up the task. ++ */ ++ if (prev->flags & PF_WQ_WORKER) { ++ struct task_struct *to_wakeup; ++ ++ to_wakeup = wq_worker_sleeping(prev); ++ if (to_wakeup) ++ try_to_wake_up_local(to_wakeup); ++ } ++ } ++ switch_count = &prev->nvcsw; ++ } ++ ++ /* ++ * Store the niffy value here for use by the next task's last_ran ++ * below to avoid losing niffies due to update_clocks being called ++ * again after this point. ++ */ ++ update_clocks(rq); ++ niffies = rq->niffies; ++ update_cpu_clock_switch(rq, prev); ++ ++ clear_tsk_need_resched(prev); ++ clear_preempt_need_resched(); ++ ++ if (idle != prev) { ++ check_deadline(prev, rq); ++ return_task(prev, rq, cpu, deactivate); ++ } ++ ++ next = earliest_deadline_task(rq, cpu, idle); ++ if (likely(next->prio != PRIO_LIMIT)) ++ clear_cpuidle_map(cpu); ++ else { ++ set_cpuidle_map(cpu); ++ update_load_avg(rq, 0); ++ } ++ ++ set_rq_task(rq, next); ++ next->last_ran = niffies; ++ ++ if (likely(prev != next)) { ++ /* ++ * Don't reschedule an idle task or deactivated tasks ++ */ ++ if (prev != idle && !deactivate) ++ resched_suitable_idle(prev); ++ if (next != idle) ++ check_siblings(rq); ++ else ++ wake_siblings(rq); ++ rq->nr_switches++; ++ rq->curr = next; ++ /* ++ * The membarrier system call requires each architecture ++ * to have a full memory barrier after updating ++ * rq->curr, before returning to user-space. For TSO ++ * (e.g. x86), the architecture must provide its own ++ * barrier in switch_mm(). For weakly ordered machines ++ * for which spin_unlock() acts as a full memory ++ * barrier, finish_lock_switch() in common code takes ++ * care of this barrier. For weakly ordered machines for ++ * which spin_unlock() acts as a RELEASE barrier (only ++ * arm64 and PowerPC), arm64 has a full barrier in ++ * switch_to(), and PowerPC has ++ * smp_mb__after_unlock_lock() before ++ * finish_lock_switch(). ++ */ ++ ++*switch_count; ++ ++ trace_sched_switch(preempt, prev, next); ++ context_switch(rq, prev, next); /* unlocks the rq */ ++ } else { ++ check_siblings(rq); ++ rq_unlock(rq); ++ do_pending_softirq(rq, next); ++ local_irq_enable(); ++ } ++} ++ ++void __noreturn do_task_dead(void) ++{ ++ /* ++ * The setting of TASK_RUNNING by try_to_wake_up() may be delayed ++ * when the following two conditions become true. ++ * - There is race condition of mmap_sem (It is acquired by ++ * exit_mm()), and ++ * - SMI occurs before setting TASK_RUNINNG. ++ * (or hypervisor of virtual machine switches to other guest) ++ * As a result, we may become TASK_RUNNING after becoming TASK_DEAD ++ * ++ * To avoid it, we have to wait for releasing tsk->pi_lock which ++ * is held by try_to_wake_up() ++ */ ++ raw_spin_lock_irq(¤t->pi_lock); ++ raw_spin_unlock_irq(¤t->pi_lock); ++ ++ /* Causes final put_task_struct in finish_task_switch(). */ ++ __set_current_state(TASK_DEAD); ++ ++ /* Tell freezer to ignore us: */ ++ current->flags |= PF_NOFREEZE; ++ __schedule(false); ++ BUG(); ++ ++ /* Avoid "noreturn function does return" - but don't continue if BUG() is a NOP: */ ++ for (;;) ++ cpu_relax(); ++} ++ ++static inline void sched_submit_work(struct task_struct *tsk) ++{ ++ if (!tsk->state || tsk_is_pi_blocked(tsk) || ++ preempt_count() || ++ signal_pending_state(tsk->state, tsk)) ++ return; ++ ++ /* ++ * If we are going to sleep and we have plugged IO queued, ++ * make sure to submit it to avoid deadlocks. ++ */ ++ if (blk_needs_flush_plug(tsk)) ++ blk_schedule_flush_plug(tsk); ++} ++ ++asmlinkage __visible void __sched schedule(void) ++{ ++ struct task_struct *tsk = current; ++ ++ sched_submit_work(tsk); ++ do { ++ preempt_disable(); ++ __schedule(false); ++ sched_preempt_enable_no_resched(); ++ } while (need_resched()); ++} ++ ++EXPORT_SYMBOL(schedule); ++ ++/* ++ * synchronize_rcu_tasks() makes sure that no task is stuck in preempted ++ * state (have scheduled out non-voluntarily) by making sure that all ++ * tasks have either left the run queue or have gone into user space. ++ * As idle tasks do not do either, they must not ever be preempted ++ * (schedule out non-voluntarily). ++ * ++ * schedule_idle() is similar to schedule_preempt_disable() except that it ++ * never enables preemption because it does not call sched_submit_work(). ++ */ ++void __sched schedule_idle(void) ++{ ++ /* ++ * As this skips calling sched_submit_work(), which the idle task does ++ * regardless because that function is a nop when the task is in a ++ * TASK_RUNNING state, make sure this isn't used someplace that the ++ * current task can be in any other state. Note, idle is always in the ++ * TASK_RUNNING state. ++ */ ++ WARN_ON_ONCE(current->state); ++ do { ++ __schedule(false); ++ } while (need_resched()); ++} ++ ++#ifdef CONFIG_CONTEXT_TRACKING ++asmlinkage __visible void __sched schedule_user(void) ++{ ++ /* ++ * If we come here after a random call to set_need_resched(), ++ * or we have been woken up remotely but the IPI has not yet arrived, ++ * we haven't yet exited the RCU idle mode. Do it here manually until ++ * we find a better solution. ++ * ++ * NB: There are buggy callers of this function. Ideally we ++ * should warn if prev_state != IN_USER, but that will trigger ++ * too frequently to make sense yet. ++ */ ++ enum ctx_state prev_state = exception_enter(); ++ schedule(); ++ exception_exit(prev_state); ++} ++#endif ++ ++/** ++ * schedule_preempt_disabled - called with preemption disabled ++ * ++ * Returns with preemption disabled. Note: preempt_count must be 1 ++ */ ++void __sched schedule_preempt_disabled(void) ++{ ++ sched_preempt_enable_no_resched(); ++ schedule(); ++ preempt_disable(); ++} ++ ++static void __sched notrace preempt_schedule_common(void) ++{ ++ do { ++ /* ++ * Because the function tracer can trace preempt_count_sub() ++ * and it also uses preempt_enable/disable_notrace(), if ++ * NEED_RESCHED is set, the preempt_enable_notrace() called ++ * by the function tracer will call this function again and ++ * cause infinite recursion. ++ * ++ * Preemption must be disabled here before the function ++ * tracer can trace. Break up preempt_disable() into two ++ * calls. One to disable preemption without fear of being ++ * traced. The other to still record the preemption latency, ++ * which can also be traced by the function tracer. ++ */ ++ preempt_disable_notrace(); ++ preempt_latency_start(1); ++ __schedule(true); ++ preempt_latency_stop(1); ++ preempt_enable_no_resched_notrace(); ++ ++ /* ++ * Check again in case we missed a preemption opportunity ++ * between schedule and now. ++ */ ++ } while (need_resched()); ++} ++ ++#ifdef CONFIG_PREEMPT ++/* ++ * this is the entry point to schedule() from in-kernel preemption ++ * off of preempt_enable. Kernel preemptions off return from interrupt ++ * occur there and call schedule directly. ++ */ ++asmlinkage __visible void __sched notrace preempt_schedule(void) ++{ ++ /* ++ * If there is a non-zero preempt_count or interrupts are disabled, ++ * we do not want to preempt the current task. Just return.. ++ */ ++ if (likely(!preemptible())) ++ return; ++ ++ preempt_schedule_common(); ++} ++NOKPROBE_SYMBOL(preempt_schedule); ++EXPORT_SYMBOL(preempt_schedule); ++ ++/** ++ * preempt_schedule_notrace - preempt_schedule called by tracing ++ * ++ * The tracing infrastructure uses preempt_enable_notrace to prevent ++ * recursion and tracing preempt enabling caused by the tracing ++ * infrastructure itself. But as tracing can happen in areas coming ++ * from userspace or just about to enter userspace, a preempt enable ++ * can occur before user_exit() is called. This will cause the scheduler ++ * to be called when the system is still in usermode. ++ * ++ * To prevent this, the preempt_enable_notrace will use this function ++ * instead of preempt_schedule() to exit user context if needed before ++ * calling the scheduler. ++ */ ++asmlinkage __visible void __sched notrace preempt_schedule_notrace(void) ++{ ++ enum ctx_state prev_ctx; ++ ++ if (likely(!preemptible())) ++ return; ++ ++ do { ++ /* ++ * Because the function tracer can trace preempt_count_sub() ++ * and it also uses preempt_enable/disable_notrace(), if ++ * NEED_RESCHED is set, the preempt_enable_notrace() called ++ * by the function tracer will call this function again and ++ * cause infinite recursion. ++ * ++ * Preemption must be disabled here before the function ++ * tracer can trace. Break up preempt_disable() into two ++ * calls. One to disable preemption without fear of being ++ * traced. The other to still record the preemption latency, ++ * which can also be traced by the function tracer. ++ */ ++ preempt_disable_notrace(); ++ preempt_latency_start(1); ++ /* ++ * Needs preempt disabled in case user_exit() is traced ++ * and the tracer calls preempt_enable_notrace() causing ++ * an infinite recursion. ++ */ ++ prev_ctx = exception_enter(); ++ __schedule(true); ++ exception_exit(prev_ctx); ++ ++ preempt_latency_stop(1); ++ preempt_enable_no_resched_notrace(); ++ } while (need_resched()); ++} ++EXPORT_SYMBOL_GPL(preempt_schedule_notrace); ++ ++#endif /* CONFIG_PREEMPT */ ++ ++/* ++ * this is the entry point to schedule() from kernel preemption ++ * off of irq context. ++ * Note, that this is called and return with irqs disabled. This will ++ * protect us against recursive calling from irq. ++ */ ++asmlinkage __visible void __sched preempt_schedule_irq(void) ++{ ++ enum ctx_state prev_state; ++ ++ /* Catch callers which need to be fixed */ ++ BUG_ON(preempt_count() || !irqs_disabled()); ++ ++ prev_state = exception_enter(); ++ ++ do { ++ preempt_disable(); ++ local_irq_enable(); ++ __schedule(true); ++ local_irq_disable(); ++ sched_preempt_enable_no_resched(); ++ } while (need_resched()); ++ ++ exception_exit(prev_state); ++} ++ ++int default_wake_function(wait_queue_entry_t *curr, unsigned mode, int wake_flags, ++ void *key) ++{ ++ return try_to_wake_up(curr->private, mode, wake_flags); ++} ++EXPORT_SYMBOL(default_wake_function); ++ ++#ifdef CONFIG_RT_MUTEXES ++ ++static inline int __rt_effective_prio(struct task_struct *pi_task, int prio) ++{ ++ if (pi_task) ++ prio = min(prio, pi_task->prio); ++ ++ return prio; ++} ++ ++static inline int rt_effective_prio(struct task_struct *p, int prio) ++{ ++ struct task_struct *pi_task = rt_mutex_get_top_task(p); ++ ++ return __rt_effective_prio(pi_task, prio); ++} ++ ++/* ++ * rt_mutex_setprio - set the current priority of a task ++ * @p: task to boost ++ * @pi_task: donor task ++ * ++ * This function changes the 'effective' priority of a task. It does ++ * not touch ->normal_prio like __setscheduler(). ++ * ++ * Used by the rt_mutex code to implement priority inheritance ++ * logic. Call site only calls if the priority of the task changed. ++ */ ++void rt_mutex_setprio(struct task_struct *p, struct task_struct *pi_task) ++{ ++ int prio, oldprio; ++ struct rq *rq; ++ ++ /* XXX used to be waiter->prio, not waiter->task->prio */ ++ prio = __rt_effective_prio(pi_task, p->normal_prio); ++ ++ /* ++ * If nothing changed; bail early. ++ */ ++ if (p->pi_top_task == pi_task && prio == p->prio) ++ return; ++ ++ rq = __task_rq_lock(p); ++ update_rq_clock(rq); ++ /* ++ * Set under pi_lock && rq->lock, such that the value can be used under ++ * either lock. ++ * ++ * Note that there is loads of tricky to make this pointer cache work ++ * right. rt_mutex_slowunlock()+rt_mutex_postunlock() work together to ++ * ensure a task is de-boosted (pi_task is set to NULL) before the ++ * task is allowed to run again (and can exit). This ensures the pointer ++ * points to a blocked task -- which guaratees the task is present. ++ */ ++ p->pi_top_task = pi_task; ++ ++ /* ++ * For FIFO/RR we only need to set prio, if that matches we're done. ++ */ ++ if (prio == p->prio) ++ goto out_unlock; ++ ++ /* ++ * Idle task boosting is a nono in general. There is one ++ * exception, when PREEMPT_RT and NOHZ is active: ++ * ++ * The idle task calls get_next_timer_interrupt() and holds ++ * the timer wheel base->lock on the CPU and another CPU wants ++ * to access the timer (probably to cancel it). We can safely ++ * ignore the boosting request, as the idle CPU runs this code ++ * with interrupts disabled and will complete the lock ++ * protected section without being interrupted. So there is no ++ * real need to boost. ++ */ ++ if (unlikely(p == rq->idle)) { ++ WARN_ON(p != rq->curr); ++ WARN_ON(p->pi_blocked_on); ++ goto out_unlock; ++ } ++ ++ trace_sched_pi_setprio(p, pi_task); ++ oldprio = p->prio; ++ p->prio = prio; ++ if (task_running(rq, p)){ ++ if (prio > oldprio) ++ resched_task(p); ++ } else if (task_queued(p)) { ++ dequeue_task(rq, p, DEQUEUE_SAVE); ++ enqueue_task(rq, p, ENQUEUE_RESTORE); ++ if (prio < oldprio) ++ try_preempt(p, rq); ++ } ++out_unlock: ++ __task_rq_unlock(rq); ++} ++#else ++static inline int rt_effective_prio(struct task_struct *p, int prio) ++{ ++ return prio; ++} ++#endif ++ ++/* ++ * Adjust the deadline for when the priority is to change, before it's ++ * changed. ++ */ ++static inline void adjust_deadline(struct task_struct *p, int new_prio) ++{ ++ p->deadline += static_deadline_diff(new_prio) - task_deadline_diff(p); ++} ++ ++void set_user_nice(struct task_struct *p, long nice) ++{ ++ int new_static, old_static; ++ unsigned long flags; ++ struct rq *rq; ++ ++ if (task_nice(p) == nice || nice < MIN_NICE || nice > MAX_NICE) ++ return; ++ new_static = NICE_TO_PRIO(nice); ++ /* ++ * We have to be careful, if called from sys_setpriority(), ++ * the task might be in the middle of scheduling on another CPU. ++ */ ++ rq = task_rq_lock(p, &flags); ++ update_rq_clock(rq); ++ ++ /* ++ * The RT priorities are set via sched_setscheduler(), but we still ++ * allow the 'normal' nice value to be set - but as expected ++ * it wont have any effect on scheduling until the task is ++ * not SCHED_NORMAL/SCHED_BATCH: ++ */ ++ if (has_rt_policy(p)) { ++ p->static_prio = new_static; ++ goto out_unlock; ++ } ++ ++ adjust_deadline(p, new_static); ++ old_static = p->static_prio; ++ p->static_prio = new_static; ++ p->prio = effective_prio(p); ++ ++ if (task_queued(p)) { ++ dequeue_task(rq, p, DEQUEUE_SAVE); ++ enqueue_task(rq, p, ENQUEUE_RESTORE); ++ if (new_static < old_static) ++ try_preempt(p, rq); ++ } else if (task_running(rq, p)) { ++ set_rq_task(rq, p); ++ if (old_static < new_static) ++ resched_task(p); ++ } ++out_unlock: ++ task_rq_unlock(rq, p, &flags); ++} ++EXPORT_SYMBOL(set_user_nice); ++ ++/* ++ * can_nice - check if a task can reduce its nice value ++ * @p: task ++ * @nice: nice value ++ */ ++int can_nice(const struct task_struct *p, const int nice) ++{ ++ /* Convert nice value [19,-20] to rlimit style value [1,40] */ ++ int nice_rlim = nice_to_rlimit(nice); ++ ++ return (nice_rlim <= task_rlimit(p, RLIMIT_NICE) || ++ capable(CAP_SYS_NICE)); ++} ++ ++#ifdef __ARCH_WANT_SYS_NICE ++ ++/* ++ * sys_nice - change the priority of the current process. ++ * @increment: priority increment ++ * ++ * sys_setpriority is a more generic, but much slower function that ++ * does similar things. ++ */ ++SYSCALL_DEFINE1(nice, int, increment) ++{ ++ long nice, retval; ++ ++ /* ++ * Setpriority might change our priority at the same moment. ++ * We don't have to worry. Conceptually one call occurs first ++ * and we have a single winner. ++ */ ++ ++ increment = clamp(increment, -NICE_WIDTH, NICE_WIDTH); ++ nice = task_nice(current) + increment; ++ ++ nice = clamp_val(nice, MIN_NICE, MAX_NICE); ++ if (increment < 0 && !can_nice(current, nice)) ++ return -EPERM; ++ ++ retval = security_task_setnice(current, nice); ++ if (retval) ++ return retval; ++ ++ set_user_nice(current, nice); ++ return 0; ++} ++ ++#endif ++ ++/** ++ * task_prio - return the priority value of a given task. ++ * @p: the task in question. ++ * ++ * Return: The priority value as seen by users in /proc. ++ * RT tasks are offset by -100. Normal tasks are centered around 1, value goes ++ * from 0 (SCHED_ISO) up to 82 (nice +19 SCHED_IDLEPRIO). ++ */ ++int task_prio(const struct task_struct *p) ++{ ++ int delta, prio = p->prio - MAX_RT_PRIO; ++ ++ /* rt tasks and iso tasks */ ++ if (prio <= 0) ++ goto out; ++ ++ /* Convert to ms to avoid overflows */ ++ delta = NS_TO_MS(p->deadline - task_rq(p)->niffies); ++ if (unlikely(delta < 0)) ++ delta = 0; ++ delta = delta * 40 / ms_longest_deadline_diff(); ++ if (delta <= 80) ++ prio += delta; ++ if (idleprio_task(p)) ++ prio += 40; ++out: ++ return prio; ++} ++ ++/** ++ * idle_cpu - is a given CPU idle currently? ++ * @cpu: the processor in question. ++ * ++ * Return: 1 if the CPU is currently idle. 0 otherwise. ++ */ ++int idle_cpu(int cpu) ++{ ++ return cpu_curr(cpu) == cpu_rq(cpu)->idle; ++} ++ ++/** ++ * idle_task - return the idle task for a given CPU. ++ * @cpu: the processor in question. ++ * ++ * Return: The idle task for the CPU @cpu. ++ */ ++struct task_struct *idle_task(int cpu) ++{ ++ return cpu_rq(cpu)->idle; ++} ++ ++/** ++ * find_process_by_pid - find a process with a matching PID value. ++ * @pid: the pid in question. ++ * ++ * The task of @pid, if found. %NULL otherwise. ++ */ ++static inline struct task_struct *find_process_by_pid(pid_t pid) ++{ ++ return pid ? find_task_by_vpid(pid) : current; ++} ++ ++/* Actually do priority change: must hold rq lock. */ ++static void __setscheduler(struct task_struct *p, struct rq *rq, int policy, ++ int prio, bool keep_boost) ++{ ++ int oldrtprio, oldprio; ++ ++ p->policy = policy; ++ oldrtprio = p->rt_priority; ++ p->rt_priority = prio; ++ p->normal_prio = normal_prio(p); ++ oldprio = p->prio; ++ /* ++ * Keep a potential priority boosting if called from ++ * sched_setscheduler(). ++ */ ++ p->prio = normal_prio(p); ++ if (keep_boost) ++ p->prio = rt_effective_prio(p, p->prio); ++ ++ if (task_running(rq, p)) { ++ set_rq_task(rq, p); ++ resched_task(p); ++ } else if (task_queued(p)) { ++ dequeue_task(rq, p, DEQUEUE_SAVE); ++ enqueue_task(rq, p, ENQUEUE_RESTORE); ++ if (p->prio < oldprio || p->rt_priority > oldrtprio) ++ try_preempt(p, rq); ++ } ++} ++ ++/* ++ * Check the target process has a UID that matches the current process's ++ */ ++static bool check_same_owner(struct task_struct *p) ++{ ++ const struct cred *cred = current_cred(), *pcred; ++ bool match; ++ ++ rcu_read_lock(); ++ pcred = __task_cred(p); ++ match = (uid_eq(cred->euid, pcred->euid) || ++ uid_eq(cred->euid, pcred->uid)); ++ rcu_read_unlock(); ++ return match; ++} ++ ++static int ++__sched_setscheduler(struct task_struct *p, int policy, ++ const struct sched_param *param, bool user, bool pi) ++{ ++ struct sched_param zero_param = { .sched_priority = 0 }; ++ unsigned long flags, rlim_rtprio = 0; ++ int retval, oldpolicy = -1; ++ int reset_on_fork; ++ struct rq *rq; ++ ++ /* The pi code expects interrupts enabled */ ++ BUG_ON(pi && in_interrupt()); ++ ++ if (is_rt_policy(policy) && !capable(CAP_SYS_NICE)) { ++ unsigned long lflags; ++ ++ if (!lock_task_sighand(p, &lflags)) ++ return -ESRCH; ++ rlim_rtprio = task_rlimit(p, RLIMIT_RTPRIO); ++ unlock_task_sighand(p, &lflags); ++ if (rlim_rtprio) ++ goto recheck; ++ /* ++ * If the caller requested an RT policy without having the ++ * necessary rights, we downgrade the policy to SCHED_ISO. ++ * We also set the parameter to zero to pass the checks. ++ */ ++ policy = SCHED_ISO; ++ param = &zero_param; ++ } ++recheck: ++ /* Double check policy once rq lock held */ ++ if (policy < 0) { ++ reset_on_fork = p->sched_reset_on_fork; ++ policy = oldpolicy = p->policy; ++ } else { ++ reset_on_fork = !!(policy & SCHED_RESET_ON_FORK); ++ policy &= ~SCHED_RESET_ON_FORK; ++ ++ if (!SCHED_RANGE(policy)) ++ return -EINVAL; ++ } ++ ++ /* ++ * Valid priorities for SCHED_FIFO and SCHED_RR are ++ * 1..MAX_USER_RT_PRIO-1, valid priority for SCHED_NORMAL and ++ * SCHED_BATCH is 0. ++ */ ++ if (param->sched_priority < 0 || ++ (p->mm && param->sched_priority > MAX_USER_RT_PRIO - 1) || ++ (!p->mm && param->sched_priority > MAX_RT_PRIO - 1)) ++ return -EINVAL; ++ if (is_rt_policy(policy) != (param->sched_priority != 0)) ++ return -EINVAL; ++ ++ /* ++ * Allow unprivileged RT tasks to decrease priority: ++ */ ++ if (user && !capable(CAP_SYS_NICE)) { ++ if (is_rt_policy(policy)) { ++ unsigned long rlim_rtprio = ++ task_rlimit(p, RLIMIT_RTPRIO); ++ ++ /* Can't set/change the rt policy */ ++ if (policy != p->policy && !rlim_rtprio) ++ return -EPERM; ++ ++ /* Can't increase priority */ ++ if (param->sched_priority > p->rt_priority && ++ param->sched_priority > rlim_rtprio) ++ return -EPERM; ++ } else { ++ switch (p->policy) { ++ /* ++ * Can only downgrade policies but not back to ++ * SCHED_NORMAL ++ */ ++ case SCHED_ISO: ++ if (policy == SCHED_ISO) ++ goto out; ++ if (policy != SCHED_NORMAL) ++ return -EPERM; ++ break; ++ case SCHED_BATCH: ++ if (policy == SCHED_BATCH) ++ goto out; ++ if (policy != SCHED_IDLEPRIO) ++ return -EPERM; ++ break; ++ case SCHED_IDLEPRIO: ++ if (policy == SCHED_IDLEPRIO) ++ goto out; ++ return -EPERM; ++ default: ++ break; ++ } ++ } ++ ++ /* Can't change other user's priorities */ ++ if (!check_same_owner(p)) ++ return -EPERM; ++ ++ /* Normal users shall not reset the sched_reset_on_fork flag: */ ++ if (p->sched_reset_on_fork && !reset_on_fork) ++ return -EPERM; ++ } ++ ++ if (user) { ++ retval = security_task_setscheduler(p); ++ if (retval) ++ return retval; ++ } ++ ++ /* ++ * Make sure no PI-waiters arrive (or leave) while we are ++ * changing the priority of the task: ++ * ++ * To be able to change p->policy safely, the runqueue lock must be ++ * held. ++ */ ++ rq = task_rq_lock(p, &flags); ++ update_rq_clock(rq); ++ ++ /* ++ * Changing the policy of the stop threads its a very bad idea: ++ */ ++ if (p == rq->stop) { ++ task_rq_unlock(rq, p, &flags); ++ return -EINVAL; ++ } ++ ++ /* ++ * If not changing anything there's no need to proceed further: ++ */ ++ if (unlikely(policy == p->policy && (!is_rt_policy(policy) || ++ param->sched_priority == p->rt_priority))) { ++ task_rq_unlock(rq, p, &flags); ++ return 0; ++ } ++ ++ /* Re-check policy now with rq lock held */ ++ if (unlikely(oldpolicy != -1 && oldpolicy != p->policy)) { ++ policy = oldpolicy = -1; ++ task_rq_unlock(rq, p, &flags); ++ goto recheck; ++ } ++ p->sched_reset_on_fork = reset_on_fork; ++ ++ __setscheduler(p, rq, policy, param->sched_priority, pi); ++ task_rq_unlock(rq, p, &flags); ++ ++ if (pi) ++ rt_mutex_adjust_pi(p); ++out: ++ return 0; ++} ++ ++/** ++ * sched_setscheduler - change the scheduling policy and/or RT priority of a thread. ++ * @p: the task in question. ++ * @policy: new policy. ++ * @param: structure containing the new RT priority. ++ * ++ * Return: 0 on success. An error code otherwise. ++ * ++ * NOTE that the task may be already dead. ++ */ ++int sched_setscheduler(struct task_struct *p, int policy, ++ const struct sched_param *param) ++{ ++ return __sched_setscheduler(p, policy, param, true, true); ++} ++ ++EXPORT_SYMBOL_GPL(sched_setscheduler); ++ ++int sched_setattr(struct task_struct *p, const struct sched_attr *attr) ++{ ++ const struct sched_param param = { .sched_priority = attr->sched_priority }; ++ int policy = attr->sched_policy; ++ ++ return __sched_setscheduler(p, policy, ¶m, true, true); ++} ++EXPORT_SYMBOL_GPL(sched_setattr); ++ ++/** ++ * sched_setscheduler_nocheck - change the scheduling policy and/or RT priority of a thread from kernelspace. ++ * @p: the task in question. ++ * @policy: new policy. ++ * @param: structure containing the new RT priority. ++ * ++ * Just like sched_setscheduler, only don't bother checking if the ++ * current context has permission. For example, this is needed in ++ * stop_machine(): we create temporary high priority worker threads, ++ * but our caller might not have that capability. ++ * ++ * Return: 0 on success. An error code otherwise. ++ */ ++int sched_setscheduler_nocheck(struct task_struct *p, int policy, ++ const struct sched_param *param) ++{ ++ return __sched_setscheduler(p, policy, param, false, true); ++} ++EXPORT_SYMBOL_GPL(sched_setscheduler_nocheck); ++ ++static int ++do_sched_setscheduler(pid_t pid, int policy, struct sched_param __user *param) ++{ ++ struct sched_param lparam; ++ struct task_struct *p; ++ int retval; ++ ++ if (!param || pid < 0) ++ return -EINVAL; ++ if (copy_from_user(&lparam, param, sizeof(struct sched_param))) ++ return -EFAULT; ++ ++ rcu_read_lock(); ++ retval = -ESRCH; ++ p = find_process_by_pid(pid); ++ if (p != NULL) ++ retval = sched_setscheduler(p, policy, &lparam); ++ rcu_read_unlock(); ++ ++ return retval; ++} ++ ++/* ++ * Mimics kernel/events/core.c perf_copy_attr(). ++ */ ++static int sched_copy_attr(struct sched_attr __user *uattr, ++ struct sched_attr *attr) ++{ ++ u32 size; ++ int ret; ++ ++ if (!access_ok(VERIFY_WRITE, uattr, SCHED_ATTR_SIZE_VER0)) ++ return -EFAULT; ++ ++ /* Zero the full structure, so that a short copy will be nice: */ ++ memset(attr, 0, sizeof(*attr)); ++ ++ ret = get_user(size, &uattr->size); ++ if (ret) ++ return ret; ++ ++ /* Bail out on silly large: */ ++ if (size > PAGE_SIZE) ++ goto err_size; ++ ++ /* ABI compatibility quirk: */ ++ if (!size) ++ size = SCHED_ATTR_SIZE_VER0; ++ ++ if (size < SCHED_ATTR_SIZE_VER0) ++ goto err_size; ++ ++ /* ++ * If we're handed a bigger struct than we know of, ++ * ensure all the unknown bits are 0 - i.e. new ++ * user-space does not rely on any kernel feature ++ * extensions we dont know about yet. ++ */ ++ if (size > sizeof(*attr)) { ++ unsigned char __user *addr; ++ unsigned char __user *end; ++ unsigned char val; ++ ++ addr = (void __user *)uattr + sizeof(*attr); ++ end = (void __user *)uattr + size; ++ ++ for (; addr < end; addr++) { ++ ret = get_user(val, addr); ++ if (ret) ++ return ret; ++ if (val) ++ goto err_size; ++ } ++ size = sizeof(*attr); ++ } ++ ++ ret = copy_from_user(attr, uattr, size); ++ if (ret) ++ return -EFAULT; ++ ++ /* ++ * XXX: Do we want to be lenient like existing syscalls; or do we want ++ * to be strict and return an error on out-of-bounds values? ++ */ ++ attr->sched_nice = clamp(attr->sched_nice, -20, 19); ++ ++ /* sched/core.c uses zero here but we already know ret is zero */ ++ return 0; ++ ++err_size: ++ put_user(sizeof(*attr), &uattr->size); ++ return -E2BIG; ++} ++ ++/* ++ * sched_setparam() passes in -1 for its policy, to let the functions ++ * it calls know not to change it. ++ */ ++#define SETPARAM_POLICY -1 ++ ++/** ++ * sys_sched_setscheduler - set/change the scheduler policy and RT priority ++ * @pid: the pid in question. ++ * @policy: new policy. ++ * @param: structure containing the new RT priority. ++ * ++ * Return: 0 on success. An error code otherwise. ++ */ ++SYSCALL_DEFINE3(sched_setscheduler, pid_t, pid, int, policy, struct sched_param __user *, param) ++{ ++ if (policy < 0) ++ return -EINVAL; ++ ++ return do_sched_setscheduler(pid, policy, param); ++} ++ ++/** ++ * sys_sched_setparam - set/change the RT priority of a thread ++ * @pid: the pid in question. ++ * @param: structure containing the new RT priority. ++ * ++ * Return: 0 on success. An error code otherwise. ++ */ ++SYSCALL_DEFINE2(sched_setparam, pid_t, pid, struct sched_param __user *, param) ++{ ++ return do_sched_setscheduler(pid, SETPARAM_POLICY, param); ++} ++ ++/** ++ * sys_sched_setattr - same as above, but with extended sched_attr ++ * @pid: the pid in question. ++ * @uattr: structure containing the extended parameters. ++ */ ++SYSCALL_DEFINE3(sched_setattr, pid_t, pid, struct sched_attr __user *, uattr, ++ unsigned int, flags) ++{ ++ struct sched_attr attr; ++ struct task_struct *p; ++ int retval; ++ ++ if (!uattr || pid < 0 || flags) ++ return -EINVAL; ++ ++ retval = sched_copy_attr(uattr, &attr); ++ if (retval) ++ return retval; ++ ++ if ((int)attr.sched_policy < 0) ++ return -EINVAL; ++ ++ rcu_read_lock(); ++ retval = -ESRCH; ++ p = find_process_by_pid(pid); ++ if (p != NULL) ++ retval = sched_setattr(p, &attr); ++ rcu_read_unlock(); ++ ++ return retval; ++} ++ ++/** ++ * sys_sched_getscheduler - get the policy (scheduling class) of a thread ++ * @pid: the pid in question. ++ * ++ * Return: On success, the policy of the thread. Otherwise, a negative error ++ * code. ++ */ ++SYSCALL_DEFINE1(sched_getscheduler, pid_t, pid) ++{ ++ struct task_struct *p; ++ int retval = -EINVAL; ++ ++ if (pid < 0) ++ goto out_nounlock; ++ ++ retval = -ESRCH; ++ rcu_read_lock(); ++ p = find_process_by_pid(pid); ++ if (p) { ++ retval = security_task_getscheduler(p); ++ if (!retval) ++ retval = p->policy; ++ } ++ rcu_read_unlock(); ++ ++out_nounlock: ++ return retval; ++} ++ ++/** ++ * sys_sched_getscheduler - get the RT priority of a thread ++ * @pid: the pid in question. ++ * @param: structure containing the RT priority. ++ * ++ * Return: On success, 0 and the RT priority is in @param. Otherwise, an error ++ * code. ++ */ ++SYSCALL_DEFINE2(sched_getparam, pid_t, pid, struct sched_param __user *, param) ++{ ++ struct sched_param lp = { .sched_priority = 0 }; ++ struct task_struct *p; ++ int retval = -EINVAL; ++ ++ if (!param || pid < 0) ++ goto out_nounlock; ++ ++ rcu_read_lock(); ++ p = find_process_by_pid(pid); ++ retval = -ESRCH; ++ if (!p) ++ goto out_unlock; ++ ++ retval = security_task_getscheduler(p); ++ if (retval) ++ goto out_unlock; ++ ++ if (has_rt_policy(p)) ++ lp.sched_priority = p->rt_priority; ++ rcu_read_unlock(); ++ ++ /* ++ * This one might sleep, we cannot do it with a spinlock held ... ++ */ ++ retval = copy_to_user(param, &lp, sizeof(*param)) ? -EFAULT : 0; ++ ++out_nounlock: ++ return retval; ++ ++out_unlock: ++ rcu_read_unlock(); ++ return retval; ++} ++ ++static int sched_read_attr(struct sched_attr __user *uattr, ++ struct sched_attr *attr, ++ unsigned int usize) ++{ ++ int ret; ++ ++ if (!access_ok(VERIFY_WRITE, uattr, usize)) ++ return -EFAULT; ++ ++ /* ++ * If we're handed a smaller struct than we know of, ++ * ensure all the unknown bits are 0 - i.e. old ++ * user-space does not get uncomplete information. ++ */ ++ if (usize < sizeof(*attr)) { ++ unsigned char *addr; ++ unsigned char *end; ++ ++ addr = (void *)attr + usize; ++ end = (void *)attr + sizeof(*attr); ++ ++ for (; addr < end; addr++) { ++ if (*addr) ++ return -EFBIG; ++ } ++ ++ attr->size = usize; ++ } ++ ++ ret = copy_to_user(uattr, attr, attr->size); ++ if (ret) ++ return -EFAULT; ++ ++ /* sched/core.c uses zero here but we already know ret is zero */ ++ return ret; ++} ++ ++/** ++ * sys_sched_getattr - similar to sched_getparam, but with sched_attr ++ * @pid: the pid in question. ++ * @uattr: structure containing the extended parameters. ++ * @size: sizeof(attr) for fwd/bwd comp. ++ * @flags: for future extension. ++ */ ++SYSCALL_DEFINE4(sched_getattr, pid_t, pid, struct sched_attr __user *, uattr, ++ unsigned int, size, unsigned int, flags) ++{ ++ struct sched_attr attr = { ++ .size = sizeof(struct sched_attr), ++ }; ++ struct task_struct *p; ++ int retval; ++ ++ if (!uattr || pid < 0 || size > PAGE_SIZE || ++ size < SCHED_ATTR_SIZE_VER0 || flags) ++ return -EINVAL; ++ ++ rcu_read_lock(); ++ p = find_process_by_pid(pid); ++ retval = -ESRCH; ++ if (!p) ++ goto out_unlock; ++ ++ retval = security_task_getscheduler(p); ++ if (retval) ++ goto out_unlock; ++ ++ attr.sched_policy = p->policy; ++ if (rt_task(p)) ++ attr.sched_priority = p->rt_priority; ++ else ++ attr.sched_nice = task_nice(p); ++ ++ rcu_read_unlock(); ++ ++ retval = sched_read_attr(uattr, &attr, size); ++ return retval; ++ ++out_unlock: ++ rcu_read_unlock(); ++ return retval; ++} ++ ++long sched_setaffinity(pid_t pid, const struct cpumask *in_mask) ++{ ++ cpumask_var_t cpus_allowed, new_mask; ++ struct task_struct *p; ++ int retval; ++ ++ rcu_read_lock(); ++ ++ p = find_process_by_pid(pid); ++ if (!p) { ++ rcu_read_unlock(); ++ return -ESRCH; ++ } ++ ++ /* Prevent p going away */ ++ get_task_struct(p); ++ rcu_read_unlock(); ++ ++ if (p->flags & PF_NO_SETAFFINITY) { ++ retval = -EINVAL; ++ goto out_put_task; ++ } ++ if (!alloc_cpumask_var(&cpus_allowed, GFP_KERNEL)) { ++ retval = -ENOMEM; ++ goto out_put_task; ++ } ++ if (!alloc_cpumask_var(&new_mask, GFP_KERNEL)) { ++ retval = -ENOMEM; ++ goto out_free_cpus_allowed; ++ } ++ retval = -EPERM; ++ if (!check_same_owner(p)) { ++ rcu_read_lock(); ++ if (!ns_capable(__task_cred(p)->user_ns, CAP_SYS_NICE)) { ++ rcu_read_unlock(); ++ goto out_unlock; ++ } ++ rcu_read_unlock(); ++ } ++ ++ retval = security_task_setscheduler(p); ++ if (retval) ++ goto out_unlock; ++ ++ cpuset_cpus_allowed(p, cpus_allowed); ++ cpumask_and(new_mask, in_mask, cpus_allowed); ++again: ++ retval = __set_cpus_allowed_ptr(p, new_mask, true); ++ ++ if (!retval) { ++ cpuset_cpus_allowed(p, cpus_allowed); ++ if (!cpumask_subset(new_mask, cpus_allowed)) { ++ /* ++ * We must have raced with a concurrent cpuset ++ * update. Just reset the cpus_allowed to the ++ * cpuset's cpus_allowed ++ */ ++ cpumask_copy(new_mask, cpus_allowed); ++ goto again; ++ } ++ } ++out_unlock: ++ free_cpumask_var(new_mask); ++out_free_cpus_allowed: ++ free_cpumask_var(cpus_allowed); ++out_put_task: ++ put_task_struct(p); ++ return retval; ++} ++ ++static int get_user_cpu_mask(unsigned long __user *user_mask_ptr, unsigned len, ++ cpumask_t *new_mask) ++{ ++ if (len < cpumask_size()) ++ cpumask_clear(new_mask); ++ else if (len > cpumask_size()) ++ len = cpumask_size(); ++ ++ return copy_from_user(new_mask, user_mask_ptr, len) ? -EFAULT : 0; ++} ++ ++ ++/** ++ * sys_sched_setaffinity - set the CPU affinity of a process ++ * @pid: pid of the process ++ * @len: length in bytes of the bitmask pointed to by user_mask_ptr ++ * @user_mask_ptr: user-space pointer to the new CPU mask ++ * ++ * Return: 0 on success. An error code otherwise. ++ */ ++SYSCALL_DEFINE3(sched_setaffinity, pid_t, pid, unsigned int, len, ++ unsigned long __user *, user_mask_ptr) ++{ ++ cpumask_var_t new_mask; ++ int retval; ++ ++ if (!alloc_cpumask_var(&new_mask, GFP_KERNEL)) ++ return -ENOMEM; ++ ++ retval = get_user_cpu_mask(user_mask_ptr, len, new_mask); ++ if (retval == 0) ++ retval = sched_setaffinity(pid, new_mask); ++ free_cpumask_var(new_mask); ++ return retval; ++} ++ ++long sched_getaffinity(pid_t pid, cpumask_t *mask) ++{ ++ struct task_struct *p; ++ unsigned long flags; ++ int retval; ++ ++ get_online_cpus(); ++ rcu_read_lock(); ++ ++ retval = -ESRCH; ++ p = find_process_by_pid(pid); ++ if (!p) ++ goto out_unlock; ++ ++ retval = security_task_getscheduler(p); ++ if (retval) ++ goto out_unlock; ++ ++ raw_spin_lock_irqsave(&p->pi_lock, flags); ++ cpumask_and(mask, &p->cpus_allowed, cpu_active_mask); ++ raw_spin_unlock_irqrestore(&p->pi_lock, flags); ++ ++out_unlock: ++ rcu_read_unlock(); ++ put_online_cpus(); ++ ++ return retval; ++} ++ ++/** ++ * sys_sched_getaffinity - get the CPU affinity of a process ++ * @pid: pid of the process ++ * @len: length in bytes of the bitmask pointed to by user_mask_ptr ++ * @user_mask_ptr: user-space pointer to hold the current CPU mask ++ * ++ * Return: 0 on success. An error code otherwise. ++ */ ++SYSCALL_DEFINE3(sched_getaffinity, pid_t, pid, unsigned int, len, ++ unsigned long __user *, user_mask_ptr) ++{ ++ int ret; ++ cpumask_var_t mask; ++ ++ if ((len * BITS_PER_BYTE) < nr_cpu_ids) ++ return -EINVAL; ++ if (len & (sizeof(unsigned long)-1)) ++ return -EINVAL; ++ ++ if (!alloc_cpumask_var(&mask, GFP_KERNEL)) ++ return -ENOMEM; ++ ++ ret = sched_getaffinity(pid, mask); ++ if (ret == 0) { ++ size_t retlen = min_t(size_t, len, cpumask_size()); ++ ++ if (copy_to_user(user_mask_ptr, mask, retlen)) ++ ret = -EFAULT; ++ else ++ ret = retlen; ++ } ++ free_cpumask_var(mask); ++ ++ return ret; ++} ++ ++/** ++ * sys_sched_yield - yield the current processor to other threads. ++ * ++ * This function yields the current CPU to other tasks. It does this by ++ * scheduling away the current task. If it still has the earliest deadline ++ * it will be scheduled again as the next task. ++ * ++ * Return: 0. ++ */ ++SYSCALL_DEFINE0(sched_yield) ++{ ++ struct rq *rq; ++ ++ if (!sched_yield_type) ++ goto out; ++ ++ local_irq_disable(); ++ rq = this_rq(); ++ rq_lock(rq); ++ ++ if (sched_yield_type > 1) ++ time_slice_expired(current, rq); ++ schedstat_inc(rq->yld_count); ++ ++ /* ++ * Since we are going to call schedule() anyway, there's ++ * no need to preempt or enable interrupts: ++ */ ++ preempt_disable(); ++ rq_unlock(rq); ++ sched_preempt_enable_no_resched(); ++ ++ schedule(); ++out: ++ return 0; ++} ++ ++#ifndef CONFIG_PREEMPT ++int __sched _cond_resched(void) ++{ ++ if (should_resched(0)) { ++ preempt_schedule_common(); ++ return 1; ++ } ++ return 0; ++} ++EXPORT_SYMBOL(_cond_resched); ++#endif ++ ++/* ++ * __cond_resched_lock() - if a reschedule is pending, drop the given lock, ++ * call schedule, and on return reacquire the lock. ++ * ++ * This works OK both with and without CONFIG_PREEMPT. We do strange low-level ++ * operations here to prevent schedule() from being called twice (once via ++ * spin_unlock(), once by hand). ++ */ ++int __cond_resched_lock(spinlock_t *lock) ++{ ++ int resched = should_resched(PREEMPT_LOCK_OFFSET); ++ int ret = 0; ++ ++ lockdep_assert_held(lock); ++ ++ if (spin_needbreak(lock) || resched) { ++ spin_unlock(lock); ++ if (resched) ++ preempt_schedule_common(); ++ else ++ cpu_relax(); ++ ret = 1; ++ spin_lock(lock); ++ } ++ return ret; ++} ++EXPORT_SYMBOL(__cond_resched_lock); ++ ++int __sched __cond_resched_softirq(void) ++{ ++ BUG_ON(!in_softirq()); ++ ++ if (should_resched(SOFTIRQ_DISABLE_OFFSET)) { ++ local_bh_enable(); ++ preempt_schedule_common(); ++ local_bh_disable(); ++ return 1; ++ } ++ return 0; ++} ++EXPORT_SYMBOL(__cond_resched_softirq); ++ ++/** ++ * yield - yield the current processor to other threads. ++ * ++ * Do not ever use this function, there's a 99% chance you're doing it wrong. ++ * ++ * The scheduler is at all times free to pick the calling task as the most ++ * eligible task to run, if removing the yield() call from your code breaks ++ * it, its already broken. ++ * ++ * Typical broken usage is: ++ * ++ * while (!event) ++ * yield(); ++ * ++ * where one assumes that yield() will let 'the other' process run that will ++ * make event true. If the current task is a SCHED_FIFO task that will never ++ * happen. Never use yield() as a progress guarantee!! ++ * ++ * If you want to use yield() to wait for something, use wait_event(). ++ * If you want to use yield() to be 'nice' for others, use cond_resched(). ++ * If you still want to use yield(), do not! ++ */ ++void __sched yield(void) ++{ ++ set_current_state(TASK_RUNNING); ++ sys_sched_yield(); ++} ++EXPORT_SYMBOL(yield); ++ ++/** ++ * yield_to - yield the current processor to another thread in ++ * your thread group, or accelerate that thread toward the ++ * processor it's on. ++ * @p: target task ++ * @preempt: whether task preemption is allowed or not ++ * ++ * It's the caller's job to ensure that the target task struct ++ * can't go away on us before we can do any checks. ++ * ++ * Return: ++ * true (>0) if we indeed boosted the target task. ++ * false (0) if we failed to boost the target. ++ * -ESRCH if there's no task to yield to. ++ */ ++int __sched yield_to(struct task_struct *p, bool preempt) ++{ ++ struct task_struct *rq_p; ++ struct rq *rq, *p_rq; ++ unsigned long flags; ++ int yielded = 0; ++ ++ local_irq_save(flags); ++ rq = this_rq(); ++ ++again: ++ p_rq = task_rq(p); ++ /* ++ * If we're the only runnable task on the rq and target rq also ++ * has only one task, there's absolutely no point in yielding. ++ */ ++ if (task_running(p_rq, p) || p->state) { ++ yielded = -ESRCH; ++ goto out_irq; ++ } ++ ++ double_rq_lock(rq, p_rq); ++ if (unlikely(task_rq(p) != p_rq)) { ++ double_rq_unlock(rq, p_rq); ++ goto again; ++ } ++ ++ yielded = 1; ++ schedstat_inc(rq->yld_count); ++ rq_p = rq->curr; ++ if (p->deadline > rq_p->deadline) ++ p->deadline = rq_p->deadline; ++ p->time_slice += rq_p->time_slice; ++ if (p->time_slice > timeslice()) ++ p->time_slice = timeslice(); ++ time_slice_expired(rq_p, rq); ++ if (preempt && rq != p_rq) ++ resched_task(p_rq->curr); ++ double_rq_unlock(rq, p_rq); ++out_irq: ++ local_irq_restore(flags); ++ ++ if (yielded > 0) ++ schedule(); ++ return yielded; ++} ++EXPORT_SYMBOL_GPL(yield_to); ++ ++int io_schedule_prepare(void) ++{ ++ int old_iowait = current->in_iowait; ++ ++ current->in_iowait = 1; ++ blk_schedule_flush_plug(current); ++ ++ return old_iowait; ++} ++ ++void io_schedule_finish(int token) ++{ ++ current->in_iowait = token; ++} ++ ++/* ++ * This task is about to go to sleep on IO. Increment rq->nr_iowait so ++ * that process accounting knows that this is a task in IO wait state. ++ * ++ * But don't do that if it is a deliberate, throttling IO wait (this task ++ * has set its backing_dev_info: the queue against which it should throttle) ++ */ ++ ++long __sched io_schedule_timeout(long timeout) ++{ ++ int token; ++ long ret; ++ ++ token = io_schedule_prepare(); ++ ret = schedule_timeout(timeout); ++ io_schedule_finish(token); ++ ++ return ret; ++} ++EXPORT_SYMBOL(io_schedule_timeout); ++ ++void io_schedule(void) ++{ ++ int token; ++ ++ token = io_schedule_prepare(); ++ schedule(); ++ io_schedule_finish(token); ++} ++EXPORT_SYMBOL(io_schedule); ++ ++/** ++ * sys_sched_get_priority_max - return maximum RT priority. ++ * @policy: scheduling class. ++ * ++ * Return: On success, this syscall returns the maximum ++ * rt_priority that can be used by a given scheduling class. ++ * On failure, a negative error code is returned. ++ */ ++SYSCALL_DEFINE1(sched_get_priority_max, int, policy) ++{ ++ int ret = -EINVAL; ++ ++ switch (policy) { ++ case SCHED_FIFO: ++ case SCHED_RR: ++ ret = MAX_USER_RT_PRIO-1; ++ break; ++ case SCHED_NORMAL: ++ case SCHED_BATCH: ++ case SCHED_ISO: ++ case SCHED_IDLEPRIO: ++ ret = 0; ++ break; ++ } ++ return ret; ++} ++ ++/** ++ * sys_sched_get_priority_min - return minimum RT priority. ++ * @policy: scheduling class. ++ * ++ * Return: On success, this syscall returns the minimum ++ * rt_priority that can be used by a given scheduling class. ++ * On failure, a negative error code is returned. ++ */ ++SYSCALL_DEFINE1(sched_get_priority_min, int, policy) ++{ ++ int ret = -EINVAL; ++ ++ switch (policy) { ++ case SCHED_FIFO: ++ case SCHED_RR: ++ ret = 1; ++ break; ++ case SCHED_NORMAL: ++ case SCHED_BATCH: ++ case SCHED_ISO: ++ case SCHED_IDLEPRIO: ++ ret = 0; ++ break; ++ } ++ return ret; ++} ++ ++/** ++ * sys_sched_rr_get_interval - return the default timeslice of a process. ++ * @pid: pid of the process. ++ * @interval: userspace pointer to the timeslice value. ++ * ++ * ++ * Return: On success, 0 and the timeslice is in @interval. Otherwise, ++ * an error code. ++ */ ++SYSCALL_DEFINE2(sched_rr_get_interval, pid_t, pid, ++ struct timespec __user *, interval) ++{ ++ struct task_struct *p; ++ unsigned int time_slice; ++ unsigned long flags; ++ struct timespec t; ++ struct rq *rq; ++ int retval; ++ ++ if (pid < 0) ++ return -EINVAL; ++ ++ retval = -ESRCH; ++ rcu_read_lock(); ++ p = find_process_by_pid(pid); ++ if (!p) ++ goto out_unlock; ++ ++ retval = security_task_getscheduler(p); ++ if (retval) ++ goto out_unlock; ++ ++ rq = task_rq_lock(p, &flags); ++ time_slice = p->policy == SCHED_FIFO ? 0 : MS_TO_NS(task_timeslice(p)); ++ task_rq_unlock(rq, p, &flags); ++ ++ rcu_read_unlock(); ++ t = ns_to_timespec(time_slice); ++ retval = copy_to_user(interval, &t, sizeof(t)) ? -EFAULT : 0; ++ return retval; ++ ++out_unlock: ++ rcu_read_unlock(); ++ return retval; ++} ++ ++void sched_show_task(struct task_struct *p) ++{ ++ unsigned long free = 0; ++ int ppid; ++ ++ if (!try_get_task_stack(p)) ++ return; ++ ++ printk(KERN_INFO "%-15.15s %c", p->comm, task_state_to_char(p)); ++ ++ if (p->state == TASK_RUNNING) ++ printk(KERN_CONT " running task "); ++#ifdef CONFIG_DEBUG_STACK_USAGE ++ free = stack_not_used(p); ++#endif ++ ppid = 0; ++ rcu_read_lock(); ++ if (pid_alive(p)) ++ ppid = task_pid_nr(rcu_dereference(p->real_parent)); ++ rcu_read_unlock(); ++ printk(KERN_CONT "%5lu %5d %6d 0x%08lx\n", free, ++ task_pid_nr(p), ppid, ++ (unsigned long)task_thread_info(p)->flags); ++ ++ print_worker_info(KERN_INFO, p); ++ show_stack(p, NULL); ++ put_task_stack(p); ++} ++ ++static inline bool ++state_filter_match(unsigned long state_filter, struct task_struct *p) ++{ ++ /* no filter, everything matches */ ++ if (!state_filter) ++ return true; ++ ++ /* filter, but doesn't match */ ++ if (!(p->state & state_filter)) ++ return false; ++ ++ /* ++ * When looking for TASK_UNINTERRUPTIBLE skip TASK_IDLE (allows ++ * TASK_KILLABLE). ++ */ ++ if (state_filter == TASK_UNINTERRUPTIBLE && p->state == TASK_IDLE) ++ return false; ++ ++ return true; ++} ++ ++void show_state_filter(unsigned long state_filter) ++{ ++ struct task_struct *g, *p; ++ ++#if BITS_PER_LONG == 32 ++ printk(KERN_INFO ++ " task PC stack pid father\n"); ++#else ++ printk(KERN_INFO ++ " task PC stack pid father\n"); ++#endif ++ rcu_read_lock(); ++ for_each_process_thread(g, p) { ++ /* ++ * reset the NMI-timeout, listing all files on a slow ++ * console might take a lot of time: ++ * Also, reset softlockup watchdogs on all CPUs, because ++ * another CPU might be blocked waiting for us to process ++ * an IPI. ++ */ ++ touch_nmi_watchdog(); ++ touch_all_softlockup_watchdogs(); ++ if (state_filter_match(state_filter, p)) ++ sched_show_task(p); ++ } ++ ++ rcu_read_unlock(); ++ /* ++ * Only show locks if all tasks are dumped: ++ */ ++ if (!state_filter) ++ debug_show_all_locks(); ++} ++ ++void dump_cpu_task(int cpu) ++{ ++ pr_info("Task dump for CPU %d:\n", cpu); ++ sched_show_task(cpu_curr(cpu)); ++} ++ ++#ifdef CONFIG_SMP ++void set_cpus_allowed_common(struct task_struct *p, const struct cpumask *new_mask) ++{ ++ cpumask_copy(&p->cpus_allowed, new_mask); ++ p->nr_cpus_allowed = cpumask_weight(new_mask); ++} ++ ++void __do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask) ++{ ++ struct rq *rq = task_rq(p); ++ ++ lockdep_assert_held(&p->pi_lock); ++ ++ cpumask_copy(&p->cpus_allowed, new_mask); ++ ++ if (task_queued(p)) { ++ /* ++ * Because __kthread_bind() calls this on blocked tasks without ++ * holding rq->lock. ++ */ ++ lockdep_assert_held(&rq->lock); ++ } ++} ++ ++/* ++ * Calling do_set_cpus_allowed from outside the scheduler code should not be ++ * called on a running or queued task. We should be holding pi_lock. ++ */ ++void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask) ++{ ++ __do_set_cpus_allowed(p, new_mask); ++ if (needs_other_cpu(p, task_cpu(p))) { ++ struct rq *rq; ++ ++ rq = __task_rq_lock(p); ++ set_task_cpu(p, valid_task_cpu(p)); ++ resched_task(p); ++ __task_rq_unlock(rq); ++ } ++} ++#endif ++ ++/** ++ * init_idle - set up an idle thread for a given CPU ++ * @idle: task in question ++ * @cpu: cpu the idle task belongs to ++ * ++ * NOTE: this function does not set the idle thread's NEED_RESCHED ++ * flag, to make booting more robust. ++ */ ++void init_idle(struct task_struct *idle, int cpu) ++{ ++ struct rq *rq = cpu_rq(cpu); ++ unsigned long flags; ++ ++ raw_spin_lock_irqsave(&idle->pi_lock, flags); ++ raw_spin_lock(&rq->lock); ++ idle->last_ran = rq->niffies; ++ time_slice_expired(idle, rq); ++ idle->state = TASK_RUNNING; ++ /* Setting prio to illegal value shouldn't matter when never queued */ ++ idle->prio = PRIO_LIMIT; ++ ++ kasan_unpoison_task_stack(idle); ++ ++#ifdef CONFIG_SMP ++ /* ++ * It's possible that init_idle() gets called multiple times on a task, ++ * in that case do_set_cpus_allowed() will not do the right thing. ++ * ++ * And since this is boot we can forgo the serialisation. ++ */ ++ set_cpus_allowed_common(idle, cpumask_of(cpu)); ++#ifdef CONFIG_SMT_NICE ++ idle->smt_bias = 0; ++#endif ++#endif ++ set_rq_task(rq, idle); ++ ++ /* Silence PROVE_RCU */ ++ rcu_read_lock(); ++ set_task_cpu(idle, cpu); ++ rcu_read_unlock(); ++ ++ rq->curr = rq->idle = idle; ++ idle->on_rq = TASK_ON_RQ_QUEUED; ++ raw_spin_unlock(&rq->lock); ++ raw_spin_unlock_irqrestore(&idle->pi_lock, flags); ++ ++ /* Set the preempt count _outside_ the spinlocks! */ ++ init_idle_preempt_count(idle, cpu); ++ ++ ftrace_graph_init_idle_task(idle, cpu); ++ vtime_init_idle(idle, cpu); ++#ifdef CONFIG_SMP ++ sprintf(idle->comm, "%s/%d", INIT_TASK_COMM, cpu); ++#endif ++} ++ ++int cpuset_cpumask_can_shrink(const struct cpumask __maybe_unused *cur, ++ const struct cpumask __maybe_unused *trial) ++{ ++ return 1; ++} ++ ++int task_can_attach(struct task_struct *p, ++ const struct cpumask *cs_cpus_allowed) ++{ ++ int ret = 0; ++ ++ /* ++ * Kthreads which disallow setaffinity shouldn't be moved ++ * to a new cpuset; we don't want to change their CPU ++ * affinity and isolating such threads by their set of ++ * allowed nodes is unnecessary. Thus, cpusets are not ++ * applicable for such threads. This prevents checking for ++ * success of set_cpus_allowed_ptr() on all attached tasks ++ * before cpus_allowed may be changed. ++ */ ++ if (p->flags & PF_NO_SETAFFINITY) ++ ret = -EINVAL; ++ ++ return ret; ++} ++ ++void resched_cpu(int cpu) ++{ ++ struct rq *rq = cpu_rq(cpu); ++ unsigned long flags; ++ ++ rq_lock_irqsave(rq, &flags); ++ resched_task(cpu_curr(cpu)); ++ rq_unlock_irqrestore(rq, &flags); ++} ++ ++#ifdef CONFIG_SMP ++#ifdef CONFIG_NO_HZ_COMMON ++void nohz_balance_enter_idle(int cpu) ++{ ++} ++ ++void select_nohz_load_balancer(int stop_tick) ++{ ++} ++ ++void set_cpu_sd_state_idle(void) {} ++ ++/* ++ * In the semi idle case, use the nearest busy CPU for migrating timers ++ * from an idle CPU. This is good for power-savings. ++ * ++ * We don't do similar optimization for completely idle system, as ++ * selecting an idle CPU will add more delays to the timers than intended ++ * (as that CPU's timer base may not be uptodate wrt jiffies etc). ++ */ ++int get_nohz_timer_target(void) ++{ ++ int i, cpu = smp_processor_id(); ++ struct sched_domain *sd; ++ ++ if (!idle_cpu(cpu) && is_housekeeping_cpu(cpu)) ++ return cpu; ++ ++ rcu_read_lock(); ++ for_each_domain(cpu, sd) { ++ for_each_cpu(i, sched_domain_span(sd)) { ++ if (cpu == i) ++ continue; ++ ++ if (!idle_cpu(i) && is_housekeeping_cpu(i)) { ++ cpu = i; ++ cpu = i; ++ goto unlock; ++ } ++ } ++ } ++ ++ if (!is_housekeeping_cpu(cpu)) ++ cpu = housekeeping_any_cpu(); ++unlock: ++ rcu_read_unlock(); ++ return cpu; ++} ++ ++/* ++ * When add_timer_on() enqueues a timer into the timer wheel of an ++ * idle CPU then this timer might expire before the next timer event ++ * which is scheduled to wake up that CPU. In case of a completely ++ * idle system the next event might even be infinite time into the ++ * future. wake_up_idle_cpu() ensures that the CPU is woken up and ++ * leaves the inner idle loop so the newly added timer is taken into ++ * account when the CPU goes back to idle and evaluates the timer ++ * wheel for the next timer event. ++ */ ++void wake_up_idle_cpu(int cpu) ++{ ++ if (cpu == smp_processor_id()) ++ return; ++ ++ if (set_nr_and_not_polling(cpu_rq(cpu)->idle)) ++ smp_sched_reschedule(cpu); ++ else ++ trace_sched_wake_idle_without_ipi(cpu); ++} ++ ++static bool wake_up_full_nohz_cpu(int cpu) ++{ ++ /* ++ * We just need the target to call irq_exit() and re-evaluate ++ * the next tick. The nohz full kick at least implies that. ++ * If needed we can still optimize that later with an ++ * empty IRQ. ++ */ ++ if (cpu_is_offline(cpu)) ++ return true; /* Don't try to wake offline CPUs. */ ++ if (tick_nohz_full_cpu(cpu)) { ++ if (cpu != smp_processor_id() || ++ tick_nohz_tick_stopped()) ++ tick_nohz_full_kick_cpu(cpu); ++ return true; ++ } ++ ++ return false; ++} ++ ++/* ++ * Wake up the specified CPU. If the CPU is going offline, it is the ++ * caller's responsibility to deal with the lost wakeup, for example, ++ * by hooking into the CPU_DEAD notifier like timers and hrtimers do. ++ */ ++void wake_up_nohz_cpu(int cpu) ++{ ++ if (!wake_up_full_nohz_cpu(cpu)) ++ wake_up_idle_cpu(cpu); ++} ++#endif /* CONFIG_NO_HZ_COMMON */ ++ ++/* ++ * Change a given task's CPU affinity. Migrate the thread to a ++ * proper CPU and schedule it away if the CPU it's executing on ++ * is removed from the allowed bitmask. ++ * ++ * NOTE: the caller must have a valid reference to the task, the ++ * task must not exit() & deallocate itself prematurely. The ++ * call is not atomic; no spinlocks may be held. ++ */ ++static int __set_cpus_allowed_ptr(struct task_struct *p, ++ const struct cpumask *new_mask, bool check) ++{ ++ const struct cpumask *cpu_valid_mask = cpu_active_mask; ++ bool queued = false, running_wrong = false, kthread; ++ struct cpumask old_mask; ++ unsigned long flags; ++ struct rq *rq; ++ int ret = 0; ++ ++ rq = task_rq_lock(p, &flags); ++ update_rq_clock(rq); ++ ++ kthread = !!(p->flags & PF_KTHREAD); ++ if (kthread) { ++ /* ++ * Kernel threads are allowed on online && !active CPUs ++ */ ++ cpu_valid_mask = cpu_online_mask; ++ } ++ ++ /* ++ * Must re-check here, to close a race against __kthread_bind(), ++ * sched_setaffinity() is not guaranteed to observe the flag. ++ */ ++ if (check && (p->flags & PF_NO_SETAFFINITY)) { ++ ret = -EINVAL; ++ goto out; ++ } ++ ++ cpumask_copy(&old_mask, &p->cpus_allowed); ++ if (cpumask_equal(&old_mask, new_mask)) ++ goto out; ++ ++ if (!cpumask_intersects(new_mask, cpu_valid_mask)) { ++ ret = -EINVAL; ++ goto out; ++ } ++ ++ queued = task_queued(p); ++ __do_set_cpus_allowed(p, new_mask); ++ ++ if (kthread) { ++ /* ++ * For kernel threads that do indeed end up on online && ++ * !active we want to ensure they are strict per-CPU threads. ++ */ ++ WARN_ON(cpumask_intersects(new_mask, cpu_online_mask) && ++ !cpumask_intersects(new_mask, cpu_active_mask) && ++ p->nr_cpus_allowed != 1); ++ } ++ ++ /* Can the task run on the task's current CPU? If so, we're done */ ++ if (cpumask_test_cpu(task_cpu(p), new_mask)) ++ goto out; ++ ++ if (task_running(rq, p)) { ++ /* Task is running on the wrong cpu now, reschedule it. */ ++ if (rq == this_rq()) { ++ set_tsk_need_resched(p); ++ running_wrong = true; ++ } else ++ resched_task(p); ++ } else { ++ int cpu = cpumask_any_and(cpu_valid_mask, new_mask); ++ ++ if (queued) { ++ /* ++ * Switch runqueue locks after dequeueing the task ++ * here while still holding the pi_lock to be holding ++ * the correct lock for enqueueing. ++ */ ++ dequeue_task(rq, p, 0); ++ rq_unlock(rq); ++ ++ rq = cpu_rq(cpu); ++ rq_lock(rq); ++ } ++ set_task_cpu(p, cpu); ++ if (queued) ++ enqueue_task(rq, p, 0); ++ } ++ if (queued) ++ try_preempt(p, rq); ++ if (running_wrong) ++ preempt_disable(); ++out: ++ task_rq_unlock(rq, p, &flags); ++ ++ if (running_wrong) { ++ __schedule(true); ++ preempt_enable(); ++ } ++ ++ return ret; ++} ++ ++int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask) ++{ ++ return __set_cpus_allowed_ptr(p, new_mask, false); ++} ++EXPORT_SYMBOL_GPL(set_cpus_allowed_ptr); ++ ++#ifdef CONFIG_HOTPLUG_CPU ++/* ++ * Run through task list and find tasks affined to the dead cpu, then remove ++ * that cpu from the list, enable cpu0 and set the zerobound flag. Must hold ++ * cpu 0 and src_cpu's runqueue locks. ++ */ ++static void bind_zero(int src_cpu) ++{ ++ struct task_struct *p, *t; ++ struct rq *rq0; ++ int bound = 0; ++ ++ if (src_cpu == 0) ++ return; ++ ++ rq0 = cpu_rq(0); ++ ++ do_each_thread(t, p) { ++ if (cpumask_test_cpu(src_cpu, &p->cpus_allowed)) { ++ bool local = (task_cpu(p) == src_cpu); ++ struct rq *rq = task_rq(p); ++ ++ /* task_running is the cpu stopper thread */ ++ if (local && task_running(rq, p)) ++ continue; ++ atomic_clear_cpu(src_cpu, &p->cpus_allowed); ++ atomic_set_cpu(0, &p->cpus_allowed); ++ p->zerobound = true; ++ bound++; ++ if (local) { ++ bool queued = task_queued(p); ++ ++ if (queued) ++ dequeue_task(rq, p, 0); ++ set_task_cpu(p, 0); ++ if (queued) ++ enqueue_task(rq0, p, 0); ++ } ++ } ++ } while_each_thread(t, p); ++ ++ if (bound) { ++ printk(KERN_INFO "Removed affinity for %d processes to cpu %d\n", ++ bound, src_cpu); ++ } ++} ++ ++/* Find processes with the zerobound flag and reenable their affinity for the ++ * CPU coming alive. */ ++static void unbind_zero(int src_cpu) ++{ ++ int unbound = 0, zerobound = 0; ++ struct task_struct *p, *t; ++ ++ if (src_cpu == 0) ++ return; ++ ++ do_each_thread(t, p) { ++ if (!p->mm) ++ p->zerobound = false; ++ if (p->zerobound) { ++ unbound++; ++ cpumask_set_cpu(src_cpu, &p->cpus_allowed); ++ /* Once every CPU affinity has been re-enabled, remove ++ * the zerobound flag */ ++ if (cpumask_subset(cpu_possible_mask, &p->cpus_allowed)) { ++ p->zerobound = false; ++ zerobound++; ++ } ++ } ++ } while_each_thread(t, p); ++ ++ if (unbound) { ++ printk(KERN_INFO "Added affinity for %d processes to cpu %d\n", ++ unbound, src_cpu); ++ } ++ if (zerobound) { ++ printk(KERN_INFO "Released forced binding to cpu0 for %d processes\n", ++ zerobound); ++ } ++} ++ ++/* ++ * Ensure that the idle task is using init_mm right before its cpu goes ++ * offline. ++ */ ++void idle_task_exit(void) ++{ ++ struct mm_struct *mm = current->active_mm; ++ ++ BUG_ON(cpu_online(smp_processor_id())); ++ ++ if (mm != &init_mm) { ++ switch_mm(mm, &init_mm, current); ++ finish_arch_post_lock_switch(); ++ } ++ mmdrop(mm); ++} ++#else /* CONFIG_HOTPLUG_CPU */ ++static void unbind_zero(int src_cpu) {} ++#endif /* CONFIG_HOTPLUG_CPU */ ++ ++void sched_set_stop_task(int cpu, struct task_struct *stop) ++{ ++ struct sched_param stop_param = { .sched_priority = STOP_PRIO }; ++ struct sched_param start_param = { .sched_priority = 0 }; ++ struct task_struct *old_stop = cpu_rq(cpu)->stop; ++ ++ if (stop) { ++ /* ++ * Make it appear like a SCHED_FIFO task, its something ++ * userspace knows about and won't get confused about. ++ * ++ * Also, it will make PI more or less work without too ++ * much confusion -- but then, stop work should not ++ * rely on PI working anyway. ++ */ ++ sched_setscheduler_nocheck(stop, SCHED_FIFO, &stop_param); ++ } ++ ++ cpu_rq(cpu)->stop = stop; ++ ++ if (old_stop) { ++ /* ++ * Reset it back to a normal scheduling policy so that ++ * it can die in pieces. ++ */ ++ sched_setscheduler_nocheck(old_stop, SCHED_NORMAL, &start_param); ++ } ++} ++ ++#if defined(CONFIG_SCHED_DEBUG) && defined(CONFIG_SYSCTL) ++ ++static struct ctl_table sd_ctl_dir[] = { ++ { ++ .procname = "sched_domain", ++ .mode = 0555, ++ }, ++ {} ++}; ++ ++static struct ctl_table sd_ctl_root[] = { ++ { ++ .procname = "kernel", ++ .mode = 0555, ++ .child = sd_ctl_dir, ++ }, ++ {} ++}; ++ ++static struct ctl_table *sd_alloc_ctl_entry(int n) ++{ ++ struct ctl_table *entry = ++ kcalloc(n, sizeof(struct ctl_table), GFP_KERNEL); ++ ++ return entry; ++} ++ ++static void sd_free_ctl_entry(struct ctl_table **tablep) ++{ ++ struct ctl_table *entry; ++ ++ /* ++ * In the intermediate directories, both the child directory and ++ * procname are dynamically allocated and could fail but the mode ++ * will always be set. In the lowest directory the names are ++ * static strings and all have proc handlers. ++ */ ++ for (entry = *tablep; entry->mode; entry++) { ++ if (entry->child) ++ sd_free_ctl_entry(&entry->child); ++ if (entry->proc_handler == NULL) ++ kfree(entry->procname); ++ } ++ ++ kfree(*tablep); ++ *tablep = NULL; ++} ++ ++#define CPU_LOAD_IDX_MAX 5 ++static int min_load_idx = 0; ++static int max_load_idx = CPU_LOAD_IDX_MAX-1; ++ ++static void ++set_table_entry(struct ctl_table *entry, ++ const char *procname, void *data, int maxlen, ++ umode_t mode, proc_handler *proc_handler, ++ bool load_idx) ++{ ++ entry->procname = procname; ++ entry->data = data; ++ entry->maxlen = maxlen; ++ entry->mode = mode; ++ entry->proc_handler = proc_handler; ++ ++ if (load_idx) { ++ entry->extra1 = &min_load_idx; ++ entry->extra2 = &max_load_idx; ++ } ++} ++ ++static struct ctl_table * ++sd_alloc_ctl_domain_table(struct sched_domain *sd) ++{ ++ struct ctl_table *table = sd_alloc_ctl_entry(14); ++ ++ if (table == NULL) ++ return NULL; ++ ++ set_table_entry(&table[0], "min_interval", &sd->min_interval, ++ sizeof(long), 0644, proc_doulongvec_minmax, false); ++ set_table_entry(&table[1], "max_interval", &sd->max_interval, ++ sizeof(long), 0644, proc_doulongvec_minmax, false); ++ set_table_entry(&table[2], "busy_idx", &sd->busy_idx, ++ sizeof(int), 0644, proc_dointvec_minmax, true); ++ set_table_entry(&table[3], "idle_idx", &sd->idle_idx, ++ sizeof(int), 0644, proc_dointvec_minmax, true); ++ set_table_entry(&table[4], "newidle_idx", &sd->newidle_idx, ++ sizeof(int), 0644, proc_dointvec_minmax, true); ++ set_table_entry(&table[5], "wake_idx", &sd->wake_idx, ++ sizeof(int), 0644, proc_dointvec_minmax, true); ++ set_table_entry(&table[6], "forkexec_idx", &sd->forkexec_idx, ++ sizeof(int), 0644, proc_dointvec_minmax, true); ++ set_table_entry(&table[7], "busy_factor", &sd->busy_factor, ++ sizeof(int), 0644, proc_dointvec_minmax, false); ++ set_table_entry(&table[8], "imbalance_pct", &sd->imbalance_pct, ++ sizeof(int), 0644, proc_dointvec_minmax, false); ++ set_table_entry(&table[9], "cache_nice_tries", ++ &sd->cache_nice_tries, ++ sizeof(int), 0644, proc_dointvec_minmax, false); ++ set_table_entry(&table[10], "flags", &sd->flags, ++ sizeof(int), 0644, proc_dointvec_minmax, false); ++ set_table_entry(&table[11], "max_newidle_lb_cost", ++ &sd->max_newidle_lb_cost, ++ sizeof(long), 0644, proc_doulongvec_minmax, false); ++ set_table_entry(&table[12], "name", sd->name, ++ CORENAME_MAX_SIZE, 0444, proc_dostring, false); ++ /* &table[13] is terminator */ ++ ++ return table; ++} ++ ++static struct ctl_table *sd_alloc_ctl_cpu_table(int cpu) ++{ ++ struct ctl_table *entry, *table; ++ struct sched_domain *sd; ++ int domain_num = 0, i; ++ char buf[32]; ++ ++ for_each_domain(cpu, sd) ++ domain_num++; ++ entry = table = sd_alloc_ctl_entry(domain_num + 1); ++ if (table == NULL) ++ return NULL; ++ ++ i = 0; ++ for_each_domain(cpu, sd) { ++ snprintf(buf, 32, "domain%d", i); ++ entry->procname = kstrdup(buf, GFP_KERNEL); ++ entry->mode = 0555; ++ entry->child = sd_alloc_ctl_domain_table(sd); ++ entry++; ++ i++; ++ } ++ return table; ++} ++ ++static cpumask_var_t sd_sysctl_cpus; ++static struct ctl_table_header *sd_sysctl_header; ++ ++void register_sched_domain_sysctl(void) ++{ ++ static struct ctl_table *cpu_entries; ++ static struct ctl_table **cpu_idx; ++ char buf[32]; ++ int i; ++ ++ if (!cpu_entries) { ++ cpu_entries = sd_alloc_ctl_entry(num_possible_cpus() + 1); ++ if (!cpu_entries) ++ return; ++ ++ WARN_ON(sd_ctl_dir[0].child); ++ sd_ctl_dir[0].child = cpu_entries; ++ } ++ ++ if (!cpu_idx) { ++ struct ctl_table *e = cpu_entries; ++ ++ cpu_idx = kcalloc(nr_cpu_ids, sizeof(struct ctl_table*), GFP_KERNEL); ++ if (!cpu_idx) ++ return; ++ ++ /* deal with sparse possible map */ ++ for_each_possible_cpu(i) { ++ cpu_idx[i] = e; ++ e++; ++ } ++ } ++ ++ if (!cpumask_available(sd_sysctl_cpus)) { ++ if (!alloc_cpumask_var(&sd_sysctl_cpus, GFP_KERNEL)) ++ return; ++ ++ /* init to possible to not have holes in @cpu_entries */ ++ cpumask_copy(sd_sysctl_cpus, cpu_possible_mask); ++ } ++ ++ for_each_cpu(i, sd_sysctl_cpus) { ++ struct ctl_table *e = cpu_idx[i]; ++ ++ if (e->child) ++ sd_free_ctl_entry(&e->child); ++ ++ if (!e->procname) { ++ snprintf(buf, 32, "cpu%d", i); ++ e->procname = kstrdup(buf, GFP_KERNEL); ++ } ++ e->mode = 0555; ++ e->child = sd_alloc_ctl_cpu_table(i); ++ ++ __cpumask_clear_cpu(i, sd_sysctl_cpus); ++ } ++ ++ WARN_ON(sd_sysctl_header); ++ sd_sysctl_header = register_sysctl_table(sd_ctl_root); ++} ++ ++void dirty_sched_domain_sysctl(int cpu) ++{ ++ if (cpumask_available(sd_sysctl_cpus)) ++ __cpumask_set_cpu(cpu, sd_sysctl_cpus); ++} ++ ++/* may be called multiple times per register */ ++void unregister_sched_domain_sysctl(void) ++{ ++ unregister_sysctl_table(sd_sysctl_header); ++ sd_sysctl_header = NULL; ++} ++#endif /* CONFIG_SYSCTL */ ++ ++void set_rq_online(struct rq *rq) ++{ ++ if (!rq->online) { ++ cpumask_set_cpu(cpu_of(rq), rq->rd->online); ++ rq->online = true; ++ } ++} ++ ++void set_rq_offline(struct rq *rq) ++{ ++ if (rq->online) { ++ int cpu = cpu_of(rq); ++ ++ cpumask_clear_cpu(cpu, rq->rd->online); ++ rq->online = false; ++ clear_cpuidle_map(cpu); ++ } ++} ++ ++/* ++ * used to mark begin/end of suspend/resume: ++ */ ++static int num_cpus_frozen; ++ ++/* ++ * Update cpusets according to cpu_active mask. If cpusets are ++ * disabled, cpuset_update_active_cpus() becomes a simple wrapper ++ * around partition_sched_domains(). ++ * ++ * If we come here as part of a suspend/resume, don't touch cpusets because we ++ * want to restore it back to its original state upon resume anyway. ++ */ ++static void cpuset_cpu_active(void) ++{ ++ if (cpuhp_tasks_frozen) { ++ /* ++ * num_cpus_frozen tracks how many CPUs are involved in suspend ++ * resume sequence. As long as this is not the last online ++ * operation in the resume sequence, just build a single sched ++ * domain, ignoring cpusets. ++ */ ++ partition_sched_domains(1, NULL, NULL); ++ if (--num_cpus_frozen) ++ return; ++ /* ++ * This is the last CPU online operation. So fall through and ++ * restore the original sched domains by considering the ++ * cpuset configurations. ++ */ ++ cpuset_force_rebuild(); ++ } ++ ++ cpuset_update_active_cpus(); ++} ++ ++static int cpuset_cpu_inactive(unsigned int cpu) ++{ ++ if (!cpuhp_tasks_frozen) { ++ cpuset_update_active_cpus(); ++ } else { ++ num_cpus_frozen++; ++ partition_sched_domains(1, NULL, NULL); ++ } ++ return 0; ++} ++ ++int sched_cpu_activate(unsigned int cpu) ++{ ++ struct rq *rq = cpu_rq(cpu); ++ unsigned long flags; ++ ++ set_cpu_active(cpu, true); ++ ++ if (sched_smp_initialized) { ++ sched_domains_numa_masks_set(cpu); ++ cpuset_cpu_active(); ++ } ++ ++ /* ++ * Put the rq online, if not already. This happens: ++ * ++ * 1) In the early boot process, because we build the real domains ++ * after all CPUs have been brought up. ++ * ++ * 2) At runtime, if cpuset_cpu_active() fails to rebuild the ++ * domains. ++ */ ++ rq_lock_irqsave(rq, &flags); ++ if (rq->rd) { ++ BUG_ON(!cpumask_test_cpu(cpu, rq->rd->span)); ++ set_rq_online(rq); ++ } ++ unbind_zero(cpu); ++ rq_unlock_irqrestore(rq, &flags); ++ ++ return 0; ++} ++ ++int sched_cpu_deactivate(unsigned int cpu) ++{ ++ int ret; ++ ++ set_cpu_active(cpu, false); ++ /* ++ * We've cleared cpu_active_mask, wait for all preempt-disabled and RCU ++ * users of this state to go away such that all new such users will ++ * observe it. ++ * ++ * Do sync before park smpboot threads to take care the rcu boost case. ++ */ ++ synchronize_rcu_mult(call_rcu, call_rcu_sched); ++ ++ if (!sched_smp_initialized) ++ return 0; ++ ++ ret = cpuset_cpu_inactive(cpu); ++ if (ret) { ++ set_cpu_active(cpu, true); ++ return ret; ++ } ++ sched_domains_numa_masks_clear(cpu); ++ return 0; ++} ++ ++int sched_cpu_starting(unsigned int __maybe_unused cpu) ++{ ++ return 0; ++} ++ ++#ifdef CONFIG_HOTPLUG_CPU ++int sched_cpu_dying(unsigned int cpu) ++{ ++ struct rq *rq = cpu_rq(cpu); ++ unsigned long flags; ++ ++ local_irq_save(flags); ++ double_rq_lock(rq, cpu_rq(0)); ++ if (rq->rd) { ++ BUG_ON(!cpumask_test_cpu(cpu, rq->rd->span)); ++ set_rq_offline(rq); ++ } ++ bind_zero(cpu); ++ double_rq_unlock(rq, cpu_rq(0)); ++ sched_start_tick(rq, cpu); ++ hrexpiry_clear(rq); ++ local_irq_restore(flags); ++ ++ return 0; ++} ++#endif ++ ++#if defined(CONFIG_SCHED_SMT) || defined(CONFIG_SCHED_MC) ++/* ++ * Cheaper version of the below functions in case support for SMT and MC is ++ * compiled in but CPUs have no siblings. ++ */ ++static bool sole_cpu_idle(struct rq *rq) ++{ ++ return rq_idle(rq); ++} ++#endif ++#ifdef CONFIG_SCHED_SMT ++static const cpumask_t *thread_cpumask(int cpu) ++{ ++ return topology_sibling_cpumask(cpu); ++} ++/* All this CPU's SMT siblings are idle */ ++static bool siblings_cpu_idle(struct rq *rq) ++{ ++ return cpumask_subset(&rq->thread_mask, &cpu_idle_map); ++} ++#endif ++#ifdef CONFIG_SCHED_MC ++static const cpumask_t *core_cpumask(int cpu) ++{ ++ return topology_core_cpumask(cpu); ++} ++/* All this CPU's shared cache siblings are idle */ ++static bool cache_cpu_idle(struct rq *rq) ++{ ++ return cpumask_subset(&rq->core_mask, &cpu_idle_map); ++} ++#endif ++ ++enum sched_domain_level { ++ SD_LV_NONE = 0, ++ SD_LV_SIBLING, ++ SD_LV_MC, ++ SD_LV_BOOK, ++ SD_LV_CPU, ++ SD_LV_NODE, ++ SD_LV_ALLNODES, ++ SD_LV_MAX ++}; ++ ++void __init sched_init_smp(void) ++{ ++ struct sched_domain *sd; ++ int cpu, other_cpu; ++#ifdef CONFIG_SCHED_SMT ++ bool smt_threads = false; ++#endif ++ cpumask_var_t non_isolated_cpus; ++ struct rq *rq; ++ ++ alloc_cpumask_var(&non_isolated_cpus, GFP_KERNEL); ++ ++ sched_init_numa(); ++ ++ /* ++ * There's no userspace yet to cause hotplug operations; hence all the ++ * cpu masks are stable and all blatant races in the below code cannot ++ * happen. ++ */ ++ mutex_lock(&sched_domains_mutex); ++ sched_init_domains(cpu_active_mask); ++ cpumask_andnot(non_isolated_cpus, cpu_possible_mask, cpu_isolated_map); ++ if (cpumask_empty(non_isolated_cpus)) ++ cpumask_set_cpu(smp_processor_id(), non_isolated_cpus); ++ mutex_unlock(&sched_domains_mutex); ++ ++ /* Move init over to a non-isolated CPU */ ++ if (set_cpus_allowed_ptr(current, non_isolated_cpus) < 0) ++ BUG(); ++ free_cpumask_var(non_isolated_cpus); ++ ++ mutex_lock(&sched_domains_mutex); ++ local_irq_disable(); ++ lock_all_rqs(); ++ /* ++ * Set up the relative cache distance of each online cpu from each ++ * other in a simple array for quick lookup. Locality is determined ++ * by the closest sched_domain that CPUs are separated by. CPUs with ++ * shared cache in SMT and MC are treated as local. Separate CPUs ++ * (within the same package or physically) within the same node are ++ * treated as not local. CPUs not even in the same domain (different ++ * nodes) are treated as very distant. ++ */ ++ for_each_online_cpu(cpu) { ++ rq = cpu_rq(cpu); ++ ++ /* First check if this cpu is in the same node */ ++ for_each_domain(cpu, sd) { ++ if (sd->level > SD_LV_MC) ++ continue; ++ /* Set locality to local node if not already found lower */ ++ for_each_cpu(other_cpu, sched_domain_span(sd)) { ++ if (rq->cpu_locality[other_cpu] > 3) ++ rq->cpu_locality[other_cpu] = 3; ++ } ++ } ++ ++ /* ++ * Each runqueue has its own function in case it doesn't have ++ * siblings of its own allowing mixed topologies. ++ */ ++#ifdef CONFIG_SCHED_MC ++ for_each_cpu(other_cpu, core_cpumask(cpu)) { ++ if (rq->cpu_locality[other_cpu] > 2) ++ rq->cpu_locality[other_cpu] = 2; ++ } ++ if (cpumask_weight(core_cpumask(cpu)) > 1) { ++ cpumask_copy(&rq->core_mask, core_cpumask(cpu)); ++ cpumask_clear_cpu(cpu, &rq->core_mask); ++ rq->cache_idle = cache_cpu_idle; ++ } ++#endif ++#ifdef CONFIG_SCHED_SMT ++ if (cpumask_weight(thread_cpumask(cpu)) > 1) { ++ cpumask_copy(&rq->thread_mask, thread_cpumask(cpu)); ++ cpumask_clear_cpu(cpu, &rq->thread_mask); ++ for_each_cpu(other_cpu, thread_cpumask(cpu)) ++ rq->cpu_locality[other_cpu] = 1; ++ rq->siblings_idle = siblings_cpu_idle; ++ smt_threads = true; ++ } ++#endif ++ } ++ for_each_possible_cpu(cpu) { ++ int total_cpus = 1, locality; ++ ++ rq = cpu_rq(cpu); ++ for (locality = 1; locality <= 4; locality++) { ++ for_each_possible_cpu(other_cpu) { ++ if (rq->cpu_locality[other_cpu] == locality) ++ rq->rq_order[total_cpus++] = cpu_rq(other_cpu); ++ } ++ } ++ } ++#ifdef CONFIG_SMT_NICE ++ if (smt_threads) { ++ check_siblings = &check_smt_siblings; ++ wake_siblings = &wake_smt_siblings; ++ smt_schedule = &smt_should_schedule; ++ } ++#endif ++ unlock_all_rqs(); ++ local_irq_enable(); ++ mutex_unlock(&sched_domains_mutex); ++ ++ for_each_online_cpu(cpu) { ++ rq = cpu_rq(cpu); ++ ++ for_each_online_cpu(other_cpu) { ++ if (other_cpu <= cpu) ++ continue; ++ printk(KERN_DEBUG "MuQSS locality CPU %d to %d: %d\n", cpu, other_cpu, rq->cpu_locality[other_cpu]); ++ } ++ } ++ ++ sched_smp_initialized = true; ++} ++#else ++void __init sched_init_smp(void) ++{ ++ sched_smp_initialized = true; ++} ++#endif /* CONFIG_SMP */ ++ ++int in_sched_functions(unsigned long addr) ++{ ++ return in_lock_functions(addr) || ++ (addr >= (unsigned long)__sched_text_start ++ && addr < (unsigned long)__sched_text_end); ++} ++ ++#ifdef CONFIG_CGROUP_SCHED ++/* task group related information */ ++struct task_group { ++ struct cgroup_subsys_state css; ++ ++ struct rcu_head rcu; ++ struct list_head list; ++ ++ struct task_group *parent; ++ struct list_head siblings; ++ struct list_head children; ++}; ++ ++/* ++ * Default task group. ++ * Every task in system belongs to this group at bootup. ++ */ ++struct task_group root_task_group; ++LIST_HEAD(task_groups); ++ ++/* Cacheline aligned slab cache for task_group */ ++static struct kmem_cache *task_group_cache __read_mostly; ++#endif /* CONFIG_CGROUP_SCHED */ ++ ++void __init sched_init(void) ++{ ++#ifdef CONFIG_SMP ++ int cpu_ids; ++#endif ++ int i; ++ struct rq *rq; ++ ++ sched_clock_init(); ++ ++ wait_bit_init(); ++ ++ prio_ratios[0] = 128; ++ for (i = 1 ; i < NICE_WIDTH ; i++) ++ prio_ratios[i] = prio_ratios[i - 1] * 11 / 10; ++ ++ skiplist_node_init(&init_task.node); ++ ++#ifdef CONFIG_SMP ++ init_defrootdomain(); ++ cpumask_clear(&cpu_idle_map); ++#else ++ uprq = &per_cpu(runqueues, 0); ++#endif ++ ++#ifdef CONFIG_CGROUP_SCHED ++ task_group_cache = KMEM_CACHE(task_group, 0); ++ ++ list_add(&root_task_group.list, &task_groups); ++ INIT_LIST_HEAD(&root_task_group.children); ++ INIT_LIST_HEAD(&root_task_group.siblings); ++#endif /* CONFIG_CGROUP_SCHED */ ++ for_each_possible_cpu(i) { ++ rq = cpu_rq(i); ++ skiplist_init(&rq->node); ++ rq->sl = new_skiplist(&rq->node); ++ raw_spin_lock_init(&rq->lock); ++ rq->nr_running = 0; ++ rq->nr_uninterruptible = 0; ++ rq->nr_switches = 0; ++ rq->clock = rq->old_clock = rq->last_niffy = rq->niffies = 0; ++ rq->last_jiffy = jiffies; ++ rq->user_ns = rq->nice_ns = rq->softirq_ns = rq->system_ns = ++ rq->iowait_ns = rq->idle_ns = 0; ++ rq->dither = 0; ++ set_rq_task(rq, &init_task); ++ rq->iso_ticks = 0; ++ rq->iso_refractory = false; ++#ifdef CONFIG_SMP ++ rq->sd = NULL; ++ rq->rd = NULL; ++ rq->online = false; ++ rq->cpu = i; ++ rq_attach_root(rq, &def_root_domain); ++#endif ++ init_rq_hrexpiry(rq); ++ atomic_set(&rq->nr_iowait, 0); ++ } ++ ++#ifdef CONFIG_SMP ++ cpu_ids = i; ++ /* ++ * Set the base locality for cpu cache distance calculation to ++ * "distant" (3). Make sure the distance from a CPU to itself is 0. ++ */ ++ for_each_possible_cpu(i) { ++ int j; ++ ++ rq = cpu_rq(i); ++#ifdef CONFIG_SCHED_SMT ++ rq->siblings_idle = sole_cpu_idle; ++#endif ++#ifdef CONFIG_SCHED_MC ++ rq->cache_idle = sole_cpu_idle; ++#endif ++ rq->cpu_locality = kmalloc(cpu_ids * sizeof(int *), GFP_ATOMIC); ++ for_each_possible_cpu(j) { ++ if (i == j) ++ rq->cpu_locality[j] = 0; ++ else ++ rq->cpu_locality[j] = 4; ++ } ++ rq->rq_order = kmalloc(cpu_ids * sizeof(struct rq *), GFP_ATOMIC); ++ rq->rq_order[0] = rq; ++ for (j = 1; j < cpu_ids; j++) ++ rq->rq_order[j] = cpu_rq(j); ++ } ++#endif ++ ++ /* ++ * The boot idle thread does lazy MMU switching as well: ++ */ ++ mmgrab(&init_mm); ++ enter_lazy_tlb(&init_mm, current); ++ ++ /* ++ * Make us the idle thread. Technically, schedule() should not be ++ * called from this thread, however somewhere below it might be, ++ * but because we are the idle thread, we just pick up running again ++ * when this runqueue becomes "idle". ++ */ ++ init_idle(current, smp_processor_id()); ++ ++#ifdef CONFIG_SMP ++ /* May be allocated at isolcpus cmdline parse time */ ++ if (cpu_isolated_map == NULL) ++ zalloc_cpumask_var(&cpu_isolated_map, GFP_NOWAIT); ++ idle_thread_set_boot_cpu(); ++#endif /* SMP */ ++ ++ init_schedstats(); ++} ++ ++#ifdef CONFIG_DEBUG_ATOMIC_SLEEP ++static inline int preempt_count_equals(int preempt_offset) ++{ ++ int nested = preempt_count() + rcu_preempt_depth(); ++ ++ return (nested == preempt_offset); ++} ++ ++void __might_sleep(const char *file, int line, int preempt_offset) ++{ ++ /* ++ * Blocking primitives will set (and therefore destroy) current->state, ++ * since we will exit with TASK_RUNNING make sure we enter with it, ++ * otherwise we will destroy state. ++ */ ++ WARN_ONCE(current->state != TASK_RUNNING && current->task_state_change, ++ "do not call blocking ops when !TASK_RUNNING; " ++ "state=%lx set at [<%p>] %pS\n", ++ current->state, ++ (void *)current->task_state_change, ++ (void *)current->task_state_change); ++ ++ ___might_sleep(file, line, preempt_offset); ++} ++EXPORT_SYMBOL(__might_sleep); ++ ++void ___might_sleep(const char *file, int line, int preempt_offset) ++{ ++ /* Ratelimiting timestamp: */ ++ static unsigned long prev_jiffy; ++ ++ unsigned long preempt_disable_ip; ++ ++ /* WARN_ON_ONCE() by default, no rate limit required: */ ++ rcu_sleep_check(); ++ ++ if ((preempt_count_equals(preempt_offset) && !irqs_disabled() && ++ !is_idle_task(current)) || ++ system_state == SYSTEM_BOOTING || system_state > SYSTEM_RUNNING || ++ oops_in_progress) ++ return; ++ ++ if (time_before(jiffies, prev_jiffy + HZ) && prev_jiffy) ++ return; ++ prev_jiffy = jiffies; ++ ++ /* Save this before calling printk(), since that will clobber it: */ ++ preempt_disable_ip = get_preempt_disable_ip(current); ++ ++ printk(KERN_ERR ++ "BUG: sleeping function called from invalid context at %s:%d\n", ++ file, line); ++ printk(KERN_ERR ++ "in_atomic(): %d, irqs_disabled(): %d, pid: %d, name: %s\n", ++ in_atomic(), irqs_disabled(), ++ current->pid, current->comm); ++ ++ if (task_stack_end_corrupted(current)) ++ printk(KERN_EMERG "Thread overran stack, or stack corrupted\n"); ++ ++ debug_show_held_locks(current); ++ if (irqs_disabled()) ++ print_irqtrace_events(current); ++ if (IS_ENABLED(CONFIG_DEBUG_PREEMPT) ++ && !preempt_count_equals(preempt_offset)) { ++ pr_err("Preemption disabled at:"); ++ print_ip_sym(preempt_disable_ip); ++ pr_cont("\n"); ++ } ++ dump_stack(); ++ add_taint(TAINT_WARN, LOCKDEP_STILL_OK); ++} ++EXPORT_SYMBOL(___might_sleep); ++#endif ++ ++#ifdef CONFIG_MAGIC_SYSRQ ++static inline void normalise_rt_tasks(void) ++{ ++ struct task_struct *g, *p; ++ unsigned long flags; ++ struct rq *rq; ++ ++ read_lock(&tasklist_lock); ++ for_each_process_thread(g, p) { ++ /* ++ * Only normalize user tasks: ++ */ ++ if (p->flags & PF_KTHREAD) ++ continue; ++ ++ if (!rt_task(p) && !iso_task(p)) ++ continue; ++ ++ rq = task_rq_lock(p, &flags); ++ __setscheduler(p, rq, SCHED_NORMAL, 0, false); ++ task_rq_unlock(rq, p, &flags); ++ } ++ read_unlock(&tasklist_lock); ++} ++ ++void normalize_rt_tasks(void) ++{ ++ normalise_rt_tasks(); ++} ++#endif /* CONFIG_MAGIC_SYSRQ */ ++ ++#if defined(CONFIG_IA64) || defined(CONFIG_KGDB_KDB) ++/* ++ * These functions are only useful for the IA64 MCA handling, or kdb. ++ * ++ * They can only be called when the whole system has been ++ * stopped - every CPU needs to be quiescent, and no scheduling ++ * activity can take place. Using them for anything else would ++ * be a serious bug, and as a result, they aren't even visible ++ * under any other configuration. ++ */ ++ ++/** ++ * curr_task - return the current task for a given CPU. ++ * @cpu: the processor in question. ++ * ++ * ONLY VALID WHEN THE WHOLE SYSTEM IS STOPPED! ++ * ++ * Return: The current task for @cpu. ++ */ ++struct task_struct *curr_task(int cpu) ++{ ++ return cpu_curr(cpu); ++} ++ ++#endif /* defined(CONFIG_IA64) || defined(CONFIG_KGDB_KDB) */ ++ ++#ifdef CONFIG_IA64 ++/** ++ * set_curr_task - set the current task for a given CPU. ++ * @cpu: the processor in question. ++ * @p: the task pointer to set. ++ * ++ * Description: This function must only be used when non-maskable interrupts ++ * are serviced on a separate stack. It allows the architecture to switch the ++ * notion of the current task on a CPU in a non-blocking manner. This function ++ * must be called with all CPU's synchronised, and interrupts disabled, the ++ * and caller must save the original value of the current task (see ++ * curr_task() above) and restore that value before reenabling interrupts and ++ * re-starting the system. ++ * ++ * ONLY VALID WHEN THE WHOLE SYSTEM IS STOPPED! ++ */ ++void ia64_set_curr_task(int cpu, struct task_struct *p) ++{ ++ cpu_curr(cpu) = p; ++} ++ ++#endif ++ ++void init_idle_bootup_task(struct task_struct *idle) ++{} ++ ++#ifdef CONFIG_SCHED_DEBUG ++__read_mostly bool sched_debug_enabled; ++ ++void proc_sched_show_task(struct task_struct *p, struct pid_namespace *ns, ++ struct seq_file *m) ++{} ++ ++void proc_sched_set_task(struct task_struct *p) ++{} ++#endif ++ ++#ifdef CONFIG_SMP ++#define SCHED_LOAD_SHIFT (10) ++#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) ++ ++unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu) ++{ ++ return SCHED_LOAD_SCALE; ++} ++ ++unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu) ++{ ++ unsigned long weight = cpumask_weight(sched_domain_span(sd)); ++ unsigned long smt_gain = sd->smt_gain; ++ ++ smt_gain /= weight; ++ ++ return smt_gain; ++} ++#endif ++ ++#ifdef CONFIG_CGROUP_SCHED ++static void sched_free_group(struct task_group *tg) ++{ ++ kmem_cache_free(task_group_cache, tg); ++} ++ ++/* allocate runqueue etc for a new task group */ ++struct task_group *sched_create_group(struct task_group *parent) ++{ ++ struct task_group *tg; ++ ++ tg = kmem_cache_alloc(task_group_cache, GFP_KERNEL | __GFP_ZERO); ++ if (!tg) ++ return ERR_PTR(-ENOMEM); ++ ++ return tg; ++} ++ ++void sched_online_group(struct task_group *tg, struct task_group *parent) ++{ ++} ++ ++/* rcu callback to free various structures associated with a task group */ ++static void sched_free_group_rcu(struct rcu_head *rhp) ++{ ++ /* Now it should be safe to free those cfs_rqs */ ++ sched_free_group(container_of(rhp, struct task_group, rcu)); ++} ++ ++void sched_destroy_group(struct task_group *tg) ++{ ++ /* Wait for possible concurrent references to cfs_rqs complete */ ++ call_rcu(&tg->rcu, sched_free_group_rcu); ++} ++ ++void sched_offline_group(struct task_group *tg) ++{ ++} ++ ++static inline struct task_group *css_tg(struct cgroup_subsys_state *css) ++{ ++ return css ? container_of(css, struct task_group, css) : NULL; ++} ++ ++static struct cgroup_subsys_state * ++cpu_cgroup_css_alloc(struct cgroup_subsys_state *parent_css) ++{ ++ struct task_group *parent = css_tg(parent_css); ++ struct task_group *tg; ++ ++ if (!parent) { ++ /* This is early initialization for the top cgroup */ ++ return &root_task_group.css; ++ } ++ ++ tg = sched_create_group(parent); ++ if (IS_ERR(tg)) ++ return ERR_PTR(-ENOMEM); ++ return &tg->css; ++} ++ ++/* Expose task group only after completing cgroup initialization */ ++static int cpu_cgroup_css_online(struct cgroup_subsys_state *css) ++{ ++ struct task_group *tg = css_tg(css); ++ struct task_group *parent = css_tg(css->parent); ++ ++ if (parent) ++ sched_online_group(tg, parent); ++ return 0; ++} ++ ++static void cpu_cgroup_css_released(struct cgroup_subsys_state *css) ++{ ++ struct task_group *tg = css_tg(css); ++ ++ sched_offline_group(tg); ++} ++ ++static void cpu_cgroup_css_free(struct cgroup_subsys_state *css) ++{ ++ struct task_group *tg = css_tg(css); ++ ++ /* ++ * Relies on the RCU grace period between css_released() and this. ++ */ ++ sched_free_group(tg); ++} ++ ++static void cpu_cgroup_fork(struct task_struct *task) ++{ ++} ++ ++static int cpu_cgroup_can_attach(struct cgroup_taskset *tset) ++{ ++ return 0; ++} ++ ++static void cpu_cgroup_attach(struct cgroup_taskset *tset) ++{ ++} ++ ++static struct cftype cpu_files[] = { ++ { } /* Terminate */ ++}; ++ ++struct cgroup_subsys cpu_cgrp_subsys = { ++ .css_alloc = cpu_cgroup_css_alloc, ++ .css_online = cpu_cgroup_css_online, ++ .css_released = cpu_cgroup_css_released, ++ .css_free = cpu_cgroup_css_free, ++ .fork = cpu_cgroup_fork, ++ .can_attach = cpu_cgroup_can_attach, ++ .attach = cpu_cgroup_attach, ++ .legacy_cftypes = cpu_files, ++ .early_init = true, ++}; ++#endif /* CONFIG_CGROUP_SCHED */ +diff -Nur a/kernel/sched/MuQSS.h b/kernel/sched/MuQSS.h +--- a/kernel/sched/MuQSS.h 1970-01-01 01:00:00.000000000 +0100 ++++ b/kernel/sched/MuQSS.h 2018-11-03 16:06:32.715529032 +0000 +@@ -0,0 +1,725 @@ ++/* SPDX-License-Identifier: GPL-2.0 */ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++#include ++#include ++#include ++ ++#ifdef CONFIG_PARAVIRT ++#include ++#endif ++ ++#include "cpuacct.h" ++ ++#ifndef MUQSS_SCHED_H ++#define MUQSS_SCHED_H ++ ++#ifdef CONFIG_SCHED_DEBUG ++# define SCHED_WARN_ON(x) WARN_ONCE(x, #x) ++#else ++# define SCHED_WARN_ON(x) ((void)(x)) ++#endif ++ ++/* task_struct::on_rq states: */ ++#define TASK_ON_RQ_QUEUED 1 ++#define TASK_ON_RQ_MIGRATING 2 ++ ++struct rq; ++ ++#ifdef CONFIG_SMP ++ ++static inline bool sched_asym_prefer(int a, int b) ++{ ++ return arch_asym_cpu_priority(a) > arch_asym_cpu_priority(b); ++} ++ ++/* ++ * We add the notion of a root-domain which will be used to define per-domain ++ * variables. Each exclusive cpuset essentially defines an island domain by ++ * fully partitioning the member cpus from any other cpuset. Whenever a new ++ * exclusive cpuset is created, we also create and attach a new root-domain ++ * object. ++ * ++ */ ++struct root_domain { ++ atomic_t refcount; ++ atomic_t rto_count; ++ struct rcu_head rcu; ++ cpumask_var_t span; ++ cpumask_var_t online; ++ ++ /* Indicate more than one runnable task for any CPU */ ++ bool overload; ++ ++ /* ++ * The bit corresponding to a CPU gets set here if such CPU has more ++ * than one runnable -deadline task (as it is below for RT tasks). ++ */ ++ cpumask_var_t dlo_mask; ++ atomic_t dlo_count; ++ /* Replace unused CFS structures with void */ ++ //struct dl_bw dl_bw; ++ //struct cpudl cpudl; ++ void *dl_bw; ++ void *cpudl; ++ ++ /* ++ * The "RT overload" flag: it gets set if a CPU has more than ++ * one runnable RT task. ++ */ ++ cpumask_var_t rto_mask; ++ //struct cpupri cpupri; ++ void *cpupri; ++ ++ unsigned long max_cpu_capacity; ++}; ++ ++extern struct root_domain def_root_domain; ++extern struct mutex sched_domains_mutex; ++ ++extern void init_defrootdomain(void); ++extern int sched_init_domains(const struct cpumask *cpu_map); ++extern void rq_attach_root(struct rq *rq, struct root_domain *rd); ++ ++static inline void cpupri_cleanup(void __maybe_unused *cpupri) ++{ ++} ++ ++static inline void cpudl_cleanup(void __maybe_unused *cpudl) ++{ ++} ++ ++static inline void init_dl_bw(void __maybe_unused *dl_bw) ++{ ++} ++ ++static inline int cpudl_init(void __maybe_unused *dl_bw) ++{ ++ return 0; ++} ++ ++static inline int cpupri_init(void __maybe_unused *cpupri) ++{ ++ return 0; ++} ++#endif /* CONFIG_SMP */ ++ ++/* ++ * This is the main, per-CPU runqueue data structure. ++ * This data should only be modified by the local cpu. ++ */ ++struct rq { ++ raw_spinlock_t lock; ++ ++ struct task_struct *curr, *idle, *stop; ++ struct mm_struct *prev_mm; ++ ++ unsigned int nr_running; ++ /* ++ * This is part of a global counter where only the total sum ++ * over all CPUs matters. A task can increase this counter on ++ * one CPU and if it got migrated afterwards it may decrease ++ * it on another CPU. Always updated under the runqueue lock: ++ */ ++ unsigned long nr_uninterruptible; ++ u64 nr_switches; ++ ++ /* Stored data about rq->curr to work outside rq lock */ ++ u64 rq_deadline; ++ int rq_prio; ++ ++ /* Best queued id for use outside lock */ ++ u64 best_key; ++ ++ unsigned long last_scheduler_tick; /* Last jiffy this RQ ticked */ ++ unsigned long last_jiffy; /* Last jiffy this RQ updated rq clock */ ++ u64 niffies; /* Last time this RQ updated rq clock */ ++ u64 last_niffy; /* Last niffies as updated by local clock */ ++ u64 last_jiffy_niffies; /* Niffies @ last_jiffy */ ++ ++ u64 load_update; /* When we last updated load */ ++ unsigned long load_avg; /* Rolling load average */ ++#ifdef CONFIG_SMT_NICE ++ struct mm_struct *rq_mm; ++ int rq_smt_bias; /* Policy/nice level bias across smt siblings */ ++#endif ++ /* Accurate timekeeping data */ ++ unsigned long user_ns, nice_ns, irq_ns, softirq_ns, system_ns, ++ iowait_ns, idle_ns; ++ atomic_t nr_iowait; ++ ++ skiplist_node node; ++ skiplist *sl; ++#ifdef CONFIG_SMP ++ struct task_struct *preempt; /* Preempt triggered on this task */ ++ struct task_struct *preempting; /* Hint only, what task is preempting */ ++ ++ int cpu; /* cpu of this runqueue */ ++ bool online; ++ ++ struct root_domain *rd; ++ struct sched_domain *sd; ++ ++ unsigned long cpu_capacity_orig; ++ ++ int *cpu_locality; /* CPU relative cache distance */ ++ struct rq **rq_order; /* RQs ordered by relative cache distance */ ++ ++#ifdef CONFIG_SCHED_SMT ++ cpumask_t thread_mask; ++ bool (*siblings_idle)(struct rq *rq); ++ /* See if all smt siblings are idle */ ++#endif /* CONFIG_SCHED_SMT */ ++#ifdef CONFIG_SCHED_MC ++ cpumask_t core_mask; ++ bool (*cache_idle)(struct rq *rq); ++ /* See if all cache siblings are idle */ ++#endif /* CONFIG_SCHED_MC */ ++#endif /* CONFIG_SMP */ ++#ifdef CONFIG_IRQ_TIME_ACCOUNTING ++ u64 prev_irq_time; ++#endif /* CONFIG_IRQ_TIME_ACCOUNTING */ ++#ifdef CONFIG_PARAVIRT ++ u64 prev_steal_time; ++#endif /* CONFIG_PARAVIRT */ ++#ifdef CONFIG_PARAVIRT_TIME_ACCOUNTING ++ u64 prev_steal_time_rq; ++#endif /* CONFIG_PARAVIRT_TIME_ACCOUNTING */ ++ ++ u64 clock, old_clock, last_tick; ++ u64 clock_task; ++ int dither; ++ ++ int iso_ticks; ++ bool iso_refractory; ++ ++#ifdef CONFIG_HIGH_RES_TIMERS ++ struct hrtimer hrexpiry_timer; ++#endif ++ ++#ifdef CONFIG_SCHEDSTATS ++ ++ /* latency stats */ ++ struct sched_info rq_sched_info; ++ unsigned long long rq_cpu_time; ++ /* could above be rq->cfs_rq.exec_clock + rq->rt_rq.rt_runtime ? */ ++ ++ /* sys_sched_yield() stats */ ++ unsigned int yld_count; ++ ++ /* schedule() stats */ ++ unsigned int sched_switch; ++ unsigned int sched_count; ++ unsigned int sched_goidle; ++ ++ /* try_to_wake_up() stats */ ++ unsigned int ttwu_count; ++ unsigned int ttwu_local; ++#endif /* CONFIG_SCHEDSTATS */ ++ ++#ifdef CONFIG_SMP ++ struct llist_head wake_list; ++#endif ++ ++#ifdef CONFIG_CPU_IDLE ++ /* Must be inspected within a rcu lock section */ ++ struct cpuidle_state *idle_state; ++#endif ++}; ++ ++#ifdef CONFIG_SMP ++struct rq *cpu_rq(int cpu); ++#endif ++ ++#ifndef CONFIG_SMP ++extern struct rq *uprq; ++#define cpu_rq(cpu) (uprq) ++#define this_rq() (uprq) ++#define raw_rq() (uprq) ++#define task_rq(p) (uprq) ++#define cpu_curr(cpu) ((uprq)->curr) ++#else /* CONFIG_SMP */ ++DECLARE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues); ++#define this_rq() this_cpu_ptr(&runqueues) ++#define raw_rq() raw_cpu_ptr(&runqueues) ++#define task_rq(p) cpu_rq(task_cpu(p)) ++#endif /* CONFIG_SMP */ ++ ++static inline int task_current(struct rq *rq, struct task_struct *p) ++{ ++ return rq->curr == p; ++} ++ ++static inline int task_running(struct rq *rq, struct task_struct *p) ++{ ++#ifdef CONFIG_SMP ++ return p->on_cpu; ++#else ++ return task_current(rq, p); ++#endif ++} ++ ++static inline void rq_lock(struct rq *rq) ++ __acquires(rq->lock) ++{ ++ raw_spin_lock(&rq->lock); ++} ++ ++static inline void rq_unlock(struct rq *rq) ++ __releases(rq->lock) ++{ ++ raw_spin_unlock(&rq->lock); ++} ++ ++static inline void rq_lock_irq(struct rq *rq) ++ __acquires(rq->lock) ++{ ++ raw_spin_lock_irq(&rq->lock); ++} ++ ++static inline void rq_unlock_irq(struct rq *rq) ++ __releases(rq->lock) ++{ ++ raw_spin_unlock_irq(&rq->lock); ++} ++ ++static inline void rq_lock_irqsave(struct rq *rq, unsigned long *flags) ++ __acquires(rq->lock) ++{ ++ raw_spin_lock_irqsave(&rq->lock, *flags); ++} ++ ++static inline void rq_unlock_irqrestore(struct rq *rq, unsigned long *flags) ++ __releases(rq->lock) ++{ ++ raw_spin_unlock_irqrestore(&rq->lock, *flags); ++} ++ ++static inline struct rq *task_rq_lock(struct task_struct *p, unsigned long *flags) ++ __acquires(p->pi_lock) ++ __acquires(rq->lock) ++{ ++ struct rq *rq; ++ ++ while (42) { ++ raw_spin_lock_irqsave(&p->pi_lock, *flags); ++ rq = task_rq(p); ++ raw_spin_lock(&rq->lock); ++ if (likely(rq == task_rq(p))) ++ break; ++ raw_spin_unlock(&rq->lock); ++ raw_spin_unlock_irqrestore(&p->pi_lock, *flags); ++ } ++ return rq; ++} ++ ++static inline void task_rq_unlock(struct rq *rq, struct task_struct *p, unsigned long *flags) ++ __releases(rq->lock) ++ __releases(p->pi_lock) ++{ ++ rq_unlock(rq); ++ raw_spin_unlock_irqrestore(&p->pi_lock, *flags); ++} ++ ++static inline struct rq *__task_rq_lock(struct task_struct *p) ++ __acquires(rq->lock) ++{ ++ struct rq *rq; ++ ++ lockdep_assert_held(&p->pi_lock); ++ ++ while (42) { ++ rq = task_rq(p); ++ raw_spin_lock(&rq->lock); ++ if (likely(rq == task_rq(p))) ++ break; ++ raw_spin_unlock(&rq->lock); ++ } ++ return rq; ++} ++ ++static inline void __task_rq_unlock(struct rq *rq) ++{ ++ rq_unlock(rq); ++} ++ ++/* ++ * {de,en}queue flags: Most not used on MuQSS. ++ * ++ * DEQUEUE_SLEEP - task is no longer runnable ++ * ENQUEUE_WAKEUP - task just became runnable ++ * ++ * SAVE/RESTORE - an otherwise spurious dequeue/enqueue, done to ensure tasks ++ * are in a known state which allows modification. Such pairs ++ * should preserve as much state as possible. ++ * ++ * MOVE - paired with SAVE/RESTORE, explicitly does not preserve the location ++ * in the runqueue. ++ * ++ * ENQUEUE_HEAD - place at front of runqueue (tail if not specified) ++ * ENQUEUE_REPLENISH - CBS (replenish runtime and postpone deadline) ++ * ENQUEUE_MIGRATED - the task was migrated during wakeup ++ * ++ */ ++ ++#define DEQUEUE_SAVE 0x02 /* matches ENQUEUE_RESTORE */ ++ ++#define ENQUEUE_RESTORE 0x02 ++ ++static inline u64 __rq_clock_broken(struct rq *rq) ++{ ++ return READ_ONCE(rq->clock); ++} ++ ++static inline u64 rq_clock(struct rq *rq) ++{ ++ lockdep_assert_held(&rq->lock); ++ ++ return rq->clock; ++} ++ ++static inline u64 rq_clock_task(struct rq *rq) ++{ ++ lockdep_assert_held(&rq->lock); ++ ++ return rq->clock_task; ++} ++ ++#ifdef CONFIG_NUMA ++enum numa_topology_type { ++ NUMA_DIRECT, ++ NUMA_GLUELESS_MESH, ++ NUMA_BACKPLANE, ++}; ++extern enum numa_topology_type sched_numa_topology_type; ++extern int sched_max_numa_distance; ++extern bool find_numa_distance(int distance); ++ ++extern void sched_init_numa(void); ++extern void sched_domains_numa_masks_set(unsigned int cpu); ++extern void sched_domains_numa_masks_clear(unsigned int cpu); ++#else ++static inline void sched_init_numa(void) { } ++static inline void sched_domains_numa_masks_set(unsigned int cpu) { } ++static inline void sched_domains_numa_masks_clear(unsigned int cpu) { } ++#endif ++ ++extern struct mutex sched_domains_mutex; ++extern struct static_key_false sched_schedstats; ++ ++#define rcu_dereference_check_sched_domain(p) \ ++ rcu_dereference_check((p), \ ++ lockdep_is_held(&sched_domains_mutex)) ++ ++#ifdef CONFIG_SMP ++ ++/* ++ * The domain tree (rq->sd) is protected by RCU's quiescent state transition. ++ * See detach_destroy_domains: synchronize_sched for details. ++ * ++ * The domain tree of any CPU may only be accessed from within ++ * preempt-disabled sections. ++ */ ++#define for_each_domain(cpu, __sd) \ ++ for (__sd = rcu_dereference_check_sched_domain(cpu_rq(cpu)->sd); \ ++ __sd; __sd = __sd->parent) ++ ++#define for_each_lower_domain(sd) for (; sd; sd = sd->child) ++ ++/** ++ * highest_flag_domain - Return highest sched_domain containing flag. ++ * @cpu: The cpu whose highest level of sched domain is to ++ * be returned. ++ * @flag: The flag to check for the highest sched_domain ++ * for the given cpu. ++ * ++ * Returns the highest sched_domain of a cpu which contains the given flag. ++ */ ++static inline struct sched_domain *highest_flag_domain(int cpu, int flag) ++{ ++ struct sched_domain *sd, *hsd = NULL; ++ ++ for_each_domain(cpu, sd) { ++ if (!(sd->flags & flag)) ++ break; ++ hsd = sd; ++ } ++ ++ return hsd; ++} ++ ++static inline struct sched_domain *lowest_flag_domain(int cpu, int flag) ++{ ++ struct sched_domain *sd; ++ ++ for_each_domain(cpu, sd) { ++ if (sd->flags & flag) ++ break; ++ } ++ ++ return sd; ++} ++ ++DECLARE_PER_CPU(struct sched_domain *, sd_llc); ++DECLARE_PER_CPU(int, sd_llc_size); ++DECLARE_PER_CPU(int, sd_llc_id); ++DECLARE_PER_CPU(struct sched_domain_shared *, sd_llc_shared); ++DECLARE_PER_CPU(struct sched_domain *, sd_numa); ++DECLARE_PER_CPU(struct sched_domain *, sd_asym); ++ ++struct sched_group_capacity { ++ atomic_t ref; ++ /* ++ * CPU capacity of this group, SCHED_CAPACITY_SCALE being max capacity ++ * for a single CPU. ++ */ ++ unsigned long capacity; ++ unsigned long min_capacity; /* Min per-CPU capacity in group */ ++ unsigned long next_update; ++ int imbalance; /* XXX unrelated to capacity but shared group state */ ++ ++#ifdef CONFIG_SCHED_DEBUG ++ int id; ++#endif ++ ++ unsigned long cpumask[0]; /* balance mask */ ++}; ++ ++struct sched_group { ++ struct sched_group *next; /* Must be a circular list */ ++ atomic_t ref; ++ ++ unsigned int group_weight; ++ struct sched_group_capacity *sgc; ++ int asym_prefer_cpu; /* cpu of highest priority in group */ ++ ++ /* ++ * The CPUs this group covers. ++ * ++ * NOTE: this field is variable length. (Allocated dynamically ++ * by attaching extra space to the end of the structure, ++ * depending on how many CPUs the kernel has booted up with) ++ */ ++ unsigned long cpumask[0]; ++}; ++ ++static inline struct cpumask *sched_group_span(struct sched_group *sg) ++{ ++ return to_cpumask(sg->cpumask); ++} ++ ++/* ++ * See build_balance_mask(). ++ */ ++static inline struct cpumask *group_balance_mask(struct sched_group *sg) ++{ ++ return to_cpumask(sg->sgc->cpumask); ++} ++ ++/** ++ * group_first_cpu - Returns the first cpu in the cpumask of a sched_group. ++ * @group: The group whose first cpu is to be returned. ++ */ ++static inline unsigned int group_first_cpu(struct sched_group *group) ++{ ++ return cpumask_first(sched_group_span(group)); ++} ++ ++ ++#if defined(CONFIG_SCHED_DEBUG) && defined(CONFIG_SYSCTL) ++void register_sched_domain_sysctl(void); ++void dirty_sched_domain_sysctl(int cpu); ++void unregister_sched_domain_sysctl(void); ++#else ++static inline void register_sched_domain_sysctl(void) ++{ ++} ++static inline void dirty_sched_domain_sysctl(int cpu) ++{ ++} ++static inline void unregister_sched_domain_sysctl(void) ++{ ++} ++#endif ++ ++extern void sched_ttwu_pending(void); ++extern void set_cpus_allowed_common(struct task_struct *p, const struct cpumask *new_mask); ++extern void set_rq_online (struct rq *rq); ++extern void set_rq_offline(struct rq *rq); ++extern bool sched_smp_initialized; ++ ++static inline void update_group_capacity(struct sched_domain *sd, int cpu) ++{ ++} ++ ++static inline void trigger_load_balance(struct rq *rq) ++{ ++} ++ ++#define sched_feat(x) 0 ++ ++#else /* CONFIG_SMP */ ++ ++static inline void sched_ttwu_pending(void) { } ++ ++#endif /* CONFIG_SMP */ ++ ++#ifdef CONFIG_CPU_IDLE ++static inline void idle_set_state(struct rq *rq, ++ struct cpuidle_state *idle_state) ++{ ++ rq->idle_state = idle_state; ++} ++ ++static inline struct cpuidle_state *idle_get_state(struct rq *rq) ++{ ++ SCHED_WARN_ON(!rcu_read_lock_held()); ++ return rq->idle_state; ++} ++#else ++static inline void idle_set_state(struct rq *rq, ++ struct cpuidle_state *idle_state) ++{ ++} ++ ++static inline struct cpuidle_state *idle_get_state(struct rq *rq) ++{ ++ return NULL; ++} ++#endif ++ ++#ifdef CONFIG_SCHED_DEBUG ++extern bool sched_debug_enabled; ++#endif ++ ++extern void schedule_idle(void); ++ ++#ifdef CONFIG_IRQ_TIME_ACCOUNTING ++struct irqtime { ++ u64 total; ++ u64 tick_delta; ++ u64 irq_start_time; ++ struct u64_stats_sync sync; ++}; ++ ++DECLARE_PER_CPU(struct irqtime, cpu_irqtime); ++ ++/* ++ * Returns the irqtime minus the softirq time computed by ksoftirqd. ++ * Otherwise ksoftirqd's sum_exec_runtime is substracted its own runtime ++ * and never move forward. ++ */ ++static inline u64 irq_time_read(int cpu) ++{ ++ struct irqtime *irqtime = &per_cpu(cpu_irqtime, cpu); ++ unsigned int seq; ++ u64 total; ++ ++ do { ++ seq = __u64_stats_fetch_begin(&irqtime->sync); ++ total = irqtime->total; ++ } while (__u64_stats_fetch_retry(&irqtime->sync, seq)); ++ ++ return total; ++} ++#endif /* CONFIG_IRQ_TIME_ACCOUNTING */ ++ ++#ifdef CONFIG_SMP ++static inline int cpu_of(struct rq *rq) ++{ ++ return rq->cpu; ++} ++#else /* CONFIG_SMP */ ++static inline int cpu_of(struct rq *rq) ++{ ++ return 0; ++} ++#endif ++ ++#ifdef CONFIG_CPU_FREQ ++DECLARE_PER_CPU(struct update_util_data *, cpufreq_update_util_data); ++ ++static inline void cpufreq_trigger(struct rq *rq, unsigned int flags) ++{ ++ struct update_util_data *data; ++ ++ data = rcu_dereference_sched(*per_cpu_ptr(&cpufreq_update_util_data, ++ cpu_of(rq))); ++ ++ if (data) ++ data->func(data, rq->niffies, flags); ++} ++#else ++static inline void cpufreq_trigger(struct rq *rq, unsigned int flag) ++{ ++} ++#endif /* CONFIG_CPU_FREQ */ ++ ++#ifdef arch_scale_freq_capacity ++#ifndef arch_scale_freq_invariant ++#define arch_scale_freq_invariant() (true) ++#endif ++#else /* arch_scale_freq_capacity */ ++#define arch_scale_freq_invariant() (false) ++#endif ++ ++/* ++ * This should only be called when current == rq->idle. Dodgy workaround for ++ * when softirqs are pending and we are in the idle loop. Setting current to ++ * resched will kick us out of the idle loop and the softirqs will be serviced ++ * on our next pass through schedule(). ++ */ ++static inline bool softirq_pending(int cpu) ++{ ++ if (likely(!local_softirq_pending())) ++ return false; ++ set_tsk_need_resched(current); ++ return true; ++} ++ ++#ifdef CONFIG_64BIT ++static inline u64 read_sum_exec_runtime(struct task_struct *t) ++{ ++ return tsk_seruntime(t); ++} ++#else ++struct rq *task_rq_lock(struct task_struct *p, unsigned long *flags); ++void task_rq_unlock(struct rq *rq, struct task_struct *p, unsigned long *flags); ++ ++static inline u64 read_sum_exec_runtime(struct task_struct *t) ++{ ++ unsigned long flags; ++ u64 ns; ++ struct rq *rq; ++ ++ rq = task_rq_lock(t, &flags); ++ ns = tsk_seruntime(t); ++ task_rq_unlock(rq, t, &flags); ++ ++ return ns; ++} ++#endif ++ ++#endif /* MUQSS_SCHED_H */ +diff -Nur a/kernel/sched/sched.h b/kernel/sched/sched.h +--- a/kernel/sched/sched.h 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/sched/sched.h 2018-11-03 16:06:32.717529096 +0000 +@@ -1,5 +1,8 @@ + /* SPDX-License-Identifier: GPL-2.0 */ + ++#ifdef CONFIG_SCHED_MUQSS ++#include "MuQSS.h" ++#else /* CONFIG_SCHED_MUQSS */ + #include + #include + #include +@@ -2103,3 +2106,29 @@ + #else /* arch_scale_freq_capacity */ + #define arch_scale_freq_invariant() (false) + #endif ++ ++static inline bool softirq_pending(int cpu) ++{ ++ return false; ++} ++ ++#ifdef CONFIG_64BIT ++static inline u64 read_sum_exec_runtime(struct task_struct *t) ++{ ++ return t->se.sum_exec_runtime; ++} ++#else ++static inline u64 read_sum_exec_runtime(struct task_struct *t) ++{ ++ u64 ns; ++ struct rq_flags rf; ++ struct rq *rq; ++ ++ rq = task_rq_lock(t, &rf); ++ ns = t->se.sum_exec_runtime; ++ task_rq_unlock(rq, t, &rf); ++ ++ return ns; ++} ++#endif ++#endif /* CONFIG_SCHED_MUQSS */ +diff -Nur a/kernel/skip_list.c b/kernel/skip_list.c +--- a/kernel/skip_list.c 1970-01-01 01:00:00.000000000 +0100 ++++ b/kernel/skip_list.c 2018-11-03 16:06:32.717529096 +0000 +@@ -0,0 +1,148 @@ ++/* ++ Copyright (C) 2011,2016 Con Kolivas. ++ ++ Code based on example originally by William Pugh. ++ ++Skip Lists are a probabilistic alternative to balanced trees, as ++described in the June 1990 issue of CACM and were invented by ++William Pugh in 1987. ++ ++A couple of comments about this implementation: ++The routine randomLevel has been hard-coded to generate random ++levels using p=0.25. It can be easily changed. ++ ++The insertion routine has been implemented so as to use the ++dirty hack described in the CACM paper: if a random level is ++generated that is more than the current maximum level, the ++current maximum level plus one is used instead. ++ ++Levels start at zero and go up to MaxLevel (which is equal to ++MaxNumberOfLevels-1). ++ ++The routines defined in this file are: ++ ++init: defines slnode ++ ++new_skiplist: returns a new, empty list ++ ++randomLevel: Returns a random level based on a u64 random seed passed to it. ++In MuQSS, the "niffy" time is used for this purpose. ++ ++insert(l,key, value): inserts the binding (key, value) into l. This operation ++occurs in O(log n) time. ++ ++delnode(slnode, l, node): deletes any binding of key from the l based on the ++actual node value. This operation occurs in O(k) time where k is the ++number of levels of the node in question (max 8). The original delete ++function occurred in O(log n) time and involved a search. ++ ++MuQSS Notes: In this implementation of skiplists, there are bidirectional ++next/prev pointers and the insert function returns a pointer to the actual ++node the value is stored. The key here is chosen by the scheduler so as to ++sort tasks according to the priority list requirements and is no longer used ++by the scheduler after insertion. The scheduler lookup, however, occurs in ++O(1) time because it is always the first item in the level 0 linked list. ++Since the task struct stores a copy of the node pointer upon skiplist_insert, ++it can also remove it much faster than the original implementation with the ++aid of prev<->next pointer manipulation and no searching. ++ ++*/ ++ ++#include ++#include ++ ++#define MaxNumberOfLevels 8 ++#define MaxLevel (MaxNumberOfLevels - 1) ++ ++void skiplist_init(skiplist_node *slnode) ++{ ++ int i; ++ ++ slnode->key = 0xFFFFFFFFFFFFFFFF; ++ slnode->level = 0; ++ slnode->value = NULL; ++ for (i = 0; i < MaxNumberOfLevels; i++) ++ slnode->next[i] = slnode->prev[i] = slnode; ++} ++ ++skiplist *new_skiplist(skiplist_node *slnode) ++{ ++ skiplist *l = kzalloc(sizeof(skiplist), GFP_ATOMIC); ++ ++ BUG_ON(!l); ++ l->header = slnode; ++ return l; ++} ++ ++void free_skiplist(skiplist *l) ++{ ++ skiplist_node *p, *q; ++ ++ p = l->header; ++ do { ++ q = p->next[0]; ++ p->next[0]->prev[0] = q->prev[0]; ++ skiplist_node_init(p); ++ p = q; ++ } while (p != l->header); ++ kfree(l); ++} ++ ++void skiplist_node_init(skiplist_node *node) ++{ ++ memset(node, 0, sizeof(skiplist_node)); ++} ++ ++static inline unsigned int randomLevel(const long unsigned int randseed) ++{ ++ return find_first_bit(&randseed, MaxLevel) / 2; ++} ++ ++void skiplist_insert(skiplist *l, skiplist_node *node, keyType key, valueType value, unsigned int randseed) ++{ ++ skiplist_node *update[MaxNumberOfLevels]; ++ skiplist_node *p, *q; ++ int k = l->level; ++ ++ p = l->header; ++ do { ++ while (q = p->next[k], q->key <= key) ++ p = q; ++ update[k] = p; ++ } while (--k >= 0); ++ ++ ++l->entries; ++ k = randomLevel(randseed); ++ if (k > l->level) { ++ k = ++l->level; ++ update[k] = l->header; ++ } ++ ++ node->level = k; ++ node->key = key; ++ node->value = value; ++ do { ++ p = update[k]; ++ node->next[k] = p->next[k]; ++ p->next[k] = node; ++ node->prev[k] = p; ++ node->next[k]->prev[k] = node; ++ } while (--k >= 0); ++} ++ ++void skiplist_delete(skiplist *l, skiplist_node *node) ++{ ++ int k, m = node->level; ++ ++ for (k = 0; k <= m; k++) { ++ node->prev[k]->next[k] = node->next[k]; ++ node->next[k]->prev[k] = node->prev[k]; ++ } ++ skiplist_node_init(node); ++ if (m == l->level) { ++ while (l->header->next[m] == l->header && l->header->prev[m] == l->header && m > 0) ++ m--; ++ l->level = m; ++ } ++ l->entries--; ++} +diff -Nur a/kernel/sysctl.c b/kernel/sysctl.c +--- a/kernel/sysctl.c 2018-11-03 16:00:51.933620936 +0000 ++++ b/kernel/sysctl.c 2018-11-03 16:12:48.444570622 +0000 +@@ -133,8 +133,14 @@ + static int __maybe_unused two __read_only = 2; + static int __maybe_unused four __read_only = 4; + static unsigned long one_ul __read_only = 1; +-static int one_hundred __read_only = 100; +-static int one_thousand __read_only = 1000; ++static int one_hundred __read_only = 100; ++static int one_thousand __read_only = 1000; ++#ifdef CONFIG_SCHED_MUQSS ++extern int rr_interval; ++extern int sched_interactive; ++extern int sched_iso_cpu; ++extern int sched_yield_type; ++#endif + #ifdef CONFIG_PRINTK + static int ten_thousand __read_only = 10000; + #endif +@@ -296,7 +302,7 @@ + { } + }; + +-#ifdef CONFIG_SCHED_DEBUG ++#if defined(CONFIG_SCHED_DEBUG) && !defined(CONFIG_SCHED_MUQSS) + static int min_sched_granularity_ns __read_only = 100000; /* 100 usecs */ + static int max_sched_granularity_ns __read_only = NSEC_PER_SEC; /* 1 second */ + static int min_wakeup_granularity_ns __read_only; /* 0 usecs */ +@@ -313,6 +319,7 @@ + #endif + + static struct ctl_table kern_table[] = { ++#ifndef CONFIG_SCHED_MUQSS + { + .procname = "sched_child_runs_first", + .data = &sysctl_sched_child_runs_first, +@@ -475,6 +482,7 @@ + .extra1 = &one, + }, + #endif ++#endif /* !CONFIG_SCHED_MUQSS */ + #ifdef CONFIG_PROVE_LOCKING + { + .procname = "prove_locking", +@@ -1073,6 +1081,44 @@ + .proc_handler = proc_dointvec, + }, + #endif ++#ifdef CONFIG_SCHED_MUQSS ++ { ++ .procname = "rr_interval", ++ .data = &rr_interval, ++ .maxlen = sizeof (int), ++ .mode = 0644, ++ .proc_handler = &proc_dointvec_minmax, ++ .extra1 = &one, ++ .extra2 = &one_thousand, ++ }, ++ { ++ .procname = "interactive", ++ .data = &sched_interactive, ++ .maxlen = sizeof(int), ++ .mode = 0644, ++ .proc_handler = &proc_dointvec_minmax, ++ .extra1 = &zero, ++ .extra2 = &one, ++ }, ++ { ++ .procname = "iso_cpu", ++ .data = &sched_iso_cpu, ++ .maxlen = sizeof (int), ++ .mode = 0644, ++ .proc_handler = &proc_dointvec_minmax, ++ .extra1 = &zero, ++ .extra2 = &one_hundred, ++ }, ++ { ++ .procname = "yield_type", ++ .data = &sched_yield_type, ++ .maxlen = sizeof (int), ++ .mode = 0644, ++ .proc_handler = &proc_dointvec_minmax, ++ .extra1 = &zero, ++ .extra2 = &two, ++ }, ++#endif + #if defined(CONFIG_S390) && defined(CONFIG_SMP) + { + .procname = "spin_retry", +diff -Nur a/kernel/time/clockevents.c b/kernel/time/clockevents.c +--- a/kernel/time/clockevents.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/time/clockevents.c 2018-11-03 16:06:32.719529160 +0000 +@@ -198,8 +198,13 @@ + + #ifdef CONFIG_GENERIC_CLOCKEVENTS_MIN_ADJUST + ++#ifdef CONFIG_SCHED_MUQSS ++/* Limit min_delta to 100us */ ++#define MIN_DELTA_LIMIT (NSEC_PER_SEC / 10000) ++#else + /* Limit min_delta to a jiffie */ + #define MIN_DELTA_LIMIT (NSEC_PER_SEC / HZ) ++#endif + + /** + * clockevents_increase_min_delta - raise minimum delta of a clock event device +diff -Nur a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c +--- a/kernel/time/posix-cpu-timers.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/time/posix-cpu-timers.c 2018-11-03 16:06:32.719529160 +0000 +@@ -818,7 +818,7 @@ + tsk_expires->virt_exp = expires; + + tsk_expires->sched_exp = check_timers_list(++timers, firing, +- tsk->se.sum_exec_runtime); ++ tsk_seruntime(tsk)); + + /* + * Check for the special case thread timers. +@@ -828,7 +828,7 @@ + unsigned long hard = task_rlimit_max(tsk, RLIMIT_RTTIME); + + if (hard != RLIM_INFINITY && +- tsk->rt.timeout > DIV_ROUND_UP(hard, USEC_PER_SEC/HZ)) { ++ tsk_rttimeout(tsk) > DIV_ROUND_UP(hard, USEC_PER_SEC/HZ)) { + /* + * At the hard limit, we just die. + * No need to calculate anything else now. +@@ -840,7 +840,7 @@ + __group_send_sig_info(SIGKILL, SEND_SIG_PRIV, tsk); + return; + } +- if (tsk->rt.timeout > DIV_ROUND_UP(soft, USEC_PER_SEC/HZ)) { ++ if (tsk_rttimeout(tsk) > DIV_ROUND_UP(soft, USEC_PER_SEC/HZ)) { + /* + * At the soft limit, send a SIGXCPU every second. + */ +@@ -1081,7 +1081,7 @@ + struct task_cputime task_sample; + + task_cputime(tsk, &task_sample.utime, &task_sample.stime); +- task_sample.sum_exec_runtime = tsk->se.sum_exec_runtime; ++ task_sample.sum_exec_runtime = tsk_seruntime(tsk); + if (task_cputime_expired(&task_sample, &tsk->cputime_expires)) + return 1; + } +diff -Nur a/kernel/time/timer.c b/kernel/time/timer.c +--- a/kernel/time/timer.c 2018-11-03 16:00:51.934620967 +0000 ++++ b/kernel/time/timer.c 2018-11-03 16:06:32.720529192 +0000 +@@ -1434,7 +1434,7 @@ + * Check, if the next hrtimer event is before the next timer wheel + * event: + */ +-static u64 cmp_next_hrtimer_event(u64 basem, u64 expires) ++static u64 cmp_next_hrtimer_event(struct timer_base *base, u64 basem, u64 expires) + { + u64 nextevt = hrtimer_get_next_event(); + +@@ -1452,6 +1452,9 @@ + if (nextevt <= basem) + return basem; + ++ if (nextevt < expires && nextevt - basem <= TICK_NSEC) ++ base->is_idle = false; ++ + /* + * Round up to the next jiffie. High resolution timers are + * off, so the hrtimers are expired in the tick and we need to +@@ -1521,7 +1524,7 @@ + } + raw_spin_unlock(&base->lock); + +- return cmp_next_hrtimer_event(basem, expires); ++ return cmp_next_hrtimer_event(base, basem, expires); + } + + /** +diff -Nur a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c +--- a/kernel/trace/trace_selftest.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/kernel/trace/trace_selftest.c 2018-11-03 16:06:32.720529192 +0000 +@@ -1041,10 +1041,15 @@ + { + /* Make this a -deadline thread */ + static const struct sched_attr attr = { ++#ifdef CONFIG_SCHED_MUQSS ++ /* No deadline on MuQSS, use RR */ ++ .sched_policy = SCHED_RR, ++#else + .sched_policy = SCHED_DEADLINE, + .sched_runtime = 100000ULL, + .sched_deadline = 10000000ULL, + .sched_period = 10000000ULL ++#endif + }; + struct wakeup_test_data *x = data; + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0002-BFQ-v8r12-20180404.patch b/sys-kernel/linux-sources-redcore-lts/files/0002-BFQ-v8r12-20180404.patch new file mode 100644 index 00000000..104325d6 --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0002-BFQ-v8r12-20180404.patch @@ -0,0 +1,4611 @@ +From 7bd365a925748767d7ed807e5498f90bae0ebc25 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Tue, 14 Nov 2017 08:28:45 +0100 +Subject: [PATCH 01/23] block, bfq-mq: turn BUG_ON on request-size into WARN_ON + +BFQ has many checks of internal and external consistency. One of them +checks that an I/O request has still sectors to serve, if it happens +to be retired without being served. If the request has no sector to +serve, a BUG_ON signals the failure and causes the kernel to +terminate. Yet, from a crash report by a user [1], this condition may +happen to hold, in apparently correct functioning, for I/O with a +CD/DVD. + +To address this issue, this commit turns the above BUG_ON into a +WARN_ON. This commit also adds a companion WARN_ON on request +insertion into the scheduler. + +[1] https://groups.google.com/d/msg/bfq-iosched/DDOTJBroBa4/VyU1zUFtCgAJ + +Reported-by: Alexandre Frade +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 4 +++- + 1 file changed, 3 insertions(+), 1 deletion(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 0c09609a6099..0fc757ae7a42 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -1540,6 +1540,8 @@ static void bfq_add_request(struct request *rq) + + BUG_ON(!RQ_BFQQ(rq)); + BUG_ON(RQ_BFQQ(rq) != bfqq); ++ WARN_ON(blk_rq_sectors(rq) == 0); ++ + elv_rb_add(&bfqq->sort_list, rq); + + /* +@@ -4962,7 +4964,7 @@ static void bfq_finish_request(struct request *rq) + rq_io_start_time_ns(rq), + rq->cmd_flags); + +- BUG_ON(blk_rq_sectors(rq) == 0 && !(rq->rq_flags & RQF_STARTED)); ++ WARN_ON(blk_rq_sectors(rq) == 0 && !(rq->rq_flags & RQF_STARTED)); + + if (likely(rq->rq_flags & RQF_STARTED)) { + unsigned long flags; + +From 1097d368a20456c88acd75b3184c68df38e8f7b8 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Sun, 12 Nov 2017 22:43:46 +0100 +Subject: [PATCH 02/23] block, bfq-sq, bfq-mq: consider also past I/O in soft + real-time detection + +BFQ privileges the I/O of soft real-time applications, such as video +players, to guarantee to these application a high bandwidth and a low +latency. In this respect, it is not easy to correctly detect when an +application is soft real-time. A particularly nasty false positive is +that of an I/O-bound application that occasionally happens to meet all +requirements to be deemed as soft real-time. After being detected as +soft real-time, such an application monopolizes the device. Fortunately, +BFQ will realize soon that the application is actually not soft +real-time and suspend every privilege. Yet, the application may happen +again to be wrongly detected as soft real-time, and so on. + +As highlighted by our tests, this problem causes BFQ to occasionally +fail to guarantee a high responsiveness, in the presence of heavy +background I/O workloads. The reason is that the background workload +happens to be detected as soft real-time, more or less frequently, +during the execution of the interactive task under test. To give an +idea, because of this problem, Libreoffice Writer occasionally takes 8 +seconds, instead of 3, to start up, if there are sequential reads and +writes in the background, on a Kingston SSDNow V300. + +This commit addresses this issue by leveraging the following facts. + +The reason why some applications are detected as soft real-time despite +all BFQ checks to avoid false positives, is simply that, during high +CPU or storage-device load, I/O-bound applications may happen to do +I/O slowly enough to meet all soft real-time requirements, and pass +all BFQ extra checks. Yet, this happens only for limited time periods: +slow-speed time intervals are usually interspersed between other time +intervals during which these applications do I/O at a very high speed. +To exploit these facts, this commit introduces a little change, in the +detection of soft real-time behavior, to systematically consider also +the recent past: the higher the speed was in the recent past, the +later next I/O should arrive for the application to be considered as +soft real-time. At the beginning of a slow-speed interval, the minimum +arrival time allowed for the next I/O usually happens to still be so +high, to fall *after* the end of the slow-speed period itself. As a +consequence, the application does not risk to be deemed as soft +real-time during the slow-speed interval. Then, during the next +high-speed interval, the application cannot, evidently, be deemed as +soft real-time (exactly because of its speed), and so on. + +This extra filtering proved to be rather effective: in the above test, +the frequency of false positives became so low that the start-up time +was 3 seconds in all iterations (apart from occasional outliers, +caused by page-cache-management issues, which are out of the scope of +this commit, and cannot be solved by an I/O scheduler). + +Signed-off-by: Paolo Valente +Signed-off-by: Angelo Ruocco +--- + block/bfq-mq-iosched.c | 115 ++++++++++++++++++++++++++++++++++--------------- + block/bfq-sq-iosched.c | 115 ++++++++++++++++++++++++++++++++++--------------- + 2 files changed, 162 insertions(+), 68 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 0fc757ae7a42..4d06d900f45e 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -3201,37 +3201,78 @@ static bool bfq_bfqq_is_slow(struct bfq_data *bfqd, struct bfq_queue *bfqq, + * whereas soft_rt_next_start is set to infinity for applications that do + * not. + * +- * Unfortunately, even a greedy application may happen to behave in an +- * isochronous way if the CPU load is high. In fact, the application may +- * stop issuing requests while the CPUs are busy serving other processes, +- * then restart, then stop again for a while, and so on. In addition, if +- * the disk achieves a low enough throughput with the request pattern +- * issued by the application (e.g., because the request pattern is random +- * and/or the device is slow), then the application may meet the above +- * bandwidth requirement too. To prevent such a greedy application to be +- * deemed as soft real-time, a further rule is used in the computation of +- * soft_rt_next_start: soft_rt_next_start must be higher than the current +- * time plus the maximum time for which the arrival of a request is waited +- * for when a sync queue becomes idle, namely bfqd->bfq_slice_idle. +- * This filters out greedy applications, as the latter issue instead their +- * next request as soon as possible after the last one has been completed +- * (in contrast, when a batch of requests is completed, a soft real-time +- * application spends some time processing data). ++ * Unfortunately, even a greedy (i.e., I/O-bound) application may ++ * happen to meet, occasionally or systematically, both the above ++ * bandwidth and isochrony requirements. This may happen at least in ++ * the following circumstances. First, if the CPU load is high. The ++ * application may stop issuing requests while the CPUs are busy ++ * serving other processes, then restart, then stop again for a while, ++ * and so on. The other circumstances are related to the storage ++ * device: the storage device is highly loaded or reaches a low-enough ++ * throughput with the I/O of the application (e.g., because the I/O ++ * is random and/or the device is slow). In all these cases, the ++ * I/O of the application may be simply slowed down enough to meet ++ * the bandwidth and isochrony requirements. To reduce the probability ++ * that greedy applications are deemed as soft real-time in these ++ * corner cases, a further rule is used in the computation of ++ * soft_rt_next_start: the return value of this function is forced to ++ * be higher than the maximum between the following two quantities. + * +- * Unfortunately, the last filter may easily generate false positives if +- * only bfqd->bfq_slice_idle is used as a reference time interval and one +- * or both the following cases occur: +- * 1) HZ is so low that the duration of a jiffy is comparable to or higher +- * than bfqd->bfq_slice_idle. This happens, e.g., on slow devices with +- * HZ=100. ++ * (a) Current time plus: (1) the maximum time for which the arrival ++ * of a request is waited for when a sync queue becomes idle, ++ * namely bfqd->bfq_slice_idle, and (2) a few extra jiffies. We ++ * postpone for a moment the reason for adding a few extra ++ * jiffies; we get back to it after next item (b). Lower-bounding ++ * the return value of this function with the current time plus ++ * bfqd->bfq_slice_idle tends to filter out greedy applications, ++ * because the latter issue their next request as soon as possible ++ * after the last one has been completed. In contrast, a soft ++ * real-time application spends some time processing data, after a ++ * batch of its requests has been completed. ++ * ++ * (b) Current value of bfqq->soft_rt_next_start. As pointed out ++ * above, greedy applications may happen to meet both the ++ * bandwidth and isochrony requirements under heavy CPU or ++ * storage-device load. In more detail, in these scenarios, these ++ * applications happen, only for limited time periods, to do I/O ++ * slowly enough to meet all the requirements described so far, ++ * including the filtering in above item (a). These slow-speed ++ * time intervals are usually interspersed between other time ++ * intervals during which these applications do I/O at a very high ++ * speed. Fortunately, exactly because of the high speed of the ++ * I/O in the high-speed intervals, the values returned by this ++ * function happen to be so high, near the end of any such ++ * high-speed interval, to be likely to fall *after* the end of ++ * the low-speed time interval that follows. These high values are ++ * stored in bfqq->soft_rt_next_start after each invocation of ++ * this function. As a consequence, if the last value of ++ * bfqq->soft_rt_next_start is constantly used to lower-bound the ++ * next value that this function may return, then, from the very ++ * beginning of a low-speed interval, bfqq->soft_rt_next_start is ++ * likely to be constantly kept so high that any I/O request ++ * issued during the low-speed interval is considered as arriving ++ * to soon for the application to be deemed as soft ++ * real-time. Then, in the high-speed interval that follows, the ++ * application will not be deemed as soft real-time, just because ++ * it will do I/O at a high speed. And so on. ++ * ++ * Getting back to the filtering in item (a), in the following two ++ * cases this filtering might be easily passed by a greedy ++ * application, if the reference quantity was just ++ * bfqd->bfq_slice_idle: ++ * 1) HZ is so low that the duration of a jiffy is comparable to or ++ * higher than bfqd->bfq_slice_idle. This happens, e.g., on slow ++ * devices with HZ=100. The time granularity may be so coarse ++ * that the approximation, in jiffies, of bfqd->bfq_slice_idle ++ * is rather lower than the exact value. + * 2) jiffies, instead of increasing at a constant rate, may stop increasing + * for a while, then suddenly 'jump' by several units to recover the lost + * increments. This seems to happen, e.g., inside virtual machines. +- * To address this issue, we do not use as a reference time interval just +- * bfqd->bfq_slice_idle, but bfqd->bfq_slice_idle plus a few jiffies. In +- * particular we add the minimum number of jiffies for which the filter +- * seems to be quite precise also in embedded systems and KVM/QEMU virtual +- * machines. ++ * To address this issue, in the filtering in (a) we do not use as a ++ * reference time interval just bfqd->bfq_slice_idle, but ++ * bfqd->bfq_slice_idle plus a few jiffies. In particular, we add the ++ * minimum number of jiffies for which the filter seems to be quite ++ * precise also in embedded systems and KVM/QEMU virtual machines. + */ + static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, + struct bfq_queue *bfqq) +@@ -3243,10 +3284,11 @@ static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, + jiffies_to_msecs(HZ * bfqq->service_from_backlogged / + bfqd->bfq_wr_max_softrt_rate)); + +- return max(bfqq->last_idle_bklogged + +- HZ * bfqq->service_from_backlogged / +- bfqd->bfq_wr_max_softrt_rate, +- jiffies + nsecs_to_jiffies(bfqq->bfqd->bfq_slice_idle) + 4); ++ return max3(bfqq->soft_rt_next_start, ++ bfqq->last_idle_bklogged + ++ HZ * bfqq->service_from_backlogged / ++ bfqd->bfq_wr_max_softrt_rate, ++ jiffies + nsecs_to_jiffies(bfqq->bfqd->bfq_slice_idle) + 4); + } + + /** +@@ -4395,10 +4437,15 @@ static void bfq_init_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, + bfqq->split_time = bfq_smallest_from_now(); + + /* +- * Set to the value for which bfqq will not be deemed as +- * soft rt when it becomes backlogged. ++ * To not forget the possibly high bandwidth consumed by a ++ * process/queue in the recent past, ++ * bfq_bfqq_softrt_next_start() returns a value at least equal ++ * to the current value of bfqq->soft_rt_next_start (see ++ * comments on bfq_bfqq_softrt_next_start). Set ++ * soft_rt_next_start to now, to mean that bfqq has consumed ++ * no bandwidth so far. + */ +- bfqq->soft_rt_next_start = bfq_greatest_from_now(); ++ bfqq->soft_rt_next_start = jiffies; + + /* first request is almost certainly seeky */ + bfqq->seek_history = 1; +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 4bbd7f4c0154..987dc255c82c 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -3089,37 +3089,78 @@ static bool bfq_bfqq_is_slow(struct bfq_data *bfqd, struct bfq_queue *bfqq, + * whereas soft_rt_next_start is set to infinity for applications that do + * not. + * +- * Unfortunately, even a greedy application may happen to behave in an +- * isochronous way if the CPU load is high. In fact, the application may +- * stop issuing requests while the CPUs are busy serving other processes, +- * then restart, then stop again for a while, and so on. In addition, if +- * the disk achieves a low enough throughput with the request pattern +- * issued by the application (e.g., because the request pattern is random +- * and/or the device is slow), then the application may meet the above +- * bandwidth requirement too. To prevent such a greedy application to be +- * deemed as soft real-time, a further rule is used in the computation of +- * soft_rt_next_start: soft_rt_next_start must be higher than the current +- * time plus the maximum time for which the arrival of a request is waited +- * for when a sync queue becomes idle, namely bfqd->bfq_slice_idle. +- * This filters out greedy applications, as the latter issue instead their +- * next request as soon as possible after the last one has been completed +- * (in contrast, when a batch of requests is completed, a soft real-time +- * application spends some time processing data). ++ * Unfortunately, even a greedy (i.e., I/O-bound) application may ++ * happen to meet, occasionally or systematically, both the above ++ * bandwidth and isochrony requirements. This may happen at least in ++ * the following circumstances. First, if the CPU load is high. The ++ * application may stop issuing requests while the CPUs are busy ++ * serving other processes, then restart, then stop again for a while, ++ * and so on. The other circumstances are related to the storage ++ * device: the storage device is highly loaded or reaches a low-enough ++ * throughput with the I/O of the application (e.g., because the I/O ++ * is random and/or the device is slow). In all these cases, the ++ * I/O of the application may be simply slowed down enough to meet ++ * the bandwidth and isochrony requirements. To reduce the probability ++ * that greedy applications are deemed as soft real-time in these ++ * corner cases, a further rule is used in the computation of ++ * soft_rt_next_start: the return value of this function is forced to ++ * be higher than the maximum between the following two quantities. + * +- * Unfortunately, the last filter may easily generate false positives if +- * only bfqd->bfq_slice_idle is used as a reference time interval and one +- * or both the following cases occur: +- * 1) HZ is so low that the duration of a jiffy is comparable to or higher +- * than bfqd->bfq_slice_idle. This happens, e.g., on slow devices with +- * HZ=100. ++ * (a) Current time plus: (1) the maximum time for which the arrival ++ * of a request is waited for when a sync queue becomes idle, ++ * namely bfqd->bfq_slice_idle, and (2) a few extra jiffies. We ++ * postpone for a moment the reason for adding a few extra ++ * jiffies; we get back to it after next item (b). Lower-bounding ++ * the return value of this function with the current time plus ++ * bfqd->bfq_slice_idle tends to filter out greedy applications, ++ * because the latter issue their next request as soon as possible ++ * after the last one has been completed. In contrast, a soft ++ * real-time application spends some time processing data, after a ++ * batch of its requests has been completed. ++ * ++ * (b) Current value of bfqq->soft_rt_next_start. As pointed out ++ * above, greedy applications may happen to meet both the ++ * bandwidth and isochrony requirements under heavy CPU or ++ * storage-device load. In more detail, in these scenarios, these ++ * applications happen, only for limited time periods, to do I/O ++ * slowly enough to meet all the requirements described so far, ++ * including the filtering in above item (a). These slow-speed ++ * time intervals are usually interspersed between other time ++ * intervals during which these applications do I/O at a very high ++ * speed. Fortunately, exactly because of the high speed of the ++ * I/O in the high-speed intervals, the values returned by this ++ * function happen to be so high, near the end of any such ++ * high-speed interval, to be likely to fall *after* the end of ++ * the low-speed time interval that follows. These high values are ++ * stored in bfqq->soft_rt_next_start after each invocation of ++ * this function. As a consequence, if the last value of ++ * bfqq->soft_rt_next_start is constantly used to lower-bound the ++ * next value that this function may return, then, from the very ++ * beginning of a low-speed interval, bfqq->soft_rt_next_start is ++ * likely to be constantly kept so high that any I/O request ++ * issued during the low-speed interval is considered as arriving ++ * to soon for the application to be deemed as soft ++ * real-time. Then, in the high-speed interval that follows, the ++ * application will not be deemed as soft real-time, just because ++ * it will do I/O at a high speed. And so on. ++ * ++ * Getting back to the filtering in item (a), in the following two ++ * cases this filtering might be easily passed by a greedy ++ * application, if the reference quantity was just ++ * bfqd->bfq_slice_idle: ++ * 1) HZ is so low that the duration of a jiffy is comparable to or ++ * higher than bfqd->bfq_slice_idle. This happens, e.g., on slow ++ * devices with HZ=100. The time granularity may be so coarse ++ * that the approximation, in jiffies, of bfqd->bfq_slice_idle ++ * is rather lower than the exact value. + * 2) jiffies, instead of increasing at a constant rate, may stop increasing + * for a while, then suddenly 'jump' by several units to recover the lost + * increments. This seems to happen, e.g., inside virtual machines. +- * To address this issue, we do not use as a reference time interval just +- * bfqd->bfq_slice_idle, but bfqd->bfq_slice_idle plus a few jiffies. In +- * particular we add the minimum number of jiffies for which the filter +- * seems to be quite precise also in embedded systems and KVM/QEMU virtual +- * machines. ++ * To address this issue, in the filtering in (a) we do not use as a ++ * reference time interval just bfqd->bfq_slice_idle, but ++ * bfqd->bfq_slice_idle plus a few jiffies. In particular, we add the ++ * minimum number of jiffies for which the filter seems to be quite ++ * precise also in embedded systems and KVM/QEMU virtual machines. + */ + static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, + struct bfq_queue *bfqq) +@@ -3131,10 +3172,11 @@ static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, + jiffies_to_msecs(HZ * bfqq->service_from_backlogged / + bfqd->bfq_wr_max_softrt_rate)); + +- return max(bfqq->last_idle_bklogged + +- HZ * bfqq->service_from_backlogged / +- bfqd->bfq_wr_max_softrt_rate, +- jiffies + nsecs_to_jiffies(bfqq->bfqd->bfq_slice_idle) + 4); ++ return max3(bfqq->soft_rt_next_start, ++ bfqq->last_idle_bklogged + ++ HZ * bfqq->service_from_backlogged / ++ bfqd->bfq_wr_max_softrt_rate, ++ jiffies + nsecs_to_jiffies(bfqq->bfqd->bfq_slice_idle) + 4); + } + + /** +@@ -4167,10 +4209,15 @@ static void bfq_init_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq, + bfqq->split_time = bfq_smallest_from_now(); + + /* +- * Set to the value for which bfqq will not be deemed as +- * soft rt when it becomes backlogged. ++ * To not forget the possibly high bandwidth consumed by a ++ * process/queue in the recent past, ++ * bfq_bfqq_softrt_next_start() returns a value at least equal ++ * to the current value of bfqq->soft_rt_next_start (see ++ * comments on bfq_bfqq_softrt_next_start). Set ++ * soft_rt_next_start to now, to mean that bfqq has consumed ++ * no bandwidth so far. + */ +- bfqq->soft_rt_next_start = bfq_greatest_from_now(); ++ bfqq->soft_rt_next_start = jiffies; + + /* first request is almost certainly seeky */ + bfqq->seek_history = 1; + +From 2a09b505660c81dbb80a5d68c9bc558c326d041f Mon Sep 17 00:00:00 2001 +From: Chiara Bruschi +Date: Thu, 7 Dec 2017 09:57:19 +0100 +Subject: [PATCH 03/23] block, bfq-mq: fix occurrences of request + prepare/finish methods' old names + +Commits 'b01f1fa3bb19' (Port of "blk-mq-sched: unify request prepare +methods") and 'cc10d2d7d2c1' (Port of "blk-mq-sched: unify request +finished methods") changed the old names of current bfq_prepare_request +and bfq_finish_request methods, but left them unchanged elsewhere in +the code (related comments, part of function name bfq_put_rq_priv_body). + +This commit fixes every occurrence of the old names of these methods +by changing them into the current names. + +Fixes: b01f1fa3bb19 (Port of "blk-mq-sched: unify request prepare methods") +Fixes: cc10d2d7d2c1 (Port of "blk-mq-sched: unify request finished methods") +Reviewed-by: Paolo Valente +Signed-off-by: Federico Motta +Signed-off-by: Chiara Bruschi +--- + block/bfq-mq-iosched.c | 38 +++++++++++++++++++------------------- + 1 file changed, 19 insertions(+), 19 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 4d06d900f45e..8f8d5eccb016 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4018,20 +4018,20 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + /* + * TESTING: reset DISP_LIST flag, because: 1) + * this rq this request has passed through +- * get_rq_private, 2) then it will have +- * put_rq_private invoked on it, and 3) in +- * put_rq_private we use this flag to check +- * that put_rq_private is not invoked on +- * requests for which get_rq_private has been +- * invoked. ++ * bfq_prepare_request, 2) then it will have ++ * bfq_finish_request invoked on it, and 3) in ++ * bfq_finish_request we use this flag to check ++ * that bfq_finish_request is not invoked on ++ * requests for which bfq_prepare_request has ++ * been invoked. + */ + rq->rq_flags &= ~RQF_DISP_LIST; + goto inc_in_driver_start_rq; + } + + /* +- * We exploit the put_rq_private hook to decrement +- * rq_in_driver, but put_rq_private will not be ++ * We exploit the bfq_finish_request hook to decrement ++ * rq_in_driver, but bfq_finish_request will not be + * invoked on this request. So, to avoid unbalance, + * just start this request, without incrementing + * rq_in_driver. As a negative consequence, +@@ -4040,14 +4040,14 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + * bfq_schedule_dispatch to be invoked uselessly. + * + * As for implementing an exact solution, the +- * put_request hook, if defined, is probably invoked +- * also on this request. So, by exploiting this hook, +- * we could 1) increment rq_in_driver here, and 2) +- * decrement it in put_request. Such a solution would +- * let the value of the counter be always accurate, +- * but it would entail using an extra interface +- * function. This cost seems higher than the benefit, +- * being the frequency of non-elevator-private ++ * bfq_finish_request hook, if defined, is probably ++ * invoked also on this request. So, by exploiting ++ * this hook, we could 1) increment rq_in_driver here, ++ * and 2) decrement it in bfq_finish_request. Such a ++ * solution would let the value of the counter be ++ * always accurate, but it would entail using an extra ++ * interface function. This cost seems higher than the ++ * benefit, being the frequency of non-elevator-private + * requests very low. + */ + goto start_rq; +@@ -4963,7 +4963,7 @@ static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd) + } + } + +-static void bfq_put_rq_priv_body(struct bfq_queue *bfqq) ++static void bfq_finish_request_body(struct bfq_queue *bfqq) + { + bfq_log_bfqq(bfqq->bfqd, bfqq, + "put_request_body: allocated %d", bfqq->allocated); +@@ -5019,7 +5019,7 @@ static void bfq_finish_request(struct request *rq) + spin_lock_irqsave(&bfqd->lock, flags); + + bfq_completed_request(bfqq, bfqd); +- bfq_put_rq_priv_body(bfqq); ++ bfq_finish_request_body(bfqq); + + spin_unlock_irqrestore(&bfqd->lock, flags); + } else { +@@ -5042,7 +5042,7 @@ static void bfq_finish_request(struct request *rq) + bfqg_stats_update_io_remove(bfqq_group(bfqq), + rq->cmd_flags); + } +- bfq_put_rq_priv_body(bfqq); ++ bfq_finish_request_body(bfqq); + } + + rq->elv.priv[0] = NULL; + +From 4df19943c3a767df453abea3d2ac3433c3326ce0 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 16 Nov 2017 18:38:13 +0100 +Subject: [PATCH 04/23] block, bfq-sq, bfq-mq: add missing rq_pos_tree update + on rq removal + +If two processes do I/O close to each other, then BFQ merges the +bfq_queues associated with these processes, to get a more sequential +I/O, and thus a higher throughput. In this respect, to detect whether +two processes are doing I/O close to each other, BFQ keeps a list of +the head-of-line I/O requests of all active bfq_queues. The list is +ordered by initial sectors, and implemented through a red-black tree +(rq_pos_tree). + +Unfortunately, the update of the rq_pos_tree was incomplete, because +the tree was not updated on the removal of the head-of-line I/O +request of a bfq_queue, in case the queue did not remain empty. This +commit adds the missing update. + +Signed-off-by: Paolo Valente +Signed-off-by: Angelo Ruocco +--- + block/bfq-mq-iosched.c | 3 +++ + block/bfq-sq-iosched.c | 3 +++ + 2 files changed, 6 insertions(+) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 8f8d5eccb016..603191c9008f 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -1729,6 +1729,9 @@ static void bfq_remove_request(struct request_queue *q, + rb_erase(&bfqq->pos_node, bfqq->pos_root); + bfqq->pos_root = NULL; + } ++ } else { ++ BUG_ON(!bfqq->next_rq); ++ bfq_pos_tree_add_move(bfqd, bfqq); + } + + if (rq->cmd_flags & REQ_META) { +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 987dc255c82c..ea90ace79e49 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -1669,6 +1669,9 @@ static void bfq_remove_request(struct request *rq) + rb_erase(&bfqq->pos_node, bfqq->pos_root); + bfqq->pos_root = NULL; + } ++ } else { ++ BUG_ON(!bfqq->next_rq); ++ bfq_pos_tree_add_move(bfqd, bfqq); + } + + if (rq->cmd_flags & REQ_META) { + +From b844e345140aaea957d84a21d2aa67588b020cd5 Mon Sep 17 00:00:00 2001 +From: Angelo Ruocco +Date: Mon, 18 Dec 2017 08:28:08 +0100 +Subject: [PATCH 05/23] block, bfq-sq, bfq-mq: check low_latency flag in + bfq_bfqq_save_state() + +A just-created bfq_queue will certainly be deemed as interactive on +the arrival of its first I/O request, if the low_latency flag is +set. Yet, if the queue is merged with another queue on the arrival of +its first I/O request, it will not have the chance to be flagged as +interactive. Nevertheless, if the queue is then split soon enough, it +has to be flagged as interactive after the split. + +To handle this early-merge scenario correctly, BFQ saves the state of +the queue, on the merge, as if the latter had already been deemed +interactive. So, if the queue is split soon, it will get +weight-raised, because the previous state of the queue is resumed on +the split. + +Unfortunately, in the act of saving the state of the newly-created +queue, BFQ doesn't check whether the low_latency flag is set, and this +causes early-merged queues to be then weight-raised, on queue splits, +even if low_latency is off. This commit addresses this problem by +adding the missing check. + +Signed-off-by: Angelo Ruocco +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 3 ++- + block/bfq-sq-iosched.c | 3 ++- + 2 files changed, 4 insertions(+), 2 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 603191c9008f..ff9776c8836a 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -2231,7 +2231,8 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq) + bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq); + bic->was_in_burst_list = !hlist_unhashed(&bfqq->burst_list_node); + if (unlikely(bfq_bfqq_just_created(bfqq) && +- !bfq_bfqq_in_large_burst(bfqq))) { ++ !bfq_bfqq_in_large_burst(bfqq) && ++ bfqq->bfqd->low_latency)) { + /* + * bfqq being merged ritgh after being created: bfqq + * would have deserved interactive weight raising, but +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index ea90ace79e49..3a2d764e760c 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -2109,7 +2109,8 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq) + bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq); + bic->was_in_burst_list = !hlist_unhashed(&bfqq->burst_list_node); + if (unlikely(bfq_bfqq_just_created(bfqq) && +- !bfq_bfqq_in_large_burst(bfqq))) { ++ !bfq_bfqq_in_large_burst(bfqq) && ++ bfqq->bfqd->low_latency)) { + /* + * bfqq being merged ritgh after being created: bfqq + * would have deserved interactive weight raising, but + +From 4cc6896fe1de2e0b4de151a6e70658f10b9ec2fa Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Fri, 27 Oct 2017 11:12:14 +0200 +Subject: [PATCH 06/23] block, bfq-sq, bfq-mq: let a queue be merged only + shortly after starting I/O + +In BFQ and CFQ, two processes are said to be cooperating if they do +I/O in such a way that the union of their I/O requests yields a +sequential I/O pattern. To get such a sequential I/O pattern out of +the non-sequential pattern of each cooperating process, BFQ and CFQ +merge the queues associated with these processes. In more detail, +cooperating processes, and thus their associated queues, usually +start, or restart, to do I/O shortly after each other. This is the +case, e.g., for the I/O threads of KVM/QEMU and of the dump +utility. Basing on this assumption, this commit allows a bfq_queue to +be merged only during a short time interval (100ms) after it starts, +or re-starts, to do I/O. This filtering provides two important +benefits. + +First, it greatly reduces the probability that two non-cooperating +processes have their queues merged by mistake, if they just happen to +do I/O close to each other for a short time interval. These spurious +merges cause loss of service guarantees. A low-weight bfq_queue may +unjustly get more than its expected share of the throughput: if such a +low-weight queue is merged with a high-weight queue, then the I/O for +the low-weight queue is served as if the queue had a high weight. This +may damage other high-weight queues unexpectedly. For instance, +because of this issue, lxterminal occasionally took 7.5 seconds to +start, instead of 6.5 seconds, when some sequential readers and +writers did I/O in the background on a FUJITSU MHX2300BT HDD. The +reason is that the bfq_queues associated with some of the readers or +the writers were merged with the high-weight queues of some processes +that had to do some urgent but little I/O. The readers then exploited +the inherited high weight for all or most of their I/O, during the +start-up of terminal. The filtering introduced by this commit +eliminated any outlier caused by spurious queue merges in our start-up +time tests. + +This filtering also provides a little boost of the throughput +sustainable by BFQ: 3-4%, depending on the CPU. The reason is that, +once a bfq_queue cannot be merged any longer, this commit makes BFQ +stop updating the data needed to handle merging for the queue. + +Signed-off-by: Paolo Valente +Signed-off-by: Angelo Ruocco +--- + block/bfq-mq-iosched.c | 64 +++++++++++++++++++++++++++++++++++++++++--------- + block/bfq-mq.h | 1 + + block/bfq-sched.c | 4 ++++ + block/bfq-sq-iosched.c | 64 +++++++++++++++++++++++++++++++++++++++++--------- + block/bfq.h | 2 ++ + 5 files changed, 113 insertions(+), 22 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index ff9776c8836a..8b17b25a3c30 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -119,6 +119,20 @@ static const int bfq_async_charge_factor = 10; + /* Default timeout values, in jiffies, approximating CFQ defaults. */ + static const int bfq_timeout = (HZ / 8); + ++/* ++ * Time limit for merging (see comments in bfq_setup_cooperator). Set ++ * to the slowest value that, in our tests, proved to be effective in ++ * removing false positives, while not causing true positives to miss ++ * queue merging. ++ * ++ * As can be deduced from the low time limit below, queue merging, if ++ * successful, happens at the very beggining of the I/O of the involved ++ * cooperating processes, as a consequence of the arrival of the very ++ * first requests from each cooperator. After that, there is very ++ * little chance to find cooperators. ++ */ ++static const unsigned long bfq_merge_time_limit = HZ/10; ++ + static struct kmem_cache *bfq_pool; + + /* Below this threshold (in ns), we consider thinktime immediate. */ +@@ -389,6 +403,13 @@ bfq_rq_pos_tree_lookup(struct bfq_data *bfqd, struct rb_root *root, + return bfqq; + } + ++static bool bfq_too_late_for_merging(struct bfq_queue *bfqq) ++{ ++ return bfqq->service_from_backlogged > 0 && ++ time_is_before_jiffies(bfqq->first_IO_time + ++ bfq_merge_time_limit); ++} ++ + static void bfq_pos_tree_add_move(struct bfq_data *bfqd, struct bfq_queue *bfqq) + { + struct rb_node **p, *parent; +@@ -399,6 +420,14 @@ static void bfq_pos_tree_add_move(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfqq->pos_root = NULL; + } + ++ /* ++ * bfqq cannot be merged any longer (see comments in ++ * bfq_setup_cooperator): no point in adding bfqq into the ++ * position tree. ++ */ ++ if (bfq_too_late_for_merging(bfqq)) ++ return; ++ + if (bfq_class_idle(bfqq)) + return; + if (!bfqq->next_rq) +@@ -2081,6 +2110,13 @@ bfq_setup_merge(struct bfq_queue *bfqq, struct bfq_queue *new_bfqq) + static bool bfq_may_be_close_cooperator(struct bfq_queue *bfqq, + struct bfq_queue *new_bfqq) + { ++ if (bfq_too_late_for_merging(new_bfqq)) { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "[%s] too late for bfq%d to be merged", ++ __func__, new_bfqq->pid); ++ return false; ++ } ++ + if (bfq_class_idle(bfqq) || bfq_class_idle(new_bfqq) || + (bfqq->ioprio_class != new_bfqq->ioprio_class)) + return false; +@@ -2149,6 +2185,23 @@ bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, + { + struct bfq_queue *in_service_bfqq, *new_bfqq; + ++ /* ++ * Prevent bfqq from being merged if it has been created too ++ * long ago. The idea is that true cooperating processes, and ++ * thus their associated bfq_queues, are supposed to be ++ * created shortly after each other. This is the case, e.g., ++ * for KVM/QEMU and dump I/O threads. Basing on this ++ * assumption, the following filtering greatly reduces the ++ * probability that two non-cooperating processes, which just ++ * happen to do close I/O for some short time interval, have ++ * their queues merged by mistake. ++ */ ++ if (bfq_too_late_for_merging(bfqq)) { ++ bfq_log_bfqq(bfqd, bfqq, ++ "would have looked for coop, but too late"); ++ return NULL; ++ } ++ + if (bfqq->new_bfqq) + return bfqq->new_bfqq; + +@@ -3338,17 +3391,6 @@ static void bfq_bfqq_expire(struct bfq_data *bfqd, + */ + slow = bfq_bfqq_is_slow(bfqd, bfqq, compensate, reason, &delta); + +- /* +- * Increase service_from_backlogged before next statement, +- * because the possible next invocation of +- * bfq_bfqq_charge_time would likely inflate +- * entity->service. In contrast, service_from_backlogged must +- * contain real service, to enable the soft real-time +- * heuristic to correctly compute the bandwidth consumed by +- * bfqq. +- */ +- bfqq->service_from_backlogged += entity->service; +- + /* + * As above explained, charge slow (typically seeky) and + * timed-out queues with the time and not the service +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 1cb05bb853d2..a5947b203ef2 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -337,6 +337,7 @@ struct bfq_queue { + unsigned long wr_start_at_switch_to_srt; + + unsigned long split_time; /* time of last split */ ++ unsigned long first_IO_time; /* time of first I/O for this queue */ + }; + + /** +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index 616c0692335a..9d261dd428e4 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -939,6 +939,10 @@ static void bfq_bfqq_served(struct bfq_queue *bfqq, int served) + struct bfq_entity *entity = &bfqq->entity; + struct bfq_service_tree *st; + ++ if (!bfqq->service_from_backlogged) ++ bfqq->first_IO_time = jiffies; ++ ++ bfqq->service_from_backlogged += served; + for_each_entity(entity) { + st = bfq_entity_service_tree(entity); + +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 3a2d764e760c..cd00a41ca35d 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -113,6 +113,20 @@ static const int bfq_async_charge_factor = 10; + /* Default timeout values, in jiffies, approximating CFQ defaults. */ + static const int bfq_timeout = (HZ / 8); + ++/* ++ * Time limit for merging (see comments in bfq_setup_cooperator). Set ++ * to the slowest value that, in our tests, proved to be effective in ++ * removing false positives, while not causing true positives to miss ++ * queue merging. ++ * ++ * As can be deduced from the low time limit below, queue merging, if ++ * successful, happens at the very beggining of the I/O of the involved ++ * cooperating processes, as a consequence of the arrival of the very ++ * first requests from each cooperator. After that, there is very ++ * little chance to find cooperators. ++ */ ++static const unsigned long bfq_merge_time_limit = HZ/10; ++ + static struct kmem_cache *bfq_pool; + + /* Below this threshold (in ns), we consider thinktime immediate. */ +@@ -351,6 +365,13 @@ bfq_rq_pos_tree_lookup(struct bfq_data *bfqd, struct rb_root *root, + return bfqq; + } + ++static bool bfq_too_late_for_merging(struct bfq_queue *bfqq) ++{ ++ return bfqq->service_from_backlogged > 0 && ++ time_is_before_jiffies(bfqq->first_IO_time + ++ bfq_merge_time_limit); ++} ++ + static void bfq_pos_tree_add_move(struct bfq_data *bfqd, struct bfq_queue *bfqq) + { + struct rb_node **p, *parent; +@@ -361,6 +382,14 @@ static void bfq_pos_tree_add_move(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfqq->pos_root = NULL; + } + ++ /* ++ * bfqq cannot be merged any longer (see comments in ++ * bfq_setup_cooperator): no point in adding bfqq into the ++ * position tree. ++ */ ++ if (bfq_too_late_for_merging(bfqq)) ++ return; ++ + if (bfq_class_idle(bfqq)) + return; + if (!bfqq->next_rq) +@@ -1960,6 +1989,13 @@ bfq_setup_merge(struct bfq_queue *bfqq, struct bfq_queue *new_bfqq) + static bool bfq_may_be_close_cooperator(struct bfq_queue *bfqq, + struct bfq_queue *new_bfqq) + { ++ if (bfq_too_late_for_merging(new_bfqq)) { ++ bfq_log_bfqq(bfqq->bfqd, bfqq, ++ "[%s] too late for bfq%d to be merged", ++ __func__, new_bfqq->pid); ++ return false; ++ } ++ + if (bfq_class_idle(bfqq) || bfq_class_idle(new_bfqq) || + (bfqq->ioprio_class != new_bfqq->ioprio_class)) + return false; +@@ -2028,6 +2064,23 @@ bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, + { + struct bfq_queue *in_service_bfqq, *new_bfqq; + ++ /* ++ * Prevent bfqq from being merged if it has been created too ++ * long ago. The idea is that true cooperating processes, and ++ * thus their associated bfq_queues, are supposed to be ++ * created shortly after each other. This is the case, e.g., ++ * for KVM/QEMU and dump I/O threads. Basing on this ++ * assumption, the following filtering greatly reduces the ++ * probability that two non-cooperating processes, which just ++ * happen to do close I/O for some short time interval, have ++ * their queues merged by mistake. ++ */ ++ if (bfq_too_late_for_merging(bfqq)) { ++ bfq_log_bfqq(bfqd, bfqq, ++ "would have looked for coop, but too late"); ++ return NULL; ++ } ++ + if (bfqq->new_bfqq) + return bfqq->new_bfqq; + +@@ -3226,17 +3279,6 @@ static void bfq_bfqq_expire(struct bfq_data *bfqd, + */ + slow = bfq_bfqq_is_slow(bfqd, bfqq, compensate, reason, &delta); + +- /* +- * Increase service_from_backlogged before next statement, +- * because the possible next invocation of +- * bfq_bfqq_charge_time would likely inflate +- * entity->service. In contrast, service_from_backlogged must +- * contain real service, to enable the soft real-time +- * heuristic to correctly compute the bandwidth consumed by +- * bfqq. +- */ +- bfqq->service_from_backlogged += entity->service; +- + /* + * As above explained, charge slow (typically seeky) and + * timed-out queues with the time and not the service +diff --git a/block/bfq.h b/block/bfq.h +index 47cd4d5a8c32..59539adc00a5 100644 +--- a/block/bfq.h ++++ b/block/bfq.h +@@ -329,6 +329,8 @@ struct bfq_queue { + unsigned long wr_start_at_switch_to_srt; + + unsigned long split_time; /* time of last split */ ++ ++ unsigned long first_IO_time; /* time of first I/O for this queue */ + }; + + /** + +From 157f39c43ab182280634cd4f6335d0187b3741a0 Mon Sep 17 00:00:00 2001 +From: Angelo Ruocco +Date: Mon, 11 Dec 2017 14:19:54 +0100 +Subject: [PATCH 07/23] block, bfq-sq, bfq-mq: remove superfluous check in + queue-merging setup + +When two or more processes do I/O in a way that the their requests are +sequential in respect to one another, BFQ merges the bfq_queues associated +with the processes. This way the overall I/O pattern becomes sequential, +and thus there is a boost in througput. +These cooperating processes usually start or restart to do I/O shortly +after each other. So, in order to avoid merging non-cooperating processes, +BFQ ensures that none of these queues has been in weight raising for too +long. + +In this respect, from commit "block, bfq-sq, bfq-mq: let a queue be merged +only shortly after being created", BFQ checks whether any queue (and not +only weight-raised ones) is doing I/O continuously from too long to be +merged. + +This new additional check makes the first one useless: a queue doing +I/O from long enough, if being weight-raised, is also a queue in +weight raising for too long to be merged. Accordingly, this commit +removes the first check. + +Signed-off-by: Angelo Ruocco +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 53 ++++---------------------------------------------- + block/bfq-sq-iosched.c | 53 ++++---------------------------------------------- + 2 files changed, 8 insertions(+), 98 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 8b17b25a3c30..f5db8613a70f 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -2140,20 +2140,6 @@ static bool bfq_may_be_close_cooperator(struct bfq_queue *bfqq, + return true; + } + +-/* +- * If this function returns true, then bfqq cannot be merged. The idea +- * is that true cooperation happens very early after processes start +- * to do I/O. Usually, late cooperations are just accidental false +- * positives. In case bfqq is weight-raised, such false positives +- * would evidently degrade latency guarantees for bfqq. +- */ +-static bool wr_from_too_long(struct bfq_queue *bfqq) +-{ +- return bfqq->wr_coeff > 1 && +- time_is_before_jiffies(bfqq->last_wr_start_finish + +- msecs_to_jiffies(100)); +-} +- + /* + * Attempt to schedule a merge of bfqq with the currently in-service + * queue or with a close queue among the scheduled queues. Return +@@ -2167,11 +2153,6 @@ static bool wr_from_too_long(struct bfq_queue *bfqq) + * to maintain. Besides, in such a critical condition as an out of memory, + * the benefits of queue merging may be little relevant, or even negligible. + * +- * Weight-raised queues can be merged only if their weight-raising +- * period has just started. In fact cooperating processes are usually +- * started together. Thus, with this filter we avoid false positives +- * that would jeopardize low-latency guarantees. +- * + * WARNING: queue merging may impair fairness among non-weight raised + * queues, for at least two reasons: 1) the original weight of a + * merged queue may change during the merged state, 2) even being the +@@ -2205,15 +2186,7 @@ bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, + if (bfqq->new_bfqq) + return bfqq->new_bfqq; + +- if (io_struct && wr_from_too_long(bfqq) && +- likely(bfqq != &bfqd->oom_bfqq)) +- bfq_log_bfqq(bfqd, bfqq, +- "would have looked for coop, but bfq%d wr", +- bfqq->pid); +- +- if (!io_struct || +- wr_from_too_long(bfqq) || +- unlikely(bfqq == &bfqd->oom_bfqq)) ++ if (!io_struct || unlikely(bfqq == &bfqd->oom_bfqq)) + return NULL; + + /* If there is only one backlogged queue, don't search. */ +@@ -2223,17 +2196,8 @@ bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, + in_service_bfqq = bfqd->in_service_queue; + + if (in_service_bfqq && in_service_bfqq != bfqq && +- wr_from_too_long(in_service_bfqq) +- && likely(in_service_bfqq == &bfqd->oom_bfqq)) +- bfq_log_bfqq(bfqd, bfqq, +- "would have tried merge with in-service-queue, but wr"); +- +- if (!in_service_bfqq || in_service_bfqq == bfqq +- || wr_from_too_long(in_service_bfqq) || +- unlikely(in_service_bfqq == &bfqd->oom_bfqq)) +- goto check_scheduled; +- +- if (bfq_rq_close_to_sector(io_struct, request, bfqd->last_position) && ++ likely(in_service_bfqq != &bfqd->oom_bfqq) && ++ bfq_rq_close_to_sector(io_struct, request, bfqd->last_position) && + bfqq->entity.parent == in_service_bfqq->entity.parent && + bfq_may_be_close_cooperator(bfqq, in_service_bfqq)) { + new_bfqq = bfq_setup_merge(bfqq, in_service_bfqq); +@@ -2245,21 +2209,12 @@ bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, + * queues. The only thing we need is that the bio/request is not + * NULL, as we need it to establish whether a cooperator exists. + */ +-check_scheduled: + new_bfqq = bfq_find_close_cooperator(bfqd, bfqq, + bfq_io_struct_pos(io_struct, request)); + + BUG_ON(new_bfqq && bfqq->entity.parent != new_bfqq->entity.parent); + +- if (new_bfqq && wr_from_too_long(new_bfqq) && +- likely(new_bfqq != &bfqd->oom_bfqq) && +- bfq_may_be_close_cooperator(bfqq, new_bfqq)) +- bfq_log_bfqq(bfqd, bfqq, +- "would have merged with bfq%d, but wr", +- new_bfqq->pid); +- +- if (new_bfqq && !wr_from_too_long(new_bfqq) && +- likely(new_bfqq != &bfqd->oom_bfqq) && ++ if (new_bfqq && likely(new_bfqq != &bfqd->oom_bfqq) && + bfq_may_be_close_cooperator(bfqq, new_bfqq)) + return bfq_setup_merge(bfqq, new_bfqq); + +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index cd00a41ca35d..d8a358e5e284 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -2019,20 +2019,6 @@ static bool bfq_may_be_close_cooperator(struct bfq_queue *bfqq, + return true; + } + +-/* +- * If this function returns true, then bfqq cannot be merged. The idea +- * is that true cooperation happens very early after processes start +- * to do I/O. Usually, late cooperations are just accidental false +- * positives. In case bfqq is weight-raised, such false positives +- * would evidently degrade latency guarantees for bfqq. +- */ +-static bool wr_from_too_long(struct bfq_queue *bfqq) +-{ +- return bfqq->wr_coeff > 1 && +- time_is_before_jiffies(bfqq->last_wr_start_finish + +- msecs_to_jiffies(100)); +-} +- + /* + * Attempt to schedule a merge of bfqq with the currently in-service + * queue or with a close queue among the scheduled queues. Return +@@ -2046,11 +2032,6 @@ static bool wr_from_too_long(struct bfq_queue *bfqq) + * to maintain. Besides, in such a critical condition as an out of memory, + * the benefits of queue merging may be little relevant, or even negligible. + * +- * Weight-raised queues can be merged only if their weight-raising +- * period has just started. In fact cooperating processes are usually +- * started together. Thus, with this filter we avoid false positives +- * that would jeopardize low-latency guarantees. +- * + * WARNING: queue merging may impair fairness among non-weight raised + * queues, for at least two reasons: 1) the original weight of a + * merged queue may change during the merged state, 2) even being the +@@ -2084,15 +2065,7 @@ bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, + if (bfqq->new_bfqq) + return bfqq->new_bfqq; + +- if (io_struct && wr_from_too_long(bfqq) && +- likely(bfqq != &bfqd->oom_bfqq)) +- bfq_log_bfqq(bfqd, bfqq, +- "would have looked for coop, but bfq%d wr", +- bfqq->pid); +- +- if (!io_struct || +- wr_from_too_long(bfqq) || +- unlikely(bfqq == &bfqd->oom_bfqq)) ++ if (!io_struct || unlikely(bfqq == &bfqd->oom_bfqq)) + return NULL; + + /* If there is only one backlogged queue, don't search. */ +@@ -2102,17 +2075,8 @@ bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, + in_service_bfqq = bfqd->in_service_queue; + + if (in_service_bfqq && in_service_bfqq != bfqq && +- bfqd->in_service_bic && wr_from_too_long(in_service_bfqq) +- && likely(in_service_bfqq == &bfqd->oom_bfqq)) +- bfq_log_bfqq(bfqd, bfqq, +- "would have tried merge with in-service-queue, but wr"); +- +- if (!in_service_bfqq || in_service_bfqq == bfqq || +- !bfqd->in_service_bic || wr_from_too_long(in_service_bfqq) || +- unlikely(in_service_bfqq == &bfqd->oom_bfqq)) +- goto check_scheduled; +- +- if (bfq_rq_close_to_sector(io_struct, request, bfqd->last_position) && ++ likely(in_service_bfqq != &bfqd->oom_bfqq) && ++ bfq_rq_close_to_sector(io_struct, request, bfqd->last_position) && + bfqq->entity.parent == in_service_bfqq->entity.parent && + bfq_may_be_close_cooperator(bfqq, in_service_bfqq)) { + new_bfqq = bfq_setup_merge(bfqq, in_service_bfqq); +@@ -2124,21 +2088,12 @@ bfq_setup_cooperator(struct bfq_data *bfqd, struct bfq_queue *bfqq, + * queues. The only thing we need is that the bio/request is not + * NULL, as we need it to establish whether a cooperator exists. + */ +-check_scheduled: + new_bfqq = bfq_find_close_cooperator(bfqd, bfqq, + bfq_io_struct_pos(io_struct, request)); + + BUG_ON(new_bfqq && bfqq->entity.parent != new_bfqq->entity.parent); + +- if (new_bfqq && wr_from_too_long(new_bfqq) && +- likely(new_bfqq != &bfqd->oom_bfqq) && +- bfq_may_be_close_cooperator(bfqq, new_bfqq)) +- bfq_log_bfqq(bfqd, bfqq, +- "would have merged with bfq%d, but wr", +- new_bfqq->pid); +- +- if (new_bfqq && !wr_from_too_long(new_bfqq) && +- likely(new_bfqq != &bfqd->oom_bfqq) && ++ if (new_bfqq && likely(new_bfqq != &bfqd->oom_bfqq) && + bfq_may_be_close_cooperator(bfqq, new_bfqq)) + return bfq_setup_merge(bfqq, new_bfqq); + + +From b82eb91d87f172aba7eb5eb98e8d5e2a621adf51 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 30 Nov 2017 17:48:28 +0100 +Subject: [PATCH 08/23] block, bfq-sq, bfq-mq: increase threshold to deem I/O + as random + +If two processes do I/O close to each other, i.e., are cooperating +processes in BFQ (and CFQ'S) nomenclature, then BFQ merges their +associated bfq_queues, so as to get sequential I/O from the union of +the I/O requests of the processes, and thus reach a higher +throughput. A merged queue is then split if its I/O stops being +sequential. In this respect, BFQ deems the I/O of a bfq_queue as +(mostly) sequential only if less than 4 I/O requests are random, out +of the last 32 requests inserted into the queue. + +Unfortunately, extensive testing (with the interleaved_io benchmark of +the S suite [1], and with real applications spawning cooperating +processes) has clearly shown that, with such a low threshold, only a +rather low I/O throughput may be reached when several cooperating +processes do I/O. In particular, the outcome of each test run was +bimodal: if queue merging occurred and was stable during the test, +then the throughput was close to the peak rate of the storage device, +otherwise the throughput was arbitrarily low (usually around 1/10 of +the peak rate with a rotational device). The probability to get the +unlucky outcomes grew with the number of cooperating processes: it was +already significant with 5 processes, and close to one with 7 or more +processes. + +The cause of the low throughput in the unlucky runs was that the +merged queues containing the I/O of these cooperating processes were +soon split, because they contained more random I/O requests than those +tolerated by the 4/32 threshold, but +- that I/O would have however allowed the storage device to reach + peak throughput or almost peak throughput; +- in contrast, the I/O of these processes, if served individually + (from separate queues) yielded a rather low throughput. + +So we repeated our tests with increasing values of the threshold, +until we found the minimum value (19) for which we obtained maximum +throughput, reliably, with at least up to 9 cooperating +processes. Then we checked that the use of that higher threshold value +did not cause any regression for any other benchmark in the suite [1]. +This commit raises the threshold to such a higher value. + +[1] https://github.com/Algodev-github/S + +Signed-off-by: Angelo Ruocco +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 2 +- + block/bfq-sq-iosched.c | 2 +- + 2 files changed, 2 insertions(+), 2 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index f5db8613a70f..cb5f49ddecb6 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -145,7 +145,7 @@ static struct kmem_cache *bfq_pool; + #define BFQQ_SEEK_THR (sector_t)(8 * 100) + #define BFQQ_SECT_THR_NONROT (sector_t)(2 * 32) + #define BFQQ_CLOSE_THR (sector_t)(8 * 1024) +-#define BFQQ_SEEKY(bfqq) (hweight32(bfqq->seek_history) > 32/8) ++#define BFQQ_SEEKY(bfqq) (hweight32(bfqq->seek_history) > 19) + + /* Min number of samples required to perform peak-rate update */ + #define BFQ_RATE_MIN_SAMPLES 32 +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index d8a358e5e284..e1c6dc651be1 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -139,7 +139,7 @@ static struct kmem_cache *bfq_pool; + #define BFQQ_SEEK_THR (sector_t)(8 * 100) + #define BFQQ_SECT_THR_NONROT (sector_t)(2 * 32) + #define BFQQ_CLOSE_THR (sector_t)(8 * 1024) +-#define BFQQ_SEEKY(bfqq) (hweight32(bfqq->seek_history) > 32/8) ++#define BFQQ_SEEKY(bfqq) (hweight32(bfqq->seek_history) > 19) + + /* Min number of samples required to perform peak-rate update */ + #define BFQ_RATE_MIN_SAMPLES 32 + +From b739dda4e4b3a1cbbc905f86f9fbb0860b068ce7 Mon Sep 17 00:00:00 2001 +From: Chiara Bruschi +Date: Mon, 11 Dec 2017 18:55:26 +0100 +Subject: [PATCH 09/23] block, bfq-sq, bfq-mq: specify usage condition of + delta_us in bfq_log_bfqq call + +Inside the function bfq_completed_request the value of a variable +called delta_us is computed as current request completion time. +delta_us is used inside a call to the function bfq_log_bfqq as divisor +in a division operation to compute a rate value, but no check makes +sure that delta_us has non-zero value. A divisor with value 0 leads +to a division error that could result in a kernel oops (therefore +unstable/unreliable system state) and consequently cause kernel panic +if resources are unavailable after the system fault. + +This commit fixes this call to bfq_log_bfqq specifying the condition +that allows delta_us to be safely used as divisor. + +Signed-off-by: Paolo Valente +Signed-off-by: Chiara Bruschi +--- + block/bfq-mq-iosched.c | 5 ++++- + block/bfq-sq-iosched.c | 5 ++++- + 2 files changed, 8 insertions(+), 2 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index cb5f49ddecb6..6ce2c0789046 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4904,9 +4904,12 @@ static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd) + bfq_log_bfqq(bfqd, bfqq, + "rq_completed: delta %uus/%luus max_size %u rate %llu/%llu", + delta_us, BFQ_MIN_TT/NSEC_PER_USEC, bfqd->last_rq_max_size, ++ delta_us > 0 ? + (USEC_PER_SEC* + (u64)((bfqd->last_rq_max_size<>BFQ_RATE_SHIFT, ++ >>BFQ_RATE_SHIFT : ++ (USEC_PER_SEC* ++ (u64)(bfqd->last_rq_max_size<>BFQ_RATE_SHIFT, + (USEC_PER_SEC*(u64)(1UL<<(BFQ_RATE_SHIFT-10)))>>BFQ_RATE_SHIFT); + + /* +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index e1c6dc651be1..eff4c4edf5a0 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -4565,9 +4565,12 @@ static void bfq_completed_request(struct request_queue *q, struct request *rq) + + bfq_log(bfqd, "rq_completed: delta %uus/%luus max_size %u rate %llu/%llu", + delta_us, BFQ_MIN_TT/NSEC_PER_USEC, bfqd->last_rq_max_size, ++ delta_us > 0 ? + (USEC_PER_SEC* + (u64)((bfqd->last_rq_max_size<>BFQ_RATE_SHIFT, ++ >>BFQ_RATE_SHIFT : ++ (USEC_PER_SEC* ++ (u64)(bfqd->last_rq_max_size<>BFQ_RATE_SHIFT, + (USEC_PER_SEC*(u64)(1UL<<(BFQ_RATE_SHIFT-10)))>>BFQ_RATE_SHIFT); + + /* + +From ae4310c13eca762644734d53074d8456c85e2dec Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Tue, 19 Dec 2017 12:07:12 +0100 +Subject: [PATCH 10/23] block, bfq-mq: limit tags for writes and async I/O + +Asynchronous I/O can easily starve synchronous I/O (both sync reads +and sync writes), by consuming all request tags. Similarly, storms of +synchronous writes, such as those that sync(2) may trigger, can starve +synchronous reads. In their turn, these two problems may also cause +BFQ to loose control on latency for interactive and soft real-time +applications. For example, on a PLEXTOR PX-256M5S SSD, LibreOffice +Writer takes 0.6 seconds to start if the device is idle, but it takes +more than 45 seconds (!) if there are sequential writes in the +background. + +This commit addresses this issue by limiting the maximum percentage of +tags that asynchronous I/O requests and synchronous write requests can +consume. In particular, this commit grants a higher threshold to +synchronous writes, to prevent the latter from being starved by +asynchronous I/O. + +According to the above test, LibreOffice Writer now starts in about +1.2 seconds on average, regardless of the background workload, and +apart from some rare outlier. To check this improvement, run, e.g., +sudo ./comm_startup_lat.sh bfq-mq 5 5 seq 10 "lowriter --terminate_after_init" +for the comm_startup_lat benchmark in the S suite [1]. + +[1] https://github.com/Algodev-github/S + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 77 ++++++++++++++++++++++++++++++++++++++++++++++++++ + block/bfq-mq.h | 12 ++++++++ + 2 files changed, 89 insertions(+) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 6ce2c0789046..f384f5566672 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -362,6 +362,82 @@ static struct request *bfq_choose_req(struct bfq_data *bfqd, + } + } + ++/* ++ * See the comments on bfq_limit_depth for the purpose of ++ * the depths set in the function. ++ */ ++static void bfq_update_depths(struct bfq_data *bfqd, struct sbitmap_queue *bt) ++{ ++ bfqd->sb_shift = bt->sb.shift; ++ ++ /* ++ * In-word depths if no bfq_queue is being weight-raised: ++ * leaving 25% of tags only for sync reads. ++ * ++ * In next formulas, right-shift the value ++ * (1U<sb_shift), instead of computing directly ++ * (1U<<(bfqd->sb_shift - something)), to be robust against ++ * any possible value of bfqd->sb_shift, without having to ++ * limit 'something'. ++ */ ++ /* no more than 50% of tags for async I/O */ ++ bfqd->word_depths[0][0] = max((1U<sb_shift)>>1, 1U); ++ /* ++ * no more than 75% of tags for sync writes (25% extra tags ++ * w.r.t. async I/O, to prevent async I/O from starving sync ++ * writes) ++ */ ++ bfqd->word_depths[0][1] = max(((1U<sb_shift) * 3)>>2, 1U); ++ ++ /* ++ * In-word depths in case some bfq_queue is being weight- ++ * raised: leaving ~63% of tags for sync reads. This is the ++ * highest percentage for which, in our tests, application ++ * start-up times didn't suffer from any regression due to tag ++ * shortage. ++ */ ++ /* no more than ~18% of tags for async I/O */ ++ bfqd->word_depths[1][0] = max(((1U<sb_shift) * 3)>>4, 1U); ++ /* no more than ~37% of tags for sync writes (~20% extra tags) */ ++ bfqd->word_depths[1][1] = max(((1U<sb_shift) * 6)>>4, 1U); ++} ++ ++/* ++ * Async I/O can easily starve sync I/O (both sync reads and sync ++ * writes), by consuming all tags. Similarly, storms of sync writes, ++ * such as those that sync(2) may trigger, can starve sync reads. ++ * Limit depths of async I/O and sync writes so as to counter both ++ * problems. ++ */ ++static void bfq_limit_depth(unsigned int op, struct blk_mq_alloc_data *data) ++{ ++ struct blk_mq_tags *tags = blk_mq_tags_from_data(data); ++ struct bfq_data *bfqd = data->q->elevator->elevator_data; ++ struct sbitmap_queue *bt; ++ ++ if (op_is_sync(op) && !op_is_write(op)) ++ return; ++ ++ if (data->flags & BLK_MQ_REQ_RESERVED) { ++ if (unlikely(!tags->nr_reserved_tags)) { ++ WARN_ON_ONCE(1); ++ return; ++ } ++ bt = &tags->breserved_tags; ++ } else ++ bt = &tags->bitmap_tags; ++ ++ if (unlikely(bfqd->sb_shift != bt->sb.shift)) ++ bfq_update_depths(bfqd, bt); ++ ++ data->shallow_depth = ++ bfqd->word_depths[!!bfqd->wr_busy_queues][op_is_sync(op)]; ++ ++ bfq_log(bfqd, "[%s] wr_busy %d sync %d depth %u", ++ __func__, bfqd->wr_busy_queues, op_is_sync(op), ++ data->shallow_depth); ++} ++ + static struct bfq_queue * + bfq_rq_pos_tree_lookup(struct bfq_data *bfqd, struct rb_root *root, + sector_t sector, struct rb_node **ret_parent, +@@ -5812,6 +5888,7 @@ static struct elv_fs_entry bfq_attrs[] = { + + static struct elevator_type iosched_bfq_mq = { + .ops.mq = { ++ .limit_depth = bfq_limit_depth, + .prepare_request = bfq_prepare_request, + .finish_request = bfq_finish_request, + .exit_icq = bfq_exit_icq, +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index a5947b203ef2..458099ee0308 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -619,6 +619,18 @@ struct bfq_data { + struct bfq_queue *bio_bfqq; + /* Extra flag used only for TESTING */ + bool bio_bfqq_set; ++ ++ /* ++ * Cached sbitmap shift, used to compute depth limits in ++ * bfq_update_depths. ++ */ ++ unsigned int sb_shift; ++ ++ /* ++ * Depth limits used in bfq_limit_depth (see comments on the ++ * function) ++ */ ++ unsigned int word_depths[2][2]; + }; + + enum bfqq_state_flags { + +From 402e5f6b59662d290ab2b3c10b0016207a63ad21 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 21 Dec 2017 15:51:39 +0100 +Subject: [PATCH 11/23] bfq-sq, bfq-mq: limit sectors served with interactive + weight raising + +To maximise responsiveness, BFQ raises the weight, and performs device +idling, for bfq_queues associated with processes deemed as +interactive. In particular, weight raising has a maximum duration, +equal to the time needed to start a large application. If a +weight-raised process goes on doing I/O beyond this maximum duration, +it loses weight-raising. + +This mechanism is evidently vulnerable to the following false +positives: I/O-bound applications that will go on doing I/O for much +longer than the duration of weight-raising. These applications have +basically no benefit from being weight-raised at the beginning of +their I/O. On the opposite end, while being weight-raised, these +applications +a) unjustly steal throughput to applications that may truly need +low latency; +b) make BFQ uselessly perform device idling; device idling results +in loss of device throughput with most flash-based storage, and may +increase latencies when used purposelessly. + +This commit adds a countermeasure to reduce both the above +problems. To introduce this countermeasure, we provide the following +extra piece of information (full details in the comments added by this +commit). During the start-up of the large application used as a +reference to set the duration of weight-raising, involved processes +transfer at most ~110K sectors each. Accordingly, a process initially +deemed as interactive has no right to be weight-raised any longer, +once transferred 110K sectors or more. + +Basing on this consideration, this commit early-ends weight-raising +for a bfq_queue if the latter happens to have received an amount of +service at least equal to 110K sectors (actually, a little bit more, +to keep a safety margin). I/O-bound applications that reach a high +throughput, such as file copy, get to this threshold much before the +allowed weight-raising period finishes. Thus this early ending of +weight-raising reduces the amount of time during which these +applications cause the problems described above. + +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 84 ++++++++++++++++++++++++++++++++++++++++++++------ + block/bfq-mq.h | 5 +++ + block/bfq-sched.c | 3 ++ + block/bfq-sq-iosched.c | 84 ++++++++++++++++++++++++++++++++++++++++++++------ + block/bfq.h | 5 +++ + 5 files changed, 163 insertions(+), 18 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index f384f5566672..63fdd16dec3c 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -162,15 +162,17 @@ static struct kmem_cache *bfq_pool; + * interactive applications automatically, using the following formula: + * duration = (R / r) * T, where r is the peak rate of the device, and + * R and T are two reference parameters. +- * In particular, R is the peak rate of the reference device (see below), +- * and T is a reference time: given the systems that are likely to be +- * installed on the reference device according to its speed class, T is +- * about the maximum time needed, under BFQ and while reading two files in +- * parallel, to load typical large applications on these systems. +- * In practice, the slower/faster the device at hand is, the more/less it +- * takes to load applications with respect to the reference device. +- * Accordingly, the longer/shorter BFQ grants weight raising to interactive +- * applications. ++ * In particular, R is the peak rate of the reference device (see ++ * below), and T is a reference time: given the systems that are ++ * likely to be installed on the reference device according to its ++ * speed class, T is about the maximum time needed, under BFQ and ++ * while reading two files in parallel, to load typical large ++ * applications on these systems (see the comments on ++ * max_service_from_wr below, for more details on how T is obtained). ++ * In practice, the slower/faster the device at hand is, the more/less ++ * it takes to load applications with respect to the reference device. ++ * Accordingly, the longer/shorter BFQ grants weight raising to ++ * interactive applications. + * + * BFQ uses four different reference pairs (R, T), depending on: + * . whether the device is rotational or non-rotational; +@@ -207,6 +209,60 @@ static int T_slow[2]; + static int T_fast[2]; + static int device_speed_thresh[2]; + ++/* ++ * BFQ uses the above-detailed, time-based weight-raising mechanism to ++ * privilege interactive tasks. This mechanism is vulnerable to the ++ * following false positives: I/O-bound applications that will go on ++ * doing I/O for much longer than the duration of weight ++ * raising. These applications have basically no benefit from being ++ * weight-raised at the beginning of their I/O. On the opposite end, ++ * while being weight-raised, these applications ++ * a) unjustly steal throughput to applications that may actually need ++ * low latency; ++ * b) make BFQ uselessly perform device idling; device idling results ++ * in loss of device throughput with most flash-based storage, and may ++ * increase latencies when used purposelessly. ++ * ++ * BFQ tries to reduce these problems, by adopting the following ++ * countermeasure. To introduce this countermeasure, we need first to ++ * finish explaining how the duration of weight-raising for ++ * interactive tasks is computed. ++ * ++ * For a bfq_queue deemed as interactive, the duration of weight ++ * raising is dynamically adjusted, as a function of the estimated ++ * peak rate of the device, so as to be equal to the time needed to ++ * execute the 'largest' interactive task we benchmarked so far. By ++ * largest task, we mean the task for which each involved process has ++ * to do more I/O than for any of the other tasks we benchmarked. This ++ * reference interactive task is the start-up of LibreOffice Writer, ++ * and in this task each process/bfq_queue needs to have at most ~110K ++ * sectors transferred. ++ * ++ * This last piece of information enables BFQ to reduce the actual ++ * duration of weight-raising for at least one class of I/O-bound ++ * applications: those doing sequential or quasi-sequential I/O. An ++ * example is file copy. In fact, once started, the main I/O-bound ++ * processes of these applications usually consume the above 110K ++ * sectors in much less time than the processes of an application that ++ * is starting, because these I/O-bound processes will greedily devote ++ * almost all their CPU cycles only to their target, ++ * throughput-friendly I/O operations. This is even more true if BFQ ++ * happens to be underestimating the device peak rate, and thus ++ * overestimating the duration of weight raising. But, according to ++ * our measurements, once transferred 110K sectors, these processes ++ * have no right to be weight-raised any longer. ++ * ++ * Basing on the last consideration, BFQ ends weight-raising for a ++ * bfq_queue if the latter happens to have received an amount of ++ * service at least equal to the following constant. The constant is ++ * set to slightly more than 110K, to have a minimum safety margin. ++ * ++ * This early ending of weight-raising reduces the amount of time ++ * during which interactive false positives cause the two problems ++ * described at the beginning of these comments. ++ */ ++static const unsigned long max_service_from_wr = 120000; ++ + #define BFQ_SERVICE_TREE_INIT ((struct bfq_service_tree) \ + { RB_ROOT, RB_ROOT, NULL, NULL, 0, 0 }) + +@@ -1361,6 +1417,7 @@ static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd, + if (old_wr_coeff == 1 && wr_or_deserves_wr) { + /* start a weight-raising period */ + if (interactive) { ++ bfqq->service_from_wr = 0; + bfqq->wr_coeff = bfqd->bfq_wr_coeff; + bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); + } else { +@@ -3980,6 +4037,15 @@ static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq) + "back to interactive wr"); + } + } ++ if (bfqq->wr_coeff > 1 && ++ bfqq->wr_cur_max_time != bfqd->bfq_wr_rt_max_time && ++ bfqq->service_from_wr > max_service_from_wr) { ++ /* see comments on max_service_from_wr */ ++ bfq_bfqq_end_wr(bfqq); ++ bfq_log_bfqq(bfqd, bfqq, ++ "[%s] too much service", ++ __func__); ++ } + } + /* + * To improve latency (for this or other queues), immediately +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 458099ee0308..9a5ce1168ff5 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -331,6 +331,11 @@ struct bfq_queue { + * last transition from idle to backlogged. + */ + unsigned long service_from_backlogged; ++ /* ++ * Cumulative service received from the @bfq_queue since its ++ * last transition to weight-raised state. ++ */ ++ unsigned long service_from_wr; + /* + * Value of wr start time when switching to soft rt + */ +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index 9d261dd428e4..4e6c5232e2fb 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -942,6 +942,9 @@ static void bfq_bfqq_served(struct bfq_queue *bfqq, int served) + if (!bfqq->service_from_backlogged) + bfqq->first_IO_time = jiffies; + ++ if (bfqq->wr_coeff > 1) ++ bfqq->service_from_wr += served; ++ + bfqq->service_from_backlogged += served; + for_each_entity(entity) { + st = bfq_entity_service_tree(entity); +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index eff4c4edf5a0..486493aafaf8 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -156,15 +156,17 @@ static struct kmem_cache *bfq_pool; + * interactive applications automatically, using the following formula: + * duration = (R / r) * T, where r is the peak rate of the device, and + * R and T are two reference parameters. +- * In particular, R is the peak rate of the reference device (see below), +- * and T is a reference time: given the systems that are likely to be +- * installed on the reference device according to its speed class, T is +- * about the maximum time needed, under BFQ and while reading two files in +- * parallel, to load typical large applications on these systems. +- * In practice, the slower/faster the device at hand is, the more/less it +- * takes to load applications with respect to the reference device. +- * Accordingly, the longer/shorter BFQ grants weight raising to interactive +- * applications. ++ * In particular, R is the peak rate of the reference device (see ++ * below), and T is a reference time: given the systems that are ++ * likely to be installed on the reference device according to its ++ * speed class, T is about the maximum time needed, under BFQ and ++ * while reading two files in parallel, to load typical large ++ * applications on these systems (see the comments on ++ * max_service_from_wr below, for more details on how T is obtained). ++ * In practice, the slower/faster the device at hand is, the more/less ++ * it takes to load applications with respect to the reference device. ++ * Accordingly, the longer/shorter BFQ grants weight raising to ++ * interactive applications. + * + * BFQ uses four different reference pairs (R, T), depending on: + * . whether the device is rotational or non-rotational; +@@ -201,6 +203,60 @@ static int T_slow[2]; + static int T_fast[2]; + static int device_speed_thresh[2]; + ++/* ++ * BFQ uses the above-detailed, time-based weight-raising mechanism to ++ * privilege interactive tasks. This mechanism is vulnerable to the ++ * following false positives: I/O-bound applications that will go on ++ * doing I/O for much longer than the duration of weight ++ * raising. These applications have basically no benefit from being ++ * weight-raised at the beginning of their I/O. On the opposite end, ++ * while being weight-raised, these applications ++ * a) unjustly steal throughput to applications that may actually need ++ * low latency; ++ * b) make BFQ uselessly perform device idling; device idling results ++ * in loss of device throughput with most flash-based storage, and may ++ * increase latencies when used purposelessly. ++ * ++ * BFQ tries to reduce these problems, by adopting the following ++ * countermeasure. To introduce this countermeasure, we need first to ++ * finish explaining how the duration of weight-raising for ++ * interactive tasks is computed. ++ * ++ * For a bfq_queue deemed as interactive, the duration of weight ++ * raising is dynamically adjusted, as a function of the estimated ++ * peak rate of the device, so as to be equal to the time needed to ++ * execute the 'largest' interactive task we benchmarked so far. By ++ * largest task, we mean the task for which each involved process has ++ * to do more I/O than for any of the other tasks we benchmarked. This ++ * reference interactive task is the start-up of LibreOffice Writer, ++ * and in this task each process/bfq_queue needs to have at most ~110K ++ * sectors transfered. ++ * ++ * This last piece of information enables BFQ to reduce the actual ++ * duration of weight-raising for at least one class of I/O-bound ++ * applications: those doing sequential or quasi-sequential I/O. An ++ * example is file copy. In fact, once started, the main I/O-bound ++ * processes of these applications usually consume the above 110K ++ * sectors in much less time than the processes of an application that ++ * is starting, because these I/O-bound processes will greedily devote ++ * almost all their CPU cycles only to their target, ++ * throughput-friendly I/O operations. This is even more true if BFQ ++ * happens to be underestimating the device peak rate, and thus ++ * overestimating the duration of weight raising. But, according to ++ * our measurements, once transferred 110K sectors, these processes ++ * have no right to be weight-raised any longer. ++ * ++ * Basing on the last consideration, BFQ ends weight-raising for a ++ * bfq_queue if the latter happens to have received an amount of ++ * service at least equal to the following constant. The constant is ++ * set to slightly more than 110K, to have a minimum safety margin. ++ * ++ * This early ending of weight-raising reduces the amount of time ++ * during which interactive false positives cause the two problems ++ * described at the beginning of these comments. ++ */ ++static const unsigned long max_service_from_wr = 120000; ++ + #define BFQ_SERVICE_TREE_INIT ((struct bfq_service_tree) \ + { RB_ROOT, RB_ROOT, NULL, NULL, 0, 0 }) + +@@ -1246,6 +1302,7 @@ static void bfq_update_bfqq_wr_on_rq_arrival(struct bfq_data *bfqd, + if (old_wr_coeff == 1 && wr_or_deserves_wr) { + /* start a weight-raising period */ + if (interactive) { ++ bfqq->service_from_wr = 0; + bfqq->wr_coeff = bfqd->bfq_wr_coeff; + bfqq->wr_cur_max_time = bfq_wr_duration(bfqd); + } else { +@@ -3794,6 +3851,15 @@ static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq) + "back to interactive wr"); + } + } ++ if (bfqq->wr_coeff > 1 && ++ bfqq->wr_cur_max_time != bfqd->bfq_wr_rt_max_time && ++ bfqq->service_from_wr > max_service_from_wr) { ++ /* see comments on max_service_from_wr */ ++ bfq_bfqq_end_wr(bfqq); ++ bfq_log_bfqq(bfqd, bfqq, ++ "[%s] too much service", ++ __func__); ++ } + } + /* + * To improve latency (for this or other queues), immediately +diff --git a/block/bfq.h b/block/bfq.h +index 59539adc00a5..0cd7a3f251a7 100644 +--- a/block/bfq.h ++++ b/block/bfq.h +@@ -323,6 +323,11 @@ struct bfq_queue { + * last transition from idle to backlogged. + */ + unsigned long service_from_backlogged; ++ /* ++ * Cumulative service received from the @bfq_queue since its ++ * last transition to weight-raised state. ++ */ ++ unsigned long service_from_wr; + /* + * Value of wr start time when switching to soft rt + */ + +From 59efebb94b2f9bac653faf62dadb45b83bd27fa7 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Thu, 4 Jan 2018 16:29:58 +0100 +Subject: [PATCH 12/23] bfq-sq, bfq-mq: put async queues for root bfq groups + too +MIME-Version: 1.0 +Content-Type: text/plain; charset=UTF-8 +Content-Transfer-Encoding: 8bit + +For each pair [device for which bfq is selected as I/O scheduler, +group in blkio/io], bfq maintains a corresponding bfq group. Each such +bfq group contains a set of async queues, with each async queue +created on demand, i.e., when some I/O request arrives for it. On +creation, an async queue gets an extra reference, to make sure that +the queue is not freed as long as its bfq group exists. Accordingly, +to allow the queue to be freed after the group exited, this extra +reference must released on group exit. + +The above holds also for a bfq root group, i.e., for the bfq group +corresponding to the root blkio/io root for a given device. Yet, by +mistake, the references to the existing async queues of a root group +are not released when the latter exits. This causes a memory leak when +the instance of bfq for a given device exits. In a similar vein, +bfqg_stats_xfer_dead is not executed for a root group. + +This commit fixes bfq_pd_offline so that the latter executes the above +missing operations for a root group too. + +Reported-by: Holger Hoffstätte +Reported-by: Guoqing Jiang +Signed-off-by: Davide Ferrari +Signed-off-by: Paolo Valente +--- + block/bfq-cgroup-included.c | 8 +++++--- + 1 file changed, 5 insertions(+), 3 deletions(-) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index 562b0ce581a7..45fefb2e2d57 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -885,13 +885,13 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) + + entity = bfqg->my_entity; + +- if (!entity) /* root group */ +- return; +- + #ifdef BFQ_MQ + spin_lock_irqsave(&bfqd->lock, flags); + #endif + ++ if (!entity) /* root group */ ++ goto put_async_queues; ++ + /* + * Empty all service_trees belonging to this group before + * deactivating the group itself. +@@ -926,6 +926,8 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) + BUG_ON(bfqg->sched_data.in_service_entity); + + __bfq_deactivate_entity(entity, false); ++ ++put_async_queues: + bfq_put_async_queues(bfqd, bfqg); + + #ifdef BFQ_MQ + +From 2dfbaaaf95054e2da3ededc0deb1ba5a4f589e53 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Mon, 8 Jan 2018 19:38:45 +0100 +Subject: [PATCH 13/23] bfq-sq, bfq-mq: release oom-queue ref to root group on + exit + +On scheduler init, a reference to the root group, and a reference to +its corresponding blkg are taken for the oom queue. Yet these +references are not released on scheduler exit, which prevents these +objects from be freed. This commit adds the missing reference +releases. + +Reported-by: Davide Ferrari +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 3 +++ + block/bfq-sq-iosched.c | 3 +++ + 2 files changed, 6 insertions(+) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 63fdd16dec3c..b82c52fabf91 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -5507,6 +5507,9 @@ static void bfq_exit_queue(struct elevator_queue *e) + + BUG_ON(hrtimer_active(&bfqd->idle_slice_timer)); + ++ /* release oom-queue reference to root group */ ++ bfqg_and_blkg_put(bfqd->root_group); ++ + #ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_deactivate_policy(bfqd->queue, &blkcg_policy_bfq); + #else +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 486493aafaf8..851af055664d 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -5052,6 +5052,9 @@ static void bfq_exit_queue(struct elevator_queue *e) + + BUG_ON(hrtimer_active(&bfqd->idle_slice_timer)); + ++ /* release oom-queue reference to root group */ ++ bfqg_put(bfqd->root_group); ++ + #ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_deactivate_policy(q, &blkcg_policy_bfq); + #else + +From 13efe00c8292d78d223e1090a7f36426e360eb38 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Mon, 8 Jan 2018 19:40:38 +0100 +Subject: [PATCH 14/23] block, bfq-sq, bfq-mq: trace get and put of bfq groups + +Signed-off-by: Paolo Valente +--- + block/bfq-cgroup-included.c | 15 +++++++++++++++ + block/bfq-mq-iosched.c | 3 ++- + 2 files changed, 17 insertions(+), 1 deletion(-) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index 45fefb2e2d57..f94743fb2e7d 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -267,6 +267,8 @@ static struct bfq_group *bfqq_group(struct bfq_queue *bfqq) + + static void bfqg_get(struct bfq_group *bfqg) + { ++ trace_printk("bfqg %p\n", bfqg); ++ + #ifdef BFQ_MQ + bfqg->ref++; + #else +@@ -280,6 +282,9 @@ static void bfqg_put(struct bfq_group *bfqg) + bfqg->ref--; + + BUG_ON(bfqg->ref < 0); ++ trace_printk("putting bfqg %p %s\n", bfqg, ++ bfqg->ref == 0 ? "and freeing it" : ""); ++ + if (bfqg->ref == 0) + kfree(bfqg); + #else +@@ -293,6 +298,7 @@ static void bfqg_and_blkg_get(struct bfq_group *bfqg) + /* see comments in bfq_bic_update_cgroup for why refcounting bfqg */ + bfqg_get(bfqg); + ++ trace_printk("getting blkg for bfqg %p\n", bfqg); + blkg_get(bfqg_to_blkg(bfqg)); + } + +@@ -300,6 +306,7 @@ static void bfqg_and_blkg_put(struct bfq_group *bfqg) + { + bfqg_put(bfqg); + ++ trace_printk("putting blkg for bfqg %p\n", bfqg); + blkg_put(bfqg_to_blkg(bfqg)); + } + #endif +@@ -382,6 +389,8 @@ static void bfq_init_entity(struct bfq_entity *entity, + * Make sure that bfqg and its associated blkg do not + * disappear before entity. + */ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "[%s] getting bfqg %p and blkg\n", __func__, bfqg); ++ + bfqg_and_blkg_get(bfqg); + #else + bfqg_get(bfqg); +@@ -475,6 +484,7 @@ static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, int node) + kfree(bfqg); + return NULL; + } ++ trace_printk("bfqg %p\n", bfqg); + + #ifdef BFQ_MQ + /* see comments in bfq_bic_update_cgroup for why refcounting */ +@@ -513,6 +523,7 @@ static void bfq_pd_init(struct blkg_policy_data *pd) + static void bfq_pd_free(struct blkg_policy_data *pd) + { + struct bfq_group *bfqg = pd_to_bfqg(pd); ++ trace_printk("bfqg %p\n", bfqg); + + bfqg_stats_exit(&bfqg->stats); + #ifdef BFQ_MQ +@@ -650,6 +661,8 @@ static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, + bfq_put_idle_entity(bfq_entity_service_tree(entity), entity); + } + #ifdef BFQ_MQ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "[%s] putting blkg and bfqg %p\n", __func__, bfqg); ++ + bfqg_and_blkg_put(bfqq_group(bfqq)); + #else + bfqg_put(bfqq_group(bfqq)); +@@ -658,6 +671,8 @@ static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, + entity->parent = bfqg->my_entity; + entity->sched_data = &bfqg->sched_data; + #ifdef BFQ_MQ ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "[%s] getting blkg and bfqg %p\n", __func__, bfqg); ++ + /* pin down bfqg and its associated blkg */ + bfqg_and_blkg_get(bfqg); + #else +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index b82c52fabf91..d5b7a6b985d7 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4385,10 +4385,11 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + if (bfqq->bfqd) + bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); + +- kmem_cache_free(bfq_pool, bfqq); + #ifdef BFQ_GROUP_IOSCHED_ENABLED ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "[%s] putting blkg and bfqg %p\n", __func__, bfqg); + bfqg_and_blkg_put(bfqg); + #endif ++ kmem_cache_free(bfq_pool, bfqq); + } + + static void bfq_put_cooperator(struct bfq_queue *bfqq) + +From 816b77fba966171974eb5ee25d81bc4e19eaf1b4 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 10 Jan 2018 09:08:22 +0100 +Subject: [PATCH 15/23] bfq-sq, bfq-mq: compile group put for oom queue only if + BFQ_GROUP_IOSCHED is set + +Commit ("bfq-sq, bfq-mq: release oom-queue ref to root group on exit") +added a missing put of the root bfq group for the oom queue. That put +has to be, and can be, performed only if CONFIG_BFQ_GROUP_IOSCHED is +defined: the function doing the put is even not defined at all if +CONFIG_BFQ_GROUP_IOSCHED is not defined. But that commit makes that +put be invoked regardless of whether CONFIG_BFQ_GROUP_IOSCHED is +defined. This commit fixes this mistake, by making that invocation be +compiled only if CONFIG_BFQ_GROUP_IOSCHED is actually defined. + +Fixes ("block, bfq: release oom-queue ref to root group on exit") +Reported-by: Jan Alexander Steffens +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 2 +- + block/bfq-sq-iosched.c | 2 +- + 2 files changed, 2 insertions(+), 2 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index d5b7a6b985d7..2581fe0f6f2f 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -5508,10 +5508,10 @@ static void bfq_exit_queue(struct elevator_queue *e) + + BUG_ON(hrtimer_active(&bfqd->idle_slice_timer)); + ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + /* release oom-queue reference to root group */ + bfqg_and_blkg_put(bfqd->root_group); + +-#ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_deactivate_policy(bfqd->queue, &blkcg_policy_bfq); + #else + spin_lock_irq(&bfqd->lock); +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index 851af055664d..c4df156b1fb4 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -5052,10 +5052,10 @@ static void bfq_exit_queue(struct elevator_queue *e) + + BUG_ON(hrtimer_active(&bfqd->idle_slice_timer)); + ++#ifdef BFQ_GROUP_IOSCHED_ENABLED + /* release oom-queue reference to root group */ + bfqg_put(bfqd->root_group); + +-#ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_deactivate_policy(q, &blkcg_policy_bfq); + #else + bfq_put_async_queues(bfqd, bfqd->root_group); + +From 643a89c659172b2c9ae16adfe03af4e3e88e1326 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Sat, 13 Jan 2018 18:48:41 +0100 +Subject: [PATCH 16/23] block, bfq-sq, bfq-mq: remove trace_printks + +Commit ("block, bfq-sq, bfq-mq: trace get and put of bfq groups") +unwisely added some invocations of the function trace_printk, which +is inappropriate in production kernels. This commit removes those +invocations. + +Signed-off-by: Paolo Valente +--- + block/bfq-cgroup-included.c | 10 ---------- + 1 file changed, 10 deletions(-) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index f94743fb2e7d..a4f8a03edfc9 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -267,8 +267,6 @@ static struct bfq_group *bfqq_group(struct bfq_queue *bfqq) + + static void bfqg_get(struct bfq_group *bfqg) + { +- trace_printk("bfqg %p\n", bfqg); +- + #ifdef BFQ_MQ + bfqg->ref++; + #else +@@ -282,9 +280,6 @@ static void bfqg_put(struct bfq_group *bfqg) + bfqg->ref--; + + BUG_ON(bfqg->ref < 0); +- trace_printk("putting bfqg %p %s\n", bfqg, +- bfqg->ref == 0 ? "and freeing it" : ""); +- + if (bfqg->ref == 0) + kfree(bfqg); + #else +@@ -298,7 +293,6 @@ static void bfqg_and_blkg_get(struct bfq_group *bfqg) + /* see comments in bfq_bic_update_cgroup for why refcounting bfqg */ + bfqg_get(bfqg); + +- trace_printk("getting blkg for bfqg %p\n", bfqg); + blkg_get(bfqg_to_blkg(bfqg)); + } + +@@ -306,7 +300,6 @@ static void bfqg_and_blkg_put(struct bfq_group *bfqg) + { + bfqg_put(bfqg); + +- trace_printk("putting blkg for bfqg %p\n", bfqg); + blkg_put(bfqg_to_blkg(bfqg)); + } + #endif +@@ -484,8 +477,6 @@ static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, int node) + kfree(bfqg); + return NULL; + } +- trace_printk("bfqg %p\n", bfqg); +- + #ifdef BFQ_MQ + /* see comments in bfq_bic_update_cgroup for why refcounting */ + bfqg_get(bfqg); +@@ -523,7 +514,6 @@ static void bfq_pd_init(struct blkg_policy_data *pd) + static void bfq_pd_free(struct blkg_policy_data *pd) + { + struct bfq_group *bfqg = pd_to_bfqg(pd); +- trace_printk("bfqg %p\n", bfqg); + + bfqg_stats_exit(&bfqg->stats); + #ifdef BFQ_MQ + +From ce050275e24fecec800f346c09d9494563e9fc8a Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Mon, 15 Jan 2018 15:07:05 +0100 +Subject: [PATCH 17/23] block, bfq-mq: add requeue-request hook + +Commit 'a6a252e64914 ("blk-mq-sched: decide how to handle flush rq via +RQF_FLUSH_SEQ")' makes all non-flush re-prepared requests for a device +be re-inserted into the active I/O scheduler for that device. As a +consequence, I/O schedulers may get the same request inserted again, +even several times, without a finish_request invoked on that request +before each re-insertion. + +This fact is the cause of the failure reported in [1]. For an I/O +scheduler, every re-insertion of the same re-prepared request is +equivalent to the insertion of a new request. For schedulers like +mq-deadline or kyber, this fact causes no harm. In contrast, it +confuses a stateful scheduler like BFQ, which keeps state for an I/O +request, until the finish_request hook is invoked on the request. In +particular, BFQ may get stuck, waiting forever for the number of +request dispatches, of the same request, to be balanced by an equal +number of request completions (while there will be one completion for +that request). In this state, BFQ may refuse to serve I/O requests +from other bfq_queues. The hang reported in [1] then follows. + +However, the above re-prepared requests undergo a requeue, thus the +requeue_request hook of the active elevator is invoked for these +requests, if set. This commit then addresses the above issue by +properly implementing the hook requeue_request in BFQ. + +[1] https://marc.info/?l=linux-block&m=151211117608676 + +Reported-by: Ivan Kozik +Reported-by: Alban Browaeys +Signed-off-by: Paolo Valente +Signed-off-by: Serena Ziviani +--- + block/bfq-mq-iosched.c | 90 ++++++++++++++++++++++++++++++++++++++++---------- + 1 file changed, 73 insertions(+), 17 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 2581fe0f6f2f..bb7ccc2f1165 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4162,9 +4162,9 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + * TESTING: reset DISP_LIST flag, because: 1) + * this rq this request has passed through + * bfq_prepare_request, 2) then it will have +- * bfq_finish_request invoked on it, and 3) in +- * bfq_finish_request we use this flag to check +- * that bfq_finish_request is not invoked on ++ * bfq_finish_requeue_request invoked on it, and 3) in ++ * bfq_finish_requeue_request we use this flag to check ++ * that bfq_finish_requeue_request is not invoked on + * requests for which bfq_prepare_request has + * been invoked. + */ +@@ -4173,8 +4173,8 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + } + + /* +- * We exploit the bfq_finish_request hook to decrement +- * rq_in_driver, but bfq_finish_request will not be ++ * We exploit the bfq_finish_requeue_request hook to decrement ++ * rq_in_driver, but bfq_finish_requeue_request will not be + * invoked on this request. So, to avoid unbalance, + * just start this request, without incrementing + * rq_in_driver. As a negative consequence, +@@ -4183,10 +4183,10 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + * bfq_schedule_dispatch to be invoked uselessly. + * + * As for implementing an exact solution, the +- * bfq_finish_request hook, if defined, is probably ++ * bfq_finish_requeue_request hook, if defined, is probably + * invoked also on this request. So, by exploiting + * this hook, we could 1) increment rq_in_driver here, +- * and 2) decrement it in bfq_finish_request. Such a ++ * and 2) decrement it in bfq_finish_requeue_request. Such a + * solution would let the value of the counter be + * always accurate, but it would entail using an extra + * interface function. This cost seems higher than the +@@ -4878,6 +4878,8 @@ static bool __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + return idle_timer_disabled; + } + ++static void bfq_prepare_request(struct request *rq, struct bio *bio); ++ + static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + bool at_head) + { +@@ -4919,6 +4921,20 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + BUG_ON(!(rq->rq_flags & RQF_GOT)); + rq->rq_flags &= ~RQF_GOT; + ++ if (!bfqq) { ++ /* ++ * This should never happen. Most likely rq is ++ * a requeued regular request, being ++ * re-inserted without being first ++ * re-prepared. Do a prepare, to avoid ++ * failure. ++ */ ++ pr_warn("Regular request associated with no queue"); ++ WARN_ON(1); ++ bfq_prepare_request(rq, rq->bio); ++ bfqq = RQ_BFQQ(rq); ++ } ++ + #if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + idle_timer_disabled = __bfq_insert_request(bfqd, rq); + /* +@@ -5110,7 +5126,7 @@ static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd) + } + } + +-static void bfq_finish_request_body(struct bfq_queue *bfqq) ++static void bfq_finish_requeue_request_body(struct bfq_queue *bfqq) + { + bfq_log_bfqq(bfqq->bfqd, bfqq, + "put_request_body: allocated %d", bfqq->allocated); +@@ -5120,7 +5136,13 @@ static void bfq_finish_request_body(struct bfq_queue *bfqq) + bfq_put_queue(bfqq); + } + +-static void bfq_finish_request(struct request *rq) ++/* ++ * Handle either a requeue or a finish for rq. The things to do are ++ * the same in both cases: all references to rq are to be dropped. In ++ * particular, rq is considered completed from the point of view of ++ * the scheduler. ++ */ ++static void bfq_finish_requeue_request(struct request *rq) + { + struct bfq_queue *bfqq; + struct bfq_data *bfqd; +@@ -5128,11 +5150,27 @@ static void bfq_finish_request(struct request *rq) + + BUG_ON(!rq); + +- if (!rq->elv.icq) ++ bfqq = RQ_BFQQ(rq); ++ ++ /* ++ * Requeue and finish hooks are invoked in blk-mq without ++ * checking whether the involved request is actually still ++ * referenced in the scheduler. To handle this fact, the ++ * following two checks make this function exit in case of ++ * spurious invocations, for which there is nothing to do. ++ * ++ * First, check whether rq has nothing to do with an elevator. ++ */ ++ if (unlikely(!(rq->rq_flags & RQF_ELVPRIV))) + return; + +- bfqq = RQ_BFQQ(rq); +- BUG_ON(!bfqq); ++ /* ++ * rq either is not associated with any icq, or is an already ++ * requeued request that has not (yet) been re-inserted into ++ * a bfq_queue. ++ */ ++ if (!rq->elv.icq || !bfqq) ++ return; + + bic = RQ_BIC(rq); + BUG_ON(!bic); +@@ -5145,7 +5183,6 @@ static void bfq_finish_request(struct request *rq) + BUG(); + } + BUG_ON(rq->rq_flags & RQF_QUEUED); +- BUG_ON(!(rq->rq_flags & RQF_ELVPRIV)); + + bfq_log_bfqq(bfqd, bfqq, + "putting rq %p with %u sects left, STARTED %d", +@@ -5166,13 +5203,14 @@ static void bfq_finish_request(struct request *rq) + spin_lock_irqsave(&bfqd->lock, flags); + + bfq_completed_request(bfqq, bfqd); +- bfq_finish_request_body(bfqq); ++ bfq_finish_requeue_request_body(bfqq); + + spin_unlock_irqrestore(&bfqd->lock, flags); + } else { + /* + * Request rq may be still/already in the scheduler, +- * in which case we need to remove it. And we cannot ++ * in which case we need to remove it (this should ++ * never happen in case of requeue). And we cannot + * defer such a check and removal, to avoid + * inconsistencies in the time interval from the end + * of this function to the start of the deferred work. +@@ -5189,9 +5227,26 @@ static void bfq_finish_request(struct request *rq) + bfqg_stats_update_io_remove(bfqq_group(bfqq), + rq->cmd_flags); + } +- bfq_finish_request_body(bfqq); ++ bfq_finish_requeue_request_body(bfqq); + } + ++ /* ++ * Reset private fields. In case of a requeue, this allows ++ * this function to correctly do nothing if it is spuriously ++ * invoked again on this same request (see the check at the ++ * beginning of the function). Probably, a better general ++ * design would be to prevent blk-mq from invoking the requeue ++ * or finish hooks of an elevator, for a request that is not ++ * referred by that elevator. ++ * ++ * Resetting the following fields would break the ++ * request-insertion logic if rq is re-inserted into a bfq ++ * internal queue, without a re-preparation. Here we assume ++ * that re-insertions of requeued requests, without ++ * re-preparation, can happen only for pass_through or at_head ++ * requests (which are not re-inserted into bfq internal ++ * queues). ++ */ + rq->elv.priv[0] = NULL; + rq->elv.priv[1] = NULL; + } +@@ -5960,7 +6015,8 @@ static struct elevator_type iosched_bfq_mq = { + .ops.mq = { + .limit_depth = bfq_limit_depth, + .prepare_request = bfq_prepare_request, +- .finish_request = bfq_finish_request, ++ .requeue_request = bfq_finish_requeue_request, ++ .finish_request = bfq_finish_requeue_request, + .exit_icq = bfq_exit_icq, + .insert_requests = bfq_insert_requests, + .dispatch_request = bfq_dispatch_request, + +From 3e4f292191cc62b3844316b9741534c3f1b36f0a Mon Sep 17 00:00:00 2001 +From: Davide Paganelli +Date: Thu, 8 Feb 2018 12:19:24 +0100 +Subject: [PATCH 18/23] block, bfq-mq, bfq-sq: make log functions print names + of calling functions + +Add the macro __func__ as a parameter to the invocations of the functions +pr_crit, blk_add_trace_msg and blk_add_cgroup_trace_msg in bfq_log* +functions, in order to include automatically in the log messages +the names of the functions that call the log functions. +The programmer can then avoid doing it. + +Signed-off-by: Davide Paganelli +Signed-off-by: Paolo Valente +--- + block/bfq-cgroup-included.c | 9 +-- + block/bfq-mq-iosched.c | 167 ++++++++++++++++++++++---------------------- + block/bfq-mq.h | 33 ++++----- + block/bfq-sched.c | 54 +++++++------- + block/bfq-sq-iosched.c | 134 +++++++++++++++++------------------ + block/bfq.h | 33 ++++----- + 6 files changed, 214 insertions(+), 216 deletions(-) + +diff --git a/block/bfq-cgroup-included.c b/block/bfq-cgroup-included.c +index a4f8a03edfc9..613f154e9da5 100644 +--- a/block/bfq-cgroup-included.c ++++ b/block/bfq-cgroup-included.c +@@ -382,7 +382,8 @@ static void bfq_init_entity(struct bfq_entity *entity, + * Make sure that bfqg and its associated blkg do not + * disappear before entity. + */ +- bfq_log_bfqq(bfqq->bfqd, bfqq, "[%s] getting bfqg %p and blkg\n", __func__, bfqg); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "getting bfqg %p and blkg\n", ++ bfqg); + + bfqg_and_blkg_get(bfqg); + #else +@@ -651,7 +652,7 @@ static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, + bfq_put_idle_entity(bfq_entity_service_tree(entity), entity); + } + #ifdef BFQ_MQ +- bfq_log_bfqq(bfqq->bfqd, bfqq, "[%s] putting blkg and bfqg %p\n", __func__, bfqg); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "putting blkg and bfqg %p\n", bfqg); + + bfqg_and_blkg_put(bfqq_group(bfqq)); + #else +@@ -661,7 +662,7 @@ static void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, + entity->parent = bfqg->my_entity; + entity->sched_data = &bfqg->sched_data; + #ifdef BFQ_MQ +- bfq_log_bfqq(bfqq->bfqd, bfqq, "[%s] getting blkg and bfqg %p\n", __func__, bfqg); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "getting blkg and bfqg %p\n", bfqg); + + /* pin down bfqg and its associated blkg */ + bfqg_and_blkg_get(bfqg); +@@ -721,7 +722,7 @@ static struct bfq_group *__bfq_bic_change_cgroup(struct bfq_data *bfqd, + if (entity->sched_data != &bfqg->sched_data) { + bic_set_bfqq(bic, NULL, 0); + bfq_log_bfqq(bfqd, async_bfqq, +- "bic_change_group: %p %d", ++ "%p %d", + async_bfqq, + async_bfqq->ref); + bfq_put_queue(async_bfqq); +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index bb7ccc2f1165..edc93b6af186 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -310,7 +310,7 @@ static struct bfq_io_cq *bfq_bic_lookup(struct bfq_data *bfqd, + static void bfq_schedule_dispatch(struct bfq_data *bfqd) + { + if (bfqd->queued != 0) { +- bfq_log(bfqd, "schedule dispatch"); ++ bfq_log(bfqd, ""); + blk_mq_run_hw_queues(bfqd->queue, true); + } + } +@@ -489,8 +489,8 @@ static void bfq_limit_depth(unsigned int op, struct blk_mq_alloc_data *data) + data->shallow_depth = + bfqd->word_depths[!!bfqd->wr_busy_queues][op_is_sync(op)]; + +- bfq_log(bfqd, "[%s] wr_busy %d sync %d depth %u", +- __func__, bfqd->wr_busy_queues, op_is_sync(op), ++ bfq_log(bfqd, "wr_busy %d sync %d depth %u", ++ bfqd->wr_busy_queues, op_is_sync(op), + data->shallow_depth); + } + +@@ -528,7 +528,7 @@ bfq_rq_pos_tree_lookup(struct bfq_data *bfqd, struct rb_root *root, + if (rb_link) + *rb_link = p; + +- bfq_log(bfqd, "rq_pos_tree_lookup %llu: returning %d", ++ bfq_log(bfqd, "%llu: returning %d", + (unsigned long long) sector, + bfqq ? bfqq->pid : 0); + +@@ -749,7 +749,7 @@ static struct request *bfq_check_fifo(struct bfq_queue *bfqq, + if (rq == last || ktime_get_ns() < rq->fifo_time) + return NULL; + +- bfq_log_bfqq(bfqq->bfqd, bfqq, "check_fifo: returned %p", rq); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "returned %p", rq); + BUG_ON(RB_EMPTY_NODE(&rq->rb_node)); + return rq; + } +@@ -842,7 +842,7 @@ static void bfq_updated_next_req(struct bfq_data *bfqd, + bfq_serv_to_charge(next_rq, bfqq)); + if (entity->budget != new_budget) { + entity->budget = new_budget; +- bfq_log_bfqq(bfqd, bfqq, "updated next rq: new budget %lu", ++ bfq_log_bfqq(bfqd, bfqq, "new budget %lu", + new_budget); + bfq_requeue_bfqq(bfqd, bfqq, false); + } +@@ -915,8 +915,7 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); + + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "[%s] bic %p wr_coeff %d start_finish %lu max_time %lu", +- __func__, ++ "bic %p wr_coeff %d start_finish %lu max_time %lu", + bic, bfqq->wr_coeff, bfqq->last_wr_start_finish, + bfqq->wr_cur_max_time); + +@@ -929,11 +928,11 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + bfq_wr_duration(bfqd))) { + switch_back_to_interactive_wr(bfqq, bfqd); + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "resume state: switching back to interactive"); ++ "switching back to interactive"); + } else { + bfqq->wr_coeff = 1; + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "resume state: switching off wr (%lu + %lu < %lu)", ++ "switching off wr (%lu + %lu < %lu)", + bfqq->last_wr_start_finish, bfqq->wr_cur_max_time, + jiffies); + } +@@ -985,7 +984,7 @@ static void bfq_add_to_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) + /* Increment burst size to take into account also bfqq */ + bfqd->burst_size++; + +- bfq_log_bfqq(bfqd, bfqq, "add_to_burst %d", bfqd->burst_size); ++ bfq_log_bfqq(bfqd, bfqq, "%d", bfqd->burst_size); + + BUG_ON(bfqd->burst_size > bfqd->bfq_large_burst_thresh); + +@@ -998,7 +997,7 @@ static void bfq_add_to_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) + * other to consider this burst as large. + */ + bfqd->large_burst = true; +- bfq_log_bfqq(bfqd, bfqq, "add_to_burst: large burst started"); ++ bfq_log_bfqq(bfqd, bfqq, "large burst started"); + + /* + * We can now mark all queues in the burst list as +@@ -1170,7 +1169,7 @@ static void bfq_handle_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfqd->large_burst = false; + bfq_reset_burst_list(bfqd, bfqq); + bfq_log_bfqq(bfqd, bfqq, +- "handle_burst: late activation or different group"); ++ "late activation or different group"); + goto end; + } + +@@ -1180,7 +1179,7 @@ static void bfq_handle_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) + * bfqq as belonging to this large burst immediately. + */ + if (bfqd->large_burst) { +- bfq_log_bfqq(bfqd, bfqq, "handle_burst: marked in burst"); ++ bfq_log_bfqq(bfqd, bfqq, "marked in burst"); + bfq_mark_bfqq_in_large_burst(bfqq); + goto end; + } +@@ -1686,7 +1685,7 @@ static void bfq_add_request(struct request *rq) + unsigned int old_wr_coeff = bfqq->wr_coeff; + bool interactive = false; + +- bfq_log_bfqq(bfqd, bfqq, "add_request: size %u %s", ++ bfq_log_bfqq(bfqd, bfqq, "size %u %s", + blk_rq_sectors(rq), rq_is_sync(rq) ? "S" : "A"); + + if (bfqq->wr_coeff > 1) /* queue is being weight-raised */ +@@ -1952,7 +1951,7 @@ static int bfq_request_merge(struct request_queue *q, struct request **req, + __rq = bfq_find_rq_fmerge(bfqd, bio, q); + if (__rq && elv_bio_merge_ok(__rq, bio)) { + *req = __rq; +- bfq_log(bfqd, "request_merge: req %p", __rq); ++ bfq_log(bfqd, "req %p", __rq); + + return ELEVATOR_FRONT_MERGE; + } +@@ -1989,7 +1988,7 @@ static void bfq_request_merged(struct request_queue *q, struct request *req, + bfqq->next_rq = next_rq; + + bfq_log_bfqq(bfqd, bfqq, +- "request_merged: req %p prev %p next_rq %p bfqq %p", ++ "req %p prev %p next_rq %p bfqq %p", + req, prev, next_rq, bfqq); + + /* +@@ -2018,7 +2017,7 @@ static void bfq_requests_merged(struct request_queue *q, struct request *rq, + goto end; + + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "requests_merged: rq %p next %p bfqq %p next_bfqq %p", ++ "rq %p next %p bfqq %p next_bfqq %p", + rq, next, bfqq, next_bfqq); + + spin_lock_irq(&bfqq->bfqd->lock); +@@ -2069,10 +2068,10 @@ static void bfq_bfqq_end_wr(struct bfq_queue *bfqq) + */ + bfqq->entity.prio_changed = 1; + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "end_wr: wrais ending at %lu, rais_max_time %u", ++ "wrais ending at %lu, rais_max_time %u", + bfqq->last_wr_start_finish, + jiffies_to_msecs(bfqq->wr_cur_max_time)); +- bfq_log_bfqq(bfqq->bfqd, bfqq, "end_wr: wr_busy %d", ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "wr_busy %d", + bfqq->bfqd->wr_busy_queues); + } + +@@ -2245,8 +2244,8 @@ static bool bfq_may_be_close_cooperator(struct bfq_queue *bfqq, + { + if (bfq_too_late_for_merging(new_bfqq)) { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "[%s] too late for bfq%d to be merged", +- __func__, new_bfqq->pid); ++ "too late for bfq%d to be merged", ++ new_bfqq->pid); + return false; + } + +@@ -2395,8 +2394,7 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq) + } + BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "[%s] bic %p wr_coeff %d start_finish %lu max_time %lu", +- __func__, ++ "bic %p wr_coeff %d start_finish %lu max_time %lu", + bic, bfqq->wr_coeff, bfqq->last_wr_start_finish, + bfqq->wr_cur_max_time); + } +@@ -2453,7 +2451,7 @@ bfq_merge_bfqqs(struct bfq_data *bfqd, struct bfq_io_cq *bic, + + } + +- bfq_log_bfqq(bfqd, new_bfqq, "merge_bfqqs: wr_busy %d", ++ bfq_log_bfqq(bfqd, new_bfqq, "wr_busy %d", + bfqd->wr_busy_queues); + + /* +@@ -2554,7 +2552,7 @@ static void bfq_set_budget_timeout(struct bfq_data *bfqd, + bfqq->budget_timeout = jiffies + + bfqd->bfq_timeout * timeout_coeff; + +- bfq_log_bfqq(bfqd, bfqq, "set budget_timeout %u", ++ bfq_log_bfqq(bfqd, bfqq, "%u", + jiffies_to_msecs(bfqd->bfq_timeout * timeout_coeff)); + } + +@@ -2620,10 +2618,10 @@ static void __bfq_set_in_service_queue(struct bfq_data *bfqd, + + bfq_set_budget_timeout(bfqd, bfqq); + bfq_log_bfqq(bfqd, bfqq, +- "set_in_service_queue, cur-budget = %d", ++ "cur-budget = %d", + bfqq->entity.budget); + } else +- bfq_log(bfqd, "set_in_service_queue: NULL"); ++ bfq_log(bfqd, "NULL"); + + bfqd->in_service_queue = bfqq; + } +@@ -2746,7 +2744,7 @@ static void bfq_reset_rate_computation(struct bfq_data *bfqd, struct request *rq + bfqd->peak_rate_samples = 0; /* full re-init on next disp. */ + + bfq_log(bfqd, +- "reset_rate_computation at end, sample %u/%u tot_sects %llu", ++ "at end, sample %u/%u tot_sects %llu", + bfqd->peak_rate_samples, bfqd->sequential_samples, + bfqd->tot_sectors_dispatched); + } +@@ -2766,7 +2764,7 @@ static void bfq_update_rate_reset(struct bfq_data *bfqd, struct request *rq) + if (bfqd->peak_rate_samples < BFQ_RATE_MIN_SAMPLES || + bfqd->delta_from_first < BFQ_RATE_MIN_INTERVAL) { + bfq_log(bfqd, +- "update_rate_reset: only resetting, delta_first %lluus samples %d", ++ "only resetting, delta_first %lluus samples %d", + bfqd->delta_from_first>>10, bfqd->peak_rate_samples); + goto reset_computation; + } +@@ -2790,7 +2788,7 @@ static void bfq_update_rate_reset(struct bfq_data *bfqd, struct request *rq) + div_u64(bfqd->delta_from_first, NSEC_PER_USEC)); + + bfq_log(bfqd, +-"update_rate_reset: tot_sects %llu delta_first %lluus rate %llu sects/s (%d)", ++"tot_sects %llu delta_first %lluus rate %llu sects/s (%d)", + bfqd->tot_sectors_dispatched, bfqd->delta_from_first>>10, + ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), + rate > 20<peak_rate) || + rate > 20<peak_rate_samples, bfqd->sequential_samples, + ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), + ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); + goto reset_computation; + } else { + bfq_log(bfqd, +- "update_rate_reset: do update, samples %u/%u rate/peak %llu/%llu", ++ "do update, samples %u/%u rate/peak %llu/%llu", + bfqd->peak_rate_samples, bfqd->sequential_samples, + ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), + ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); +@@ -2868,7 +2866,7 @@ static void bfq_update_rate_reset(struct bfq_data *bfqd, struct request *rq) + rate /= divisor; /* smoothing constant alpha = 1/divisor */ + + bfq_log(bfqd, +- "update_rate_reset: divisor %d tmp_peak_rate %llu tmp_rate %u", ++ "divisor %d tmp_peak_rate %llu tmp_rate %u", + divisor, + ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT), + (u32)((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT)); +@@ -2922,7 +2920,7 @@ static void bfq_update_peak_rate(struct bfq_data *bfqd, struct request *rq) + + if (bfqd->peak_rate_samples == 0) { /* first dispatch */ + bfq_log(bfqd, +- "update_peak_rate: goto reset, samples %d", ++ "goto reset, samples %d", + bfqd->peak_rate_samples) ; + bfq_reset_rate_computation(bfqd, rq); + goto update_last_values; /* will add one sample */ +@@ -2943,7 +2941,7 @@ static void bfq_update_peak_rate(struct bfq_data *bfqd, struct request *rq) + if (now_ns - bfqd->last_dispatch > 100*NSEC_PER_MSEC && + bfqd->rq_in_driver == 0) { + bfq_log(bfqd, +-"update_peak_rate: jumping to updating&resetting delta_last %lluus samples %d", ++"jumping to updating&resetting delta_last %lluus samples %d", + (now_ns - bfqd->last_dispatch)>>10, + bfqd->peak_rate_samples) ; + goto update_rate_and_reset; +@@ -2969,7 +2967,7 @@ static void bfq_update_peak_rate(struct bfq_data *bfqd, struct request *rq) + bfqd->delta_from_first = now_ns - bfqd->first_dispatch; + + bfq_log(bfqd, +- "update_peak_rate: added samples %u/%u tot_sects %llu delta_first %lluus", ++ "added samples %u/%u tot_sects %llu delta_first %lluus", + bfqd->peak_rate_samples, bfqd->sequential_samples, + bfqd->tot_sectors_dispatched, + bfqd->delta_from_first>>10); +@@ -2985,12 +2983,12 @@ static void bfq_update_peak_rate(struct bfq_data *bfqd, struct request *rq) + bfqd->last_dispatch = now_ns; + + bfq_log(bfqd, +- "update_peak_rate: delta_first %lluus last_pos %llu peak_rate %llu", ++ "delta_first %lluus last_pos %llu peak_rate %llu", + (now_ns - bfqd->first_dispatch)>>10, + (unsigned long long) bfqd->last_position, + ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); + bfq_log(bfqd, +- "update_peak_rate: samples at end %d", bfqd->peak_rate_samples); ++ "samples at end %d", bfqd->peak_rate_samples); + } + + /* +@@ -3088,11 +3086,11 @@ static void __bfq_bfqq_recalc_budget(struct bfq_data *bfqd, + */ + budget = 2 * min_budget; + +- bfq_log_bfqq(bfqd, bfqq, "recalc_budg: last budg %d, budg left %d", ++ bfq_log_bfqq(bfqd, bfqq, "last budg %d, budg left %d", + bfqq->entity.budget, bfq_bfqq_budget_left(bfqq)); +- bfq_log_bfqq(bfqd, bfqq, "recalc_budg: last max_budg %d, min budg %d", ++ bfq_log_bfqq(bfqd, bfqq, "last max_budg %d, min budg %d", + budget, bfq_min_budget(bfqd)); +- bfq_log_bfqq(bfqd, bfqq, "recalc_budg: sync %d, seeky %d", ++ bfq_log_bfqq(bfqd, bfqq, "sync %d, seeky %d", + bfq_bfqq_sync(bfqq), BFQQ_SEEKY(bfqd->in_service_queue)); + + if (bfq_bfqq_sync(bfqq) && bfqq->wr_coeff == 1) { +@@ -3294,7 +3292,7 @@ static bool bfq_bfqq_is_slow(struct bfq_data *bfqd, struct bfq_queue *bfqq, + else /* charge at least one seek */ + *delta_ms = bfq_slice_idle / NSEC_PER_MSEC; + +- bfq_log(bfqd, "bfq_bfqq_is_slow: too short %u", delta_usecs); ++ bfq_log(bfqd, "too short %u", delta_usecs); + + return slow; + } +@@ -3317,11 +3315,11 @@ static bool bfq_bfqq_is_slow(struct bfq_data *bfqd, struct bfq_queue *bfqq, + * peak rate. + */ + slow = bfqq->entity.service < bfqd->bfq_max_budget / 2; +- bfq_log(bfqd, "bfq_bfqq_is_slow: relative rate %d/%d", ++ bfq_log(bfqd, "relative rate %d/%d", + bfqq->entity.service, bfqd->bfq_max_budget); + } + +- bfq_log_bfqq(bfqd, bfqq, "bfq_bfqq_is_slow: slow %d", slow); ++ bfq_log_bfqq(bfqd, bfqq, "slow %d", slow); + + return slow; + } +@@ -3423,7 +3421,7 @@ static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, + struct bfq_queue *bfqq) + { + bfq_log_bfqq(bfqd, bfqq, +-"softrt_next_start: service_blkg %lu soft_rate %u sects/sec interval %u", ++"service_blkg %lu soft_rate %u sects/sec interval %u", + bfqq->service_from_backlogged, + bfqd->bfq_wr_max_softrt_rate, + jiffies_to_msecs(HZ * bfqq->service_from_backlogged / +@@ -3602,7 +3600,7 @@ static bool bfq_bfqq_budget_timeout(struct bfq_queue *bfqq) + static bool bfq_may_expire_for_budg_timeout(struct bfq_queue *bfqq) + { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "may_budget_timeout: wait_request %d left %d timeout %d", ++ "wait_request %d left %d timeout %d", + bfq_bfqq_wait_request(bfqq), + bfq_bfqq_budget_left(bfqq) >= bfqq->entity.budget / 3, + bfq_bfqq_budget_timeout(bfqq)); +@@ -3863,11 +3861,11 @@ static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq) + * either boosts the throughput (without issues), or is + * necessary to preserve service guarantees. + */ +- bfq_log_bfqq(bfqd, bfqq, "may_idle: sync %d idling_boosts_thr %d", ++ bfq_log_bfqq(bfqd, bfqq, "sync %d idling_boosts_thr %d", + bfq_bfqq_sync(bfqq), idling_boosts_thr); + + bfq_log_bfqq(bfqd, bfqq, +- "may_idle: wr_busy %d boosts %d IO-bound %d guar %d", ++ "wr_busy %d boosts %d IO-bound %d guar %d", + bfqd->wr_busy_queues, + idling_boosts_thr_without_issues, + bfq_bfqq_IO_bound(bfqq), +@@ -3907,7 +3905,7 @@ static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) + if (!bfqq) + goto new_queue; + +- bfq_log_bfqq(bfqd, bfqq, "select_queue: already in-service queue"); ++ bfq_log_bfqq(bfqd, bfqq, "already in-service queue"); + + if (bfq_may_expire_for_budg_timeout(bfqq) && + !bfq_bfqq_wait_request(bfqq) && +@@ -3983,14 +3981,14 @@ static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) + new_queue: + bfqq = bfq_set_in_service_queue(bfqd); + if (bfqq) { +- bfq_log_bfqq(bfqd, bfqq, "select_queue: checking new queue"); ++ bfq_log_bfqq(bfqd, bfqq, "checking new queue"); + goto check_queue; + } + keep_queue: + if (bfqq) +- bfq_log_bfqq(bfqd, bfqq, "select_queue: returned this queue"); ++ bfq_log_bfqq(bfqd, bfqq, "returned this queue"); + else +- bfq_log(bfqd, "select_queue: no queue returned"); ++ bfq_log(bfqd, "no queue returned"); + + return bfqq; + } +@@ -4043,8 +4041,7 @@ static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq) + /* see comments on max_service_from_wr */ + bfq_bfqq_end_wr(bfqq); + bfq_log_bfqq(bfqd, bfqq, +- "[%s] too much service", +- __func__); ++ "too much service"); + } + } + /* +@@ -4122,7 +4119,7 @@ static bool bfq_has_work(struct blk_mq_hw_ctx *hctx) + { + struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; + +- bfq_log(bfqd, "has_work, dispatch_non_empty %d busy_queues %d", ++ bfq_log(bfqd, "dispatch_non_empty %d busy_queues %d", + !list_empty_careful(&bfqd->dispatch), bfqd->busy_queues > 0); + + /* +@@ -4146,7 +4143,7 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + rq->rq_flags &= ~RQF_DISP_LIST; + + bfq_log(bfqd, +- "dispatch requests: picked %p from dispatch list", rq); ++ "picked %p from dispatch list", rq); + bfqq = RQ_BFQQ(rq); + + if (bfqq) { +@@ -4196,7 +4193,7 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + goto start_rq; + } + +- bfq_log(bfqd, "dispatch requests: %d busy queues", bfqd->busy_queues); ++ bfq_log(bfqd, "%d busy queues", bfqd->busy_queues); + + if (bfqd->busy_queues == 0) + goto exit; +@@ -4236,13 +4233,13 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + rq->rq_flags |= RQF_STARTED; + if (bfqq) + bfq_log_bfqq(bfqd, bfqq, +- "dispatched %s request %p, rq_in_driver %d", ++ "%s request %p, rq_in_driver %d", + bfq_bfqq_sync(bfqq) ? "sync" : "async", + rq, + bfqd->rq_in_driver); + else + bfq_log(bfqd, +- "dispatched request %p from dispatch list, rq_in_driver %d", ++ "request %p from dispatch list, rq_in_driver %d", + rq, bfqd->rq_in_driver); + } else + bfq_log(bfqd, +@@ -4339,7 +4336,7 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + BUG_ON(bfqq->ref <= 0); + + if (bfqq->bfqd) +- bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p %d", bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "%p %d", bfqq, bfqq->ref); + + bfqq->ref--; + if (bfqq->ref) +@@ -4383,10 +4380,10 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + } + + if (bfqq->bfqd) +- bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "%p freed", bfqq); + + #ifdef BFQ_GROUP_IOSCHED_ENABLED +- bfq_log_bfqq(bfqq->bfqd, bfqq, "[%s] putting blkg and bfqg %p\n", __func__, bfqg); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "putting blkg and bfqg %p\n", bfqg); + bfqg_and_blkg_put(bfqg); + #endif + kmem_cache_free(bfq_pool, bfqq); +@@ -4418,7 +4415,7 @@ static void bfq_exit_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfq_schedule_dispatch(bfqd); + } + +- bfq_log_bfqq(bfqd, bfqq, "exit_bfqq: %p, %d", bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqd, bfqq, "%p, %d", bfqq, bfqq->ref); + + bfq_put_cooperator(bfqq); + +@@ -4502,7 +4499,7 @@ static void bfq_set_next_ioprio_data(struct bfq_queue *bfqq, + bfqq->entity.new_weight = bfq_ioprio_to_weight(bfqq->new_ioprio); + bfqq->entity.prio_changed = 1; + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "set_next_ioprio_data: bic_class %d prio %d class %d", ++ "bic_class %d prio %d class %d", + ioprio_class, bfqq->new_ioprio, bfqq->new_ioprio_class); + } + +@@ -4529,7 +4526,7 @@ static void bfq_check_ioprio_change(struct bfq_io_cq *bic, struct bio *bio) + bfqq = bfq_get_queue(bfqd, bio, BLK_RW_ASYNC, bic); + bic_set_bfqq(bic, bfqq, false); + bfq_log_bfqq(bfqd, bfqq, +- "check_ioprio_change: bfqq %p %d", ++ "bfqq %p %d", + bfqq, bfqq->ref); + } + +@@ -4667,14 +4664,14 @@ static struct bfq_queue *bfq_get_queue(struct bfq_data *bfqd, + * guarantee that this queue is not freed + * until its group goes away. + */ +- bfq_log_bfqq(bfqd, bfqq, "get_queue, bfqq not in async: %p, %d", ++ bfq_log_bfqq(bfqd, bfqq, "bfqq not in async: %p, %d", + bfqq, bfqq->ref); + *async_bfqq = bfqq; + } + + out: + bfqq->ref++; /* get a process reference to this queue */ +- bfq_log_bfqq(bfqd, bfqq, "get_queue, at end: %p, %d", bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqd, bfqq, "at end: %p, %d", bfqq, bfqq->ref); + rcu_read_unlock(); + return bfqq; + } +@@ -4733,7 +4730,7 @@ static void bfq_update_has_short_ttime(struct bfq_data *bfqd, + bfqq->ttime.ttime_mean > bfqd->bfq_slice_idle)) + has_short_ttime = false; + +- bfq_log_bfqq(bfqd, bfqq, "update_has_short_ttime: has_short_ttime %d", ++ bfq_log_bfqq(bfqd, bfqq, "has_short_ttime %d", + has_short_ttime); + + if (has_short_ttime) +@@ -4759,7 +4756,7 @@ static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, + bfq_update_io_seektime(bfqd, bfqq, rq); + + bfq_log_bfqq(bfqd, bfqq, +- "rq_enqueued: has_short_ttime=%d (seeky %d)", ++ "has_short_ttime=%d (seeky %d)", + bfq_bfqq_has_short_ttime(bfqq), BFQQ_SEEKY(bfqq)); + + bfqq->last_request_pos = blk_rq_pos(rq) + blk_rq_sectors(rq); +@@ -4818,7 +4815,7 @@ static bool __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + + assert_spin_locked(&bfqd->lock); + +- bfq_log_bfqq(bfqd, bfqq, "__insert_req: rq %p bfqq %p", rq, bfqq); ++ bfq_log_bfqq(bfqd, bfqq, "rq %p bfqq %p", rq, bfqq); + + /* + * An unplug may trigger a requeue of a request from the device +@@ -4837,9 +4834,9 @@ static bool __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + new_bfqq->allocated++; + bfqq->allocated--; + bfq_log_bfqq(bfqd, bfqq, +- "insert_request: new allocated %d", bfqq->allocated); ++ "new allocated %d", bfqq->allocated); + bfq_log_bfqq(bfqd, new_bfqq, +- "insert_request: new_bfqq new allocated %d", ++ "new_bfqq new allocated %d", + bfqq->allocated); + + new_bfqq->ref++; +@@ -4911,11 +4908,11 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + rq->rq_flags |= RQF_DISP_LIST; + if (bfqq) + bfq_log_bfqq(bfqd, bfqq, +- "insert_request %p in disp: at_head %d", ++ "%p in disp: at_head %d", + rq, at_head); + else + bfq_log(bfqd, +- "insert_request %p in disp: at_head %d", ++ "%p in disp: at_head %d", + rq, at_head); + } else { + BUG_ON(!(rq->rq_flags & RQF_GOT)); +@@ -5033,7 +5030,7 @@ static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd) + bfqq->dispatched--; + + bfq_log_bfqq(bfqd, bfqq, +- "completed_requests: new disp %d, new rq_in_driver %d", ++ "new disp %d, new rq_in_driver %d", + bfqq->dispatched, bfqd->rq_in_driver); + + if (!bfqq->dispatched && !bfq_bfqq_busy(bfqq)) { +@@ -5061,7 +5058,7 @@ static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd) + delta_us = div_u64(now_ns - bfqd->last_completion, NSEC_PER_USEC); + + bfq_log_bfqq(bfqd, bfqq, +- "rq_completed: delta %uus/%luus max_size %u rate %llu/%llu", ++ "delta %uus/%luus max_size %u rate %llu/%llu", + delta_us, BFQ_MIN_TT/NSEC_PER_USEC, bfqd->last_rq_max_size, + delta_us > 0 ? + (USEC_PER_SEC* +@@ -5129,7 +5126,7 @@ static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd) + static void bfq_finish_requeue_request_body(struct bfq_queue *bfqq) + { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "put_request_body: allocated %d", bfqq->allocated); ++ "allocated %d", bfqq->allocated); + BUG_ON(!bfqq->allocated); + bfqq->allocated--; + +@@ -5406,10 +5403,10 @@ static void bfq_prepare_request(struct request *rq, struct bio *bio) + + bfqq->allocated++; + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "get_request: new allocated %d", bfqq->allocated); ++ "new allocated %d", bfqq->allocated); + + bfqq->ref++; +- bfq_log_bfqq(bfqd, bfqq, "get_request %p: bfqq %p, %d", rq, bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqd, bfqq, "%p: bfqq %p, %d", rq, bfqq, bfqq->ref); + + rq->elv.priv[0] = bic; + rq->elv.priv[1] = bfqq; +@@ -5493,7 +5490,7 @@ static enum hrtimer_restart bfq_idle_slice_timer(struct hrtimer *timer) + idle_slice_timer); + struct bfq_queue *bfqq = bfqd->in_service_queue; + +- bfq_log(bfqd, "slice_timer expired"); ++ bfq_log(bfqd, "expired"); + + /* + * Theoretical race here: the in-service queue can be NULL or +@@ -5515,10 +5512,10 @@ static void __bfq_put_async_bfqq(struct bfq_data *bfqd, + struct bfq_group *root_group = bfqd->root_group; + struct bfq_queue *bfqq = *bfqq_ptr; + +- bfq_log(bfqd, "put_async_bfqq: %p", bfqq); ++ bfq_log(bfqd, "%p", bfqq); + if (bfqq) { + bfq_bfqq_move(bfqd, bfqq, root_group); +- bfq_log_bfqq(bfqd, bfqq, "put_async_bfqq: putting %p, %d", ++ bfq_log_bfqq(bfqd, bfqq, "putting %p, %d", + bfqq, bfqq->ref); + bfq_put_queue(bfqq); + *bfqq_ptr = NULL; +@@ -5547,7 +5544,7 @@ static void bfq_exit_queue(struct elevator_queue *e) + struct bfq_data *bfqd = e->elevator_data; + struct bfq_queue *bfqq, *n; + +- bfq_log(bfqd, "exit_queue: starting ..."); ++ bfq_log(bfqd, "starting ..."); + + hrtimer_cancel(&bfqd->idle_slice_timer); + +@@ -5575,7 +5572,7 @@ static void bfq_exit_queue(struct elevator_queue *e) + spin_unlock_irq(&bfqd->lock); + #endif + +- bfq_log(bfqd, "exit_queue: finished ..."); ++ bfq_log(bfqd, "finished ..."); + kfree(bfqd); + } + +diff --git a/block/bfq-mq.h b/block/bfq-mq.h +index 9a5ce1168ff5..e2ae11bf8f76 100644 +--- a/block/bfq-mq.h ++++ b/block/bfq-mq.h +@@ -712,34 +712,34 @@ static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); + static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ +- pr_crit("%s bfq%d%c %s " fmt "\n", \ ++ pr_crit("%s bfq%d%c %s [%s] " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ + (bfqq)->pid, \ + bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- bfqq_group(bfqq)->blkg_path, ##args); \ ++ bfqq_group(bfqq)->blkg_path, __func__, ##args); \ + } while (0) + + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ +- pr_crit("%s %s " fmt "\n", \ ++ pr_crit("%s %s [%s] " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ +- bfqg->blkg_path, ##args); \ ++ bfqg->blkg_path, __func__, ##args); \ + } while (0) + + #else /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ +- pr_crit("%s bfq%d%c " fmt "\n", \ ++ pr_crit("%s bfq%d%c [%s] " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ + (bfqq)->pid, bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- ##args) ++ __func__, ##args) + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) + + #endif /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log(bfqd, fmt, args...) \ +- pr_crit("%s bfq " fmt "\n", \ ++ pr_crit("%s bfq [%s] " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ +- ##args) ++ __func__, ##args) + + #else /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ + +@@ -762,28 +762,29 @@ static struct bfq_group *bfqq_group(struct bfq_queue *bfqq); + static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) do { \ +- blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s " fmt, \ ++ blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s [%s] " fmt, \ + (bfqq)->pid, \ + bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- bfqq_group(bfqq)->blkg_path, ##args); \ ++ bfqq_group(bfqq)->blkg_path, __func__, ##args); \ + } while (0) + + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ +- blk_add_trace_msg((bfqd)->queue, "%s " fmt, bfqg->blkg_path, ##args);\ ++ blk_add_trace_msg((bfqd)->queue, "%s [%s] " fmt, bfqg->blkg_path, \ ++ __func__, ##args);\ + } while (0) + + #else /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ +- blk_add_trace_msg((bfqd)->queue, "bfq%d%c " fmt, (bfqq)->pid, \ ++ blk_add_trace_msg((bfqd)->queue, "bfq%d%c [%s] " fmt, (bfqq)->pid, \ + bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- ##args) ++ __func__, ##args) + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) + + #endif /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log(bfqd, fmt, args...) \ +- blk_add_trace_msg((bfqd)->queue, "bfq " fmt, ##args) ++ blk_add_trace_msg((bfqd)->queue, "bfq [%s] " fmt, __func__, ##args) + + #endif /* CONFIG_BLK_DEV_IO_TRACE */ + #endif /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ +@@ -938,7 +939,7 @@ bfq_entity_service_tree(struct bfq_entity *entity) + + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "entity_service_tree %p %d", ++ "%p %d", + sched_data->service_tree + idx, idx); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + else { +@@ -946,7 +947,7 @@ bfq_entity_service_tree(struct bfq_entity *entity) + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "entity_service_tree %p %d", ++ "%p %d", + sched_data->service_tree + idx, idx); + } + #endif +diff --git a/block/bfq-sched.c b/block/bfq-sched.c +index 4e6c5232e2fb..ead34c30a7c2 100644 +--- a/block/bfq-sched.c ++++ b/block/bfq-sched.c +@@ -119,7 +119,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "update_next_in_service: chose without lookup"); ++ "chose without lookup"); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = +@@ -127,7 +127,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data*)bfqg->bfqd, bfqg, +- "update_next_in_service: chose without lookup"); ++ "chose without lookup"); + } + #endif + } +@@ -148,7 +148,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + bfqq = bfq_entity_to_bfqq(next_in_service); + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "update_next_in_service: chosen this queue"); ++ "chosen this queue"); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + else { + struct bfq_group *bfqg = +@@ -156,7 +156,7 @@ static bool bfq_update_next_in_service(struct bfq_sched_data *sd, + struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "update_next_in_service: chosen this entity"); ++ "chosen this entity"); + } + #endif + return parent_sched_may_change; +@@ -331,10 +331,10 @@ static void bfq_calc_finish(struct bfq_entity *entity, unsigned long service) + + if (bfqq) { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "calc_finish: serv %lu, w %d", ++ "serv %lu, w %d", + service, entity->weight); + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "calc_finish: start %llu, finish %llu, delta %llu", ++ "start %llu, finish %llu, delta %llu", + start, finish, delta); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { +@@ -342,10 +342,10 @@ static void bfq_calc_finish(struct bfq_entity *entity, unsigned long service) + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "calc_finish group: serv %lu, w %d", ++ "group: serv %lu, w %d", + service, entity->weight); + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "calc_finish group: start %llu, finish %llu, delta %llu", ++ "group: start %llu, finish %llu, delta %llu", + start, finish, delta); + #endif + } +@@ -484,7 +484,7 @@ static void bfq_update_active_node(struct rb_node *node) + + if (bfqq) { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "update_active_node: new min_start %llu", ++ "new min_start %llu", + ((entity->min_start>>10)*1000)>>12); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { +@@ -492,7 +492,7 @@ static void bfq_update_active_node(struct rb_node *node) + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "update_active_node: new min_start %llu", ++ "new min_start %llu", + ((entity->min_start>>10)*1000)>>12); + #endif + } +@@ -620,7 +620,7 @@ static void bfq_get_entity(struct bfq_entity *entity) + + if (bfqq) { + bfqq->ref++; +- bfq_log_bfqq(bfqq->bfqd, bfqq, "get_entity: %p %d", ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "%p %d", + bfqq, bfqq->ref); + } + } +@@ -748,7 +748,7 @@ static void bfq_forget_entity(struct bfq_service_tree *st, + entity->on_st = false; + st->wsum -= entity->weight; + if (bfqq && !is_in_service) { +- bfq_log_bfqq(bfqq->bfqd, bfqq, "forget_entity (before): %p %d", ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "(before): %p %d", + bfqq, bfqq->ref); + bfq_put_queue(bfqq); + } +@@ -1008,7 +1008,7 @@ static void bfq_bfqq_charge_time(struct bfq_data *bfqd, struct bfq_queue *bfqq, + tot_serv_to_charge = entity->service; + + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "charge_time: %lu/%u ms, %d/%d/%d sectors", ++ "%lu/%u ms, %d/%d/%d sectors", + time_ms, timeout_ms, entity->service, + tot_serv_to_charge, entity->budget); + +@@ -1080,7 +1080,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + + if (bfqq) { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "update_fin_time_enqueue: new queue finish %llu", ++ "new queue finish %llu", + ((entity->finish>>10)*1000)>>12); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { +@@ -1088,7 +1088,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "update_fin_time_enqueue: new group finish %llu", ++ "new group finish %llu", + ((entity->finish>>10)*1000)>>12); + #endif + } +@@ -1098,7 +1098,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + + if (bfqq) { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "update_fin_time_enqueue: queue %seligible in st %p", ++ "queue %seligible in st %p", + entity->start <= st->vtime ? "" : "non ", st); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + } else { +@@ -1106,7 +1106,7 @@ static void bfq_update_fin_time_enqueue(struct bfq_entity *entity, + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "update_fin_time_enqueue: group %seligible in st %p", ++ "group %seligible in st %p", + entity->start <= st->vtime ? "" : "non ", st); + #endif + } +@@ -1550,7 +1550,7 @@ static u64 bfq_calc_vtime_jump(struct bfq_service_tree *st) + + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "calc_vtime_jump: new value %llu", ++ "new value %llu", + ((root_entity->min_start>>10)*1000)>>12); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + else { +@@ -1559,7 +1559,7 @@ static u64 bfq_calc_vtime_jump(struct bfq_service_tree *st) + entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "calc_vtime_jump: new value %llu", ++ "new value %llu", + ((root_entity->min_start>>10)*1000)>>12); + } + #endif +@@ -1677,7 +1677,7 @@ __bfq_lookup_next_entity(struct bfq_service_tree *st, bool in_service) + bfqq = bfq_entity_to_bfqq(entity); + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "__lookup_next: start %llu vtime %llu st %p", ++ "start %llu vtime %llu st %p", + ((entity->start>>10)*1000)>>12, + ((new_vtime>>10)*1000)>>12, st); + #ifdef BFQ_GROUP_IOSCHED_ENABLED +@@ -1686,7 +1686,7 @@ __bfq_lookup_next_entity(struct bfq_service_tree *st, bool in_service) + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "__lookup_next: start %llu vtime %llu (%llu) st %p", ++ "start %llu vtime %llu (%llu) st %p", + ((entity->start>>10)*1000)>>12, + ((st->vtime>>10)*1000)>>12, + ((new_vtime>>10)*1000)>>12, st); +@@ -1821,14 +1821,14 @@ static struct bfq_queue *bfq_get_next_queue(struct bfq_data *bfqd) + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg(bfqd, bfqg, +- "get_next_queue: lookup in this group"); ++ "lookup in this group"); + if (!sd->next_in_service) +- pr_crit("get_next_queue: lookup in this group"); ++ pr_crit("lookup in this group"); + } else { + bfq_log_bfqg(bfqd, bfqd->root_group, +- "get_next_queue: lookup in root group"); ++ "lookup in root group"); + if (!sd->next_in_service) +- pr_crit("get_next_queue: lookup in root group"); ++ pr_crit("lookup in root group"); + } + #endif + +@@ -1903,7 +1903,7 @@ static struct bfq_queue *bfq_get_next_queue(struct bfq_data *bfqd) + bfqq = bfq_entity_to_bfqq(entity); + if (bfqq) + bfq_log_bfqq(bfqd, bfqq, +- "get_next_queue: this queue, finish %llu", ++ "this queue, finish %llu", + (((entity->finish>>10)*1000)>>10)>>2); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + else { +@@ -1911,7 +1911,7 @@ static struct bfq_queue *bfq_get_next_queue(struct bfq_data *bfqd) + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg(bfqd, bfqg, +- "get_next_queue: this entity, finish %llu", ++ "this entity, finish %llu", + (((entity->finish>>10)*1000)>>10)>>2); + } + #endif +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index c4df156b1fb4..e49e8ac882b3 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -281,7 +281,7 @@ static void bfq_schedule_dispatch(struct bfq_data *bfqd); + static void bfq_schedule_dispatch(struct bfq_data *bfqd) + { + if (bfqd->queued != 0) { +- bfq_log(bfqd, "schedule dispatch"); ++ bfq_log(bfqd, ""); + kblockd_schedule_work(&bfqd->unplug_work); + } + } +@@ -414,7 +414,7 @@ bfq_rq_pos_tree_lookup(struct bfq_data *bfqd, struct rb_root *root, + if (rb_link) + *rb_link = p; + +- bfq_log(bfqd, "rq_pos_tree_lookup %llu: returning %d", ++ bfq_log(bfqd, "%llu: returning %d", + (unsigned long long) sector, + bfqq ? bfqq->pid : 0); + +@@ -635,7 +635,7 @@ static struct request *bfq_check_fifo(struct bfq_queue *bfqq, + if (rq == last || ktime_get_ns() < rq->fifo_time) + return NULL; + +- bfq_log_bfqq(bfqq->bfqd, bfqq, "check_fifo: returned %p", rq); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "returned %p", rq); + BUG_ON(RB_EMPTY_NODE(&rq->rb_node)); + return rq; + } +@@ -728,7 +728,7 @@ static void bfq_updated_next_req(struct bfq_data *bfqd, + bfq_serv_to_charge(next_rq, bfqq)); + if (entity->budget != new_budget) { + entity->budget = new_budget; +- bfq_log_bfqq(bfqd, bfqq, "updated next rq: new budget %lu", ++ bfq_log_bfqq(bfqd, bfqq, "new budget %lu", + new_budget); + bfq_requeue_bfqq(bfqd, bfqq, false); + } +@@ -800,8 +800,7 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + BUG_ON(time_is_after_jiffies(bfqq->last_wr_start_finish)); + + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "[%s] bic %p wr_coeff %d start_finish %lu max_time %lu", +- __func__, ++ "bic %p wr_coeff %d start_finish %lu max_time %lu", + bic, bfqq->wr_coeff, bfqq->last_wr_start_finish, + bfqq->wr_cur_max_time); + +@@ -814,11 +813,11 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd, + bfq_wr_duration(bfqd))) { + switch_back_to_interactive_wr(bfqq, bfqd); + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "resume state: switching back to interactive"); ++ "switching back to interactive"); + } else { + bfqq->wr_coeff = 1; + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "resume state: switching off wr (%lu + %lu < %lu)", ++ "switching off wr (%lu + %lu < %lu)", + bfqq->last_wr_start_finish, bfqq->wr_cur_max_time, + jiffies); + } +@@ -870,7 +869,7 @@ static void bfq_add_to_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) + /* Increment burst size to take into account also bfqq */ + bfqd->burst_size++; + +- bfq_log_bfqq(bfqd, bfqq, "add_to_burst %d", bfqd->burst_size); ++ bfq_log_bfqq(bfqd, bfqq, "%d", bfqd->burst_size); + + BUG_ON(bfqd->burst_size > bfqd->bfq_large_burst_thresh); + +@@ -883,7 +882,7 @@ static void bfq_add_to_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) + * other to consider this burst as large. + */ + bfqd->large_burst = true; +- bfq_log_bfqq(bfqd, bfqq, "add_to_burst: large burst started"); ++ bfq_log_bfqq(bfqd, bfqq, "large burst started"); + + /* + * We can now mark all queues in the burst list as +@@ -1055,7 +1054,7 @@ static void bfq_handle_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfqd->large_burst = false; + bfq_reset_burst_list(bfqd, bfqq); + bfq_log_bfqq(bfqd, bfqq, +- "handle_burst: late activation or different group"); ++ "late activation or different group"); + goto end; + } + +@@ -1065,7 +1064,7 @@ static void bfq_handle_burst(struct bfq_data *bfqd, struct bfq_queue *bfqq) + * bfqq as belonging to this large burst immediately. + */ + if (bfqd->large_burst) { +- bfq_log_bfqq(bfqd, bfqq, "handle_burst: marked in burst"); ++ bfq_log_bfqq(bfqd, bfqq, "marked in burst"); + bfq_mark_bfqq_in_large_burst(bfqq); + goto end; + } +@@ -1572,7 +1571,7 @@ static void bfq_add_request(struct request *rq) + unsigned int old_wr_coeff = bfqq->wr_coeff; + bool interactive = false; + +- bfq_log_bfqq(bfqd, bfqq, "add_request: size %u %s", ++ bfq_log_bfqq(bfqd, bfqq, "size %u %s", + blk_rq_sectors(rq), rq_is_sync(rq) ? "S" : "A"); + + if (bfqq->wr_coeff > 1) /* queue is being weight-raised */ +@@ -1870,10 +1869,10 @@ static void bfq_bfqq_end_wr(struct bfq_queue *bfqq) + */ + bfqq->entity.prio_changed = 1; + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "end_wr: wrais ending at %lu, rais_max_time %u", ++ "wrais ending at %lu, rais_max_time %u", + bfqq->last_wr_start_finish, + jiffies_to_msecs(bfqq->wr_cur_max_time)); +- bfq_log_bfqq(bfqq->bfqd, bfqq, "end_wr: wr_busy %d", ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "wr_busy %d", + bfqq->bfqd->wr_busy_queues); + } + +@@ -2048,8 +2047,8 @@ static bool bfq_may_be_close_cooperator(struct bfq_queue *bfqq, + { + if (bfq_too_late_for_merging(new_bfqq)) { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "[%s] too late for bfq%d to be merged", +- __func__, new_bfqq->pid); ++ "too late for bfq%d to be merged", ++ new_bfqq->pid); + return false; + } + +@@ -2258,7 +2257,7 @@ bfq_merge_bfqqs(struct bfq_data *bfqd, struct bfq_io_cq *bic, + + } + +- bfq_log_bfqq(bfqd, new_bfqq, "merge_bfqqs: wr_busy %d", ++ bfq_log_bfqq(bfqd, new_bfqq, "wr_busy %d", + bfqd->wr_busy_queues); + + /* +@@ -2359,7 +2358,7 @@ static void bfq_set_budget_timeout(struct bfq_data *bfqd, + bfqq->budget_timeout = jiffies + + bfqd->bfq_timeout * timeout_coeff; + +- bfq_log_bfqq(bfqd, bfqq, "set budget_timeout %u", ++ bfq_log_bfqq(bfqd, bfqq, "%u", + jiffies_to_msecs(bfqd->bfq_timeout * timeout_coeff)); + } + +@@ -2427,10 +2426,10 @@ static void __bfq_set_in_service_queue(struct bfq_data *bfqd, + + bfq_set_budget_timeout(bfqd, bfqq); + bfq_log_bfqq(bfqd, bfqq, +- "set_in_service_queue, cur-budget = %d", ++ "cur-budget = %d", + bfqq->entity.budget); + } else +- bfq_log(bfqd, "set_in_service_queue: NULL"); ++ bfq_log(bfqd, "NULL"); + + bfqd->in_service_queue = bfqq; + } +@@ -2559,7 +2558,7 @@ static void bfq_reset_rate_computation(struct bfq_data *bfqd, struct request *rq + bfqd->peak_rate_samples = 0; /* full re-init on next disp. */ + + bfq_log(bfqd, +- "reset_rate_computation at end, sample %u/%u tot_sects %llu", ++ "at end, sample %u/%u tot_sects %llu", + bfqd->peak_rate_samples, bfqd->sequential_samples, + bfqd->tot_sectors_dispatched); + } +@@ -2579,7 +2578,7 @@ static void bfq_update_rate_reset(struct bfq_data *bfqd, struct request *rq) + if (bfqd->peak_rate_samples < BFQ_RATE_MIN_SAMPLES || + bfqd->delta_from_first < BFQ_RATE_MIN_INTERVAL) { + bfq_log(bfqd, +- "update_rate_reset: only resetting, delta_first %lluus samples %d", ++ "only resetting, delta_first %lluus samples %d", + bfqd->delta_from_first>>10, bfqd->peak_rate_samples); + goto reset_computation; + } +@@ -2603,7 +2602,7 @@ static void bfq_update_rate_reset(struct bfq_data *bfqd, struct request *rq) + div_u64(bfqd->delta_from_first, NSEC_PER_USEC)); + + bfq_log(bfqd, +-"update_rate_reset: tot_sects %llu delta_first %lluus rate %llu sects/s (%d)", ++"tot_sects %llu delta_first %lluus rate %llu sects/s (%d)", + bfqd->tot_sectors_dispatched, bfqd->delta_from_first>>10, + ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), + rate > 20<peak_rate) || + rate > 20<peak_rate_samples, bfqd->sequential_samples, + ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), + ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); + goto reset_computation; + } else { + bfq_log(bfqd, +- "update_rate_reset: do update, samples %u/%u rate/peak %llu/%llu", ++ "do update, samples %u/%u rate/peak %llu/%llu", + bfqd->peak_rate_samples, bfqd->sequential_samples, + ((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT), + ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); +@@ -2681,7 +2680,7 @@ static void bfq_update_rate_reset(struct bfq_data *bfqd, struct request *rq) + rate /= divisor; /* smoothing constant alpha = 1/divisor */ + + bfq_log(bfqd, +- "update_rate_reset: divisor %d tmp_peak_rate %llu tmp_rate %u", ++ "divisor %d tmp_peak_rate %llu tmp_rate %u", + divisor, + ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT), + (u32)((USEC_PER_SEC*(u64)rate)>>BFQ_RATE_SHIFT)); +@@ -2735,7 +2734,7 @@ static void bfq_update_peak_rate(struct bfq_data *bfqd, struct request *rq) + + if (bfqd->peak_rate_samples == 0) { /* first dispatch */ + bfq_log(bfqd, +- "update_peak_rate: goto reset, samples %d", ++ "goto reset, samples %d", + bfqd->peak_rate_samples) ; + bfq_reset_rate_computation(bfqd, rq); + goto update_last_values; /* will add one sample */ +@@ -2756,7 +2755,7 @@ static void bfq_update_peak_rate(struct bfq_data *bfqd, struct request *rq) + if (now_ns - bfqd->last_dispatch > 100*NSEC_PER_MSEC && + bfqd->rq_in_driver == 0) { + bfq_log(bfqd, +-"update_peak_rate: jumping to updating&resetting delta_last %lluus samples %d", ++"jumping to updating&resetting delta_last %lluus samples %d", + (now_ns - bfqd->last_dispatch)>>10, + bfqd->peak_rate_samples) ; + goto update_rate_and_reset; +@@ -2782,7 +2781,7 @@ static void bfq_update_peak_rate(struct bfq_data *bfqd, struct request *rq) + bfqd->delta_from_first = now_ns - bfqd->first_dispatch; + + bfq_log(bfqd, +- "update_peak_rate: added samples %u/%u tot_sects %llu delta_first %lluus", ++ "added samples %u/%u tot_sects %llu delta_first %lluus", + bfqd->peak_rate_samples, bfqd->sequential_samples, + bfqd->tot_sectors_dispatched, + bfqd->delta_from_first>>10); +@@ -2798,12 +2797,12 @@ static void bfq_update_peak_rate(struct bfq_data *bfqd, struct request *rq) + bfqd->last_dispatch = now_ns; + + bfq_log(bfqd, +- "update_peak_rate: delta_first %lluus last_pos %llu peak_rate %llu", ++ "delta_first %lluus last_pos %llu peak_rate %llu", + (now_ns - bfqd->first_dispatch)>>10, + (unsigned long long) bfqd->last_position, + ((USEC_PER_SEC*(u64)bfqd->peak_rate)>>BFQ_RATE_SHIFT)); + bfq_log(bfqd, +- "update_peak_rate: samples at end %d", bfqd->peak_rate_samples); ++ "samples at end %d", bfqd->peak_rate_samples); + } + + /* +@@ -2900,11 +2899,11 @@ static void __bfq_bfqq_recalc_budget(struct bfq_data *bfqd, + */ + budget = 2 * min_budget; + +- bfq_log_bfqq(bfqd, bfqq, "recalc_budg: last budg %d, budg left %d", ++ bfq_log_bfqq(bfqd, bfqq, "last budg %d, budg left %d", + bfqq->entity.budget, bfq_bfqq_budget_left(bfqq)); +- bfq_log_bfqq(bfqd, bfqq, "recalc_budg: last max_budg %d, min budg %d", ++ bfq_log_bfqq(bfqd, bfqq, "last max_budg %d, min budg %d", + budget, bfq_min_budget(bfqd)); +- bfq_log_bfqq(bfqd, bfqq, "recalc_budg: sync %d, seeky %d", ++ bfq_log_bfqq(bfqd, bfqq, "sync %d, seeky %d", + bfq_bfqq_sync(bfqq), BFQQ_SEEKY(bfqd->in_service_queue)); + + if (bfq_bfqq_sync(bfqq) && bfqq->wr_coeff == 1) { +@@ -3106,7 +3105,7 @@ static bool bfq_bfqq_is_slow(struct bfq_data *bfqd, struct bfq_queue *bfqq, + else /* charge at least one seek */ + *delta_ms = bfq_slice_idle / NSEC_PER_MSEC; + +- bfq_log(bfqd, "bfq_bfqq_is_slow: too short %u", delta_usecs); ++ bfq_log(bfqd, "too short %u", delta_usecs); + + return slow; + } +@@ -3129,11 +3128,11 @@ static bool bfq_bfqq_is_slow(struct bfq_data *bfqd, struct bfq_queue *bfqq, + * peak rate. + */ + slow = bfqq->entity.service < bfqd->bfq_max_budget / 2; +- bfq_log(bfqd, "bfq_bfqq_is_slow: relative rate %d/%d", ++ bfq_log(bfqd, "relative rate %d/%d", + bfqq->entity.service, bfqd->bfq_max_budget); + } + +- bfq_log_bfqq(bfqd, bfqq, "bfq_bfqq_is_slow: slow %d", slow); ++ bfq_log_bfqq(bfqd, bfqq, "slow %d", slow); + + return slow; + } +@@ -3235,7 +3234,7 @@ static unsigned long bfq_bfqq_softrt_next_start(struct bfq_data *bfqd, + struct bfq_queue *bfqq) + { + bfq_log_bfqq(bfqd, bfqq, +-"softrt_next_start: service_blkg %lu soft_rate %u sects/sec interval %u", ++"service_blkg %lu soft_rate %u sects/sec interval %u", + bfqq->service_from_backlogged, + bfqd->bfq_wr_max_softrt_rate, + jiffies_to_msecs(HZ * bfqq->service_from_backlogged / +@@ -3414,7 +3413,7 @@ static bool bfq_bfqq_budget_timeout(struct bfq_queue *bfqq) + static bool bfq_may_expire_for_budg_timeout(struct bfq_queue *bfqq) + { + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "may_budget_timeout: wait_request %d left %d timeout %d", ++ "wait_request %d left %d timeout %d", + bfq_bfqq_wait_request(bfqq), + bfq_bfqq_budget_left(bfqq) >= bfqq->entity.budget / 3, + bfq_bfqq_budget_timeout(bfqq)); +@@ -3675,11 +3674,11 @@ static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq) + * either boosts the throughput (without issues), or is + * necessary to preserve service guarantees. + */ +- bfq_log_bfqq(bfqd, bfqq, "may_idle: sync %d idling_boosts_thr %d", ++ bfq_log_bfqq(bfqd, bfqq, "sync %d idling_boosts_thr %d", + bfq_bfqq_sync(bfqq), idling_boosts_thr); + + bfq_log_bfqq(bfqd, bfqq, +- "may_idle: wr_busy %d boosts %d IO-bound %d guar %d", ++ "wr_busy %d boosts %d IO-bound %d guar %d", + bfqd->wr_busy_queues, + idling_boosts_thr_without_issues, + bfq_bfqq_IO_bound(bfqq), +@@ -3719,7 +3718,7 @@ static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) + if (!bfqq) + goto new_queue; + +- bfq_log_bfqq(bfqd, bfqq, "select_queue: already in-service queue"); ++ bfq_log_bfqq(bfqd, bfqq, "already in-service queue"); + + if (bfq_may_expire_for_budg_timeout(bfqq) && + !hrtimer_active(&bfqd->idle_slice_timer) && +@@ -3797,14 +3796,14 @@ static struct bfq_queue *bfq_select_queue(struct bfq_data *bfqd) + new_queue: + bfqq = bfq_set_in_service_queue(bfqd); + if (bfqq) { +- bfq_log_bfqq(bfqd, bfqq, "select_queue: checking new queue"); ++ bfq_log_bfqq(bfqd, bfqq, "checking new queue"); + goto check_queue; + } + keep_queue: + if (bfqq) +- bfq_log_bfqq(bfqd, bfqq, "select_queue: returned this queue"); ++ bfq_log_bfqq(bfqd, bfqq, "returned this queue"); + else +- bfq_log(bfqd, "select_queue: no queue returned"); ++ bfq_log(bfqd, "no queue returned"); + + return bfqq; + } +@@ -3857,8 +3856,7 @@ static void bfq_update_wr_data(struct bfq_data *bfqd, struct bfq_queue *bfqq) + /* see comments on max_service_from_wr */ + bfq_bfqq_end_wr(bfqq); + bfq_log_bfqq(bfqd, bfqq, +- "[%s] too much service", +- __func__); ++ "too much service"); + } + } + /* +@@ -3987,7 +3985,7 @@ static int bfq_dispatch_requests(struct request_queue *q, int force) + struct bfq_data *bfqd = q->elevator->elevator_data; + struct bfq_queue *bfqq; + +- bfq_log(bfqd, "dispatch requests: %d busy queues", bfqd->busy_queues); ++ bfq_log(bfqd, "%d busy queues", bfqd->busy_queues); + + if (bfqd->busy_queues == 0) + return 0; +@@ -4021,7 +4019,7 @@ static int bfq_dispatch_requests(struct request_queue *q, int force) + if (!bfq_dispatch_request(bfqd, bfqq)) + return 0; + +- bfq_log_bfqq(bfqd, bfqq, "dispatched %s request", ++ bfq_log_bfqq(bfqd, bfqq, "%s request", + bfq_bfqq_sync(bfqq) ? "sync" : "async"); + + BUG_ON(bfqq->next_rq == NULL && +@@ -4044,7 +4042,7 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + + BUG_ON(bfqq->ref <= 0); + +- bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p %d", bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "%p %d", bfqq, bfqq->ref); + bfqq->ref--; + if (bfqq->ref) + return; +@@ -4086,7 +4084,7 @@ static void bfq_put_queue(struct bfq_queue *bfqq) + bfqq->bfqd->burst_size--; + } + +- bfq_log_bfqq(bfqq->bfqd, bfqq, "put_queue: %p freed", bfqq); ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "%p freed", bfqq); + + kmem_cache_free(bfq_pool, bfqq); + #ifdef BFQ_GROUP_IOSCHED_ENABLED +@@ -4120,7 +4118,7 @@ static void bfq_exit_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq) + bfq_schedule_dispatch(bfqd); + } + +- bfq_log_bfqq(bfqd, bfqq, "exit_bfqq: %p, %d", bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqd, bfqq, "%p, %d", bfqq, bfqq->ref); + + bfq_put_cooperator(bfqq); + +@@ -4200,7 +4198,7 @@ static void bfq_set_next_ioprio_data(struct bfq_queue *bfqq, + bfqq->entity.new_weight = bfq_ioprio_to_weight(bfqq->new_ioprio); + bfqq->entity.prio_changed = 1; + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "set_next_ioprio_data: bic_class %d prio %d class %d", ++ "bic_class %d prio %d class %d", + ioprio_class, bfqq->new_ioprio, bfqq->new_ioprio_class); + } + +@@ -4227,7 +4225,7 @@ static void bfq_check_ioprio_change(struct bfq_io_cq *bic, struct bio *bio) + bfqq = bfq_get_queue(bfqd, bio, BLK_RW_ASYNC, bic); + bic_set_bfqq(bic, bfqq, false); + bfq_log_bfqq(bfqd, bfqq, +- "check_ioprio_change: bfqq %p %d", ++ "bfqq %p %d", + bfqq, bfqq->ref); + } + +@@ -4362,14 +4360,14 @@ static struct bfq_queue *bfq_get_queue(struct bfq_data *bfqd, + * guarantee that this queue is not freed + * until its group goes away. + */ +- bfq_log_bfqq(bfqd, bfqq, "get_queue, bfqq not in async: %p, %d", ++ bfq_log_bfqq(bfqd, bfqq, "bfqq not in async: %p, %d", + bfqq, bfqq->ref); + *async_bfqq = bfqq; + } + + out: + bfqq->ref++; /* get a process reference to this queue */ +- bfq_log_bfqq(bfqd, bfqq, "get_queue, at end: %p, %d", bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqd, bfqq, "at end: %p, %d", bfqq, bfqq->ref); + rcu_read_unlock(); + return bfqq; + } +@@ -4428,7 +4426,7 @@ static void bfq_update_has_short_ttime(struct bfq_data *bfqd, + bic->ttime.ttime_mean > bfqd->bfq_slice_idle)) + has_short_ttime = false; + +- bfq_log_bfqq(bfqd, bfqq, "update_has_short_ttime: has_short_ttime %d", ++ bfq_log_bfqq(bfqd, bfqq, "has_short_ttime %d", + has_short_ttime); + + if (has_short_ttime) +@@ -4454,7 +4452,7 @@ static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq, + bfq_update_io_seektime(bfqd, bfqq, rq); + + bfq_log_bfqq(bfqd, bfqq, +- "rq_enqueued: has_short_ttime=%d (seeky %d)", ++ "has_short_ttime=%d (seeky %d)", + bfq_bfqq_has_short_ttime(bfqq), BFQQ_SEEKY(bfqq)); + + bfqq->last_request_pos = blk_rq_pos(rq) + blk_rq_sectors(rq); +@@ -4629,7 +4627,7 @@ static void bfq_completed_request(struct request_queue *q, struct request *rq) + */ + delta_us = div_u64(now_ns - bfqd->last_completion, NSEC_PER_USEC); + +- bfq_log(bfqd, "rq_completed: delta %uus/%luus max_size %u rate %llu/%llu", ++ bfq_log(bfqd, "delta %uus/%luus max_size %u rate %llu/%llu", + delta_us, BFQ_MIN_TT/NSEC_PER_USEC, bfqd->last_rq_max_size, + delta_us > 0 ? + (USEC_PER_SEC* +@@ -4750,7 +4748,7 @@ static void bfq_put_request(struct request *rq) + rq->elv.priv[0] = NULL; + rq->elv.priv[1] = NULL; + +- bfq_log_bfqq(bfqq->bfqd, bfqq, "put_request %p, %d", ++ bfq_log_bfqq(bfqq->bfqd, bfqq, "%p, %d", + bfqq, bfqq->ref); + bfq_put_queue(bfqq); + } +@@ -4816,7 +4814,7 @@ static int bfq_set_request(struct request_queue *q, struct request *rq, + bic_set_bfqq(bic, bfqq, is_sync); + if (split && is_sync) { + bfq_log_bfqq(bfqd, bfqq, +- "set_request: was_in_list %d " ++ "was_in_list %d " + "was_in_large_burst %d " + "large burst in progress %d", + bic->was_in_burst_list, +@@ -4826,12 +4824,12 @@ static int bfq_set_request(struct request_queue *q, struct request *rq, + if ((bic->was_in_burst_list && bfqd->large_burst) || + bic->saved_in_large_burst) { + bfq_log_bfqq(bfqd, bfqq, +- "set_request: marking in " ++ "marking in " + "large burst"); + bfq_mark_bfqq_in_large_burst(bfqq); + } else { + bfq_log_bfqq(bfqd, bfqq, +- "set_request: clearing in " ++ "clearing in " + "large burst"); + bfq_clear_bfqq_in_large_burst(bfqq); + if (bic->was_in_burst_list) +@@ -4888,7 +4886,7 @@ static int bfq_set_request(struct request_queue *q, struct request *rq, + + bfqq->allocated[rw]++; + bfqq->ref++; +- bfq_log_bfqq(bfqd, bfqq, "set_request: bfqq %p, %d", bfqq, bfqq->ref); ++ bfq_log_bfqq(bfqd, bfqq, "bfqq %p, %d", bfqq, bfqq->ref); + + rq->elv.priv[0] = bic; + rq->elv.priv[1] = bfqq; +@@ -4962,7 +4960,7 @@ static enum hrtimer_restart bfq_idle_slice_timer(struct hrtimer *timer) + * case we just expire a queue too early. + */ + if (bfqq) { +- bfq_log_bfqq(bfqd, bfqq, "slice_timer expired"); ++ bfq_log_bfqq(bfqd, bfqq, "expired"); + bfq_clear_bfqq_wait_request(bfqq); + + if (bfq_bfqq_budget_timeout(bfqq)) +@@ -5005,10 +5003,10 @@ static void __bfq_put_async_bfqq(struct bfq_data *bfqd, + struct bfq_group *root_group = bfqd->root_group; + struct bfq_queue *bfqq = *bfqq_ptr; + +- bfq_log(bfqd, "put_async_bfqq: %p", bfqq); ++ bfq_log(bfqd, "%p", bfqq); + if (bfqq) { + bfq_bfqq_move(bfqd, bfqq, root_group); +- bfq_log_bfqq(bfqd, bfqq, "put_async_bfqq: putting %p, %d", ++ bfq_log_bfqq(bfqd, bfqq, "putting %p, %d", + bfqq, bfqq->ref); + bfq_put_queue(bfqq); + *bfqq_ptr = NULL; +diff --git a/block/bfq.h b/block/bfq.h +index 0cd7a3f251a7..4d2fe7f77af1 100644 +--- a/block/bfq.h ++++ b/block/bfq.h +@@ -698,37 +698,37 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + \ + assert_spin_locked((bfqd)->queue->queue_lock); \ + blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ +- pr_crit("%s bfq%d%c %s " fmt "\n", \ ++ pr_crit("%s bfq%d%c %s [%s] " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ + (bfqq)->pid, \ + bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- __pbuf, ##args); \ ++ __pbuf, __func__, ##args); \ + } while (0) + + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ + char __pbuf[128]; \ + \ + blkg_path(bfqg_to_blkg(bfqg), __pbuf, sizeof(__pbuf)); \ +- pr_crit("%s %s " fmt "\n", \ ++ pr_crit("%s %s [%s] " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ +- __pbuf, ##args); \ ++ __pbuf, __func__, ##args); \ + } while (0) + + #else /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ +- pr_crit("%s bfq%d%c " fmt "\n", \ ++ pr_crit("%s bfq%d%c [%s] " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ + (bfqq)->pid, bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- ##args) ++ __func__, ##args) + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) + + #endif /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log(bfqd, fmt, args...) \ +- pr_crit("%s bfq " fmt "\n", \ ++ pr_crit("%s bfq [%s] " fmt "\n", \ + checked_dev_name((bfqd)->queue->backing_dev_info->dev), \ +- ##args) ++ __func__, ##args) + + #else /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ + +@@ -755,31 +755,32 @@ static struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg); + \ + assert_spin_locked((bfqd)->queue->queue_lock); \ + blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \ +- blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s " fmt, \ ++ blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s [%s] " fmt, \ + (bfqq)->pid, \ + bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- __pbuf, ##args); \ ++ __pbuf, __func__, ##args); \ + } while (0) + + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do { \ + char __pbuf[128]; \ + \ + blkg_path(bfqg_to_blkg(bfqg), __pbuf, sizeof(__pbuf)); \ +- blk_add_trace_msg((bfqd)->queue, "%s " fmt, __pbuf, ##args); \ ++ blk_add_trace_msg((bfqd)->queue, "%s [%s] " fmt, __pbuf, \ ++ __func__, ##args); \ + } while (0) + + #else /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log_bfqq(bfqd, bfqq, fmt, args...) \ +- blk_add_trace_msg((bfqd)->queue, "bfq%d%c " fmt, (bfqq)->pid, \ ++ blk_add_trace_msg((bfqd)->queue, "bfq%d%c [%s] " fmt, (bfqq)->pid, \ + bfq_bfqq_sync((bfqq)) ? 'S' : 'A', \ +- ##args) ++ __func__, ##args) + #define bfq_log_bfqg(bfqd, bfqg, fmt, args...) do {} while (0) + + #endif /* BFQ_GROUP_IOSCHED_ENABLED */ + + #define bfq_log(bfqd, fmt, args...) \ +- blk_add_trace_msg((bfqd)->queue, "bfq " fmt, ##args) ++ blk_add_trace_msg((bfqd)->queue, "bfq [%s] " fmt, __func__, ##args) + + #endif /* CONFIG_BLK_DEV_IO_TRACE */ + #endif /* CONFIG_BFQ_REDIRECT_TO_CONSOLE */ +@@ -928,7 +929,7 @@ bfq_entity_service_tree(struct bfq_entity *entity) + + if (bfqq) + bfq_log_bfqq(bfqq->bfqd, bfqq, +- "entity_service_tree %p %d", ++ "%p %d", + sched_data->service_tree + idx, idx); + #ifdef BFQ_GROUP_IOSCHED_ENABLED + else { +@@ -936,7 +937,7 @@ bfq_entity_service_tree(struct bfq_entity *entity) + container_of(entity, struct bfq_group, entity); + + bfq_log_bfqg((struct bfq_data *)bfqg->bfqd, bfqg, +- "entity_service_tree %p %d", ++ "%p %d", + sched_data->service_tree + idx, idx); + } + #endif + +From 673a457e8a54d1c4b66e61b1a50956ba0b8c6a60 Mon Sep 17 00:00:00 2001 +From: Davide Paganelli +Date: Thu, 8 Feb 2018 11:49:58 +0100 +Subject: [PATCH 19/23] block, bfq-mq, bfq-sq: make bfq_bfqq_expire print + expiration reason + +Improve readability of the log messages related to the expiration +reasons of the function bfq_bfqq_expire. +Change the printing of the number that represents the reason for +expiration with an actual textual description of the reason. + +Signed-off-by: Davide Paganelli +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 10 ++++++++-- + block/bfq-sq-iosched.c | 10 ++++++++-- + 2 files changed, 16 insertions(+), 4 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index edc93b6af186..9268dd47a4e5 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -133,6 +133,12 @@ static const int bfq_timeout = (HZ / 8); + */ + static const unsigned long bfq_merge_time_limit = HZ/10; + ++#define MAX_LENGTH_REASON_NAME 25 ++ ++static const char reason_name[][MAX_LENGTH_REASON_NAME] = {"TOO_IDLE", ++"BUDGET_TIMEOUT", "BUDGET_EXHAUSTED", "NO_MORE_REQUESTS", ++"PREEMPTED"}; ++ + static struct kmem_cache *bfq_pool; + + /* Below this threshold (in ns), we consider thinktime immediate. */ +@@ -3553,8 +3559,8 @@ static void bfq_bfqq_expire(struct bfq_data *bfqd, + } + + bfq_log_bfqq(bfqd, bfqq, +- "expire (%d, slow %d, num_disp %d, short_ttime %d, weight %d)", +- reason, slow, bfqq->dispatched, ++ "expire (%s, slow %d, num_disp %d, short_ttime %d, weight %d)", ++ reason_name[reason], slow, bfqq->dispatched, + bfq_bfqq_has_short_ttime(bfqq), entity->weight); + + /* +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index e49e8ac882b3..f95deaab49a1 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -127,6 +127,12 @@ static const int bfq_timeout = (HZ / 8); + */ + static const unsigned long bfq_merge_time_limit = HZ/10; + ++#define MAX_LENGTH_REASON_NAME 25 ++ ++static const char reason_name[][MAX_LENGTH_REASON_NAME] = {"TOO_IDLE", ++"BUDGET_TIMEOUT", "BUDGET_EXHAUSTED", "NO_MORE_REQUESTS", ++"PREEMPTED"}; ++ + static struct kmem_cache *bfq_pool; + + /* Below this threshold (in ns), we consider thinktime immediate. */ +@@ -3366,8 +3372,8 @@ static void bfq_bfqq_expire(struct bfq_data *bfqd, + } + + bfq_log_bfqq(bfqd, bfqq, +- "expire (%d, slow %d, num_disp %d, short_ttime %d, weight %d)", +- reason, slow, bfqq->dispatched, ++ "expire (%s, slow %d, num_disp %d, short_ttime %d, weight %d)", ++ reason_name[reason], slow, bfqq->dispatched, + bfq_bfqq_has_short_ttime(bfqq), entity->weight); + + /* + +From 62e80623fbb58367c3f667dab22fea0804001f3b Mon Sep 17 00:00:00 2001 +From: Melzani Alessandro +Date: Mon, 26 Feb 2018 22:21:59 +0100 +Subject: [PATCH 20/23] bfq-mq: port of "block, bfq: remove batches of + confusing ifdefs" + +Commit a33801e8b473 ("block, bfq: move debug blkio stats behind +CONFIG_DEBUG_BLK_CGROUP") introduced two batches of confusing ifdefs: +one reported in [1], plus a similar one in another function. This +commit removes both batches, in the way suggested in [1]. + +[1] https://www.spinics.net/lists/linux-block/msg20043.html + +Fixes: a33801e8b473 ("block, bfq: move debug blkio stats behind CONFIG_DEBUG_BLK_CGROUP") + +Signed-off-by: Alessandro Melzani +--- + block/bfq-mq-iosched.c | 128 ++++++++++++++++++++++++++++--------------------- + 1 file changed, 73 insertions(+), 55 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 9268dd47a4e5..5a211620f316 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -4256,35 +4256,17 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + return rq; + } + +-static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) +-{ +- struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; +- struct request *rq; +-#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) +- struct bfq_queue *in_serv_queue, *bfqq; +- bool waiting_rq, idle_timer_disabled; +-#endif + +- spin_lock_irq(&bfqd->lock); +- +-#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) +- in_serv_queue = bfqd->in_service_queue; +- waiting_rq = in_serv_queue && bfq_bfqq_wait_request(in_serv_queue); +- +- rq = __bfq_dispatch_request(hctx); +- +- idle_timer_disabled = +- waiting_rq && !bfq_bfqq_wait_request(in_serv_queue); +- +-#else +- rq = __bfq_dispatch_request(hctx); +-#endif +- spin_unlock_irq(&bfqd->lock); ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) ++static void bfq_update_dispatch_stats(struct request_queue *q, ++ struct request *rq, ++ struct bfq_queue *in_serv_queue, ++ bool idle_timer_disabled) ++{ ++ struct bfq_queue *bfqq = rq ? RQ_BFQQ(rq) : NULL; + +-#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) +- bfqq = rq ? RQ_BFQQ(rq) : NULL; + if (!idle_timer_disabled && !bfqq) +- return rq; ++ return; + + /* + * rq and bfqq are guaranteed to exist until this function +@@ -4299,7 +4281,7 @@ static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + * In addition, the following queue lock guarantees that + * bfqq_group(bfqq) exists as well. + */ +- spin_lock_irq(hctx->queue->queue_lock); ++ spin_lock_irq(q->queue_lock); + if (idle_timer_disabled) + /* + * Since the idle timer has been disabled, +@@ -4318,8 +4300,35 @@ static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) + bfqg_stats_set_start_empty_time(bfqg); + bfqg_stats_update_io_remove(bfqg, rq->cmd_flags); + } +- spin_unlock_irq(hctx->queue->queue_lock); ++ spin_unlock_irq(q->queue_lock); ++} ++#else ++static inline void bfq_update_dispatch_stats(struct request_queue *q, ++ struct request *rq, ++ struct bfq_queue *in_serv_queue, ++ bool idle_timer_disabled) {} + #endif ++static struct request *bfq_dispatch_request(struct blk_mq_hw_ctx *hctx) ++{ ++ struct bfq_data *bfqd = hctx->queue->elevator->elevator_data; ++ struct request *rq; ++ struct bfq_queue *in_serv_queue; ++ bool waiting_rq, idle_timer_disabled; ++ ++ spin_lock_irq(&bfqd->lock); ++ ++ in_serv_queue = bfqd->in_service_queue; ++ waiting_rq = in_serv_queue && bfq_bfqq_wait_request(in_serv_queue); ++ ++ rq = __bfq_dispatch_request(hctx); ++ ++ idle_timer_disabled = ++ waiting_rq && !bfq_bfqq_wait_request(in_serv_queue); ++ ++ spin_unlock_irq(&bfqd->lock); ++ ++ bfq_update_dispatch_stats(hctx->queue, rq, in_serv_queue, ++ idle_timer_disabled); + + return rq; + } +@@ -4881,6 +4890,38 @@ static bool __bfq_insert_request(struct bfq_data *bfqd, struct request *rq) + return idle_timer_disabled; + } + ++#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) ++static void bfq_update_insert_stats(struct request_queue *q, ++ struct bfq_queue *bfqq, ++ bool idle_timer_disabled, ++ unsigned int cmd_flags) ++{ ++ if (!bfqq) ++ return; ++ ++ /* ++ * bfqq still exists, because it can disappear only after ++ * either it is merged with another queue, or the process it ++ * is associated with exits. But both actions must be taken by ++ * the same process currently executing this flow of ++ * instructions. ++ * ++ * In addition, the following queue lock guarantees that ++ * bfqq_group(bfqq) exists as well. ++ */ ++ spin_lock_irq(q->queue_lock); ++ bfqg_stats_update_io_add(bfqq_group(bfqq), bfqq, cmd_flags); ++ if (idle_timer_disabled) ++ bfqg_stats_update_idle_time(bfqq_group(bfqq)); ++ spin_unlock_irq(q->queue_lock); ++} ++#else ++static inline void bfq_update_insert_stats(struct request_queue *q, ++ struct bfq_queue *bfqq, ++ bool idle_timer_disabled, ++ unsigned int cmd_flags) {} ++#endif ++ + static void bfq_prepare_request(struct request *rq, struct bio *bio); + + static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, +@@ -4889,10 +4930,8 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + struct request_queue *q = hctx->queue; + struct bfq_data *bfqd = q->elevator->elevator_data; + struct bfq_queue *bfqq = RQ_BFQQ(rq); +-#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + bool idle_timer_disabled = false; + unsigned int cmd_flags; +-#endif + + spin_lock_irq(&bfqd->lock); + if (blk_mq_sched_try_insert_merge(q, rq)) { +@@ -4938,7 +4977,6 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + bfqq = RQ_BFQQ(rq); + } + +-#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) + idle_timer_disabled = __bfq_insert_request(bfqd, rq); + /* + * Update bfqq, because, if a queue merge has occurred +@@ -4946,9 +4984,6 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + * redirected into a new queue. + */ + bfqq = RQ_BFQQ(rq); +-#else +- __bfq_insert_request(bfqd, rq); +-#endif + + if (rq_mergeable(rq)) { + elv_rqhash_add(q, rq); +@@ -4956,34 +4991,17 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, + q->last_merge = rq; + } + } +-#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) ++ + /* + * Cache cmd_flags before releasing scheduler lock, because rq + * may disappear afterwards (for example, because of a request + * merge). + */ + cmd_flags = rq->cmd_flags; +-#endif ++ + spin_unlock_irq(&bfqd->lock); +-#if defined(BFQ_GROUP_IOSCHED_ENABLED) && defined(CONFIG_DEBUG_BLK_CGROUP) +- if (!bfqq) +- return; +- /* +- * bfqq still exists, because it can disappear only after +- * either it is merged with another queue, or the process it +- * is associated with exits. But both actions must be taken by +- * the same process currently executing this flow of +- * instruction. +- * +- * In addition, the following queue lock guarantees that +- * bfqq_group(bfqq) exists as well. +- */ +- spin_lock_irq(q->queue_lock); +- bfqg_stats_update_io_add(bfqq_group(bfqq), bfqq, cmd_flags); +- if (idle_timer_disabled) +- bfqg_stats_update_idle_time(bfqq_group(bfqq)); +- spin_unlock_irq(q->queue_lock); +-#endif ++ bfq_update_insert_stats(q, bfqq, idle_timer_disabled, ++ cmd_flags); + } + + static void bfq_insert_requests(struct blk_mq_hw_ctx *hctx, + +From 0d0d05632872b226f4fae5e56af8736a4c24bf57 Mon Sep 17 00:00:00 2001 +From: Melzani Alessandro +Date: Mon, 26 Feb 2018 22:43:30 +0100 +Subject: [PATCH 21/23] bfq-sq, bfq-mq: port of "bfq: Use icq_to_bic() + consistently" + +Some code uses icq_to_bic() to convert an io_cq pointer to a +bfq_io_cq pointer while other code uses a direct cast. Convert +the code that uses a direct cast such that it uses icq_to_bic(). + +Signed-off-by: Alessandro Melzani +--- + block/bfq-mq-iosched.c | 2 +- + block/bfq-sq-iosched.c | 2 +- + 2 files changed, 2 insertions(+), 2 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 5a211620f316..7b1269558c47 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -272,7 +272,7 @@ static const unsigned long max_service_from_wr = 120000; + #define BFQ_SERVICE_TREE_INIT ((struct bfq_service_tree) \ + { RB_ROOT, RB_ROOT, NULL, NULL, 0, 0 }) + +-#define RQ_BIC(rq) ((struct bfq_io_cq *) (rq)->elv.priv[0]) ++#define RQ_BIC(rq) icq_to_bic((rq)->elv.priv[0]) + #define RQ_BFQQ(rq) ((rq)->elv.priv[1]) + + /** +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index f95deaab49a1..c4aff8d55fc4 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -266,7 +266,7 @@ static const unsigned long max_service_from_wr = 120000; + #define BFQ_SERVICE_TREE_INIT ((struct bfq_service_tree) \ + { RB_ROOT, RB_ROOT, NULL, NULL, 0, 0 }) + +-#define RQ_BIC(rq) ((struct bfq_io_cq *) (rq)->elv.priv[0]) ++#define RQ_BIC(rq) icq_to_bic((rq)->elv.priv[0]) + #define RQ_BFQQ(rq) ((rq)->elv.priv[1]) + + static void bfq_schedule_dispatch(struct bfq_data *bfqd); + +From 4cb5de6add7d6ad0d25d73cb95dc871305db1522 Mon Sep 17 00:00:00 2001 +From: Melzani Alessandro +Date: Mon, 26 Feb 2018 22:59:30 +0100 +Subject: [PATCH 22/23] bfq-sq, bfq-mq: port of "block, bfq: fix error handle + in bfq_init" + +if elv_register fail, bfq_pool should be free. + +Signed-off-by: Alessandro Melzani +--- + block/bfq-mq-iosched.c | 4 +++- + block/bfq-sq-iosched.c | 4 +++- + 2 files changed, 6 insertions(+), 2 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 7b1269558c47..964e88c2ce59 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -6129,7 +6129,7 @@ static int __init bfq_init(void) + + ret = elv_register(&iosched_bfq_mq); + if (ret) +- goto err_pol_unreg; ++ goto slab_kill; + + #ifdef BFQ_GROUP_IOSCHED_ENABLED + strcat(msg, " (with cgroups support)"); +@@ -6138,6 +6138,8 @@ static int __init bfq_init(void) + + return 0; + ++slab_kill: ++ bfq_slab_kill(); + err_pol_unreg: + #ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_policy_unregister(&blkcg_policy_bfq); +diff --git a/block/bfq-sq-iosched.c b/block/bfq-sq-iosched.c +index c4aff8d55fc4..7f0cf1f01ffc 100644 +--- a/block/bfq-sq-iosched.c ++++ b/block/bfq-sq-iosched.c +@@ -5590,7 +5590,7 @@ static int __init bfq_init(void) + + ret = elv_register(&iosched_bfq); + if (ret) +- goto err_pol_unreg; ++ goto slab_kill; + + #ifdef BFQ_GROUP_IOSCHED_ENABLED + strcat(msg, " (with cgroups support)"); +@@ -5599,6 +5599,8 @@ static int __init bfq_init(void) + + return 0; + ++slab_kill: ++ bfq_slab_kill(); + err_pol_unreg: + #ifdef BFQ_GROUP_IOSCHED_ENABLED + blkcg_policy_unregister(&blkcg_policy_bfq); + +From 1f77c173aaa87ffb22c9f062a6449245d14311e4 Mon Sep 17 00:00:00 2001 +From: Paolo Valente +Date: Wed, 4 Apr 2018 11:28:16 +0200 +Subject: [PATCH 23/23] block, bfq-sq, bfq-mq: lower-bound the estimated peak + rate to 1 + +If a storage device handled by BFQ happens to be slower than 7.5 KB/s +for a certain amount of time (in the order of a second), then the +estimated peak rate of the device, maintained in BFQ, becomes equal to +0. The reason is the limited precision with which the rate is +represented (details on the range of representable values in the +comments introduced by this commit). This leads to a division-by-zero +error where the estimated peak rate is used as divisor. Such a type of +failure has been reported in [1]. + +This commit addresses this issue by: +1. Lower-bounding the estimated peak rate to 1 +2. Adding and improving comments on the range of rates representable + +[1] https://www.spinics.net/lists/kernel/msg2739205.html + +Signed-off-by: Konstantin Khlebnikov +Signed-off-by: Paolo Valente +--- + block/bfq-mq-iosched.c | 25 ++++++++++++++++++++++++- + block/bfq-mq.h | 7 ++++++- + block/bfq-sq-iosched.c | 25 ++++++++++++++++++++++++- + block/bfq.h | 7 ++++++- + 4 files changed, 60 insertions(+), 4 deletions(-) + +diff --git a/block/bfq-mq-iosched.c b/block/bfq-mq-iosched.c +index 964e88c2ce59..03efd90c5d20 100644 +--- a/block/bfq-mq-iosched.c ++++ b/block/bfq-mq-iosched.c +@@ -160,7 +160,20 @@ static struct kmem_cache *bfq_pool; + /* Target observation time interval for a peak-rate update (ns) */ + #define BFQ_RATE_REF_INTERVAL NSEC_PER_SEC + +-/* Shift used for peak rate fixed precision calculations. */ ++/* ++ * Shift used for peak-rate fixed precision calculations. ++ * With ++ * - the current shift: 16 positions ++ * - the current type used to store rate: u32 ++ * - the current unit of measure for rate: [sectors/usec], or, more precisely, ++ * [(sectors/usec) / 2^BFQ_RATE_SHIFT] to take into account the shift, ++ * the range of rates that can be stored is ++ * [1 / 2^BFQ_RATE_SHIFT, 2^(32 - BFQ_RATE_SHIFT)] sectors/usec = ++ * [1 / 2^16, 2^16] sectors/usec = [15e-6, 65536] sectors/usec = ++ * [15, 65G] sectors/sec ++ * Which, assuming a sector size of 512B, corresponds to a range of ++ * [7.5K, 33T] B/sec ++ */ + #define BFQ_RATE_SHIFT 16 + + /* +@@ -2881,6 +2894,16 @@ static void bfq_update_rate_reset(struct bfq_data *bfqd, struct request *rq) + BUG_ON(bfqd->peak_rate > 20<peak_rate += rate; ++ ++ /* ++ * For a very slow device, bfqd->peak_rate can reach 0 (see ++ * the minimum representable values reported in the comments ++ * on BFQ_RATE_SHIFT). Push to 1 if this happens, to avoid ++ * divisions by zero where bfqd->peak_rate is used as a ++ * divisor. ++ */ ++ bfqd->peak_rate = max_t(u32, 1, bfqd->peak_rate); ++ + update_thr_responsiveness_params(bfqd); + BUG_ON(bfqd->peak_rate > 20<peak_rate > 20<peak_rate += rate; ++ ++ /* ++ * For a very slow device, bfqd->peak_rate can reach 0 (see ++ * the minimum representable values reported in the comments ++ * on BFQ_RATE_SHIFT). Push to 1 if this happens, to avoid ++ * divisions by zero where bfqd->peak_rate is used as a ++ * divisor. ++ */ ++ bfqd->peak_rate = max_t(u32, 1, bfqd->peak_rate); ++ + update_thr_responsiveness_params(bfqd); + BUG_ON(bfqd->peak_rate > 20< +Date: Sat, 29 Oct 2016 11:20:37 +1100 +Subject: [PATCH 02/16] Make preemptible kernel default. + +Make full preempt default on all arches. +--- + arch/arc/configs/tb10x_defconfig | 2 +- + arch/arm/configs/bcm2835_defconfig | 2 +- + arch/arm/configs/imx_v6_v7_defconfig | 2 +- + arch/arm/configs/mps2_defconfig | 2 +- + arch/arm/configs/mxs_defconfig | 2 +- + arch/blackfin/configs/BF518F-EZBRD_defconfig | 2 +- + arch/blackfin/configs/BF526-EZBRD_defconfig | 2 +- + arch/blackfin/configs/BF527-EZKIT-V2_defconfig | 2 +- + arch/blackfin/configs/BF527-EZKIT_defconfig | 2 +- + arch/blackfin/configs/BF527-TLL6527M_defconfig | 2 +- + arch/blackfin/configs/BF533-EZKIT_defconfig | 2 +- + arch/blackfin/configs/BF533-STAMP_defconfig | 2 +- + arch/blackfin/configs/BF537-STAMP_defconfig | 2 +- + arch/blackfin/configs/BF538-EZKIT_defconfig | 2 +- + arch/blackfin/configs/BF548-EZKIT_defconfig | 2 +- + arch/blackfin/configs/BF561-ACVILON_defconfig | 2 +- + arch/blackfin/configs/BF561-EZKIT-SMP_defconfig | 2 +- + arch/blackfin/configs/BF561-EZKIT_defconfig | 2 +- + arch/blackfin/configs/BF609-EZKIT_defconfig | 2 +- + arch/blackfin/configs/BlackStamp_defconfig | 2 +- + arch/blackfin/configs/CM-BF527_defconfig | 2 +- + arch/blackfin/configs/PNAV-10_defconfig | 2 +- + arch/blackfin/configs/SRV1_defconfig | 2 +- + arch/blackfin/configs/TCM-BF518_defconfig | 2 +- + arch/mips/configs/fuloong2e_defconfig | 3 ++- + arch/mips/configs/gpr_defconfig | 3 ++- + arch/mips/configs/ip22_defconfig | 3 ++- + arch/mips/configs/ip28_defconfig | 3 ++- + arch/mips/configs/jazz_defconfig | 3 ++- + arch/mips/configs/mtx1_defconfig | 3 ++- + arch/mips/configs/nlm_xlr_defconfig | 2 +- + arch/mips/configs/pic32mzda_defconfig | 2 +- + arch/mips/configs/pistachio_defconfig | 2 +- + arch/mips/configs/pnx8335_stb225_defconfig | 2 +- + arch/mips/configs/rm200_defconfig | 3 ++- + arch/parisc/configs/712_defconfig | 2 +- + arch/parisc/configs/c3000_defconfig | 2 +- + arch/parisc/configs/default_defconfig | 2 +- + arch/powerpc/configs/c2k_defconfig | 2 +- + arch/powerpc/configs/ppc6xx_defconfig | 2 +- + arch/score/configs/spct6600_defconfig | 2 +- + arch/sh/configs/se7712_defconfig | 2 +- + arch/sh/configs/se7721_defconfig | 2 +- + arch/sh/configs/titan_defconfig | 2 +- + arch/sparc/configs/sparc64_defconfig | 2 +- + arch/tile/configs/tilegx_defconfig | 2 +- + arch/tile/configs/tilepro_defconfig | 2 +- + arch/x86/configs/i386_defconfig | 2 +- + arch/x86/configs/x86_64_defconfig | 2 +- + kernel/Kconfig.preempt | 7 ++++--- + 50 files changed, 60 insertions(+), 52 deletions(-) + +diff --git a/arch/arc/configs/tb10x_defconfig b/arch/arc/configs/tb10x_defconfig +index f30182549395..42910f628869 100644 +--- a/arch/arc/configs/tb10x_defconfig ++++ b/arch/arc/configs/tb10x_defconfig +@@ -28,7 +28,7 @@ CONFIG_ARC_PLAT_TB10X=y + CONFIG_ARC_CACHE_LINE_SHIFT=5 + CONFIG_HZ=250 + CONFIG_ARC_BUILTIN_DTB_NAME="abilis_tb100_dvk" +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + # CONFIG_COMPACTION is not set + CONFIG_NET=y + CONFIG_PACKET=y +diff --git a/arch/arm/configs/bcm2835_defconfig b/arch/arm/configs/bcm2835_defconfig +index 43dab4890ad3..44a52166ca5e 100644 +--- a/arch/arm/configs/bcm2835_defconfig ++++ b/arch/arm/configs/bcm2835_defconfig +@@ -29,7 +29,7 @@ CONFIG_MODULE_UNLOAD=y + CONFIG_ARCH_MULTI_V6=y + CONFIG_ARCH_BCM=y + CONFIG_ARCH_BCM2835=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_AEABI=y + CONFIG_KSM=y + CONFIG_CLEANCACHE=y +diff --git a/arch/arm/configs/imx_v6_v7_defconfig b/arch/arm/configs/imx_v6_v7_defconfig +index 32acac9ab81a..1482bb312987 100644 +--- a/arch/arm/configs/imx_v6_v7_defconfig ++++ b/arch/arm/configs/imx_v6_v7_defconfig +@@ -47,7 +47,7 @@ CONFIG_PCI_MSI=y + CONFIG_PCI_IMX6=y + CONFIG_SMP=y + CONFIG_ARM_PSCI=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_AEABI=y + CONFIG_HIGHMEM=y + CONFIG_CMA=y +diff --git a/arch/arm/configs/mps2_defconfig b/arch/arm/configs/mps2_defconfig +index 0bcdec7cc169..10ceaefa51e0 100644 +--- a/arch/arm/configs/mps2_defconfig ++++ b/arch/arm/configs/mps2_defconfig +@@ -18,7 +18,7 @@ CONFIG_ARCH_MPS2=y + CONFIG_SET_MEM_PARAM=y + CONFIG_DRAM_BASE=0x21000000 + CONFIG_DRAM_SIZE=0x1000000 +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + # CONFIG_ATAGS is not set + CONFIG_ZBOOT_ROM_TEXT=0x0 + CONFIG_ZBOOT_ROM_BSS=0x0 +diff --git a/arch/arm/configs/mxs_defconfig b/arch/arm/configs/mxs_defconfig +index e5822ab01b7d..3e77e02f678f 100644 +--- a/arch/arm/configs/mxs_defconfig ++++ b/arch/arm/configs/mxs_defconfig +@@ -27,7 +27,7 @@ CONFIG_BLK_DEV_INTEGRITY=y + # CONFIG_ARCH_MULTI_V7 is not set + CONFIG_ARCH_MXS=y + # CONFIG_ARM_THUMB is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_AEABI=y + CONFIG_NET=y + CONFIG_PACKET=y +diff --git a/arch/blackfin/configs/BF518F-EZBRD_defconfig b/arch/blackfin/configs/BF518F-EZBRD_defconfig +index 99c00d835f47..39b91dfa55b5 100644 +--- a/arch/blackfin/configs/BF518F-EZBRD_defconfig ++++ b/arch/blackfin/configs/BF518F-EZBRD_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF518=y + CONFIG_IRQ_TIMER0=12 + # CONFIG_CYCLES_CLOCKSOURCE is not set +diff --git a/arch/blackfin/configs/BF526-EZBRD_defconfig b/arch/blackfin/configs/BF526-EZBRD_defconfig +index e66ba31ef84d..675cadb3a0c4 100644 +--- a/arch/blackfin/configs/BF526-EZBRD_defconfig ++++ b/arch/blackfin/configs/BF526-EZBRD_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF526=y + CONFIG_IRQ_TIMER0=12 + CONFIG_BFIN526_EZBRD=y +diff --git a/arch/blackfin/configs/BF527-EZKIT-V2_defconfig b/arch/blackfin/configs/BF527-EZKIT-V2_defconfig +index 0207c588c19f..4c517c443af5 100644 +--- a/arch/blackfin/configs/BF527-EZKIT-V2_defconfig ++++ b/arch/blackfin/configs/BF527-EZKIT-V2_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF527=y + CONFIG_BF_REV_0_2=y + CONFIG_BFIN527_EZKIT_V2=y +diff --git a/arch/blackfin/configs/BF527-EZKIT_defconfig b/arch/blackfin/configs/BF527-EZKIT_defconfig +index 99c131ba7d90..bf8df3e6cf02 100644 +--- a/arch/blackfin/configs/BF527-EZKIT_defconfig ++++ b/arch/blackfin/configs/BF527-EZKIT_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF527=y + CONFIG_BF_REV_0_1=y + CONFIG_IRQ_USB_INT0=11 +diff --git a/arch/blackfin/configs/BF527-TLL6527M_defconfig b/arch/blackfin/configs/BF527-TLL6527M_defconfig +index cdeb51856f26..0220b3b15c53 100644 +--- a/arch/blackfin/configs/BF527-TLL6527M_defconfig ++++ b/arch/blackfin/configs/BF527-TLL6527M_defconfig +@@ -21,7 +21,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_LBDAF is not set + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF527=y + CONFIG_BF_REV_0_2=y + CONFIG_BFIN527_TLL6527M=y +diff --git a/arch/blackfin/configs/BF533-EZKIT_defconfig b/arch/blackfin/configs/BF533-EZKIT_defconfig +index ed7d2c096739..6023e3fd2c48 100644 +--- a/arch/blackfin/configs/BF533-EZKIT_defconfig ++++ b/arch/blackfin/configs/BF533-EZKIT_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BFIN533_EZKIT=y + CONFIG_TIMER0=11 + CONFIG_CLKIN_HZ=27000000 +diff --git a/arch/blackfin/configs/BF533-STAMP_defconfig b/arch/blackfin/configs/BF533-STAMP_defconfig +index 0c241f4d28d7..f5cd0f18b711 100644 +--- a/arch/blackfin/configs/BF533-STAMP_defconfig ++++ b/arch/blackfin/configs/BF533-STAMP_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_TIMER0=11 + CONFIG_HIGH_RES_TIMERS=y + CONFIG_NOMMU_INITIAL_TRIM_EXCESS=0 +diff --git a/arch/blackfin/configs/BF537-STAMP_defconfig b/arch/blackfin/configs/BF537-STAMP_defconfig +index e5360b30e39a..48085fde7f9e 100644 +--- a/arch/blackfin/configs/BF537-STAMP_defconfig ++++ b/arch/blackfin/configs/BF537-STAMP_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF537=y + CONFIG_HIGH_RES_TIMERS=y + CONFIG_NOMMU_INITIAL_TRIM_EXCESS=0 +diff --git a/arch/blackfin/configs/BF538-EZKIT_defconfig b/arch/blackfin/configs/BF538-EZKIT_defconfig +index 60f6fb86125c..12deeaaef3cb 100644 +--- a/arch/blackfin/configs/BF538-EZKIT_defconfig ++++ b/arch/blackfin/configs/BF538-EZKIT_defconfig +@@ -21,7 +21,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF538=y + CONFIG_IRQ_TIMER0=12 + CONFIG_IRQ_TIMER1=12 +diff --git a/arch/blackfin/configs/BF548-EZKIT_defconfig b/arch/blackfin/configs/BF548-EZKIT_defconfig +index 38cb17d218d4..6a68ffc55b5a 100644 +--- a/arch/blackfin/configs/BF548-EZKIT_defconfig ++++ b/arch/blackfin/configs/BF548-EZKIT_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF548_std=y + CONFIG_IRQ_TIMER0=11 + # CONFIG_CYCLES_CLOCKSOURCE is not set +diff --git a/arch/blackfin/configs/BF561-ACVILON_defconfig b/arch/blackfin/configs/BF561-ACVILON_defconfig +index 78f6bc79f910..e9f3ba783a4e 100644 +--- a/arch/blackfin/configs/BF561-ACVILON_defconfig ++++ b/arch/blackfin/configs/BF561-ACVILON_defconfig +@@ -20,7 +20,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_LBDAF is not set + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF561=y + CONFIG_BF_REV_0_5=y + CONFIG_IRQ_TIMER0=10 +diff --git a/arch/blackfin/configs/BF561-EZKIT-SMP_defconfig b/arch/blackfin/configs/BF561-EZKIT-SMP_defconfig +index fac8bb578249..89b75a6c3fab 100644 +--- a/arch/blackfin/configs/BF561-EZKIT-SMP_defconfig ++++ b/arch/blackfin/configs/BF561-EZKIT-SMP_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF561=y + CONFIG_SMP=y + CONFIG_IRQ_TIMER0=10 +diff --git a/arch/blackfin/configs/BF561-EZKIT_defconfig b/arch/blackfin/configs/BF561-EZKIT_defconfig +index 2a2e4d0cebc1..67b3d2f419ba 100644 +--- a/arch/blackfin/configs/BF561-EZKIT_defconfig ++++ b/arch/blackfin/configs/BF561-EZKIT_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF561=y + CONFIG_IRQ_TIMER0=10 + CONFIG_CLKIN_HZ=30000000 +diff --git a/arch/blackfin/configs/BF609-EZKIT_defconfig b/arch/blackfin/configs/BF609-EZKIT_defconfig +index 3ce77f07208a..8cc75d4218fb 100644 +--- a/arch/blackfin/configs/BF609-EZKIT_defconfig ++++ b/arch/blackfin/configs/BF609-EZKIT_defconfig +@@ -20,7 +20,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF609=y + CONFIG_PINT1_ASSIGN=0x01010000 + CONFIG_PINT2_ASSIGN=0x07000101 +diff --git a/arch/blackfin/configs/BlackStamp_defconfig b/arch/blackfin/configs/BlackStamp_defconfig +index f4a9200e1ab1..9faf0ec7007f 100644 +--- a/arch/blackfin/configs/BlackStamp_defconfig ++++ b/arch/blackfin/configs/BlackStamp_defconfig +@@ -17,7 +17,7 @@ CONFIG_MODULE_UNLOAD=y + CONFIG_MODULE_FORCE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF532=y + CONFIG_BF_REV_0_5=y + CONFIG_BLACKSTAMP=y +diff --git a/arch/blackfin/configs/CM-BF527_defconfig b/arch/blackfin/configs/CM-BF527_defconfig +index 1902bb05d086..4a1ad4fd7bb2 100644 +--- a/arch/blackfin/configs/CM-BF527_defconfig ++++ b/arch/blackfin/configs/CM-BF527_defconfig +@@ -19,7 +19,7 @@ CONFIG_MODULES=y + CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF527=y + CONFIG_BF_REV_0_1=y + CONFIG_IRQ_TIMER0=12 +diff --git a/arch/blackfin/configs/PNAV-10_defconfig b/arch/blackfin/configs/PNAV-10_defconfig +index c7926812971c..9d787e28bbe8 100644 +--- a/arch/blackfin/configs/PNAV-10_defconfig ++++ b/arch/blackfin/configs/PNAV-10_defconfig +@@ -15,7 +15,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF537=y + CONFIG_IRQ_TIMER0=12 + CONFIG_PNAV10=y +diff --git a/arch/blackfin/configs/SRV1_defconfig b/arch/blackfin/configs/SRV1_defconfig +index 23fdc57d657a..225df32dc9a8 100644 +--- a/arch/blackfin/configs/SRV1_defconfig ++++ b/arch/blackfin/configs/SRV1_defconfig +@@ -13,7 +13,7 @@ CONFIG_MMAP_ALLOW_UNINITIALIZED=y + CONFIG_MODULES=y + CONFIG_MODULE_UNLOAD=y + # CONFIG_IOSCHED_DEADLINE is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF537=y + CONFIG_IRQ_TIMER0=12 + CONFIG_BOOT_LOAD=0x400000 +diff --git a/arch/blackfin/configs/TCM-BF518_defconfig b/arch/blackfin/configs/TCM-BF518_defconfig +index e28959479fe0..425c24e43c34 100644 +--- a/arch/blackfin/configs/TCM-BF518_defconfig ++++ b/arch/blackfin/configs/TCM-BF518_defconfig +@@ -23,7 +23,7 @@ CONFIG_MODULE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + # CONFIG_IOSCHED_DEADLINE is not set + # CONFIG_IOSCHED_CFQ is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BF518=y + CONFIG_BF_REV_0_1=y + CONFIG_BFIN518F_TCM=y +diff --git a/arch/mips/configs/fuloong2e_defconfig b/arch/mips/configs/fuloong2e_defconfig +index 499f51498ecb..f7cb39b0662c 100644 +--- a/arch/mips/configs/fuloong2e_defconfig ++++ b/arch/mips/configs/fuloong2e_defconfig +@@ -2,7 +2,8 @@ CONFIG_MACH_LOONGSON64=y + CONFIG_64BIT=y + CONFIG_NO_HZ=y + CONFIG_HIGH_RES_TIMERS=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y ++CONFIG_EXPERIMENTAL=y + CONFIG_LOCALVERSION="-fuloong2e" + # CONFIG_LOCALVERSION_AUTO is not set + CONFIG_SYSVIPC=y +diff --git a/arch/mips/configs/gpr_defconfig b/arch/mips/configs/gpr_defconfig +index 55438fc9991e..db03ef4f737d 100644 +--- a/arch/mips/configs/gpr_defconfig ++++ b/arch/mips/configs/gpr_defconfig +@@ -1,7 +1,8 @@ + CONFIG_MIPS_ALCHEMY=y + CONFIG_MIPS_GPR=y + CONFIG_HIGH_RES_TIMERS=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y ++CONFIG_EXPERIMENTAL=y + # CONFIG_LOCALVERSION_AUTO is not set + CONFIG_SYSVIPC=y + CONFIG_POSIX_MQUEUE=y +diff --git a/arch/mips/configs/ip22_defconfig b/arch/mips/configs/ip22_defconfig +index 83e8fe2064aa..93e7b167433b 100644 +--- a/arch/mips/configs/ip22_defconfig ++++ b/arch/mips/configs/ip22_defconfig +@@ -3,7 +3,8 @@ CONFIG_CPU_R5000=y + CONFIG_NO_HZ=y + CONFIG_HIGH_RES_TIMERS=y + CONFIG_HZ_1000=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y ++CONFIG_EXPERIMENTAL=y + CONFIG_SYSVIPC=y + CONFIG_IKCONFIG=y + CONFIG_IKCONFIG_PROC=y +diff --git a/arch/mips/configs/ip28_defconfig b/arch/mips/configs/ip28_defconfig +index d0a4c2cfacf8..6f0600e99c25 100644 +--- a/arch/mips/configs/ip28_defconfig ++++ b/arch/mips/configs/ip28_defconfig +@@ -1,6 +1,7 @@ + CONFIG_SGI_IP28=y + CONFIG_ARC_CONSOLE=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y ++CONFIG_EXPERIMENTAL=y + CONFIG_SYSVIPC=y + CONFIG_IKCONFIG=y + CONFIG_IKCONFIG_PROC=y +diff --git a/arch/mips/configs/jazz_defconfig b/arch/mips/configs/jazz_defconfig +index 9ad1c94376c8..1d62ce7ff5dc 100644 +--- a/arch/mips/configs/jazz_defconfig ++++ b/arch/mips/configs/jazz_defconfig +@@ -1,6 +1,7 @@ + CONFIG_MACH_JAZZ=y + CONFIG_OLIVETTI_M700=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y ++CONFIG_EXPERIMENTAL=y + CONFIG_SYSVIPC=y + CONFIG_POSIX_MQUEUE=y + CONFIG_BSD_PROCESS_ACCT=y +diff --git a/arch/mips/configs/mtx1_defconfig b/arch/mips/configs/mtx1_defconfig +index c3d0d0a6e044..aa3426d5f7d7 100644 +--- a/arch/mips/configs/mtx1_defconfig ++++ b/arch/mips/configs/mtx1_defconfig +@@ -1,6 +1,7 @@ + CONFIG_MIPS_ALCHEMY=y + CONFIG_MIPS_MTX1=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y ++CONFIG_EXPERIMENTAL=y + # CONFIG_LOCALVERSION_AUTO is not set + CONFIG_SYSVIPC=y + CONFIG_POSIX_MQUEUE=y +diff --git a/arch/mips/configs/nlm_xlr_defconfig b/arch/mips/configs/nlm_xlr_defconfig +index 1e18fd7de209..b514e91e5426 100644 +--- a/arch/mips/configs/nlm_xlr_defconfig ++++ b/arch/mips/configs/nlm_xlr_defconfig +@@ -5,7 +5,7 @@ CONFIG_DEFAULT_MMAP_MIN_ADDR=65536 + CONFIG_SMP=y + CONFIG_NO_HZ=y + CONFIG_HIGH_RES_TIMERS=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_KEXEC=y + CONFIG_CROSS_COMPILE="" + # CONFIG_LOCALVERSION_AUTO is not set +diff --git a/arch/mips/configs/pic32mzda_defconfig b/arch/mips/configs/pic32mzda_defconfig +index 52192c632ae8..96b087498dab 100644 +--- a/arch/mips/configs/pic32mzda_defconfig ++++ b/arch/mips/configs/pic32mzda_defconfig +@@ -1,7 +1,7 @@ + CONFIG_MACH_PIC32=y + CONFIG_DTB_PIC32_MZDA_SK=y + CONFIG_HZ_100=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + # CONFIG_SECCOMP is not set + CONFIG_SYSVIPC=y + CONFIG_NO_HZ=y +diff --git a/arch/mips/configs/pistachio_defconfig b/arch/mips/configs/pistachio_defconfig +index b22a3cf149b6..cfffca3d37f4 100644 +--- a/arch/mips/configs/pistachio_defconfig ++++ b/arch/mips/configs/pistachio_defconfig +@@ -5,7 +5,7 @@ CONFIG_MIPS_CPS=y + CONFIG_DEFAULT_MMAP_MIN_ADDR=32768 + CONFIG_ZSMALLOC=y + CONFIG_NR_CPUS=4 +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + # CONFIG_LOCALVERSION_AUTO is not set + CONFIG_DEFAULT_HOSTNAME="localhost" + CONFIG_SYSVIPC=y +diff --git a/arch/mips/configs/pnx8335_stb225_defconfig b/arch/mips/configs/pnx8335_stb225_defconfig +index 81b5eb89446c..19f8cea849a1 100644 +--- a/arch/mips/configs/pnx8335_stb225_defconfig ++++ b/arch/mips/configs/pnx8335_stb225_defconfig +@@ -3,7 +3,7 @@ CONFIG_CPU_LITTLE_ENDIAN=y + CONFIG_NO_HZ=y + CONFIG_HIGH_RES_TIMERS=y + CONFIG_HZ_128=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + # CONFIG_SECCOMP is not set + # CONFIG_LOCALVERSION_AUTO is not set + # CONFIG_SWAP is not set +diff --git a/arch/mips/configs/rm200_defconfig b/arch/mips/configs/rm200_defconfig +index 99679e514042..2ced507a8ba7 100644 +--- a/arch/mips/configs/rm200_defconfig ++++ b/arch/mips/configs/rm200_defconfig +@@ -2,7 +2,8 @@ CONFIG_SNI_RM=y + CONFIG_CPU_LITTLE_ENDIAN=y + CONFIG_ARC_CONSOLE=y + CONFIG_HZ_1000=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y ++CONFIG_EXPERIMENTAL=y + CONFIG_SYSVIPC=y + CONFIG_POSIX_MQUEUE=y + CONFIG_BSD_PROCESS_ACCT=y +diff --git a/arch/parisc/configs/712_defconfig b/arch/parisc/configs/712_defconfig +index ccc109761f44..a6a5b0b7a9c9 100644 +--- a/arch/parisc/configs/712_defconfig ++++ b/arch/parisc/configs/712_defconfig +@@ -13,7 +13,7 @@ CONFIG_MODULES=y + CONFIG_MODULE_UNLOAD=y + CONFIG_MODULE_FORCE_UNLOAD=y + CONFIG_PA7100LC=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_GSC_LASI=y + # CONFIG_PDC_CHASSIS is not set + CONFIG_BINFMT_MISC=m +diff --git a/arch/parisc/configs/c3000_defconfig b/arch/parisc/configs/c3000_defconfig +index 8d41a73bd71b..b8e0a6662ff9 100644 +--- a/arch/parisc/configs/c3000_defconfig ++++ b/arch/parisc/configs/c3000_defconfig +@@ -13,7 +13,7 @@ CONFIG_MODULES=y + CONFIG_MODULE_UNLOAD=y + CONFIG_MODULE_FORCE_UNLOAD=y + CONFIG_PA8X00=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + # CONFIG_GSC is not set + CONFIG_PCI=y + CONFIG_PCI_LBA=y +diff --git a/arch/parisc/configs/default_defconfig b/arch/parisc/configs/default_defconfig +index 52c9050a7c5c..8d86d2e989f4 100644 +--- a/arch/parisc/configs/default_defconfig ++++ b/arch/parisc/configs/default_defconfig +@@ -14,7 +14,7 @@ CONFIG_MODULE_UNLOAD=y + CONFIG_MODULE_FORCE_UNLOAD=y + # CONFIG_BLK_DEV_BSG is not set + CONFIG_PA7100LC=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_IOMMU_CCIO=y + CONFIG_GSC_LASI=y + CONFIG_GSC_WAX=y +diff --git a/arch/powerpc/configs/c2k_defconfig b/arch/powerpc/configs/c2k_defconfig +index f1552af9eecc..f8505e6ec7b3 100644 +--- a/arch/powerpc/configs/c2k_defconfig ++++ b/arch/powerpc/configs/c2k_defconfig +@@ -29,7 +29,7 @@ CONFIG_CPU_FREQ_GOV_POWERSAVE=m + CONFIG_CPU_FREQ_GOV_ONDEMAND=m + CONFIG_GEN_RTC=y + CONFIG_HIGHMEM=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BINFMT_MISC=y + CONFIG_PM=y + CONFIG_PCI_MSI=y +diff --git a/arch/powerpc/configs/ppc6xx_defconfig b/arch/powerpc/configs/ppc6xx_defconfig +index da0e8d535eb8..c016af41ab4f 100644 +--- a/arch/powerpc/configs/ppc6xx_defconfig ++++ b/arch/powerpc/configs/ppc6xx_defconfig +@@ -74,7 +74,7 @@ CONFIG_QE_GPIO=y + CONFIG_MCU_MPC8349EMITX=y + CONFIG_HIGHMEM=y + CONFIG_HZ_1000=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_BINFMT_MISC=y + CONFIG_HIBERNATION=y + CONFIG_PM_DEBUG=y +diff --git a/arch/score/configs/spct6600_defconfig b/arch/score/configs/spct6600_defconfig +index b2d8802f43b4..46434ca1fa10 100644 +--- a/arch/score/configs/spct6600_defconfig ++++ b/arch/score/configs/spct6600_defconfig +@@ -1,5 +1,5 @@ + CONFIG_HZ_100=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_EXPERIMENTAL=y + # CONFIG_LOCALVERSION_AUTO is not set + CONFIG_SYSVIPC=y +diff --git a/arch/sh/configs/se7712_defconfig b/arch/sh/configs/se7712_defconfig +index 5a1097641247..eb5fbf554e7f 100644 +--- a/arch/sh/configs/se7712_defconfig ++++ b/arch/sh/configs/se7712_defconfig +@@ -23,7 +23,7 @@ CONFIG_FLATMEM_MANUAL=y + CONFIG_SH_SOLUTION_ENGINE=y + CONFIG_SH_PCLK_FREQ=66666666 + CONFIG_HEARTBEAT=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_CMDLINE_OVERWRITE=y + CONFIG_CMDLINE="console=ttySC0,115200 root=/dev/sda1" + CONFIG_NET=y +diff --git a/arch/sh/configs/se7721_defconfig b/arch/sh/configs/se7721_defconfig +index 9c0ef13bee10..cbaa65c8bf9e 100644 +--- a/arch/sh/configs/se7721_defconfig ++++ b/arch/sh/configs/se7721_defconfig +@@ -23,7 +23,7 @@ CONFIG_FLATMEM_MANUAL=y + CONFIG_SH_7721_SOLUTION_ENGINE=y + CONFIG_SH_PCLK_FREQ=33333333 + CONFIG_HEARTBEAT=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_CMDLINE_OVERWRITE=y + CONFIG_CMDLINE="console=ttySC0,115200 root=/dev/sda2" + CONFIG_NET=y +diff --git a/arch/sh/configs/titan_defconfig b/arch/sh/configs/titan_defconfig +index ceb48e9b70f4..1a69eda6610c 100644 +--- a/arch/sh/configs/titan_defconfig ++++ b/arch/sh/configs/titan_defconfig +@@ -20,7 +20,7 @@ CONFIG_SH_TITAN=y + CONFIG_SH_PCLK_FREQ=30000000 + CONFIG_SH_DMA=y + CONFIG_SH_DMA_API=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_CMDLINE_OVERWRITE=y + CONFIG_CMDLINE="console=ttySC1,38400N81 root=/dev/nfs ip=:::::eth1:autoconf rw" + CONFIG_PCI=y +diff --git a/arch/sparc/configs/sparc64_defconfig b/arch/sparc/configs/sparc64_defconfig +index 4d4e1cc6402f..04bea1d28ba7 100644 +--- a/arch/sparc/configs/sparc64_defconfig ++++ b/arch/sparc/configs/sparc64_defconfig +@@ -22,7 +22,7 @@ CONFIG_NO_HZ=y + CONFIG_HIGH_RES_TIMERS=y + CONFIG_NUMA=y + CONFIG_DEFAULT_MMAP_MIN_ADDR=8192 +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_SUN_LDOMS=y + CONFIG_PCI=y + CONFIG_PCI_MSI=y +diff --git a/arch/tile/configs/tilegx_defconfig b/arch/tile/configs/tilegx_defconfig +index 9f94435cc44f..aa78ee6cd5eb 100644 +--- a/arch/tile/configs/tilegx_defconfig ++++ b/arch/tile/configs/tilegx_defconfig +@@ -47,7 +47,7 @@ CONFIG_CFQ_GROUP_IOSCHED=y + CONFIG_NR_CPUS=100 + CONFIG_HZ_100=y + # CONFIG_COMPACTION is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_TILE_PCI_IO=y + CONFIG_PCI_DEBUG=y + # CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS is not set +diff --git a/arch/tile/configs/tilepro_defconfig b/arch/tile/configs/tilepro_defconfig +index 1c5bd4f8ffca..38005862062c 100644 +--- a/arch/tile/configs/tilepro_defconfig ++++ b/arch/tile/configs/tilepro_defconfig +@@ -44,7 +44,7 @@ CONFIG_KARMA_PARTITION=y + CONFIG_CFQ_GROUP_IOSCHED=y + CONFIG_HZ_100=y + # CONFIG_COMPACTION is not set +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_PCI_DEBUG=y + # CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS is not set + CONFIG_BINFMT_MISC=y +diff --git a/arch/x86/configs/i386_defconfig b/arch/x86/configs/i386_defconfig +index 0eb9f92f3717..e5890ae917e5 100644 +--- a/arch/x86/configs/i386_defconfig ++++ b/arch/x86/configs/i386_defconfig +@@ -41,7 +41,7 @@ CONFIG_SMP=y + CONFIG_X86_GENERIC=y + CONFIG_HPET_TIMER=y + CONFIG_SCHED_SMT=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS=y + CONFIG_X86_MCE=y + CONFIG_X86_REBOOTFIXUPS=y +diff --git a/arch/x86/configs/x86_64_defconfig b/arch/x86/configs/x86_64_defconfig +index 4a4b16e56d35..7452dcadda74 100644 +--- a/arch/x86/configs/x86_64_defconfig ++++ b/arch/x86/configs/x86_64_defconfig +@@ -40,7 +40,7 @@ CONFIG_SMP=y + CONFIG_CALGARY_IOMMU=y + CONFIG_NR_CPUS=64 + CONFIG_SCHED_SMT=y +-CONFIG_PREEMPT_VOLUNTARY=y ++CONFIG_PREEMPT=y + CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS=y + CONFIG_X86_MCE=y + CONFIG_MICROCODE=y +diff --git a/kernel/Kconfig.preempt b/kernel/Kconfig.preempt +index 3f9c97419f02..1dc79ec7ad09 100644 +--- a/kernel/Kconfig.preempt ++++ b/kernel/Kconfig.preempt +@@ -1,7 +1,7 @@ + + choice + prompt "Preemption Model" +- default PREEMPT_NONE ++ default PREEMPT + + config PREEMPT_NONE + bool "No Forced Preemption (Server)" +@@ -17,7 +17,7 @@ config PREEMPT_NONE + latencies. + + config PREEMPT_VOLUNTARY +- bool "Voluntary Kernel Preemption (Desktop)" ++ bool "Voluntary Kernel Preemption (Nothing)" + help + This option reduces the latency of the kernel by adding more + "explicit preemption points" to the kernel code. These new +@@ -31,7 +31,8 @@ config PREEMPT_VOLUNTARY + applications to run more 'smoothly' even when the system is + under load. + +- Select this if you are building a kernel for a desktop system. ++ Select this for no system in particular (choose Preemptible ++ instead on a desktop if you know what's good for you). + + config PREEMPT + bool "Preemptible Kernel (Low-Latency Desktop)" +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0003-Expose-vmsplit-for-our-poor-32-bit-users.patch b/sys-kernel/linux-sources-redcore-lts/files/0003-Expose-vmsplit-for-our-poor-32-bit-users.patch new file mode 100644 index 00000000..b7897dbe --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0003-Expose-vmsplit-for-our-poor-32-bit-users.patch @@ -0,0 +1,48 @@ +From 44fc740a3ff85d378c28a416a076cc7e019d7b8c Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Fri, 12 May 2017 13:07:37 +1000 +Subject: [PATCH 03/16] Expose vmsplit for our poor 32 bit users. + +--- + arch/x86/Kconfig | 12 ++++++------ + 1 file changed, 6 insertions(+), 6 deletions(-) + +diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig +index e06a7b4e1dc4..931aba4fc567 100644 +--- a/arch/x86/Kconfig ++++ b/arch/x86/Kconfig +@@ -1361,7 +1361,7 @@ config HIGHMEM64G + endchoice + + choice +- prompt "Memory split" if EXPERT ++ prompt "Memory split" + default VMSPLIT_3G + depends on X86_32 + ---help--- +@@ -1381,17 +1381,17 @@ choice + option alone! + + config VMSPLIT_3G +- bool "3G/1G user/kernel split" ++ bool "Default 896MB lowmem (3G/1G user/kernel split)" + config VMSPLIT_3G_OPT + depends on !X86_PAE +- bool "3G/1G user/kernel split (for full 1G low memory)" ++ bool "1GB lowmem (3G/1G user/kernel split)" + config VMSPLIT_2G +- bool "2G/2G user/kernel split" ++ bool "2GB lowmem (2G/2G user/kernel split)" + config VMSPLIT_2G_OPT + depends on !X86_PAE +- bool "2G/2G user/kernel split (for full 2G low memory)" ++ bool "2GB lowmem (2G/2G user/kernel split)" + config VMSPLIT_1G +- bool "1G/3G user/kernel split" ++ bool "3GB lowmem (1G/3G user/kernel split)" + endchoice + + config PAGE_OFFSET +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0004-Create-highres-timeout-variants-of-schedule_timeout-.patch b/sys-kernel/linux-sources-redcore-lts/files/0004-Create-highres-timeout-variants-of-schedule_timeout-.patch new file mode 100644 index 00000000..3c182fbe --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0004-Create-highres-timeout-variants-of-schedule_timeout-.patch @@ -0,0 +1,153 @@ +From d27b58b0707ac311be5a51594fc6f22ed1d109e5 Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Sat, 12 Aug 2017 11:53:39 +1000 +Subject: [PATCH 04/16] Create highres timeout variants of schedule_timeout + functions. + +--- + include/linux/freezer.h | 1 + + include/linux/sched.h | 31 +++++++++++++++++++-- + kernel/time/hrtimer.c | 71 +++++++++++++++++++++++++++++++++++++++++++++++++ + 3 files changed, 101 insertions(+), 2 deletions(-) + +diff --git a/include/linux/freezer.h b/include/linux/freezer.h +index 3995df1d068f..f8645e8f2444 100644 +--- a/include/linux/freezer.h ++++ b/include/linux/freezer.h +@@ -297,6 +297,7 @@ static inline void set_freezable(void) {} + #define wait_event_freezekillable_unsafe(wq, condition) \ + wait_event_killable(wq, condition) + ++#define pm_freezing (false) + #endif /* !CONFIG_FREEZER */ + + #endif /* FREEZER_H_INCLUDED */ +diff --git a/include/linux/sched.h b/include/linux/sched.h +index 35dc91a0e2ed..38852ebfa864 100644 +--- a/include/linux/sched.h ++++ b/include/linux/sched.h +@@ -173,13 +173,40 @@ extern cpumask_var_t cpu_isolated_map; + + extern void scheduler_tick(void); + +-#define MAX_SCHEDULE_TIMEOUT LONG_MAX +- ++#define MAX_SCHEDULE_TIMEOUT LONG_MAX + extern long schedule_timeout(long timeout); + extern long schedule_timeout_interruptible(long timeout); + extern long schedule_timeout_killable(long timeout); + extern long schedule_timeout_uninterruptible(long timeout); + extern long schedule_timeout_idle(long timeout); ++ ++#ifdef CONFIG_HIGH_RES_TIMERS ++extern long schedule_msec_hrtimeout(long timeout); ++extern long schedule_min_hrtimeout(void); ++extern long schedule_msec_hrtimeout_interruptible(long timeout); ++extern long schedule_msec_hrtimeout_uninterruptible(long timeout); ++#else ++static inline long schedule_msec_hrtimeout(long timeout) ++{ ++ return schedule_timeout(msecs_to_jiffies(timeout)); ++} ++ ++static inline long schedule_min_hrtimeout(void) ++{ ++ return schedule_timeout(1); ++} ++ ++static inline long schedule_msec_hrtimeout_interruptible(long timeout) ++{ ++ return schedule_timeout_interruptible(msecs_to_jiffies(timeout)); ++} ++ ++static inline long schedule_msec_hrtimeout_uninterruptible(long timeout) ++{ ++ return schedule_timeout_uninterruptible(msecs_to_jiffies(timeout)); ++} ++#endif ++ + asmlinkage void schedule(void); + extern void schedule_preempt_disabled(void); + +diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c +index 88f75f92ef36..13227cf2814c 100644 +--- a/kernel/time/hrtimer.c ++++ b/kernel/time/hrtimer.c +@@ -1787,3 +1787,74 @@ int __sched schedule_hrtimeout(ktime_t *expires, + return schedule_hrtimeout_range(expires, 0, mode); + } + EXPORT_SYMBOL_GPL(schedule_hrtimeout); ++ ++/* ++ * As per schedule_hrtimeout but taskes a millisecond value and returns how ++ * many milliseconds are left. ++ */ ++long __sched schedule_msec_hrtimeout(long timeout) ++{ ++ struct hrtimer_sleeper t; ++ int delta, secs, jiffs; ++ ktime_t expires; ++ ++ if (!timeout) { ++ __set_current_state(TASK_RUNNING); ++ return 0; ++ } ++ ++ jiffs = msecs_to_jiffies(timeout); ++ /* ++ * If regular timer resolution is adequate or hrtimer resolution is not ++ * (yet) better than Hz, as would occur during startup, use regular ++ * timers. ++ */ ++ if (jiffs > 4 || hrtimer_resolution >= NSEC_PER_SEC / HZ) ++ return schedule_timeout(jiffs); ++ ++ secs = timeout / 1000; ++ delta = (timeout % 1000) * NSEC_PER_MSEC; ++ expires = ktime_set(secs, delta); ++ ++ hrtimer_init_on_stack(&t.timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); ++ hrtimer_set_expires_range_ns(&t.timer, expires, delta); ++ ++ hrtimer_init_sleeper(&t, current); ++ ++ hrtimer_start_expires(&t.timer, HRTIMER_MODE_REL); ++ ++ if (likely(t.task)) ++ schedule(); ++ ++ hrtimer_cancel(&t.timer); ++ destroy_hrtimer_on_stack(&t.timer); ++ ++ __set_current_state(TASK_RUNNING); ++ ++ expires = hrtimer_expires_remaining(&t.timer); ++ timeout = ktime_to_ms(expires); ++ return timeout < 0 ? 0 : timeout; ++} ++ ++EXPORT_SYMBOL(schedule_msec_hrtimeout); ++ ++long __sched schedule_min_hrtimeout(void) ++{ ++ return schedule_msec_hrtimeout(1); ++} ++ ++EXPORT_SYMBOL(schedule_min_hrtimeout); ++ ++long __sched schedule_msec_hrtimeout_interruptible(long timeout) ++{ ++ __set_current_state(TASK_INTERRUPTIBLE); ++ return schedule_msec_hrtimeout(timeout); ++} ++EXPORT_SYMBOL(schedule_msec_hrtimeout_interruptible); ++ ++long __sched schedule_msec_hrtimeout_uninterruptible(long timeout) ++{ ++ __set_current_state(TASK_UNINTERRUPTIBLE); ++ return schedule_msec_hrtimeout(timeout); ++} ++EXPORT_SYMBOL(schedule_msec_hrtimeout_uninterruptible); +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0005-Special-case-calls-of-schedule_timeout-1-to-use-the-.patch b/sys-kernel/linux-sources-redcore-lts/files/0005-Special-case-calls-of-schedule_timeout-1-to-use-the-.patch new file mode 100644 index 00000000..3c889719 --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0005-Special-case-calls-of-schedule_timeout-1-to-use-the-.patch @@ -0,0 +1,50 @@ +From 5da7d1778b96c514394334c92de9b3d8d71f4a29 Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Sat, 5 Nov 2016 09:27:36 +1100 +Subject: [PATCH 05/16] Special case calls of schedule_timeout(1) to use the + min hrtimeout of 1ms, working around low Hz resolutions. + +--- + kernel/time/timer.c | 17 +++++++++++++++-- + 1 file changed, 15 insertions(+), 2 deletions(-) + +diff --git a/kernel/time/timer.c b/kernel/time/timer.c +index 9c18e16059a3..dd4d1b193286 100644 +--- a/kernel/time/timer.c ++++ b/kernel/time/timer.c +@@ -1741,6 +1741,19 @@ signed long __sched schedule_timeout(signed long timeout) + + expire = timeout + jiffies; + ++#ifdef CONFIG_HIGH_RES_TIMERS ++ if (timeout == 1 && hrtimer_resolution < NSEC_PER_SEC / HZ) { ++ /* ++ * Special case 1 as being a request for the minimum timeout ++ * and use highres timers to timeout after 1ms to workaround ++ * the granularity of low Hz tick timers. ++ */ ++ if (!schedule_min_hrtimeout()) ++ return 0; ++ goto out_timeout; ++ } ++#endif ++ + setup_timer_on_stack(&timer, process_timeout, (unsigned long)current); + __mod_timer(&timer, expire, false); + schedule(); +@@ -1748,10 +1761,10 @@ signed long __sched schedule_timeout(signed long timeout) + + /* Remove the timer from the object tracker */ + destroy_timer_on_stack(&timer); +- ++out_timeout: + timeout = expire - jiffies; + +- out: ++out: + return timeout < 0 ? 0 : timeout; + } + EXPORT_SYMBOL(schedule_timeout); +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0006-Convert-msleep-to-use-hrtimers-when-active.patch b/sys-kernel/linux-sources-redcore-lts/files/0006-Convert-msleep-to-use-hrtimers-when-active.patch new file mode 100644 index 00000000..2f065652 --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0006-Convert-msleep-to-use-hrtimers-when-active.patch @@ -0,0 +1,54 @@ +From 9df803c28bb8ccb2588c0ccaf857b9e673175fed Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Fri, 4 Nov 2016 09:25:54 +1100 +Subject: [PATCH 06/16] Convert msleep to use hrtimers when active. + +--- + kernel/time/timer.c | 24 ++++++++++++++++++++++-- + 1 file changed, 22 insertions(+), 2 deletions(-) + +diff --git a/kernel/time/timer.c b/kernel/time/timer.c +index dd4d1b193286..c68cb9307f64 100644 +--- a/kernel/time/timer.c ++++ b/kernel/time/timer.c +@@ -1884,7 +1884,19 @@ void __init init_timers(void) + */ + void msleep(unsigned int msecs) + { +- unsigned long timeout = msecs_to_jiffies(msecs) + 1; ++ int jiffs = msecs_to_jiffies(msecs); ++ unsigned long timeout; ++ ++ /* ++ * Use high resolution timers where the resolution of tick based ++ * timers is inadequate. ++ */ ++ if (jiffs < 5 && hrtimer_resolution < NSEC_PER_SEC / HZ) { ++ while (msecs) ++ msecs = schedule_msec_hrtimeout_uninterruptible(msecs); ++ return; ++ } ++ timeout = msecs_to_jiffies(msecs) + 1; + + while (timeout) + timeout = schedule_timeout_uninterruptible(timeout); +@@ -1898,7 +1910,15 @@ EXPORT_SYMBOL(msleep); + */ + unsigned long msleep_interruptible(unsigned int msecs) + { +- unsigned long timeout = msecs_to_jiffies(msecs) + 1; ++ int jiffs = msecs_to_jiffies(msecs); ++ unsigned long timeout; ++ ++ if (jiffs < 5 && hrtimer_resolution < NSEC_PER_SEC / HZ) { ++ while (msecs && !signal_pending(current)) ++ msecs = schedule_msec_hrtimeout_interruptible(msecs); ++ return msecs; ++ } ++ timeout = msecs_to_jiffies(msecs) + 1; + + while (timeout && !signal_pending(current)) + timeout = schedule_timeout_interruptible(timeout); +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0007-Replace-all-schedule-timeout-1-with-schedule_min_hrt.patch b/sys-kernel/linux-sources-redcore-lts/files/0007-Replace-all-schedule-timeout-1-with-schedule_min_hrt.patch new file mode 100644 index 00000000..ff071da8 --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0007-Replace-all-schedule-timeout-1-with-schedule_min_hrt.patch @@ -0,0 +1,529 @@ +diff -Nur a/drivers/block/swim.c b/drivers/block/swim.c +--- a/drivers/block/swim.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/block/swim.c 2018-11-03 16:30:39.471807304 +0000 +@@ -332,7 +332,7 @@ + if (swim_readbit(base, MOTOR_ON)) + break; + current->state = TASK_INTERRUPTIBLE; +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + } + } else if (action == OFF) { + swim_action(base, MOTOR_OFF); +@@ -351,7 +351,7 @@ + if (!swim_readbit(base, DISK_IN)) + break; + current->state = TASK_INTERRUPTIBLE; +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + } + swim_select(base, RELAX); + } +@@ -375,7 +375,7 @@ + for (wait = 0; wait < HZ; wait++) { + + current->state = TASK_INTERRUPTIBLE; +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + + swim_select(base, RELAX); + if (!swim_readbit(base, STEP)) +diff -Nur a/drivers/bluetooth/hci_qca.c b/drivers/bluetooth/hci_qca.c +--- a/drivers/bluetooth/hci_qca.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/bluetooth/hci_qca.c 2018-11-03 16:31:56.065260061 +0000 +@@ -880,7 +880,7 @@ + * then host can communicate with new baudrate to controller + */ + set_current_state(TASK_UNINTERRUPTIBLE); +- schedule_timeout(msecs_to_jiffies(BAUDRATE_SETTLE_TIMEOUT_MS)); ++ schedule_msec_hrtimeout((BAUDRATE_SETTLE_TIMEOUT_MS)); + set_current_state(TASK_RUNNING); + + return 0; +diff -Nur a/drivers/char/ipmi/ipmi_msghandler.c b/drivers/char/ipmi/ipmi_msghandler.c +--- a/drivers/char/ipmi/ipmi_msghandler.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/char/ipmi/ipmi_msghandler.c 2018-11-03 16:30:39.473807368 +0000 +@@ -2953,7 +2953,7 @@ + /* Current message first, to preserve order */ + while (intf->curr_msg && !list_empty(&intf->waiting_rcv_msgs)) { + /* Wait for the message to clear out. */ +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + } + + /* No need for locks, the interface is down. */ +diff -Nur a/drivers/char/ipmi/ipmi_ssif.c b/drivers/char/ipmi/ipmi_ssif.c +--- a/drivers/char/ipmi/ipmi_ssif.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/char/ipmi/ipmi_ssif.c 2018-11-03 16:30:39.473807368 +0000 +@@ -1200,7 +1200,7 @@ + + /* make sure the driver is not looking for flags any more. */ + while (ssif_info->ssif_state != SSIF_NORMAL) +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + + ssif_info->stopping = true; + del_timer_sync(&ssif_info->retry_timer); +diff -Nur a/drivers/char/snsc.c b/drivers/char/snsc.c +--- a/drivers/char/snsc.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/char/snsc.c 2018-11-03 16:30:39.474807400 +0000 +@@ -198,7 +198,7 @@ + add_wait_queue(&sd->sd_rq, &wait); + spin_unlock_irqrestore(&sd->sd_rlock, flags); + +- schedule_timeout(msecs_to_jiffies(SCDRV_TIMEOUT)); ++ schedule_msec_hrtimeout((SCDRV_TIMEOUT)); + + remove_wait_queue(&sd->sd_rq, &wait); + if (signal_pending(current)) { +@@ -294,7 +294,7 @@ + add_wait_queue(&sd->sd_wq, &wait); + spin_unlock_irqrestore(&sd->sd_wlock, flags); + +- schedule_timeout(msecs_to_jiffies(SCDRV_TIMEOUT)); ++ schedule_msec_hrtimeout((SCDRV_TIMEOUT)); + + remove_wait_queue(&sd->sd_wq, &wait); + if (signal_pending(current)) { +diff -Nur a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c +--- a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c 2018-11-03 16:30:39.474807400 +0000 +@@ -235,7 +235,7 @@ + DRM_ERROR("SVGA device lockup.\n"); + break; + } +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + if (interruptible && signal_pending(current)) { + ret = -ERESTARTSYS; + break; +diff -Nur a/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c b/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c +--- a/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c 2018-11-03 16:30:39.474807400 +0000 +@@ -202,7 +202,7 @@ + break; + } + if (lazy) +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + else if ((++count & 0x0F) == 0) { + /** + * FIXME: Use schedule_hr_timeout here for +diff -Nur a/drivers/media/pci/ivtv/ivtv-ioctl.c b/drivers/media/pci/ivtv/ivtv-ioctl.c +--- a/drivers/media/pci/ivtv/ivtv-ioctl.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/media/pci/ivtv/ivtv-ioctl.c 2018-11-03 16:30:39.475807432 +0000 +@@ -1154,7 +1154,7 @@ + TASK_UNINTERRUPTIBLE); + if ((read_reg(IVTV_REG_DEC_LINE_FIELD) >> 16) < 100) + break; +- schedule_timeout(msecs_to_jiffies(25)); ++ schedule_msec_hrtimeout((25)); + } + finish_wait(&itv->vsync_waitq, &wait); + mutex_lock(&itv->serialize_lock); +diff -Nur a/drivers/media/pci/ivtv/ivtv-streams.c b/drivers/media/pci/ivtv/ivtv-streams.c +--- a/drivers/media/pci/ivtv/ivtv-streams.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/media/pci/ivtv/ivtv-streams.c 2018-11-03 16:30:39.475807432 +0000 +@@ -834,7 +834,7 @@ + while (!test_bit(IVTV_F_I_EOS, &itv->i_flags) && + time_before(jiffies, + then + msecs_to_jiffies(2000))) { +- schedule_timeout(msecs_to_jiffies(10)); ++ schedule_msec_hrtimeout((10)); + } + + /* To convert jiffies to ms, we must multiply by 1000 +diff -Nur a/drivers/mfd/ucb1x00-core.c b/drivers/mfd/ucb1x00-core.c +--- a/drivers/mfd/ucb1x00-core.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/mfd/ucb1x00-core.c 2018-11-03 16:30:39.476807464 +0000 +@@ -253,7 +253,7 @@ + break; + /* yield to other processes */ + set_current_state(TASK_INTERRUPTIBLE); +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + } + + return UCB_ADC_DAT(val); +diff -Nur a/drivers/misc/sgi-xp/xpc_channel.c b/drivers/misc/sgi-xp/xpc_channel.c +--- a/drivers/misc/sgi-xp/xpc_channel.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/misc/sgi-xp/xpc_channel.c 2018-11-03 16:30:39.476807464 +0000 +@@ -837,7 +837,7 @@ + + atomic_inc(&ch->n_on_msg_allocate_wq); + prepare_to_wait(&ch->msg_allocate_wq, &wait, TASK_INTERRUPTIBLE); +- ret = schedule_timeout(1); ++ ret = schedule_min_hrtimeout(); + finish_wait(&ch->msg_allocate_wq, &wait); + atomic_dec(&ch->n_on_msg_allocate_wq); + +diff -Nur a/drivers/net/caif/caif_hsi.c b/drivers/net/caif/caif_hsi.c +--- a/drivers/net/caif/caif_hsi.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/net/caif/caif_hsi.c 2018-11-03 16:30:39.477807497 +0000 +@@ -940,7 +940,7 @@ + break; + + set_current_state(TASK_INTERRUPTIBLE); +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + retry--; + } + +diff -Nur a/drivers/net/can/usb/peak_usb/pcan_usb.c b/drivers/net/can/usb/peak_usb/pcan_usb.c +--- a/drivers/net/can/usb/peak_usb/pcan_usb.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/net/can/usb/peak_usb/pcan_usb.c 2018-11-03 16:30:39.477807497 +0000 +@@ -250,7 +250,7 @@ + } else { + /* the PCAN-USB needs time to init */ + set_current_state(TASK_INTERRUPTIBLE); +- schedule_timeout(msecs_to_jiffies(PCAN_USB_STARTUP_TIMEOUT)); ++ schedule_msec_hrtimeout((PCAN_USB_STARTUP_TIMEOUT)); + } + + return err; +diff -Nur a/drivers/net/usb/lan78xx.c b/drivers/net/usb/lan78xx.c +--- a/drivers/net/usb/lan78xx.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/net/usb/lan78xx.c 2018-11-03 16:30:39.478807529 +0000 +@@ -2567,7 +2567,7 @@ + while (!skb_queue_empty(&dev->rxq) && + !skb_queue_empty(&dev->txq) && + !skb_queue_empty(&dev->done)) { +- schedule_timeout(msecs_to_jiffies(UNLINK_TIMEOUT_MS)); ++ schedule_msec_hrtimeout((UNLINK_TIMEOUT_MS)); + set_current_state(TASK_UNINTERRUPTIBLE); + netif_dbg(dev, ifdown, dev->net, + "waited for %d urb completions\n", temp); +diff -Nur a/drivers/net/usb/usbnet.c b/drivers/net/usb/usbnet.c +--- a/drivers/net/usb/usbnet.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/net/usb/usbnet.c 2018-11-03 16:30:39.479807561 +0000 +@@ -772,7 +772,7 @@ + spin_lock_irqsave(&q->lock, flags); + while (!skb_queue_empty(q)) { + spin_unlock_irqrestore(&q->lock, flags); +- schedule_timeout(msecs_to_jiffies(UNLINK_TIMEOUT_MS)); ++ schedule_msec_hrtimeout((UNLINK_TIMEOUT_MS)); + set_current_state(TASK_UNINTERRUPTIBLE); + spin_lock_irqsave(&q->lock, flags); + } +diff -Nur a/drivers/ntb/test/ntb_perf.c b/drivers/ntb/test/ntb_perf.c +--- a/drivers/ntb/test/ntb_perf.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/ntb/test/ntb_perf.c 2018-11-03 16:30:39.479807561 +0000 +@@ -310,7 +310,7 @@ + if (unlikely((jiffies - last_sleep) > 5 * HZ)) { + last_sleep = jiffies; + set_current_state(TASK_INTERRUPTIBLE); +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + } + + if (unlikely(kthread_should_stop())) +diff -Nur a/drivers/scsi/fnic/fnic_scsi.c b/drivers/scsi/fnic/fnic_scsi.c +--- a/drivers/scsi/fnic/fnic_scsi.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/scsi/fnic/fnic_scsi.c 2018-11-03 16:30:39.480807592 +0000 +@@ -217,7 +217,7 @@ + + /* wait for io cmpl */ + while (atomic_read(&fnic->in_flight)) +- schedule_timeout(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout((1)); + + spin_lock_irqsave(&fnic->wq_copy_lock[0], flags); + +@@ -2255,7 +2255,7 @@ + } + } + +- schedule_timeout(msecs_to_jiffies(2 * fnic->config.ed_tov)); ++ schedule_msec_hrtimeout((2 * fnic->config.ed_tov)); + + /* walk again to check, if IOs are still pending in fw */ + if (fnic_is_abts_pending(fnic, lr_sc)) +diff -Nur a/drivers/scsi/snic/snic_scsi.c b/drivers/scsi/snic/snic_scsi.c +--- a/drivers/scsi/snic/snic_scsi.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/scsi/snic/snic_scsi.c 2018-11-03 16:30:39.481807625 +0000 +@@ -2354,7 +2354,7 @@ + + /* Wait for all the IOs that are entered in Qcmd */ + while (atomic_read(&snic->ios_inflight)) +- schedule_timeout(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout((1)); + + ret = snic_issue_hba_reset(snic, sc); + if (ret) { +diff -Nur a/drivers/staging/comedi/drivers/ni_mio_common.c b/drivers/staging/comedi/drivers/ni_mio_common.c +--- a/drivers/staging/comedi/drivers/ni_mio_common.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/comedi/drivers/ni_mio_common.c 2018-11-03 16:30:39.483807688 +0000 +@@ -4657,7 +4657,7 @@ + if ((status & NI67XX_CAL_STATUS_BUSY) == 0) + break; + set_current_state(TASK_INTERRUPTIBLE); +- if (schedule_timeout(1)) ++ if (schedule_min_hrtimeout()) + return -EIO; + } + if (i == timeout) { +diff -Nur a/drivers/staging/lustre/lnet/lnet/lib-eq.c b/drivers/staging/lustre/lnet/lnet/lib-eq.c +--- a/drivers/staging/lustre/lnet/lnet/lib-eq.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/lustre/lnet/lnet/lib-eq.c 2018-11-03 16:30:39.483807688 +0000 +@@ -329,7 +329,7 @@ + schedule(); + } else { + now = jiffies; +- schedule_timeout(msecs_to_jiffies(tms)); ++ schedule_msec_hrtimeout((tms)); + tms -= jiffies_to_msecs(jiffies - now); + if (tms < 0) /* no more wait but may have new event */ + tms = 0; +diff -Nur a/drivers/staging/rts5208/rtsx.c b/drivers/staging/rts5208/rtsx.c +--- a/drivers/staging/rts5208/rtsx.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/rts5208/rtsx.c 2018-11-03 16:30:39.483807688 +0000 +@@ -524,7 +524,7 @@ + + for (;;) { + set_current_state(TASK_INTERRUPTIBLE); +- schedule_timeout(msecs_to_jiffies(POLLING_INTERVAL)); ++ schedule_msec_hrtimeout((POLLING_INTERVAL)); + + /* lock the device pointers */ + mutex_lock(&dev->dev_mutex); +diff -Nur a/drivers/staging/speakup/speakup_acntpc.c b/drivers/staging/speakup/speakup_acntpc.c +--- a/drivers/staging/speakup/speakup_acntpc.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/speakup/speakup_acntpc.c 2018-11-03 16:30:39.484807721 +0000 +@@ -206,7 +206,7 @@ + full_time_val = full_time->u.n.value; + spin_unlock_irqrestore(&speakup_info.spinlock, flags); + if (synth_full()) { +- schedule_timeout(msecs_to_jiffies(full_time_val)); ++ schedule_msec_hrtimeout((full_time_val)); + continue; + } + set_current_state(TASK_RUNNING); +@@ -234,7 +234,7 @@ + jiffy_delta_val = jiffy_delta->u.n.value; + delay_time_val = delay_time->u.n.value; + spin_unlock_irqrestore(&speakup_info.spinlock, flags); +- schedule_timeout(msecs_to_jiffies(delay_time_val)); ++ schedule_msec_hrtimeout(delay_time_val); + jiff_max = jiffies + jiffy_delta_val; + } + } +diff -Nur a/drivers/staging/speakup/speakup_apollo.c b/drivers/staging/speakup/speakup_apollo.c +--- a/drivers/staging/speakup/speakup_apollo.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/speakup/speakup_apollo.c 2018-11-03 16:30:39.484807721 +0000 +@@ -174,7 +174,7 @@ + if (!synth->io_ops->synth_out(synth, ch)) { + synth->io_ops->tiocmset(0, UART_MCR_RTS); + synth->io_ops->tiocmset(UART_MCR_RTS, 0); +- schedule_timeout(msecs_to_jiffies(full_time_val)); ++ schedule_msec_hrtimeout(full_time_val); + continue; + } + if (time_after_eq(jiffies, jiff_max) && (ch == SPACE)) { +diff -Nur a/drivers/staging/speakup/speakup_decext.c b/drivers/staging/speakup/speakup_decext.c +--- a/drivers/staging/speakup/speakup_decext.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/speakup/speakup_decext.c 2018-11-03 16:30:39.484807721 +0000 +@@ -185,7 +185,7 @@ + if (ch == '\n') + ch = 0x0D; + if (synth_full() || !synth->io_ops->synth_out(synth, ch)) { +- schedule_timeout(msecs_to_jiffies(delay_time_val)); ++ schedule_msec_hrtimeout(delay_time_val); + continue; + } + set_current_state(TASK_RUNNING); +diff -Nur a/drivers/staging/speakup/speakup_decpc.c b/drivers/staging/speakup/speakup_decpc.c +--- a/drivers/staging/speakup/speakup_decpc.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/speakup/speakup_decpc.c 2018-11-03 16:30:39.484807721 +0000 +@@ -403,7 +403,7 @@ + if (ch == '\n') + ch = 0x0D; + if (dt_sendchar(ch)) { +- schedule_timeout(msecs_to_jiffies(delay_time_val)); ++ schedule_msec_hrtimeout((delay_time_val)); + continue; + } + set_current_state(TASK_RUNNING); +diff -Nur a/drivers/staging/speakup/speakup_dectlk.c b/drivers/staging/speakup/speakup_dectlk.c +--- a/drivers/staging/speakup/speakup_dectlk.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/speakup/speakup_dectlk.c 2018-11-03 16:30:39.485807753 +0000 +@@ -253,7 +253,7 @@ + if (ch == '\n') + ch = 0x0D; + if (synth_full_val || !synth->io_ops->synth_out(synth, ch)) { +- schedule_timeout(msecs_to_jiffies(delay_time_val)); ++ schedule_msec_hrtimeout(delay_time_val); + continue; + } + set_current_state(TASK_RUNNING); +diff -Nur a/drivers/staging/speakup/speakup_dtlk.c b/drivers/staging/speakup/speakup_dtlk.c +--- a/drivers/staging/speakup/speakup_dtlk.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/speakup/speakup_dtlk.c 2018-11-03 16:30:39.485807753 +0000 +@@ -220,7 +220,7 @@ + delay_time_val = delay_time->u.n.value; + spin_unlock_irqrestore(&speakup_info.spinlock, flags); + if (synth_full()) { +- schedule_timeout(msecs_to_jiffies(delay_time_val)); ++ schedule_msec_hrtimeout((delay_time_val)); + continue; + } + set_current_state(TASK_RUNNING); +@@ -236,7 +236,7 @@ + delay_time_val = delay_time->u.n.value; + jiffy_delta_val = jiffy_delta->u.n.value; + spin_unlock_irqrestore(&speakup_info.spinlock, flags); +- schedule_timeout(msecs_to_jiffies(delay_time_val)); ++ schedule_msec_hrtimeout((delay_time_val)); + jiff_max = jiffies + jiffy_delta_val; + } + } +diff -Nur a/drivers/staging/speakup/speakup_keypc.c b/drivers/staging/speakup/speakup_keypc.c +--- a/drivers/staging/speakup/speakup_keypc.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/speakup/speakup_keypc.c 2018-11-03 16:30:39.485807753 +0000 +@@ -208,7 +208,7 @@ + full_time_val = full_time->u.n.value; + spin_unlock_irqrestore(&speakup_info.spinlock, flags); + if (synth_full()) { +- schedule_timeout(msecs_to_jiffies(full_time_val)); ++ schedule_msec_hrtimeout((full_time_val)); + continue; + } + set_current_state(TASK_RUNNING); +@@ -241,7 +241,7 @@ + jiffy_delta_val = jiffy_delta->u.n.value; + delay_time_val = delay_time->u.n.value; + spin_unlock_irqrestore(&speakup_info.spinlock, flags); +- schedule_timeout(msecs_to_jiffies(delay_time_val)); ++ schedule_msec_hrtimeout((delay_time_val)); + jiff_max = jiffies+jiffy_delta_val; + } + } +diff -Nur a/drivers/staging/speakup/synth.c b/drivers/staging/speakup/synth.c +--- a/drivers/staging/speakup/synth.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/speakup/synth.c 2018-11-03 16:30:39.486807785 +0000 +@@ -92,7 +92,7 @@ + if (ch == '\n') + ch = synth->procspeech; + if (!synth->io_ops->synth_out(synth, ch)) { +- schedule_timeout(msecs_to_jiffies(full_time_val)); ++ schedule_msec_hrtimeout(full_time_val); + continue; + } + if (time_after_eq(jiffies, jiff_max) && (ch == SPACE)) { +diff -Nur a/drivers/staging/unisys/visornic/visornic_main.c b/drivers/staging/unisys/visornic/visornic_main.c +--- a/drivers/staging/unisys/visornic/visornic_main.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/staging/unisys/visornic/visornic_main.c 2018-11-03 16:30:39.486807785 +0000 +@@ -556,7 +556,7 @@ + } + set_current_state(TASK_INTERRUPTIBLE); + spin_unlock_irqrestore(&devdata->priv_lock, flags); +- wait += schedule_timeout(msecs_to_jiffies(10)); ++ wait += schedule_msec_hrtimeout((10)); + spin_lock_irqsave(&devdata->priv_lock, flags); + } + +@@ -567,7 +567,7 @@ + while (1) { + set_current_state(TASK_INTERRUPTIBLE); + spin_unlock_irqrestore(&devdata->priv_lock, flags); +- schedule_timeout(msecs_to_jiffies(10)); ++ schedule_msec_hrtimeout((10)); + spin_lock_irqsave(&devdata->priv_lock, flags); + if (atomic_read(&devdata->usage)) + break; +@@ -721,7 +721,7 @@ + } + set_current_state(TASK_INTERRUPTIBLE); + spin_unlock_irqrestore(&devdata->priv_lock, flags); +- wait += schedule_timeout(msecs_to_jiffies(10)); ++ wait += schedule_msec_hrtimeout((10)); + spin_lock_irqsave(&devdata->priv_lock, flags); + } + +diff -Nur a/drivers/target/target_core_user.c b/drivers/target/target_core_user.c +--- a/drivers/target/target_core_user.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/target/target_core_user.c 2018-11-03 16:30:39.487807817 +0000 +@@ -808,10 +808,9 @@ + pr_debug("sleeping for ring space\n"); + mutex_unlock(&udev->cmdr_lock); + if (udev->cmd_time_out) +- ret = schedule_timeout( +- msecs_to_jiffies(udev->cmd_time_out)); ++ ret = schedule_msec_hrtimeout(udev->cmd_time_out); + else +- ret = schedule_timeout(msecs_to_jiffies(TCMU_TIME_OUT)); ++ ret = schedule_msec_hrtimeout(TCMU_TIME_OUT); + finish_wait(&udev->wait_cmdr, &__wait); + if (!ret) { + pr_warn("tcmu: command timed out\n"); +diff -Nur a/drivers/video/fbdev/omap/hwa742.c b/drivers/video/fbdev/omap/hwa742.c +--- a/drivers/video/fbdev/omap/hwa742.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/video/fbdev/omap/hwa742.c 2018-11-03 16:30:39.487807817 +0000 +@@ -926,7 +926,7 @@ + if (hwa742_read_reg(HWA742_PLL_DIV_REG) & (1 << 7)) + break; + set_current_state(TASK_UNINTERRUPTIBLE); +- schedule_timeout(msecs_to_jiffies(5)); ++ schedule_msec_hrtimeout((5)); + } + hwa742_set_update_mode(hwa742.update_mode_before_suspend); + } +diff -Nur a/drivers/video/fbdev/pxafb.c b/drivers/video/fbdev/pxafb.c +--- a/drivers/video/fbdev/pxafb.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/drivers/video/fbdev/pxafb.c 2018-11-03 16:30:39.488807849 +0000 +@@ -1286,7 +1286,7 @@ + mutex_unlock(&fbi->ctrlr_lock); + + set_current_state(TASK_INTERRUPTIBLE); +- schedule_timeout(msecs_to_jiffies(30)); ++ schedule_msec_hrtimeout((30)); + } + + pr_debug("%s(): task ending\n", __func__); +diff -Nur a/fs/afs/vlocation.c b/fs/afs/vlocation.c +--- a/fs/afs/vlocation.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/fs/afs/vlocation.c 2018-11-03 16:30:39.488807849 +0000 +@@ -129,7 +129,7 @@ + if (vl->upd_busy_cnt > 1) { + /* second+ BUSY - sleep a little bit */ + set_current_state(TASK_UNINTERRUPTIBLE); +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + } + continue; + } +diff -Nur a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c +--- a/fs/btrfs/extent-tree.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/fs/btrfs/extent-tree.c 2018-11-03 16:30:39.491807945 +0000 +@@ -6106,7 +6106,7 @@ + + if (flush != BTRFS_RESERVE_NO_FLUSH && + btrfs_transaction_in_commit(fs_info)) +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + + if (delalloc_lock) + mutex_lock(&inode->delalloc_mutex); +diff -Nur a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c +--- a/fs/btrfs/inode-map.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/fs/btrfs/inode-map.c 2018-11-03 16:30:39.492807977 +0000 +@@ -89,7 +89,7 @@ + btrfs_release_path(path); + root->ino_cache_progress = last; + up_read(&fs_info->commit_root_sem); +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + goto again; + } else + continue; +diff -Nur a/sound/usb/line6/pcm.c b/sound/usb/line6/pcm.c +--- a/sound/usb/line6/pcm.c 2018-10-10 07:54:28.000000000 +0100 ++++ b/sound/usb/line6/pcm.c 2018-11-03 16:30:39.492807977 +0000 +@@ -131,7 +131,7 @@ + if (!alive) + break; + set_current_state(TASK_UNINTERRUPTIBLE); +- schedule_timeout(1); ++ schedule_min_hrtimeout(); + } while (--timeout > 0); + if (alive) + dev_err(line6pcm->line6->ifcdev, diff --git a/sys-kernel/linux-sources-redcore-lts/files/0008-Replace-all-calls-to-schedule_timeout_interruptible-.patch b/sys-kernel/linux-sources-redcore-lts/files/0008-Replace-all-calls-to-schedule_timeout_interruptible-.patch new file mode 100644 index 00000000..f9f274ce --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0008-Replace-all-calls-to-schedule_timeout_interruptible-.patch @@ -0,0 +1,311 @@ +From 3ef5df78c2f425115b87f0f2f59fd189c0f1bbe3 Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Mon, 20 Feb 2017 13:30:07 +1100 +Subject: [PATCH 08/16] Replace all calls to schedule_timeout_interruptible of + potentially under 50ms to use schedule_msec_hrtimeout_interruptible. + +--- + drivers/hwmon/fam15h_power.c | 2 +- + drivers/iio/light/tsl2563.c | 6 +----- + drivers/media/i2c/msp3400-driver.c | 4 ++-- + drivers/media/pci/ivtv/ivtv-gpio.c | 6 +++--- + drivers/media/radio/radio-mr800.c | 2 +- + drivers/media/radio/radio-tea5777.c | 2 +- + drivers/media/radio/tea575x.c | 2 +- + drivers/parport/ieee1284.c | 2 +- + drivers/parport/ieee1284_ops.c | 2 +- + drivers/platform/x86/intel_ips.c | 8 ++++---- + net/core/pktgen.c | 2 +- + sound/soc/codecs/wm8350.c | 12 ++++++------ + sound/soc/codecs/wm8900.c | 2 +- + sound/soc/codecs/wm9713.c | 4 ++-- + 14 files changed, 26 insertions(+), 30 deletions(-) + +diff --git a/drivers/hwmon/fam15h_power.c b/drivers/hwmon/fam15h_power.c +index 9545a346044f..c24cf1302ec7 100644 +--- a/drivers/hwmon/fam15h_power.c ++++ b/drivers/hwmon/fam15h_power.c +@@ -237,7 +237,7 @@ static ssize_t power1_average_show(struct device *dev, + prev_ptsc[cu] = data->cpu_sw_pwr_ptsc[cu]; + } + +- leftover = schedule_timeout_interruptible(msecs_to_jiffies(data->power_period)); ++ leftover = schedule_msec_hrtimeout_interruptible((data->power_period)); + if (leftover) + return 0; + +diff --git a/drivers/iio/light/tsl2563.c b/drivers/iio/light/tsl2563.c +index 7599693f7fe9..452090739138 100644 +--- a/drivers/iio/light/tsl2563.c ++++ b/drivers/iio/light/tsl2563.c +@@ -282,11 +282,7 @@ static void tsl2563_wait_adc(struct tsl2563_chip *chip) + default: + delay = 402; + } +- /* +- * TODO: Make sure that we wait at least required delay but why we +- * have to extend it one tick more? +- */ +- schedule_timeout_interruptible(msecs_to_jiffies(delay) + 2); ++ schedule_msec_hrtimeout_interruptible(delay + 1); + } + + static int tsl2563_adjust_gainlevel(struct tsl2563_chip *chip, u16 adc) +diff --git a/drivers/media/i2c/msp3400-driver.c b/drivers/media/i2c/msp3400-driver.c +index 3db966db83eb..f0fab7676f72 100644 +--- a/drivers/media/i2c/msp3400-driver.c ++++ b/drivers/media/i2c/msp3400-driver.c +@@ -179,7 +179,7 @@ static int msp_read(struct i2c_client *client, int dev, int addr) + break; + dev_warn(&client->dev, "I/O error #%d (read 0x%02x/0x%02x)\n", err, + dev, addr); +- schedule_timeout_interruptible(msecs_to_jiffies(10)); ++ schedule_msec_hrtimeout_interruptible((10)); + } + if (err == 3) { + dev_warn(&client->dev, "resetting chip, sound will go off.\n"); +@@ -220,7 +220,7 @@ static int msp_write(struct i2c_client *client, int dev, int addr, int val) + break; + dev_warn(&client->dev, "I/O error #%d (write 0x%02x/0x%02x)\n", err, + dev, addr); +- schedule_timeout_interruptible(msecs_to_jiffies(10)); ++ schedule_msec_hrtimeout_interruptible((10)); + } + if (err == 3) { + dev_warn(&client->dev, "resetting chip, sound will go off.\n"); +diff --git a/drivers/media/pci/ivtv/ivtv-gpio.c b/drivers/media/pci/ivtv/ivtv-gpio.c +index f752f3993687..23372af61ebf 100644 +--- a/drivers/media/pci/ivtv/ivtv-gpio.c ++++ b/drivers/media/pci/ivtv/ivtv-gpio.c +@@ -117,7 +117,7 @@ void ivtv_reset_ir_gpio(struct ivtv *itv) + curout = (curout & ~0xF) | 1; + write_reg(curout, IVTV_REG_GPIO_OUT); + /* We could use something else for smaller time */ +- schedule_timeout_interruptible(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout_interruptible((1)); + curout |= 2; + write_reg(curout, IVTV_REG_GPIO_OUT); + curdir &= ~0x80; +@@ -137,11 +137,11 @@ int ivtv_reset_tuner_gpio(void *dev, int component, int cmd, int value) + curout = read_reg(IVTV_REG_GPIO_OUT); + curout &= ~(1 << itv->card->xceive_pin); + write_reg(curout, IVTV_REG_GPIO_OUT); +- schedule_timeout_interruptible(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout_interruptible((1)); + + curout |= 1 << itv->card->xceive_pin; + write_reg(curout, IVTV_REG_GPIO_OUT); +- schedule_timeout_interruptible(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout_interruptible((1)); + return 0; + } + +diff --git a/drivers/media/radio/radio-mr800.c b/drivers/media/radio/radio-mr800.c +index c9f59129af79..cb6f8394a5c2 100644 +--- a/drivers/media/radio/radio-mr800.c ++++ b/drivers/media/radio/radio-mr800.c +@@ -378,7 +378,7 @@ static int vidioc_s_hw_freq_seek(struct file *file, void *priv, + retval = -ENODATA; + break; + } +- if (schedule_timeout_interruptible(msecs_to_jiffies(10))) { ++ if (schedule_msec_hrtimeout_interruptible((10))) { + retval = -ERESTARTSYS; + break; + } +diff --git a/drivers/media/radio/radio-tea5777.c b/drivers/media/radio/radio-tea5777.c +index 04ed1a5d1177..d593d28dc286 100644 +--- a/drivers/media/radio/radio-tea5777.c ++++ b/drivers/media/radio/radio-tea5777.c +@@ -245,7 +245,7 @@ static int radio_tea5777_update_read_reg(struct radio_tea5777 *tea, int wait) + } + + if (wait) { +- if (schedule_timeout_interruptible(msecs_to_jiffies(wait))) ++ if (schedule_msec_hrtimeout_interruptible((wait))) + return -ERESTARTSYS; + } + +diff --git a/drivers/media/radio/tea575x.c b/drivers/media/radio/tea575x.c +index 4dc2067bce14..29f4416fb9ae 100644 +--- a/drivers/media/radio/tea575x.c ++++ b/drivers/media/radio/tea575x.c +@@ -416,7 +416,7 @@ int snd_tea575x_s_hw_freq_seek(struct file *file, struct snd_tea575x *tea, + for (;;) { + if (time_after(jiffies, timeout)) + break; +- if (schedule_timeout_interruptible(msecs_to_jiffies(10))) { ++ if (schedule_msec_hrtimeout_interruptible((10))) { + /* some signal arrived, stop search */ + tea->val &= ~TEA575X_BIT_SEARCH; + snd_tea575x_set_freq(tea); +diff --git a/drivers/parport/ieee1284.c b/drivers/parport/ieee1284.c +index 74cc6dd982d2..c22c4d5f08d0 100644 +--- a/drivers/parport/ieee1284.c ++++ b/drivers/parport/ieee1284.c +@@ -215,7 +215,7 @@ int parport_wait_peripheral(struct parport *port, + /* parport_wait_event didn't time out, but the + * peripheral wasn't actually ready either. + * Wait for another 10ms. */ +- schedule_timeout_interruptible(msecs_to_jiffies(10)); ++ schedule_msec_hrtimeout_interruptible((10)); + } + } + +diff --git a/drivers/parport/ieee1284_ops.c b/drivers/parport/ieee1284_ops.c +index 5d41dda6da4e..34705f6b423f 100644 +--- a/drivers/parport/ieee1284_ops.c ++++ b/drivers/parport/ieee1284_ops.c +@@ -537,7 +537,7 @@ size_t parport_ieee1284_ecp_read_data (struct parport *port, + /* Yield the port for a while. */ + if (count && dev->port->irq != PARPORT_IRQ_NONE) { + parport_release (dev); +- schedule_timeout_interruptible(msecs_to_jiffies(40)); ++ schedule_msec_hrtimeout_interruptible((40)); + parport_claim_or_block (dev); + } + else +diff --git a/drivers/platform/x86/intel_ips.c b/drivers/platform/x86/intel_ips.c +index 58dcee562d64..b661b7c071bb 100644 +--- a/drivers/platform/x86/intel_ips.c ++++ b/drivers/platform/x86/intel_ips.c +@@ -813,7 +813,7 @@ static int ips_adjust(void *data) + ips_gpu_lower(ips); + + sleep: +- schedule_timeout_interruptible(msecs_to_jiffies(IPS_ADJUST_PERIOD)); ++ schedule_msec_hrtimeout_interruptible((IPS_ADJUST_PERIOD)); + } while (!kthread_should_stop()); + + dev_dbg(&ips->dev->dev, "ips-adjust thread stopped\n"); +@@ -992,7 +992,7 @@ static int ips_monitor(void *data) + seqno_timestamp = get_jiffies_64(); + + old_cpu_power = thm_readl(THM_CEC); +- schedule_timeout_interruptible(msecs_to_jiffies(IPS_SAMPLE_PERIOD)); ++ schedule_msec_hrtimeout_interruptible((IPS_SAMPLE_PERIOD)); + + /* Collect an initial average */ + for (i = 0; i < IPS_SAMPLE_COUNT; i++) { +@@ -1019,7 +1019,7 @@ static int ips_monitor(void *data) + mchp_samples[i] = mchp; + } + +- schedule_timeout_interruptible(msecs_to_jiffies(IPS_SAMPLE_PERIOD)); ++ schedule_msec_hrtimeout_interruptible((IPS_SAMPLE_PERIOD)); + if (kthread_should_stop()) + break; + } +@@ -1046,7 +1046,7 @@ static int ips_monitor(void *data) + * us to reduce the sample frequency if the CPU and GPU are idle. + */ + old_cpu_power = thm_readl(THM_CEC); +- schedule_timeout_interruptible(msecs_to_jiffies(IPS_SAMPLE_PERIOD)); ++ schedule_msec_hrtimeout_interruptible((IPS_SAMPLE_PERIOD)); + last_sample_period = IPS_SAMPLE_PERIOD; + + setup_deferrable_timer_on_stack(&timer, monitor_timeout, +diff --git a/net/core/pktgen.c b/net/core/pktgen.c +index 6e1e10ff433a..be5d6f7142e4 100644 +--- a/net/core/pktgen.c ++++ b/net/core/pktgen.c +@@ -1992,7 +1992,7 @@ static void pktgen_mark_device(const struct pktgen_net *pn, const char *ifname) + mutex_unlock(&pktgen_thread_lock); + pr_debug("%s: waiting for %s to disappear....\n", + __func__, ifname); +- schedule_timeout_interruptible(msecs_to_jiffies(msec_per_try)); ++ schedule_msec_hrtimeout_interruptible((msec_per_try)); + mutex_lock(&pktgen_thread_lock); + + if (++i >= max_tries) { +diff --git a/sound/soc/codecs/wm8350.c b/sound/soc/codecs/wm8350.c +index 2efc5b41ad0f..3e3248c48c6b 100644 +--- a/sound/soc/codecs/wm8350.c ++++ b/sound/soc/codecs/wm8350.c +@@ -236,10 +236,10 @@ static void wm8350_pga_work(struct work_struct *work) + out2->ramp == WM8350_RAMP_UP) { + /* delay is longer over 0dB as increases are larger */ + if (i >= WM8350_OUTn_0dB) +- schedule_timeout_interruptible(msecs_to_jiffies ++ schedule_msec_hrtimeout_interruptible( + (2)); + else +- schedule_timeout_interruptible(msecs_to_jiffies ++ schedule_msec_hrtimeout_interruptible( + (1)); + } else + udelay(50); /* doesn't matter if we delay longer */ +@@ -1123,7 +1123,7 @@ static int wm8350_set_bias_level(struct snd_soc_codec *codec, + (platform->dis_out4 << 6)); + + /* wait for discharge */ +- schedule_timeout_interruptible(msecs_to_jiffies ++ schedule_msec_hrtimeout_interruptible( + (platform-> + cap_discharge_msecs)); + +@@ -1139,7 +1139,7 @@ static int wm8350_set_bias_level(struct snd_soc_codec *codec, + WM8350_VBUFEN); + + /* wait for vmid */ +- schedule_timeout_interruptible(msecs_to_jiffies ++ schedule_msec_hrtimeout_interruptible( + (platform-> + vmid_charge_msecs)); + +@@ -1190,7 +1190,7 @@ static int wm8350_set_bias_level(struct snd_soc_codec *codec, + wm8350_reg_write(wm8350, WM8350_POWER_MGMT_1, pm1); + + /* wait */ +- schedule_timeout_interruptible(msecs_to_jiffies ++ schedule_msec_hrtimeout_interruptible( + (platform-> + vmid_discharge_msecs)); + +@@ -1208,7 +1208,7 @@ static int wm8350_set_bias_level(struct snd_soc_codec *codec, + pm1 | WM8350_OUTPUT_DRAIN_EN); + + /* wait */ +- schedule_timeout_interruptible(msecs_to_jiffies ++ schedule_msec_hrtimeout_interruptible( + (platform->drain_msecs)); + + pm1 &= ~WM8350_BIASEN; +diff --git a/sound/soc/codecs/wm8900.c b/sound/soc/codecs/wm8900.c +index c77b49a29311..fc50456e90a9 100644 +--- a/sound/soc/codecs/wm8900.c ++++ b/sound/soc/codecs/wm8900.c +@@ -1112,7 +1112,7 @@ static int wm8900_set_bias_level(struct snd_soc_codec *codec, + /* Need to let things settle before stopping the clock + * to ensure that restart works, see "Stopping the + * master clock" in the datasheet. */ +- schedule_timeout_interruptible(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout_interruptible((1)); + snd_soc_write(codec, WM8900_REG_POWER2, + WM8900_REG_POWER2_SYSCLK_ENA); + break; +diff --git a/sound/soc/codecs/wm9713.c b/sound/soc/codecs/wm9713.c +index 7e4822185feb..0c85a207446a 100644 +--- a/sound/soc/codecs/wm9713.c ++++ b/sound/soc/codecs/wm9713.c +@@ -199,7 +199,7 @@ static int wm9713_voice_shutdown(struct snd_soc_dapm_widget *w, + + /* Gracefully shut down the voice interface. */ + snd_soc_update_bits(codec, AC97_HANDSET_RATE, 0x0f00, 0x0200); +- schedule_timeout_interruptible(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout_interruptible((1)); + snd_soc_update_bits(codec, AC97_HANDSET_RATE, 0x0f00, 0x0f00); + snd_soc_update_bits(codec, AC97_EXTENDED_MID, 0x1000, 0x1000); + +@@ -868,7 +868,7 @@ static int wm9713_set_pll(struct snd_soc_codec *codec, + wm9713->pll_in = freq_in; + + /* wait 10ms AC97 link frames for the link to stabilise */ +- schedule_timeout_interruptible(msecs_to_jiffies(10)); ++ schedule_msec_hrtimeout_interruptible((10)); + return 0; + } + +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0009-Replace-all-calls-to-schedule_timeout_uninterruptibl.patch b/sys-kernel/linux-sources-redcore-lts/files/0009-Replace-all-calls-to-schedule_timeout_uninterruptibl.patch new file mode 100644 index 00000000..c910f3df --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0009-Replace-all-calls-to-schedule_timeout_uninterruptibl.patch @@ -0,0 +1,160 @@ +From 6044370cf4bbc5e05f5d78f5772c1d88e3153603 Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Mon, 20 Feb 2017 13:30:32 +1100 +Subject: [PATCH 09/16] Replace all calls to schedule_timeout_uninterruptible + of potentially under 50ms to use schedule_msec_hrtimeout_uninterruptible + +--- + drivers/media/pci/cx18/cx18-gpio.c | 4 ++-- + drivers/net/wireless/intel/ipw2x00/ipw2100.c | 4 ++-- + drivers/rtc/rtc-wm8350.c | 6 +++--- + drivers/scsi/lpfc/lpfc_scsi.c | 2 +- + sound/pci/maestro3.c | 4 ++-- + sound/soc/codecs/rt5631.c | 4 ++-- + sound/soc/soc-dapm.c | 2 +- + 7 files changed, 13 insertions(+), 13 deletions(-) + +diff --git a/drivers/media/pci/cx18/cx18-gpio.c b/drivers/media/pci/cx18/cx18-gpio.c +index 012859e6dc7b..206bd08265a5 100644 +--- a/drivers/media/pci/cx18/cx18-gpio.c ++++ b/drivers/media/pci/cx18/cx18-gpio.c +@@ -90,11 +90,11 @@ static void gpio_reset_seq(struct cx18 *cx, u32 active_lo, u32 active_hi, + + /* Assert */ + gpio_update(cx, mask, ~active_lo); +- schedule_timeout_uninterruptible(msecs_to_jiffies(assert_msecs)); ++ schedule_msec_hrtimeout_uninterruptible((assert_msecs)); + + /* Deassert */ + gpio_update(cx, mask, ~active_hi); +- schedule_timeout_uninterruptible(msecs_to_jiffies(recovery_msecs)); ++ schedule_msec_hrtimeout_uninterruptible((recovery_msecs)); + } + + /* +diff --git a/drivers/net/wireless/intel/ipw2x00/ipw2100.c b/drivers/net/wireless/intel/ipw2x00/ipw2100.c +index 19c442cb93e4..448f41782060 100644 +--- a/drivers/net/wireless/intel/ipw2x00/ipw2100.c ++++ b/drivers/net/wireless/intel/ipw2x00/ipw2100.c +@@ -830,7 +830,7 @@ static int ipw2100_hw_send_command(struct ipw2100_priv *priv, + * doesn't seem to have as many firmware restart cycles... + * + * As a test, we're sticking in a 1/100s delay here */ +- schedule_timeout_uninterruptible(msecs_to_jiffies(10)); ++ schedule_msec_hrtimeout_uninterruptible((10)); + + return 0; + +@@ -1281,7 +1281,7 @@ static int ipw2100_start_adapter(struct ipw2100_priv *priv) + IPW_DEBUG_FW("Waiting for f/w initialization to complete...\n"); + i = 5000; + do { +- schedule_timeout_uninterruptible(msecs_to_jiffies(40)); ++ schedule_msec_hrtimeout_uninterruptible((40)); + /* Todo... wait for sync command ... */ + + read_register(priv->net_dev, IPW_REG_INTA, &inta); +diff --git a/drivers/rtc/rtc-wm8350.c b/drivers/rtc/rtc-wm8350.c +index 483c7993516b..fddbaa475066 100644 +--- a/drivers/rtc/rtc-wm8350.c ++++ b/drivers/rtc/rtc-wm8350.c +@@ -119,7 +119,7 @@ static int wm8350_rtc_settime(struct device *dev, struct rtc_time *tm) + /* Wait until confirmation of stopping */ + do { + rtc_ctrl = wm8350_reg_read(wm8350, WM8350_RTC_TIME_CONTROL); +- schedule_timeout_uninterruptible(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout_uninterruptible((1)); + } while (--retries && !(rtc_ctrl & WM8350_RTC_STS)); + + if (!retries) { +@@ -202,7 +202,7 @@ static int wm8350_rtc_stop_alarm(struct wm8350 *wm8350) + /* Wait until confirmation of stopping */ + do { + rtc_ctrl = wm8350_reg_read(wm8350, WM8350_RTC_TIME_CONTROL); +- schedule_timeout_uninterruptible(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout_uninterruptible((1)); + } while (retries-- && !(rtc_ctrl & WM8350_RTC_ALMSTS)); + + if (!(rtc_ctrl & WM8350_RTC_ALMSTS)) +@@ -225,7 +225,7 @@ static int wm8350_rtc_start_alarm(struct wm8350 *wm8350) + /* Wait until confirmation */ + do { + rtc_ctrl = wm8350_reg_read(wm8350, WM8350_RTC_TIME_CONTROL); +- schedule_timeout_uninterruptible(msecs_to_jiffies(1)); ++ schedule_msec_hrtimeout_uninterruptible((1)); + } while (retries-- && rtc_ctrl & WM8350_RTC_ALMSTS); + + if (rtc_ctrl & WM8350_RTC_ALMSTS) +diff --git a/drivers/scsi/lpfc/lpfc_scsi.c b/drivers/scsi/lpfc/lpfc_scsi.c +index 1a6f122bb25d..c0db66302a3e 100644 +--- a/drivers/scsi/lpfc/lpfc_scsi.c ++++ b/drivers/scsi/lpfc/lpfc_scsi.c +@@ -5131,7 +5131,7 @@ lpfc_reset_flush_io_context(struct lpfc_vport *vport, uint16_t tgt_id, + tgt_id, lun_id, context); + later = msecs_to_jiffies(2 * vport->cfg_devloss_tmo * 1000) + jiffies; + while (time_after(later, jiffies) && cnt) { +- schedule_timeout_uninterruptible(msecs_to_jiffies(20)); ++ schedule_msec_hrtimeout_uninterruptible((20)); + cnt = lpfc_sli_sum_iocb(vport, tgt_id, lun_id, context); + } + if (cnt) { +diff --git a/sound/pci/maestro3.c b/sound/pci/maestro3.c +index 8f20dec97843..944ce63431b0 100644 +--- a/sound/pci/maestro3.c ++++ b/sound/pci/maestro3.c +@@ -2016,7 +2016,7 @@ static void snd_m3_ac97_reset(struct snd_m3 *chip) + outw(0, io + GPIO_DATA); + outw(dir | GPO_PRIMARY_AC97, io + GPIO_DIRECTION); + +- schedule_timeout_uninterruptible(msecs_to_jiffies(delay1)); ++ schedule_msec_hrtimeout_uninterruptible((delay1)); + + outw(GPO_PRIMARY_AC97, io + GPIO_DATA); + udelay(5); +@@ -2024,7 +2024,7 @@ static void snd_m3_ac97_reset(struct snd_m3 *chip) + outw(IO_SRAM_ENABLE | SERIAL_AC_LINK_ENABLE, io + RING_BUS_CTRL_A); + outw(~0, io + GPIO_MASK); + +- schedule_timeout_uninterruptible(msecs_to_jiffies(delay2)); ++ schedule_msec_hrtimeout_uninterruptible((delay2)); + + if (! snd_m3_try_read_vendor(chip)) + break; +diff --git a/sound/soc/codecs/rt5631.c b/sound/soc/codecs/rt5631.c +index 55b04c55fb4b..2ed02ad6ac41 100644 +--- a/sound/soc/codecs/rt5631.c ++++ b/sound/soc/codecs/rt5631.c +@@ -419,7 +419,7 @@ static void onebit_depop_mute_stage(struct snd_soc_codec *codec, int enable) + hp_zc = snd_soc_read(codec, RT5631_INT_ST_IRQ_CTRL_2); + snd_soc_write(codec, RT5631_INT_ST_IRQ_CTRL_2, hp_zc & 0xf7ff); + if (enable) { +- schedule_timeout_uninterruptible(msecs_to_jiffies(10)); ++ schedule_msec_hrtimeout_uninterruptible((10)); + /* config one-bit depop parameter */ + rt5631_write_index(codec, RT5631_SPK_INTL_CTRL, 0x307f); + snd_soc_update_bits(codec, RT5631_HP_OUT_VOL, +@@ -529,7 +529,7 @@ static void depop_seq_mute_stage(struct snd_soc_codec *codec, int enable) + hp_zc = snd_soc_read(codec, RT5631_INT_ST_IRQ_CTRL_2); + snd_soc_write(codec, RT5631_INT_ST_IRQ_CTRL_2, hp_zc & 0xf7ff); + if (enable) { +- schedule_timeout_uninterruptible(msecs_to_jiffies(10)); ++ schedule_msec_hrtimeout_uninterruptible((10)); + + /* config depop sequence parameter */ + rt5631_write_index(codec, RT5631_SPK_INTL_CTRL, 0x302f); +diff --git a/sound/soc/soc-dapm.c b/sound/soc/soc-dapm.c +index dcef67a9bd48..11c2bb48c8f2 100644 +--- a/sound/soc/soc-dapm.c ++++ b/sound/soc/soc-dapm.c +@@ -134,7 +134,7 @@ static void dapm_assert_locked(struct snd_soc_dapm_context *dapm) + static void pop_wait(u32 pop_time) + { + if (pop_time) +- schedule_timeout_uninterruptible(msecs_to_jiffies(pop_time)); ++ schedule_msec_hrtimeout_uninterruptible((pop_time)); + } + + static void pop_dbg(struct device *dev, u32 pop_time, const char *fmt, ...) +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0010-Don-t-use-hrtimer-overlay-when-pm_freezing-since-som.patch b/sys-kernel/linux-sources-redcore-lts/files/0010-Don-t-use-hrtimer-overlay-when-pm_freezing-since-som.patch new file mode 100644 index 00000000..260bb98d --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0010-Don-t-use-hrtimer-overlay-when-pm_freezing-since-som.patch @@ -0,0 +1,69 @@ +From 071486de633698dcdd163295173ce4663ec9158c Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Mon, 20 Feb 2017 13:32:58 +1100 +Subject: [PATCH 10/16] Don't use hrtimer overlay when pm_freezing since some + drivers still don't correctly use freezable timeouts. + +--- + kernel/time/hrtimer.c | 2 +- + kernel/time/timer.c | 9 +++++---- + 2 files changed, 6 insertions(+), 5 deletions(-) + +diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c +index 13227cf2814c..66456c72bace 100644 +--- a/kernel/time/hrtimer.c ++++ b/kernel/time/hrtimer.c +@@ -1809,7 +1809,7 @@ long __sched schedule_msec_hrtimeout(long timeout) + * (yet) better than Hz, as would occur during startup, use regular + * timers. + */ +- if (jiffs > 4 || hrtimer_resolution >= NSEC_PER_SEC / HZ) ++ if (jiffs > 4 || hrtimer_resolution >= NSEC_PER_SEC / HZ || pm_freezing) + return schedule_timeout(jiffs); + + secs = timeout / 1000; +diff --git a/kernel/time/timer.c b/kernel/time/timer.c +index c68cb9307f64..2f2c96b03efe 100644 +--- a/kernel/time/timer.c ++++ b/kernel/time/timer.c +@@ -44,6 +44,7 @@ + #include + #include + #include ++#include + + #include + #include +@@ -1891,12 +1892,12 @@ void msleep(unsigned int msecs) + * Use high resolution timers where the resolution of tick based + * timers is inadequate. + */ +- if (jiffs < 5 && hrtimer_resolution < NSEC_PER_SEC / HZ) { ++ if (jiffs < 5 && hrtimer_resolution < NSEC_PER_SEC / HZ && !pm_freezing) { + while (msecs) + msecs = schedule_msec_hrtimeout_uninterruptible(msecs); + return; + } +- timeout = msecs_to_jiffies(msecs) + 1; ++ timeout = jiffs + 1; + + while (timeout) + timeout = schedule_timeout_uninterruptible(timeout); +@@ -1913,12 +1914,12 @@ unsigned long msleep_interruptible(unsigned int msecs) + int jiffs = msecs_to_jiffies(msecs); + unsigned long timeout; + +- if (jiffs < 5 && hrtimer_resolution < NSEC_PER_SEC / HZ) { ++ if (jiffs < 5 && hrtimer_resolution < NSEC_PER_SEC / HZ && !pm_freezing) { + while (msecs && !signal_pending(current)) + msecs = schedule_msec_hrtimeout_interruptible(msecs); + return msecs; + } +- timeout = msecs_to_jiffies(msecs) + 1; ++ timeout = jiffs + 1; + + while (timeout && !signal_pending(current)) + timeout = schedule_timeout_interruptible(timeout); +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0011-Make-hrtimer-granularity-and-minimum-hrtimeout-confi.patch b/sys-kernel/linux-sources-redcore-lts/files/0011-Make-hrtimer-granularity-and-minimum-hrtimeout-confi.patch new file mode 100644 index 00000000..5ac20300 --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0011-Make-hrtimer-granularity-and-minimum-hrtimeout-confi.patch @@ -0,0 +1,136 @@ +diff -Nur a/kernel/sysctl.c b/kernel/sysctl.c +--- a/kernel/sysctl.c 2018-11-03 17:03:07.433069521 +0000 ++++ b/kernel/sysctl.c 2018-11-03 17:02:11.020267246 +0000 +@@ -141,7 +141,9 @@ + extern int sched_iso_cpu; + extern int sched_yield_type; + #endif +-#ifdef CONFIG_PRINTK ++extern int hrtimer_granularity_us; ++extern int hrtimeout_min_us; ++#if defined(CONFIG_PRINTK) || defined(CONFIG_SCHED_MUQSS) + static int ten_thousand __read_only = 10000; + #endif + #ifdef CONFIG_PERF_EVENTS +@@ -1119,6 +1121,24 @@ + .extra2 = &two, + }, + #endif ++ { ++ .procname = "hrtimer_granularity_us", ++ .data = &hrtimer_granularity_us, ++ .maxlen = sizeof(int), ++ .mode = 0644, ++ .proc_handler = &proc_dointvec_minmax, ++ .extra1 = &one, ++ .extra2 = &ten_thousand, ++ }, ++ { ++ .procname = "hrtimeout_min_us", ++ .data = &hrtimeout_min_us, ++ .maxlen = sizeof(int), ++ .mode = 0644, ++ .proc_handler = &proc_dointvec_minmax, ++ .extra1 = &one, ++ .extra2 = &ten_thousand, ++ }, + #if defined(CONFIG_S390) && defined(CONFIG_SMP) + { + .procname = "spin_retry", +diff -Nur a/kernel/time/clockevents.c b/kernel/time/clockevents.c +--- a/kernel/time/clockevents.c 2018-11-03 17:03:07.433069521 +0000 ++++ b/kernel/time/clockevents.c 2018-11-03 16:58:17.283800909 +0000 +@@ -198,13 +198,9 @@ + + #ifdef CONFIG_GENERIC_CLOCKEVENTS_MIN_ADJUST + +-#ifdef CONFIG_SCHED_MUQSS ++int __read_mostly hrtimer_granularity_us = 100; + /* Limit min_delta to 100us */ +-#define MIN_DELTA_LIMIT (NSEC_PER_SEC / 10000) +-#else +-/* Limit min_delta to a jiffie */ +-#define MIN_DELTA_LIMIT (NSEC_PER_SEC / HZ) +-#endif ++#define MIN_DELTA_LIMIT (hrtimer_granularity_us * NSEC_PER_USEC) + + /** + * clockevents_increase_min_delta - raise minimum delta of a clock event device +diff -Nur a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c +--- a/kernel/time/hrtimer.c 2018-11-03 17:04:16.448274547 +0000 ++++ b/kernel/time/hrtimer.c 2018-11-03 16:58:17.283800909 +0000 +@@ -1803,7 +1803,7 @@ + long __sched schedule_msec_hrtimeout(long timeout) + { + struct hrtimer_sleeper t; +- int delta, secs, jiffs; ++ int delta, jiffs; + ktime_t expires; + + if (!timeout) { +@@ -1820,9 +1820,8 @@ + if (jiffs > 4 || hrtimer_resolution >= NSEC_PER_SEC / HZ || pm_freezing) + return schedule_timeout(jiffs); + +- secs = timeout / 1000; + delta = (timeout % 1000) * NSEC_PER_MSEC; +- expires = ktime_set(secs, delta); ++ expires = ktime_set(0, delta); + + hrtimer_init_on_stack(&t.timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + hrtimer_set_expires_range_ns(&t.timer, expires, delta); +@@ -1846,9 +1845,53 @@ + + EXPORT_SYMBOL(schedule_msec_hrtimeout); + ++#define USECS_PER_SEC 1000000 ++extern int hrtimer_granularity_us; ++ ++static inline long schedule_usec_hrtimeout(long timeout) ++{ ++ struct hrtimer_sleeper t; ++ ktime_t expires; ++ int delta; ++ ++ if (!timeout) { ++ __set_current_state(TASK_RUNNING); ++ return 0; ++ } ++ ++ if (hrtimer_resolution >= NSEC_PER_SEC / HZ) ++ return schedule_timeout(usecs_to_jiffies(timeout)); ++ ++ if (timeout < hrtimer_granularity_us) ++ timeout = hrtimer_granularity_us; ++ delta = (timeout % USECS_PER_SEC) * NSEC_PER_USEC; ++ expires = ktime_set(0, delta); ++ ++ hrtimer_init_on_stack(&t.timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); ++ hrtimer_set_expires_range_ns(&t.timer, expires, delta); ++ ++ hrtimer_init_sleeper(&t, current); ++ ++ hrtimer_start_expires(&t.timer, HRTIMER_MODE_REL); ++ ++ if (likely(t.task)) ++ schedule(); ++ ++ hrtimer_cancel(&t.timer); ++ destroy_hrtimer_on_stack(&t.timer); ++ ++ __set_current_state(TASK_RUNNING); ++ ++ expires = hrtimer_expires_remaining(&t.timer); ++ timeout = ktime_to_us(expires); ++ return timeout < 0 ? 0 : timeout; ++} ++ ++int __read_mostly hrtimeout_min_us = 1000; ++ + long __sched schedule_min_hrtimeout(void) + { +- return schedule_msec_hrtimeout(1); ++ return usecs_to_jiffies(schedule_usec_hrtimeout(hrtimeout_min_us)); + } + + EXPORT_SYMBOL(schedule_min_hrtimeout); diff --git a/sys-kernel/linux-sources-redcore-lts/files/0012-Reinstate-default-Hz-of-100-in-combination-with-MuQS.patch b/sys-kernel/linux-sources-redcore-lts/files/0012-Reinstate-default-Hz-of-100-in-combination-with-MuQS.patch new file mode 100644 index 00000000..99b28d65 --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0012-Reinstate-default-Hz-of-100-in-combination-with-MuQS.patch @@ -0,0 +1,81 @@ +From 9e47a80f690080c12ce607158b96c305707543b8 Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Wed, 7 Dec 2016 21:23:01 +1100 +Subject: [PATCH 12/16] Reinstate default Hz of 100 in combination with MuQSS + and -ck patches. + +--- + kernel/Kconfig.hz | 25 ++++++++++++++++++------- + 1 file changed, 18 insertions(+), 7 deletions(-) + +diff --git a/kernel/Kconfig.hz b/kernel/Kconfig.hz +index 2a202a846757..1806fcac8f14 100644 +--- a/kernel/Kconfig.hz ++++ b/kernel/Kconfig.hz +@@ -4,7 +4,8 @@ + + choice + prompt "Timer frequency" +- default HZ_250 ++ default HZ_100 if SCHED_MUQSS ++ default HZ_250_NODEF if !SCHED_MUQSS + help + Allows the configuration of the timer frequency. It is customary + to have the timer interrupt run at 1000 Hz but 100 Hz may be more +@@ -19,11 +20,18 @@ choice + config HZ_100 + bool "100 HZ" + help ++ 100 Hz is a suitable choice in combination with MuQSS which does ++ not rely on ticks for rescheduling interrupts, and is not Hz limited ++ for timeouts and sleeps from both the kernel and userspace. ++ This allows us to benefit from the lower overhead and higher ++ throughput of fewer timer ticks. ++ ++ Non-MuQSS kernels: + 100 Hz is a typical choice for servers, SMP and NUMA systems + with lots of processors that may show reduced performance if + too many timer interrupts are occurring. + +- config HZ_250 ++ config HZ_250_NODEF + bool "250 HZ" + help + 250 Hz is a good compromise choice allowing server performance +@@ -31,7 +39,10 @@ choice + on SMP and NUMA systems. If you are going to be using NTSC video + or multimedia, selected 300Hz instead. + +- config HZ_300 ++ 250 Hz is the default choice for the mainline scheduler but not ++ advantageous in combination with MuQSS. ++ ++ config HZ_300_NODEF + bool "300 HZ" + help + 300 Hz is a good compromise choice allowing server performance +@@ -39,7 +50,7 @@ choice + on SMP and NUMA systems and exactly dividing by both PAL and + NTSC frame rates for video and multimedia work. + +- config HZ_1000 ++ config HZ_1000_NODEF + bool "1000 HZ" + help + 1000 Hz is the preferred choice for desktop systems and other +@@ -50,9 +61,9 @@ endchoice + config HZ + int + default 100 if HZ_100 +- default 250 if HZ_250 +- default 300 if HZ_300 +- default 1000 if HZ_1000 ++ default 250 if HZ_250_NODEF ++ default 300 if HZ_300_NODEF ++ default 1000 if HZ_1000_NODEF + + config SCHED_HRTICK + def_bool HIGH_RES_TIMERS +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0013-Make-threaded-IRQs-optionally-the-default-which-can-.patch b/sys-kernel/linux-sources-redcore-lts/files/0013-Make-threaded-IRQs-optionally-the-default-which-can-.patch new file mode 100644 index 00000000..63ec9fdf --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0013-Make-threaded-IRQs-optionally-the-default-which-can-.patch @@ -0,0 +1,61 @@ +From 5902b315d4061ebbe73a62c52e6d3b618066cebc Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Wed, 7 Dec 2016 21:13:16 +1100 +Subject: [PATCH 13/16] Make threaded IRQs optionally the default which can be + disabled. + +--- + kernel/irq/Kconfig | 14 ++++++++++++++ + kernel/irq/manage.c | 10 ++++++++++ + 2 files changed, 24 insertions(+) + +diff --git a/kernel/irq/Kconfig b/kernel/irq/Kconfig +index a117adf7084b..0984c54fd4e9 100644 +--- a/kernel/irq/Kconfig ++++ b/kernel/irq/Kconfig +@@ -111,6 +111,20 @@ config IRQ_DOMAIN_DEBUG + config IRQ_FORCED_THREADING + bool + ++config FORCE_IRQ_THREADING ++ bool "Make IRQ threading compulsory" ++ depends on IRQ_FORCED_THREADING ++ default y ++ ---help--- ++ ++ Make IRQ threading mandatory for any IRQ handlers that support it ++ instead of being optional and requiring the threadirqs kernel ++ parameter. Instead they can be optionally disabled with the ++ nothreadirqs kernel parameter. ++ ++ Enable if you are building for a desktop or low latency system, ++ otherwise say N. ++ + config SPARSE_IRQ + bool "Support sparse irq numbering" if MAY_HAVE_SPARSE_IRQ + ---help--- +diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c +index 4bff6a10ae8e..5a6df0dd23c4 100644 +--- a/kernel/irq/manage.c ++++ b/kernel/irq/manage.c +@@ -24,7 +24,17 @@ + #include "internals.h" + + #ifdef CONFIG_IRQ_FORCED_THREADING ++#ifdef CONFIG_FORCE_IRQ_THREADING ++__read_mostly bool force_irqthreads = true; ++#else + __read_mostly bool force_irqthreads; ++#endif ++static int __init setup_noforced_irqthreads(char *arg) ++{ ++ force_irqthreads = false; ++ return 0; ++} ++early_param("nothreadirqs", setup_noforced_irqthreads); + + static int __init setup_forced_irqthreads(char *arg) + { +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0014-Swap-sucks.patch b/sys-kernel/linux-sources-redcore-lts/files/0014-Swap-sucks.patch new file mode 100644 index 00000000..6bf5bcda --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0014-Swap-sucks.patch @@ -0,0 +1,25 @@ +From ed0ab4c80fcb6fa4abb4f2f897e591df6eaa2d0e Mon Sep 17 00:00:00 2001 +From: Con Kolivas +Date: Sat, 12 Aug 2017 12:02:04 +1000 +Subject: [PATCH 14/16] Swap sucks. + +--- + mm/vmscan.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/mm/vmscan.c b/mm/vmscan.c +index eb2f0315b8c0..67d03efab288 100644 +--- a/mm/vmscan.c ++++ b/mm/vmscan.c +@@ -149,7 +149,7 @@ struct scan_control { + /* + * From 0 .. 100. Higher means more swappy. + */ +-int vm_swappiness = 60; ++int vm_swappiness = 33; + /* + * The total number of pages which are beyond the high watermark within all + * zones. +-- +2.11.0 + diff --git a/sys-kernel/linux-sources-redcore-lts/files/0015-Enable-BFQ-io-scheduler-by-default.patch b/sys-kernel/linux-sources-redcore-lts/files/0015-Enable-BFQ-io-scheduler-by-default.patch deleted file mode 100644 index d12753be..00000000 --- a/sys-kernel/linux-sources-redcore-lts/files/0015-Enable-BFQ-io-scheduler-by-default.patch +++ /dev/null @@ -1,38 +0,0 @@ -From 0e7ab31fb218e2a18fbecd19c24dfaae14c88afd Mon Sep 17 00:00:00 2001 -From: Con Kolivas -Date: Mon, 20 Nov 2017 18:02:03 +1100 -Subject: [PATCH 15/18] Enable BFQ io scheduler by default. - ---- - block/Kconfig.iosched | 2 +- - drivers/scsi/Kconfig | 1 + - 2 files changed, 2 insertions(+), 1 deletion(-) - -diff --git a/block/Kconfig.iosched b/block/Kconfig.iosched -index a4a8914bf7a4..2d9be91e8e87 100644 ---- a/block/Kconfig.iosched -+++ b/block/Kconfig.iosched -@@ -82,7 +82,7 @@ config MQ_IOSCHED_KYBER - - config IOSCHED_BFQ - tristate "BFQ I/O scheduler" -- default n -+ default y - ---help--- - BFQ I/O scheduler for BLK-MQ. BFQ distributes the bandwidth of - of the device among all processes according to their weights, -diff --git a/drivers/scsi/Kconfig b/drivers/scsi/Kconfig -index 8a739b74cfb7..9e939ee76e72 100644 ---- a/drivers/scsi/Kconfig -+++ b/drivers/scsi/Kconfig -@@ -50,6 +50,7 @@ config SCSI_NETLINK - config SCSI_MQ_DEFAULT - bool "SCSI: use blk-mq I/O path by default" - depends on SCSI -+ default y - ---help--- - This option enables the new blk-mq based I/O path for SCSI - devices by default. With the option the scsi_mod.use_blk_mq --- -2.14.1 - diff --git a/sys-kernel/linux-sources-redcore-lts/files/0015-MuQSS.c-needs-irq_regs.h-to-use-get_irq_regs.patch b/sys-kernel/linux-sources-redcore-lts/files/0015-MuQSS.c-needs-irq_regs.h-to-use-get_irq_regs.patch new file mode 100644 index 00000000..bfa509a5 --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0015-MuQSS.c-needs-irq_regs.h-to-use-get_irq_regs.patch @@ -0,0 +1,19 @@ +diff --git a/kernel/sched/MuQSS.c b/kernel/sched/MuQSS.c +index e84d700709ff6..16364915cff53 100644 +--- a/kernel/sched/MuQSS.c ++++ b/kernel/sched/MuQSS.c +@@ -70,6 +70,14 @@ + + #include "MuQSS.h" + ++/* needing to include irq_regs.h, "because reasons"... ++ * implicit declaration of function ‘get_irq_regs’; ++ * did you mean ‘get_ibs_caps’? ++ * [-Werror=implicit-function-declaration] ++ * ^ this is because autodetect is not flawless ++ */ ++#include ++ + #define rt_prio(prio) unlikely((prio) < MAX_RT_PRIO) + #define rt_task(p) rt_prio((p)->prio) + #define batch_task(p) (unlikely((p)->policy == SCHED_BATCH)) diff --git a/sys-kernel/linux-sources-redcore-lts/files/0016-unfuck-MuQSS-on-linux-4_14_15+.patch b/sys-kernel/linux-sources-redcore-lts/files/0016-unfuck-MuQSS-on-linux-4_14_15+.patch new file mode 100644 index 00000000..f7dc1d1c --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/0016-unfuck-MuQSS-on-linux-4_14_15+.patch @@ -0,0 +1,48 @@ +diff --git a/kernel/sched/MuQSS.c b/kernel/sched/MuQSS.c +index e84d700709ff6..b0be7fcfe41f9 100644 +--- a/kernel/sched/MuQSS.c ++++ b/kernel/sched/MuQSS.c +@@ -55,6 +55,7 @@ + #include + #include + #include ++#include + + #include + #include +@@ -1959,7 +1960,11 @@ try_to_wake_up(struct task_struct *p, unsigned int state, int wake_flags) + p->state = TASK_WAKING; + + if (p->in_iowait) { ++#if LINUX_VERSION_CODE < KERNEL_VERSION(4, 14, 15) + delayacct_blkio_end(); ++#else ++ delayacct_blkio_end(p); ++#endif + atomic_dec(&task_rq(p)->nr_iowait); + } + +@@ -1970,7 +1975,11 @@ try_to_wake_up(struct task_struct *p, unsigned int state, int wake_flags) + #else /* CONFIG_SMP */ + + if (p->in_iowait) { ++#if LINUX_VERSION_CODE < KERNEL_VERSION(4, 14, 15) + delayacct_blkio_end(); ++#else ++ delayacct_blkio_end(p); ++#endif + atomic_dec(&task_rq(p)->nr_iowait); + } + +@@ -2022,7 +2031,11 @@ static void try_to_wake_up_local(struct task_struct *p) + + if (!task_on_rq_queued(p)) { + if (p->in_iowait) { ++#if LINUX_VERSION_CODE < KERNEL_VERSION(4, 14, 15) + delayacct_blkio_end(); ++#else ++ delayacct_blkio_end(p); ++#endif + atomic_dec(&rq->nr_iowait); + } + ttwu_activate(rq, p); diff --git a/sys-kernel/linux-sources-redcore-lts/files/redcore-lts-amd64.config b/sys-kernel/linux-sources-redcore-lts/files/redcore-lts-amd64.config index 73c7d194..f41bc39d 100644 --- a/sys-kernel/linux-sources-redcore-lts/files/redcore-lts-amd64.config +++ b/sys-kernel/linux-sources-redcore-lts/files/redcore-lts-amd64.config @@ -1,6 +1,6 @@ # # Automatically generated file; DO NOT EDIT. -# Linux/x86 4.14.65-redcore-lts Kernel Configuration +# Linux/x86 4.14.75-redcore-lts Kernel Configuration # CONFIG_64BIT=y CONFIG_X86_64=y @@ -50,6 +50,7 @@ CONFIG_THREAD_INFO_IN_TASK=y # # General setup # +CONFIG_SCHED_MUQSS=y CONFIG_INIT_ENV_ARG_LIMIT=32 CONFIG_CROSS_COMPILE="" # CONFIG_COMPILE_TEST is not set @@ -98,6 +99,7 @@ CONFIG_GENERIC_MSI_IRQ=y CONFIG_GENERIC_MSI_IRQ_DOMAIN=y # CONFIG_IRQ_DOMAIN_DEBUG is not set CONFIG_IRQ_FORCED_THREADING=y +CONFIG_FORCE_IRQ_THREADING=y CONFIG_SPARSE_IRQ=y # CONFIG_GENERIC_IRQ_DEBUGFS is not set CONFIG_CLOCKSOURCE_WATCHDOG=y @@ -113,11 +115,9 @@ CONFIG_GENERIC_CMOS_UPDATE=y # Timers subsystem # CONFIG_TICK_ONESHOT=y -CONFIG_NO_HZ_COMMON=y -# CONFIG_HZ_PERIODIC is not set +CONFIG_HZ_PERIODIC=y # CONFIG_NO_HZ_IDLE is not set -CONFIG_NO_HZ_FULL=y -# CONFIG_NO_HZ_FULL_ALL is not set +# CONFIG_NO_HZ_FULL is not set CONFIG_NO_HZ=y CONFIG_HIGH_RES_TIMERS=y @@ -125,6 +125,7 @@ CONFIG_HIGH_RES_TIMERS=y # CPU/Task time and stats accounting # CONFIG_VIRT_CPU_ACCOUNTING=y +# CONFIG_TICK_CPU_ACCOUNTING is not set CONFIG_VIRT_CPU_ACCOUNTING_GEN=y CONFIG_IRQ_TIME_ACCOUNTING=y CONFIG_BSD_PROCESS_ACCT=y @@ -146,7 +147,6 @@ CONFIG_RCU_STALL_COMMON=y CONFIG_RCU_NEED_SEGCBLIST=y CONFIG_CONTEXT_TRACKING=y # CONFIG_CONTEXT_TRACKING_FORCE is not set -CONFIG_RCU_NOCB_CPU=y CONFIG_BUILD_BIN2C=y CONFIG_IKCONFIG=y CONFIG_IKCONFIG_PROC=y @@ -157,8 +157,6 @@ CONFIG_HAVE_UNSTABLE_SCHED_CLOCK=y CONFIG_ARCH_SUPPORTS_NUMA_BALANCING=y CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH=y CONFIG_ARCH_SUPPORTS_INT128=y -CONFIG_NUMA_BALANCING=y -CONFIG_NUMA_BALANCING_DEFAULT_ENABLED=y CONFIG_CGROUPS=y CONFIG_PAGE_COUNTER=y CONFIG_MEMCG=y @@ -168,9 +166,6 @@ CONFIG_BLK_CGROUP=y # CONFIG_DEBUG_BLK_CGROUP is not set CONFIG_CGROUP_WRITEBACK=y CONFIG_CGROUP_SCHED=y -CONFIG_FAIR_GROUP_SCHED=y -CONFIG_CFS_BANDWIDTH=y -CONFIG_RT_GROUP_SCHED=y CONFIG_CGROUP_PIDS=y # CONFIG_CGROUP_RDMA is not set CONFIG_CGROUP_FREEZER=y @@ -178,7 +173,6 @@ CONFIG_CGROUP_HUGETLB=y CONFIG_CPUSETS=y CONFIG_PROC_PID_CPUSET=y CONFIG_CGROUP_DEVICE=y -CONFIG_CGROUP_CPUACCT=y CONFIG_CGROUP_PERF=y CONFIG_CGROUP_BPF=y # CONFIG_CGROUP_DEBUG is not set @@ -190,7 +184,6 @@ CONFIG_IPC_NS=y CONFIG_USER_NS=y CONFIG_PID_NS=y CONFIG_NET_NS=y -CONFIG_SCHED_AUTOGROUP=y # CONFIG_SYSFS_DEPRECATED is not set CONFIG_RELAY=y CONFIG_BLK_DEV_INITRD=y @@ -306,6 +299,7 @@ CONFIG_HAVE_PERF_REGS=y CONFIG_HAVE_PERF_USER_STACK_DUMP=y CONFIG_HAVE_ARCH_JUMP_LABEL=y CONFIG_HAVE_RCU_TABLE_FREE=y +CONFIG_HAVE_RCU_TABLE_INVALIDATE=y CONFIG_ARCH_HAVE_NMI_SAFE_CMPXCHG=y CONFIG_HAVE_ALIGNED_STRUCT_PAGE=y CONFIG_HAVE_CMPXCHG_LOCAL=y @@ -438,10 +432,15 @@ CONFIG_IOSCHED_NOOP=y CONFIG_IOSCHED_DEADLINE=y CONFIG_IOSCHED_CFQ=y CONFIG_CFQ_GROUP_IOSCHED=y +CONFIG_IOSCHED_BFQ_SQ=y +CONFIG_BFQ_SQ_GROUP_IOSCHED=y # CONFIG_DEFAULT_DEADLINE is not set -CONFIG_DEFAULT_CFQ=y +# CONFIG_DEFAULT_CFQ is not set +CONFIG_DEFAULT_BFQ_SQ=y # CONFIG_DEFAULT_NOOP is not set -CONFIG_DEFAULT_IOSCHED="cfq" +CONFIG_DEFAULT_IOSCHED="bfq-sq" +CONFIG_MQ_IOSCHED_BFQ=y +CONFIG_MQ_BFQ_GROUP_IOSCHED=y CONFIG_MQ_IOSCHED_DEADLINE=y # CONFIG_MQ_IOSCHED_KYBER is not set CONFIG_IOSCHED_BFQ=y @@ -515,6 +514,7 @@ CONFIG_IOMMU_HELPER=y CONFIG_MAXSMP=y CONFIG_NR_CPUS=8192 CONFIG_SCHED_SMT=y +CONFIG_SMT_NICE=y CONFIG_SCHED_MC=y CONFIG_SCHED_MC_PRIO=y # CONFIG_PREEMPT_NONE is not set @@ -655,11 +655,11 @@ CONFIG_EFI=y CONFIG_EFI_STUB=y CONFIG_EFI_MIXED=y CONFIG_SECCOMP=y -# CONFIG_HZ_100 is not set -# CONFIG_HZ_250 is not set -# CONFIG_HZ_300 is not set -CONFIG_HZ_1000=y -CONFIG_HZ=1000 +CONFIG_HZ_100=y +# CONFIG_HZ_250_NODEF is not set +# CONFIG_HZ_300_NODEF is not set +# CONFIG_HZ_1000_NODEF is not set +CONFIG_HZ=100 CONFIG_SCHED_HRTICK=y CONFIG_KEXEC=y # CONFIG_CRASH_DUMP is not set @@ -4083,6 +4083,7 @@ CONFIG_HSI_BOARDINFO=y CONFIG_HSI_CHAR=m CONFIG_PPS=m # CONFIG_PPS_DEBUG is not set +# CONFIG_NTP_PPS is not set # # PPS clients support diff --git a/sys-kernel/linux-sources-redcore-lts/files/uksm-for-linux-hardened.patch b/sys-kernel/linux-sources-redcore-lts/files/uksm-for-linux-hardened.patch deleted file mode 100644 index f0596117..00000000 --- a/sys-kernel/linux-sources-redcore-lts/files/uksm-for-linux-hardened.patch +++ /dev/null @@ -1,6919 +0,0 @@ -diff -Nur a/Documentation/vm/00-INDEX b/Documentation/vm/00-INDEX ---- a/Documentation/vm/00-INDEX 2018-05-25 15:18:02.000000000 +0100 -+++ b/Documentation/vm/00-INDEX 2018-05-26 19:30:55.783140311 +0100 -@@ -20,6 +20,8 @@ - - description of the idle page tracking feature. - ksm.txt - - how to use the Kernel Samepage Merging feature. -+uksm.txt -+ - Introduction to Ultra KSM - numa - - information about NUMA specific code in the Linux vm. - numa_memory_policy.txt -diff -Nur a/Documentation/vm/uksm.txt b/Documentation/vm/uksm.txt ---- a/Documentation/vm/uksm.txt 1970-01-01 01:00:00.000000000 +0100 -+++ b/Documentation/vm/uksm.txt 2018-05-26 19:30:55.783140311 +0100 -@@ -0,0 +1,61 @@ -+The Ultra Kernel Samepage Merging feature -+---------------------------------------------- -+/* -+ * Ultra KSM. Copyright (C) 2011-2012 Nai Xia -+ * -+ * This is an improvement upon KSM. Some basic data structures and routines -+ * are borrowed from ksm.c . -+ * -+ * Its new features: -+ * 1. Full system scan: -+ * It automatically scans all user processes' anonymous VMAs. Kernel-user -+ * interaction to submit a memory area to KSM is no longer needed. -+ * -+ * 2. Rich area detection: -+ * It automatically detects rich areas containing abundant duplicated -+ * pages based. Rich areas are given a full scan speed. Poor areas are -+ * sampled at a reasonable speed with very low CPU consumption. -+ * -+ * 3. Ultra Per-page scan speed improvement: -+ * A new hash algorithm is proposed. As a result, on a machine with -+ * Core(TM)2 Quad Q9300 CPU in 32-bit mode and 800MHZ DDR2 main memory, it -+ * can scan memory areas that does not contain duplicated pages at speed of -+ * 627MB/sec ~ 2445MB/sec and can merge duplicated areas at speed of -+ * 477MB/sec ~ 923MB/sec. -+ * -+ * 4. Thrashing area avoidance: -+ * Thrashing area(an VMA that has frequent Ksm page break-out) can be -+ * filtered out. My benchmark shows it's more efficient than KSM's per-page -+ * hash value based volatile page detection. -+ * -+ * -+ * 5. Misc changes upon KSM: -+ * * It has a fully x86-opitmized memcmp dedicated for 4-byte-aligned page -+ * comparison. It's much faster than default C version on x86. -+ * * rmap_item now has an struct *page member to loosely cache a -+ * address-->page mapping, which reduces too much time-costly -+ * follow_page(). -+ * * The VMA creation/exit procedures are hooked to let the Ultra KSM know. -+ * * try_to_merge_two_pages() now can revert a pte if it fails. No break_ -+ * ksm is needed for this case. -+ * -+ * 6. Full Zero Page consideration(contributed by Figo Zhang) -+ * Now uksmd consider full zero pages as special pages and merge them to an -+ * special unswappable uksm zero page. -+ */ -+ -+ChangeLog: -+ -+2012-05-05 The creation of this Doc -+2012-05-08 UKSM 0.1.1.1 libc crash bug fix, api clean up, doc clean up. -+2012-05-28 UKSM 0.1.1.2 bug fix release -+2012-06-26 UKSM 0.1.2-beta1 first beta release for 0.1.2 -+2012-07-2 UKSM 0.1.2-beta2 -+2012-07-10 UKSM 0.1.2-beta3 -+2012-07-26 UKSM 0.1.2 Fine grained speed control, more scan optimization. -+2012-10-13 UKSM 0.1.2.1 Bug fixes. -+2012-12-31 UKSM 0.1.2.2 Minor bug fixes. -+2014-07-02 UKSM 0.1.2.3 Fix a " __this_cpu_read() in preemptible bug". -+2015-04-22 UKSM 0.1.2.4 Fix a race condition that can sometimes trigger anonying warnings. -+2016-09-10 UKSM 0.1.2.5 Fix a bug in dedup ratio calculation. -+2017-02-26 UKSM 0.1.2.6 Fix a bug in hugetlbpage handling and a race bug with page migration. -diff -Nur a/fs/exec.c b/fs/exec.c ---- a/fs/exec.c 2018-05-26 19:24:34.831782903 +0100 -+++ b/fs/exec.c 2018-05-26 19:31:18.404873956 +0100 -@@ -63,6 +63,7 @@ - #include - #include - #include -+#include - - #include - #include -@@ -1377,6 +1378,7 @@ - /* An exec changes our domain. We are no longer part of the thread - group */ - current->self_exec_id++; -+ - flush_signal_handlers(current, 0); - } - EXPORT_SYMBOL(setup_new_exec); -diff -Nur a/fs/proc/meminfo.c b/fs/proc/meminfo.c ---- a/fs/proc/meminfo.c 2018-05-25 15:18:02.000000000 +0100 -+++ b/fs/proc/meminfo.c 2018-05-26 19:30:55.784140344 +0100 -@@ -118,6 +118,10 @@ - global_zone_page_state(NR_KERNEL_STACK_KB)); - show_val_kb(m, "PageTables: ", - global_zone_page_state(NR_PAGETABLE)); -+#ifdef CONFIG_UKSM -+ show_val_kb(m, "KsmZeroPages: ", -+ global_zone_page_state(NR_UKSM_ZERO_PAGES)); -+#endif - #ifdef CONFIG_QUICKLIST - show_val_kb(m, "Quicklists: ", quicklist_total_size()); - #endif -diff -Nur a/include/asm-generic/pgtable.h b/include/asm-generic/pgtable.h ---- a/include/asm-generic/pgtable.h 2018-05-25 15:18:02.000000000 +0100 -+++ b/include/asm-generic/pgtable.h 2018-05-26 19:30:55.784140344 +0100 -@@ -789,12 +789,25 @@ - extern void untrack_pfn_moved(struct vm_area_struct *vma); - #endif - -+#ifdef CONFIG_UKSM -+static inline int is_uksm_zero_pfn(unsigned long pfn) -+{ -+ extern unsigned long uksm_zero_pfn; -+ return pfn == uksm_zero_pfn; -+} -+#else -+static inline int is_uksm_zero_pfn(unsigned long pfn) -+{ -+ return 0; -+} -+#endif -+ - #ifdef __HAVE_COLOR_ZERO_PAGE - static inline int is_zero_pfn(unsigned long pfn) - { - extern unsigned long zero_pfn; - unsigned long offset_from_zero_pfn = pfn - zero_pfn; -- return offset_from_zero_pfn <= (zero_page_mask >> PAGE_SHIFT); -+ return offset_from_zero_pfn <= (zero_page_mask >> PAGE_SHIFT) || is_uksm_zero_pfn(pfn); - } - - #define my_zero_pfn(addr) page_to_pfn(ZERO_PAGE(addr)) -@@ -803,7 +816,7 @@ - static inline int is_zero_pfn(unsigned long pfn) - { - extern unsigned long zero_pfn; -- return pfn == zero_pfn; -+ return (pfn == zero_pfn) || (is_uksm_zero_pfn(pfn)); - } - - static inline unsigned long my_zero_pfn(unsigned long addr) -diff -Nur a/include/linux/ksm.h b/include/linux/ksm.h ---- a/include/linux/ksm.h 2018-05-25 15:18:02.000000000 +0100 -+++ b/include/linux/ksm.h 2018-05-26 19:30:55.784140344 +0100 -@@ -21,21 +21,6 @@ - #ifdef CONFIG_KSM - int ksm_madvise(struct vm_area_struct *vma, unsigned long start, - unsigned long end, int advice, unsigned long *vm_flags); --int __ksm_enter(struct mm_struct *mm); --void __ksm_exit(struct mm_struct *mm); -- --static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm) --{ -- if (test_bit(MMF_VM_MERGEABLE, &oldmm->flags)) -- return __ksm_enter(mm); -- return 0; --} -- --static inline void ksm_exit(struct mm_struct *mm) --{ -- if (test_bit(MMF_VM_MERGEABLE, &mm->flags)) -- __ksm_exit(mm); --} - - static inline struct stable_node *page_stable_node(struct page *page) - { -@@ -65,6 +50,33 @@ - void rmap_walk_ksm(struct page *page, struct rmap_walk_control *rwc); - void ksm_migrate_page(struct page *newpage, struct page *oldpage); - -+#ifdef CONFIG_KSM_LEGACY -+int __ksm_enter(struct mm_struct *mm); -+void __ksm_exit(struct mm_struct *mm); -+static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm) -+{ -+ if (test_bit(MMF_VM_MERGEABLE, &oldmm->flags)) -+ return __ksm_enter(mm); -+ return 0; -+} -+ -+static inline void ksm_exit(struct mm_struct *mm) -+{ -+ if (test_bit(MMF_VM_MERGEABLE, &mm->flags)) -+ __ksm_exit(mm); -+} -+ -+#elif defined(CONFIG_UKSM) -+static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm) -+{ -+ return 0; -+} -+ -+static inline void ksm_exit(struct mm_struct *mm) -+{ -+} -+#endif /* !CONFIG_UKSM */ -+ - #else /* !CONFIG_KSM */ - - static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm) -@@ -106,4 +118,6 @@ - #endif /* CONFIG_MMU */ - #endif /* !CONFIG_KSM */ - -+#include -+ - #endif /* __LINUX_KSM_H */ -diff -Nur a/include/linux/mm_types.h b/include/linux/mm_types.h ---- a/include/linux/mm_types.h 2018-05-25 15:18:02.000000000 +0100 -+++ b/include/linux/mm_types.h 2018-05-26 19:30:55.784140344 +0100 -@@ -337,6 +337,9 @@ - struct mempolicy *vm_policy; /* NUMA policy for the VMA */ - #endif - struct vm_userfaultfd_ctx vm_userfaultfd_ctx; -+#ifdef CONFIG_UKSM -+ struct vma_slot *uksm_vma_slot; -+#endif - } __randomize_layout; - - struct core_thread { -diff -Nur a/include/linux/mmzone.h b/include/linux/mmzone.h ---- a/include/linux/mmzone.h 2018-05-25 15:18:02.000000000 +0100 -+++ b/include/linux/mmzone.h 2018-05-26 19:30:55.785140376 +0100 -@@ -148,6 +148,9 @@ - NR_ZSPAGES, /* allocated in zsmalloc */ - #endif - NR_FREE_CMA_PAGES, -+#ifdef CONFIG_UKSM -+ NR_UKSM_ZERO_PAGES, -+#endif - NR_VM_ZONE_STAT_ITEMS }; - - enum node_stat_item { -@@ -872,7 +875,7 @@ - } - - /** -- * is_highmem - helper function to quickly check if a struct zone is a -+ * is_highmem - helper function to quickly check if a struct zone is a - * highmem zone or not. This is an attempt to keep references - * to ZONE_{DMA/NORMAL/HIGHMEM/etc} in general code to a minimum. - * @zone - pointer to struct zone variable -diff -Nur a/include/linux/sradix-tree.h b/include/linux/sradix-tree.h ---- a/include/linux/sradix-tree.h 1970-01-01 01:00:00.000000000 +0100 -+++ b/include/linux/sradix-tree.h 2018-05-26 19:30:55.785140376 +0100 -@@ -0,0 +1,77 @@ -+#ifndef _LINUX_SRADIX_TREE_H -+#define _LINUX_SRADIX_TREE_H -+ -+ -+#define INIT_SRADIX_TREE(root, mask) \ -+do { \ -+ (root)->height = 0; \ -+ (root)->gfp_mask = (mask); \ -+ (root)->rnode = NULL; \ -+} while (0) -+ -+#define ULONG_BITS (sizeof(unsigned long) * 8) -+#define SRADIX_TREE_INDEX_BITS (8 /* CHAR_BIT */ * sizeof(unsigned long)) -+//#define SRADIX_TREE_MAP_SHIFT 6 -+//#define SRADIX_TREE_MAP_SIZE (1UL << SRADIX_TREE_MAP_SHIFT) -+//#define SRADIX_TREE_MAP_MASK (SRADIX_TREE_MAP_SIZE-1) -+ -+struct sradix_tree_node { -+ unsigned int height; /* Height from the bottom */ -+ unsigned int count; -+ unsigned int fulls; /* Number of full sublevel trees */ -+ struct sradix_tree_node *parent; -+ void *stores[0]; -+}; -+ -+/* A simple radix tree implementation */ -+struct sradix_tree_root { -+ unsigned int height; -+ struct sradix_tree_node *rnode; -+ -+ /* Where found to have available empty stores in its sublevels */ -+ struct sradix_tree_node *enter_node; -+ unsigned int shift; -+ unsigned int stores_size; -+ unsigned int mask; -+ unsigned long min; /* The first hole index */ -+ unsigned long num; -+ //unsigned long *height_to_maxindex; -+ -+ /* How the node is allocated and freed. */ -+ struct sradix_tree_node *(*alloc)(void); -+ void (*free)(struct sradix_tree_node *node); -+ -+ /* When a new node is added and removed */ -+ void (*extend)(struct sradix_tree_node *parent, struct sradix_tree_node *child); -+ void (*assign)(struct sradix_tree_node *node, unsigned int index, void *item); -+ void (*rm)(struct sradix_tree_node *node, unsigned int offset); -+}; -+ -+struct sradix_tree_path { -+ struct sradix_tree_node *node; -+ int offset; -+}; -+ -+static inline -+void init_sradix_tree_root(struct sradix_tree_root *root, unsigned long shift) -+{ -+ root->height = 0; -+ root->rnode = NULL; -+ root->shift = shift; -+ root->stores_size = 1UL << shift; -+ root->mask = root->stores_size - 1; -+} -+ -+ -+extern void *sradix_tree_next(struct sradix_tree_root *root, -+ struct sradix_tree_node *node, unsigned long index, -+ int (*iter)(void *, unsigned long)); -+ -+extern int sradix_tree_enter(struct sradix_tree_root *root, void **item, int num); -+ -+extern void sradix_tree_delete_from_leaf(struct sradix_tree_root *root, -+ struct sradix_tree_node *node, unsigned long index); -+ -+extern void *sradix_tree_lookup(struct sradix_tree_root *root, unsigned long index); -+ -+#endif /* _LINUX_SRADIX_TREE_H */ -diff -Nur a/include/linux/uksm.h b/include/linux/uksm.h ---- a/include/linux/uksm.h 1970-01-01 01:00:00.000000000 +0100 -+++ b/include/linux/uksm.h 2018-05-26 19:30:55.785140376 +0100 -@@ -0,0 +1,149 @@ -+#ifndef __LINUX_UKSM_H -+#define __LINUX_UKSM_H -+/* -+ * Memory merging support. -+ * -+ * This code enables dynamic sharing of identical pages found in different -+ * memory areas, even if they are not shared by fork(). -+ */ -+ -+/* if !CONFIG_UKSM this file should not be compiled at all. */ -+#ifdef CONFIG_UKSM -+ -+#include -+#include -+#include -+#include -+#include -+ -+extern unsigned long zero_pfn __read_mostly; -+extern unsigned long uksm_zero_pfn __read_mostly; -+extern struct page *empty_uksm_zero_page; -+ -+/* must be done before linked to mm */ -+extern void uksm_vma_add_new(struct vm_area_struct *vma); -+extern void uksm_remove_vma(struct vm_area_struct *vma); -+ -+#define UKSM_SLOT_NEED_SORT (1 << 0) -+#define UKSM_SLOT_NEED_RERAND (1 << 1) -+#define UKSM_SLOT_SCANNED (1 << 2) /* It's scanned in this round */ -+#define UKSM_SLOT_FUL_SCANNED (1 << 3) -+#define UKSM_SLOT_IN_UKSM (1 << 4) -+ -+struct vma_slot { -+ struct sradix_tree_node *snode; -+ unsigned long sindex; -+ -+ struct list_head slot_list; -+ unsigned long fully_scanned_round; -+ unsigned long dedup_num; -+ unsigned long pages_scanned; -+ unsigned long this_sampled; -+ unsigned long last_scanned; -+ unsigned long pages_to_scan; -+ struct scan_rung *rung; -+ struct page **rmap_list_pool; -+ unsigned int *pool_counts; -+ unsigned long pool_size; -+ struct vm_area_struct *vma; -+ struct mm_struct *mm; -+ unsigned long ctime_j; -+ unsigned long pages; -+ unsigned long flags; -+ unsigned long pages_cowed; /* pages cowed this round */ -+ unsigned long pages_merged; /* pages merged this round */ -+ unsigned long pages_bemerged; -+ -+ /* when it has page merged in this eval round */ -+ struct list_head dedup_list; -+}; -+ -+static inline void uksm_unmap_zero_page(pte_t pte) -+{ -+ if (pte_pfn(pte) == uksm_zero_pfn) -+ __dec_zone_page_state(empty_uksm_zero_page, NR_UKSM_ZERO_PAGES); -+} -+ -+static inline void uksm_map_zero_page(pte_t pte) -+{ -+ if (pte_pfn(pte) == uksm_zero_pfn) -+ __inc_zone_page_state(empty_uksm_zero_page, NR_UKSM_ZERO_PAGES); -+} -+ -+static inline void uksm_cow_page(struct vm_area_struct *vma, struct page *page) -+{ -+ if (vma->uksm_vma_slot && PageKsm(page)) -+ vma->uksm_vma_slot->pages_cowed++; -+} -+ -+static inline void uksm_cow_pte(struct vm_area_struct *vma, pte_t pte) -+{ -+ if (vma->uksm_vma_slot && pte_pfn(pte) == uksm_zero_pfn) -+ vma->uksm_vma_slot->pages_cowed++; -+} -+ -+static inline int uksm_flags_can_scan(unsigned long vm_flags) -+{ -+#ifdef VM_SAO -+ if (vm_flags & VM_SAO) -+ return 0; -+#endif -+ -+ return !(vm_flags & (VM_PFNMAP | VM_IO | VM_DONTEXPAND | -+ VM_HUGETLB | VM_MIXEDMAP | VM_SHARED -+ | VM_MAYSHARE | VM_GROWSUP | VM_GROWSDOWN)); -+} -+ -+static inline void uksm_vm_flags_mod(unsigned long *vm_flags_p) -+{ -+ if (uksm_flags_can_scan(*vm_flags_p)) -+ *vm_flags_p |= VM_MERGEABLE; -+} -+ -+/* -+ * Just a wrapper for BUG_ON for where ksm_zeropage must not be. TODO: it will -+ * be removed when uksm zero page patch is stable enough. -+ */ -+static inline void uksm_bugon_zeropage(pte_t pte) -+{ -+ BUG_ON(pte_pfn(pte) == uksm_zero_pfn); -+} -+#else -+static inline void uksm_vma_add_new(struct vm_area_struct *vma) -+{ -+} -+ -+static inline void uksm_remove_vma(struct vm_area_struct *vma) -+{ -+} -+ -+static inline void uksm_unmap_zero_page(pte_t pte) -+{ -+} -+ -+static inline void uksm_map_zero_page(pte_t pte) -+{ -+} -+ -+static inline void uksm_cow_page(struct vm_area_struct *vma, struct page *page) -+{ -+} -+ -+static inline void uksm_cow_pte(struct vm_area_struct *vma, pte_t pte) -+{ -+} -+ -+static inline int uksm_flags_can_scan(unsigned long vm_flags) -+{ -+ return 0; -+} -+ -+static inline void uksm_vm_flags_mod(unsigned long *vm_flags_p) -+{ -+} -+ -+static inline void uksm_bugon_zeropage(pte_t pte) -+{ -+} -+#endif /* !CONFIG_UKSM */ -+#endif /* __LINUX_UKSM_H */ -diff -Nur a/kernel/fork.c b/kernel/fork.c ---- a/kernel/fork.c 2018-05-26 19:24:34.840783196 +0100 -+++ b/kernel/fork.c 2018-05-26 19:30:55.785140376 +0100 -@@ -655,7 +655,7 @@ - goto fail_nomem; - charge = len; - } -- tmp = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL); -+ tmp = kmem_cache_zalloc(vm_area_cachep, GFP_KERNEL); - if (!tmp) - goto fail_nomem; - *tmp = *mpnt; -@@ -714,7 +714,7 @@ - __vma_link_rb(mm, tmp, rb_link, rb_parent); - rb_link = &tmp->vm_rb.rb_right; - rb_parent = &tmp->vm_rb; -- -+ uksm_vma_add_new(tmp); - mm->map_count++; - if (!(tmp->vm_flags & VM_WIPEONFORK)) - retval = copy_page_range(mm, oldmm, mpnt); -diff -Nur a/lib/Makefile b/lib/Makefile ---- a/lib/Makefile 2018-05-25 15:18:02.000000000 +0100 -+++ b/lib/Makefile 2018-05-26 19:30:55.786140408 +0100 -@@ -18,7 +18,7 @@ - KCOV_INSTRUMENT_dynamic_debug.o := n - - lib-y := ctype.o string.o vsprintf.o cmdline.o \ -- rbtree.o radix-tree.o dump_stack.o timerqueue.o\ -+ rbtree.o radix-tree.o sradix-tree.o dump_stack.o timerqueue.o\ - idr.o int_sqrt.o extable.o \ - sha1.o chacha20.o irq_regs.o argv_split.o \ - flex_proportions.o ratelimit.o show_mem.o \ -diff -Nur a/lib/sradix-tree.c b/lib/sradix-tree.c ---- a/lib/sradix-tree.c 1970-01-01 01:00:00.000000000 +0100 -+++ b/lib/sradix-tree.c 2018-05-26 19:30:55.786140408 +0100 -@@ -0,0 +1,476 @@ -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+ -+static inline int sradix_node_full(struct sradix_tree_root *root, struct sradix_tree_node *node) -+{ -+ return node->fulls == root->stores_size || -+ (node->height == 1 && node->count == root->stores_size); -+} -+ -+/* -+ * Extend a sradix tree so it can store key @index. -+ */ -+static int sradix_tree_extend(struct sradix_tree_root *root, unsigned long index) -+{ -+ struct sradix_tree_node *node; -+ unsigned int height; -+ -+ if (unlikely(root->rnode == NULL)) { -+ if (!(node = root->alloc())) -+ return -ENOMEM; -+ -+ node->height = 1; -+ root->rnode = node; -+ root->height = 1; -+ } -+ -+ /* Figure out what the height should be. */ -+ height = root->height; -+ index >>= root->shift * height; -+ -+ while (index) { -+ index >>= root->shift; -+ height++; -+ } -+ -+ while (height > root->height) { -+ unsigned int newheight; -+ -+ if (!(node = root->alloc())) -+ return -ENOMEM; -+ -+ /* Increase the height. */ -+ node->stores[0] = root->rnode; -+ root->rnode->parent = node; -+ if (root->extend) -+ root->extend(node, root->rnode); -+ -+ newheight = root->height + 1; -+ node->height = newheight; -+ node->count = 1; -+ if (sradix_node_full(root, root->rnode)) -+ node->fulls = 1; -+ -+ root->rnode = node; -+ root->height = newheight; -+ } -+ -+ return 0; -+} -+ -+/* -+ * Search the next item from the current node, that is not NULL -+ * and can satify root->iter(). -+ */ -+void *sradix_tree_next(struct sradix_tree_root *root, -+ struct sradix_tree_node *node, unsigned long index, -+ int (*iter)(void *item, unsigned long height)) -+{ -+ unsigned long offset; -+ void *item; -+ -+ if (unlikely(node == NULL)) { -+ node = root->rnode; -+ for (offset = 0; offset < root->stores_size; offset++) { -+ item = node->stores[offset]; -+ if (item && (!iter || iter(item, node->height))) -+ break; -+ } -+ -+ if (unlikely(offset >= root->stores_size)) -+ return NULL; -+ -+ if (node->height == 1) -+ return item; -+ else -+ goto go_down; -+ } -+ -+ while (node) { -+ offset = (index & root->mask) + 1; -+ for (; offset < root->stores_size; offset++) { -+ item = node->stores[offset]; -+ if (item && (!iter || iter(item, node->height))) -+ break; -+ } -+ -+ if (offset < root->stores_size) -+ break; -+ -+ node = node->parent; -+ index >>= root->shift; -+ } -+ -+ if (!node) -+ return NULL; -+ -+ while (node->height > 1) { -+go_down: -+ node = item; -+ for (offset = 0; offset < root->stores_size; offset++) { -+ item = node->stores[offset]; -+ if (item && (!iter || iter(item, node->height))) -+ break; -+ } -+ -+ if (unlikely(offset >= root->stores_size)) -+ return NULL; -+ } -+ -+ BUG_ON(offset > root->stores_size); -+ -+ return item; -+} -+ -+/* -+ * Blindly insert the item to the tree. Typically, we reuse the -+ * first empty store item. -+ */ -+int sradix_tree_enter(struct sradix_tree_root *root, void **item, int num) -+{ -+ unsigned long index; -+ unsigned int height; -+ struct sradix_tree_node *node, *tmp = NULL; -+ int offset, offset_saved; -+ void **store = NULL; -+ int error, i, j, shift; -+ -+go_on: -+ index = root->min; -+ -+ if (root->enter_node && !sradix_node_full(root, root->enter_node)) { -+ node = root->enter_node; -+ BUG_ON((index >> (root->shift * root->height))); -+ } else { -+ node = root->rnode; -+ if (node == NULL || (index >> (root->shift * root->height)) -+ || sradix_node_full(root, node)) { -+ error = sradix_tree_extend(root, index); -+ if (error) -+ return error; -+ -+ node = root->rnode; -+ } -+ } -+ -+ -+ height = node->height; -+ shift = (height - 1) * root->shift; -+ offset = (index >> shift) & root->mask; -+ while (shift > 0) { -+ offset_saved = offset; -+ for (; offset < root->stores_size; offset++) { -+ store = &node->stores[offset]; -+ tmp = *store; -+ -+ if (!tmp || !sradix_node_full(root, tmp)) -+ break; -+ } -+ BUG_ON(offset >= root->stores_size); -+ -+ if (offset != offset_saved) { -+ index += (offset - offset_saved) << shift; -+ index &= ~((1UL << shift) - 1); -+ } -+ -+ if (!tmp) { -+ if (!(tmp = root->alloc())) -+ return -ENOMEM; -+ -+ tmp->height = shift / root->shift; -+ *store = tmp; -+ tmp->parent = node; -+ node->count++; -+// if (root->extend) -+// root->extend(node, tmp); -+ } -+ -+ node = tmp; -+ shift -= root->shift; -+ offset = (index >> shift) & root->mask; -+ } -+ -+ BUG_ON(node->height != 1); -+ -+ -+ store = &node->stores[offset]; -+ for (i = 0, j = 0; -+ j < root->stores_size - node->count && -+ i < root->stores_size - offset && j < num; i++) { -+ if (!store[i]) { -+ store[i] = item[j]; -+ if (root->assign) -+ root->assign(node, index + i, item[j]); -+ j++; -+ } -+ } -+ -+ node->count += j; -+ root->num += j; -+ num -= j; -+ -+ while (sradix_node_full(root, node)) { -+ node = node->parent; -+ if (!node) -+ break; -+ -+ node->fulls++; -+ } -+ -+ if (unlikely(!node)) { -+ /* All nodes are full */ -+ root->min = 1 << (root->height * root->shift); -+ root->enter_node = NULL; -+ } else { -+ root->min = index + i - 1; -+ root->min |= (1UL << (node->height - 1)) - 1; -+ root->min++; -+ root->enter_node = node; -+ } -+ -+ if (num) { -+ item += j; -+ goto go_on; -+ } -+ -+ return 0; -+} -+ -+ -+/** -+ * sradix_tree_shrink - shrink height of a sradix tree to minimal -+ * @root sradix tree root -+ * -+ */ -+static inline void sradix_tree_shrink(struct sradix_tree_root *root) -+{ -+ /* try to shrink tree height */ -+ while (root->height > 1) { -+ struct sradix_tree_node *to_free = root->rnode; -+ -+ /* -+ * The candidate node has more than one child, or its child -+ * is not at the leftmost store, we cannot shrink. -+ */ -+ if (to_free->count != 1 || !to_free->stores[0]) -+ break; -+ -+ root->rnode = to_free->stores[0]; -+ root->rnode->parent = NULL; -+ root->height--; -+ if (unlikely(root->enter_node == to_free)) -+ root->enter_node = NULL; -+ root->free(to_free); -+ } -+} -+ -+/* -+ * Del the item on the known leaf node and index -+ */ -+void sradix_tree_delete_from_leaf(struct sradix_tree_root *root, -+ struct sradix_tree_node *node, unsigned long index) -+{ -+ unsigned int offset; -+ struct sradix_tree_node *start, *end; -+ -+ BUG_ON(node->height != 1); -+ -+ start = node; -+ while (node && !(--node->count)) -+ node = node->parent; -+ -+ end = node; -+ if (!node) { -+ root->rnode = NULL; -+ root->height = 0; -+ root->min = 0; -+ root->num = 0; -+ root->enter_node = NULL; -+ } else { -+ offset = (index >> (root->shift * (node->height - 1))) & root->mask; -+ if (root->rm) -+ root->rm(node, offset); -+ node->stores[offset] = NULL; -+ root->num--; -+ if (root->min > index) { -+ root->min = index; -+ root->enter_node = node; -+ } -+ } -+ -+ if (start != end) { -+ do { -+ node = start; -+ start = start->parent; -+ if (unlikely(root->enter_node == node)) -+ root->enter_node = end; -+ root->free(node); -+ } while (start != end); -+ -+ /* -+ * Note that shrink may free "end", so enter_node still need to -+ * be checked inside. -+ */ -+ sradix_tree_shrink(root); -+ } else if (node->count == root->stores_size - 1) { -+ /* It WAS a full leaf node. Update the ancestors */ -+ node = node->parent; -+ while (node) { -+ node->fulls--; -+ if (node->fulls != root->stores_size - 1) -+ break; -+ -+ node = node->parent; -+ } -+ } -+} -+ -+void *sradix_tree_lookup(struct sradix_tree_root *root, unsigned long index) -+{ -+ unsigned int height, offset; -+ struct sradix_tree_node *node; -+ int shift; -+ -+ node = root->rnode; -+ if (node == NULL || (index >> (root->shift * root->height))) -+ return NULL; -+ -+ height = root->height; -+ shift = (height - 1) * root->shift; -+ -+ do { -+ offset = (index >> shift) & root->mask; -+ node = node->stores[offset]; -+ if (!node) -+ return NULL; -+ -+ shift -= root->shift; -+ } while (shift >= 0); -+ -+ return node; -+} -+ -+/* -+ * Return the item if it exists, otherwise create it in place -+ * and return the created item. -+ */ -+void *sradix_tree_lookup_create(struct sradix_tree_root *root, -+ unsigned long index, void *(*item_alloc)(void)) -+{ -+ unsigned int height, offset; -+ struct sradix_tree_node *node, *tmp; -+ void *item; -+ int shift, error; -+ -+ if (root->rnode == NULL || (index >> (root->shift * root->height))) { -+ if (item_alloc) { -+ error = sradix_tree_extend(root, index); -+ if (error) -+ return NULL; -+ } else { -+ return NULL; -+ } -+ } -+ -+ node = root->rnode; -+ height = root->height; -+ shift = (height - 1) * root->shift; -+ -+ do { -+ offset = (index >> shift) & root->mask; -+ if (!node->stores[offset]) { -+ if (!(tmp = root->alloc())) -+ return NULL; -+ -+ tmp->height = shift / root->shift; -+ node->stores[offset] = tmp; -+ tmp->parent = node; -+ node->count++; -+ node = tmp; -+ } else { -+ node = node->stores[offset]; -+ } -+ -+ shift -= root->shift; -+ } while (shift > 0); -+ -+ BUG_ON(node->height != 1); -+ offset = index & root->mask; -+ if (node->stores[offset]) { -+ return node->stores[offset]; -+ } else if (item_alloc) { -+ if (!(item = item_alloc())) -+ return NULL; -+ -+ node->stores[offset] = item; -+ -+ /* -+ * NOTE: we do NOT call root->assign here, since this item is -+ * newly created by us having no meaning. Caller can call this -+ * if it's necessary to do so. -+ */ -+ -+ node->count++; -+ root->num++; -+ -+ while (sradix_node_full(root, node)) { -+ node = node->parent; -+ if (!node) -+ break; -+ -+ node->fulls++; -+ } -+ -+ if (unlikely(!node)) { -+ /* All nodes are full */ -+ root->min = 1 << (root->height * root->shift); -+ } else { -+ if (root->min == index) { -+ root->min |= (1UL << (node->height - 1)) - 1; -+ root->min++; -+ root->enter_node = node; -+ } -+ } -+ -+ return item; -+ } else { -+ return NULL; -+ } -+ -+} -+ -+int sradix_tree_delete(struct sradix_tree_root *root, unsigned long index) -+{ -+ unsigned int height, offset; -+ struct sradix_tree_node *node; -+ int shift; -+ -+ node = root->rnode; -+ if (node == NULL || (index >> (root->shift * root->height))) -+ return -ENOENT; -+ -+ height = root->height; -+ shift = (height - 1) * root->shift; -+ -+ do { -+ offset = (index >> shift) & root->mask; -+ node = node->stores[offset]; -+ if (!node) -+ return -ENOENT; -+ -+ shift -= root->shift; -+ } while (shift > 0); -+ -+ offset = index & root->mask; -+ if (!node->stores[offset]) -+ return -ENOENT; -+ -+ sradix_tree_delete_from_leaf(root, node, index); -+ -+ return 0; -+} -diff -Nur a/mm/Kconfig b/mm/Kconfig ---- a/mm/Kconfig 2018-05-26 19:24:34.846783391 +0100 -+++ b/mm/Kconfig 2018-05-26 19:30:55.786140408 +0100 -@@ -315,6 +315,32 @@ - See Documentation/vm/ksm.txt for more information: KSM is inactive - until a program has madvised that an area is MADV_MERGEABLE, and - root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set). -+choice -+ prompt "Choose UKSM/KSM strategy" -+ default UKSM -+ depends on KSM -+ help -+ This option allows to select a UKSM/KSM stragety. -+ -+config UKSM -+ bool "Ultra-KSM for page merging" -+ depends on KSM -+ help -+ UKSM is inspired by the Linux kernel project \u2014 KSM(Kernel Same -+ page Merging), but with a fundamentally rewritten core algorithm. With -+ an advanced algorithm, UKSM now can transparently scans all anonymously -+ mapped user space applications with an significantly improved scan speed -+ and CPU efficiency. Since KVM is friendly to KSM, KVM can also benefit from -+ UKSM. Now UKSM has its first stable release and first real world enterprise user. -+ For more information, please goto its project page. -+ (www.kerneldedup.org) -+ -+config KSM_LEGACY -+ bool "Legacy KSM implementation" -+ depends on KSM -+ help -+ The legacy KSM implementation from Red Hat. -+endchoice - - config DEFAULT_MMAP_MIN_ADDR - int "Low address space to protect from user allocation" -diff -Nur a/mm/Makefile b/mm/Makefile ---- a/mm/Makefile 2018-05-25 15:18:02.000000000 +0100 -+++ b/mm/Makefile 2018-05-26 19:30:55.786140408 +0100 -@@ -65,7 +65,8 @@ - obj-$(CONFIG_SPARSEMEM_VMEMMAP) += sparse-vmemmap.o - obj-$(CONFIG_SLOB) += slob.o - obj-$(CONFIG_MMU_NOTIFIER) += mmu_notifier.o --obj-$(CONFIG_KSM) += ksm.o -+obj-$(CONFIG_KSM_LEGACY) += ksm.o -+obj-$(CONFIG_UKSM) += uksm.o - obj-$(CONFIG_PAGE_POISONING) += page_poison.o - obj-$(CONFIG_SLAB) += slab.o - obj-$(CONFIG_SLUB) += slub.o -diff -Nur a/mm/memory.c b/mm/memory.c ---- a/mm/memory.c 2018-05-25 15:18:02.000000000 +0100 -+++ b/mm/memory.c 2018-05-26 19:30:55.787140441 +0100 -@@ -129,6 +129,25 @@ - - unsigned long highest_memmap_pfn __read_mostly; - -+#ifdef CONFIG_UKSM -+unsigned long uksm_zero_pfn __read_mostly; -+EXPORT_SYMBOL_GPL(uksm_zero_pfn); -+struct page *empty_uksm_zero_page; -+ -+static int __init setup_uksm_zero_page(void) -+{ -+ empty_uksm_zero_page = alloc_pages(__GFP_ZERO & ~__GFP_MOVABLE, 0); -+ if (!empty_uksm_zero_page) -+ panic("Oh boy, that early out of memory?"); -+ -+ SetPageReserved(empty_uksm_zero_page); -+ uksm_zero_pfn = page_to_pfn(empty_uksm_zero_page); -+ -+ return 0; -+} -+core_initcall(setup_uksm_zero_page); -+#endif -+ - /* - * CONFIG_MMU architectures set up ZERO_PAGE in their paging_init() - */ -@@ -140,6 +159,7 @@ - core_initcall(init_zero_pfn); - - -+ - #if defined(SPLIT_RSS_COUNTING) - - void sync_mm_rss(struct mm_struct *mm) -@@ -1035,6 +1055,9 @@ - get_page(page); - page_dup_rmap(page, false); - rss[mm_counter(page)]++; -+ -+ /* Should return NULL in vm_normal_page() */ -+ uksm_bugon_zeropage(pte); - } else if (pte_devmap(pte)) { - page = pte_page(pte); - -@@ -1048,6 +1071,8 @@ - page_dup_rmap(page, false); - rss[mm_counter(page)]++; - } -+ } else { -+ uksm_map_zero_page(pte); - } - - out_set_pte: -@@ -1317,8 +1342,10 @@ - ptent = ptep_get_and_clear_full(mm, addr, pte, - tlb->fullmm); - tlb_remove_tlb_entry(tlb, pte, addr); -- if (unlikely(!page)) -+ if (unlikely(!page)) { -+ uksm_unmap_zero_page(ptent); - continue; -+ } - - if (!PageAnon(page)) { - if (pte_dirty(ptent)) { -@@ -2318,8 +2345,10 @@ - clear_page(kaddr); - kunmap_atomic(kaddr); - flush_dcache_page(dst); -- } else -+ } else { - copy_user_highpage(dst, src, va, vma); -+ uksm_cow_page(vma, src); -+ } - } - - static gfp_t __get_fault_gfp_mask(struct vm_area_struct *vma) -@@ -2468,6 +2497,7 @@ - vmf->address); - if (!new_page) - goto oom; -+ uksm_cow_pte(vma, vmf->orig_pte); - } else { - new_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, - vmf->address); -@@ -2494,7 +2524,9 @@ - mm_counter_file(old_page)); - inc_mm_counter_fast(mm, MM_ANONPAGES); - } -+ uksm_bugon_zeropage(vmf->orig_pte); - } else { -+ uksm_unmap_zero_page(vmf->orig_pte); - inc_mm_counter_fast(mm, MM_ANONPAGES); - } - flush_cache_page(vma, vmf->address, pte_pfn(vmf->orig_pte)); -diff -Nur a/mm/mmap.c b/mm/mmap.c ---- a/mm/mmap.c 2018-05-26 19:24:34.847783423 +0100 -+++ b/mm/mmap.c 2018-05-26 19:30:55.788140473 +0100 -@@ -45,6 +45,7 @@ - #include - #include - #include -+#include - - #include - #include -@@ -173,6 +174,7 @@ - if (vma->vm_file) - fput(vma->vm_file); - mpol_put(vma_policy(vma)); -+ uksm_remove_vma(vma); - kmem_cache_free(vm_area_cachep, vma); - return next; - } -@@ -699,9 +701,16 @@ - long adjust_next = 0; - int remove_next = 0; - -+/* -+ * to avoid deadlock, ksm_remove_vma must be done before any spin_lock is -+ * acquired -+ */ -+ uksm_remove_vma(vma); -+ - if (next && !insert) { - struct vm_area_struct *exporter = NULL, *importer = NULL; - -+ uksm_remove_vma(next); - if (end >= next->vm_end) { - /* - * vma expands, overlapping all the next, and -@@ -834,6 +843,7 @@ - end_changed = true; - } - vma->vm_pgoff = pgoff; -+ - if (adjust_next) { - next->vm_start += adjust_next << PAGE_SHIFT; - next->vm_pgoff += adjust_next; -@@ -939,6 +949,7 @@ - if (remove_next == 2) { - remove_next = 1; - end = next->vm_end; -+ uksm_remove_vma(next); - goto again; - } - else if (next) -@@ -965,10 +976,14 @@ - */ - VM_WARN_ON(mm->highest_vm_end != vm_end_gap(vma)); - } -+ } else { -+ if (next && !insert) -+ uksm_vma_add_new(next); - } - if (insert && file) - uprobe_mmap(insert); - -+ uksm_vma_add_new(vma); - validate_mm(mm); - - return 0; -@@ -1385,6 +1400,9 @@ - vm_flags |= calc_vm_prot_bits(prot, pkey) | calc_vm_flag_bits(flags) | - mm->def_flags | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC; - -+ /* If uksm is enabled, we add VM_MERGEABLE to new VMAs. */ -+ uksm_vm_flags_mod(&vm_flags); -+ - if (flags & MAP_LOCKED) - if (!can_do_mlock()) - return -EPERM; -@@ -1724,6 +1742,7 @@ - allow_write_access(file); - } - file = vma->vm_file; -+ uksm_vma_add_new(vma); - out: - perf_event_mmap(vma); - -@@ -1765,6 +1784,7 @@ - if (vm_flags & VM_DENYWRITE) - allow_write_access(file); - free_vma: -+ uksm_remove_vma(vma); - kmem_cache_free(vm_area_cachep, vma); - unacct_error: - if (charged) -@@ -2589,6 +2609,8 @@ - else - err = vma_adjust(vma, vma->vm_start, addr, vma->vm_pgoff, new); - -+ uksm_vma_add_new(new); -+ - /* Success. */ - if (!err) - return 0; -@@ -2881,6 +2903,7 @@ - if ((flags & (~VM_EXEC)) != 0) - return -EINVAL; - flags |= VM_DATA_DEFAULT_FLAGS | VM_ACCOUNT | mm->def_flags; -+ uksm_vm_flags_mod(&flags); - - error = get_unmapped_area(NULL, addr, len, 0, MAP_FIXED); - if (offset_in_page(error)) -@@ -2938,6 +2961,7 @@ - vma->vm_flags = flags; - vma->vm_page_prot = vm_get_page_prot(flags); - vma_link(mm, vma, prev, rb_link, rb_parent); -+ uksm_vma_add_new(vma); - out: - perf_event_mmap(vma); - mm->total_vm += len >> PAGE_SHIFT; -@@ -3015,6 +3039,12 @@ - up_write(&mm->mmap_sem); - } - -+ /* -+ * Taking write lock on mmap_sem does not harm others, -+ * but it's crucial for uksm to avoid races. -+ */ -+ down_write(&mm->mmap_sem); -+ - if (mm->locked_vm) { - vma = mm->mmap; - while (vma) { -@@ -3049,6 +3079,11 @@ - vma = remove_vma(vma); - } - vm_unacct_memory(nr_accounted); -+ -+ mm->mmap = NULL; -+ mm->mm_rb = RB_ROOT; -+ vmacache_invalidate(mm); -+ up_write(&mm->mmap_sem); - } - - /* Insert vm structure into process list sorted by address -@@ -3158,6 +3193,7 @@ - new_vma->vm_ops->open(new_vma); - vma_link(mm, new_vma, prev, rb_link, rb_parent); - *need_rmap_locks = false; -+ uksm_vma_add_new(new_vma); - } - return new_vma; - -@@ -3308,6 +3344,7 @@ - vm_stat_account(mm, vma->vm_flags, len >> PAGE_SHIFT); - - perf_event_mmap(vma); -+ uksm_vma_add_new(vma); - - return vma; - -diff -Nur a/mm/rmap.c b/mm/rmap.c ---- a/mm/rmap.c 2018-05-25 15:18:02.000000000 +0100 -+++ b/mm/rmap.c 2018-05-26 19:30:55.788140473 +0100 -@@ -1013,9 +1013,9 @@ - - /** - * __page_set_anon_rmap - set up new anonymous rmap -- * @page: Page to add to rmap -+ * @page: Page to add to rmap - * @vma: VM area to add page to. -- * @address: User virtual address of the mapping -+ * @address: User virtual address of the mapping - * @exclusive: the page is exclusively owned by the current process - */ - static void __page_set_anon_rmap(struct page *page, -diff -Nur a/mm/uksm.c b/mm/uksm.c ---- a/mm/uksm.c 1970-01-01 01:00:00.000000000 +0100 -+++ b/mm/uksm.c 2018-05-26 19:30:55.791140570 +0100 -@@ -0,0 +1,5584 @@ -+/* -+ * Ultra KSM. Copyright (C) 2011-2012 Nai Xia -+ * -+ * This is an improvement upon KSM. Some basic data structures and routines -+ * are borrowed from ksm.c . -+ * -+ * Its new features: -+ * 1. Full system scan: -+ * It automatically scans all user processes' anonymous VMAs. Kernel-user -+ * interaction to submit a memory area to KSM is no longer needed. -+ * -+ * 2. Rich area detection: -+ * It automatically detects rich areas containing abundant duplicated -+ * pages based. Rich areas are given a full scan speed. Poor areas are -+ * sampled at a reasonable speed with very low CPU consumption. -+ * -+ * 3. Ultra Per-page scan speed improvement: -+ * A new hash algorithm is proposed. As a result, on a machine with -+ * Core(TM)2 Quad Q9300 CPU in 32-bit mode and 800MHZ DDR2 main memory, it -+ * can scan memory areas that does not contain duplicated pages at speed of -+ * 627MB/sec ~ 2445MB/sec and can merge duplicated areas at speed of -+ * 477MB/sec ~ 923MB/sec. -+ * -+ * 4. Thrashing area avoidance: -+ * Thrashing area(an VMA that has frequent Ksm page break-out) can be -+ * filtered out. My benchmark shows it's more efficient than KSM's per-page -+ * hash value based volatile page detection. -+ * -+ * -+ * 5. Misc changes upon KSM: -+ * * It has a fully x86-opitmized memcmp dedicated for 4-byte-aligned page -+ * comparison. It's much faster than default C version on x86. -+ * * rmap_item now has an struct *page member to loosely cache a -+ * address-->page mapping, which reduces too much time-costly -+ * follow_page(). -+ * * The VMA creation/exit procedures are hooked to let the Ultra KSM know. -+ * * try_to_merge_two_pages() now can revert a pte if it fails. No break_ -+ * ksm is needed for this case. -+ * -+ * 6. Full Zero Page consideration(contributed by Figo Zhang) -+ * Now uksmd consider full zero pages as special pages and merge them to an -+ * special unswappable uksm zero page. -+ */ -+ -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+#include -+ -+#include -+#include "internal.h" -+ -+#ifdef CONFIG_X86 -+#undef memcmp -+ -+#ifdef CONFIG_X86_32 -+#define memcmp memcmpx86_32 -+/* -+ * Compare 4-byte-aligned address s1 and s2, with length n -+ */ -+int memcmpx86_32(void *s1, void *s2, size_t n) -+{ -+ size_t num = n / 4; -+ register int res; -+ -+ __asm__ __volatile__ -+ ( -+ "testl %3,%3\n\t" -+ "repe; cmpsd\n\t" -+ "je 1f\n\t" -+ "sbbl %0,%0\n\t" -+ "orl $1,%0\n" -+ "1:" -+ : "=&a" (res), "+&S" (s1), "+&D" (s2), "+&c" (num) -+ : "0" (0) -+ : "cc"); -+ -+ return res; -+} -+ -+/* -+ * Check the page is all zero ? -+ */ -+static int is_full_zero(const void *s1, size_t len) -+{ -+ unsigned char same; -+ -+ len /= 4; -+ -+ __asm__ __volatile__ -+ ("repe; scasl;" -+ "sete %0" -+ : "=qm" (same), "+D" (s1), "+c" (len) -+ : "a" (0) -+ : "cc"); -+ -+ return same; -+} -+ -+ -+#elif defined(CONFIG_X86_64) -+#define memcmp memcmpx86_64 -+/* -+ * Compare 8-byte-aligned address s1 and s2, with length n -+ */ -+int memcmpx86_64(void *s1, void *s2, size_t n) -+{ -+ size_t num = n / 8; -+ register int res; -+ -+ __asm__ __volatile__ -+ ( -+ "testq %q3,%q3\n\t" -+ "repe; cmpsq\n\t" -+ "je 1f\n\t" -+ "sbbq %q0,%q0\n\t" -+ "orq $1,%q0\n" -+ "1:" -+ : "=&a" (res), "+&S" (s1), "+&D" (s2), "+&c" (num) -+ : "0" (0) -+ : "cc"); -+ -+ return res; -+} -+ -+static int is_full_zero(const void *s1, size_t len) -+{ -+ unsigned char same; -+ -+ len /= 8; -+ -+ __asm__ __volatile__ -+ ("repe; scasq;" -+ "sete %0" -+ : "=qm" (same), "+D" (s1), "+c" (len) -+ : "a" (0) -+ : "cc"); -+ -+ return same; -+} -+ -+#endif -+#else -+static int is_full_zero(const void *s1, size_t len) -+{ -+ unsigned long *src = s1; -+ int i; -+ -+ len /= sizeof(*src); -+ -+ for (i = 0; i < len; i++) { -+ if (src[i]) -+ return 0; -+ } -+ -+ return 1; -+} -+#endif -+ -+#define UKSM_RUNG_ROUND_FINISHED (1 << 0) -+#define TIME_RATIO_SCALE 10000 -+ -+#define SLOT_TREE_NODE_SHIFT 8 -+#define SLOT_TREE_NODE_STORE_SIZE (1UL << SLOT_TREE_NODE_SHIFT) -+struct slot_tree_node { -+ unsigned long size; -+ struct sradix_tree_node snode; -+ void *stores[SLOT_TREE_NODE_STORE_SIZE]; -+}; -+ -+static struct kmem_cache *slot_tree_node_cachep; -+ -+static struct sradix_tree_node *slot_tree_node_alloc(void) -+{ -+ struct slot_tree_node *p; -+ -+ p = kmem_cache_zalloc(slot_tree_node_cachep, GFP_KERNEL | -+ __GFP_NORETRY | __GFP_NOWARN); -+ if (!p) -+ return NULL; -+ -+ return &p->snode; -+} -+ -+static void slot_tree_node_free(struct sradix_tree_node *node) -+{ -+ struct slot_tree_node *p; -+ -+ p = container_of(node, struct slot_tree_node, snode); -+ kmem_cache_free(slot_tree_node_cachep, p); -+} -+ -+static void slot_tree_node_extend(struct sradix_tree_node *parent, -+ struct sradix_tree_node *child) -+{ -+ struct slot_tree_node *p, *c; -+ -+ p = container_of(parent, struct slot_tree_node, snode); -+ c = container_of(child, struct slot_tree_node, snode); -+ -+ p->size += c->size; -+} -+ -+void slot_tree_node_assign(struct sradix_tree_node *node, -+ unsigned int index, void *item) -+{ -+ struct vma_slot *slot = item; -+ struct slot_tree_node *cur; -+ -+ slot->snode = node; -+ slot->sindex = index; -+ -+ while (node) { -+ cur = container_of(node, struct slot_tree_node, snode); -+ cur->size += slot->pages; -+ node = node->parent; -+ } -+} -+ -+void slot_tree_node_rm(struct sradix_tree_node *node, unsigned int offset) -+{ -+ struct vma_slot *slot; -+ struct slot_tree_node *cur; -+ unsigned long pages; -+ -+ if (node->height == 1) { -+ slot = node->stores[offset]; -+ pages = slot->pages; -+ } else { -+ cur = container_of(node->stores[offset], -+ struct slot_tree_node, snode); -+ pages = cur->size; -+ } -+ -+ while (node) { -+ cur = container_of(node, struct slot_tree_node, snode); -+ cur->size -= pages; -+ node = node->parent; -+ } -+} -+ -+unsigned long slot_iter_index; -+int slot_iter(void *item, unsigned long height) -+{ -+ struct slot_tree_node *node; -+ struct vma_slot *slot; -+ -+ if (height == 1) { -+ slot = item; -+ if (slot_iter_index < slot->pages) { -+ /*in this one*/ -+ return 1; -+ } else { -+ slot_iter_index -= slot->pages; -+ return 0; -+ } -+ -+ } else { -+ node = container_of(item, struct slot_tree_node, snode); -+ if (slot_iter_index < node->size) { -+ /*in this one*/ -+ return 1; -+ } else { -+ slot_iter_index -= node->size; -+ return 0; -+ } -+ } -+} -+ -+ -+static inline void slot_tree_init_root(struct sradix_tree_root *root) -+{ -+ init_sradix_tree_root(root, SLOT_TREE_NODE_SHIFT); -+ root->alloc = slot_tree_node_alloc; -+ root->free = slot_tree_node_free; -+ root->extend = slot_tree_node_extend; -+ root->assign = slot_tree_node_assign; -+ root->rm = slot_tree_node_rm; -+} -+ -+void slot_tree_init(void) -+{ -+ slot_tree_node_cachep = kmem_cache_create("slot_tree_node", -+ sizeof(struct slot_tree_node), 0, -+ SLAB_PANIC | SLAB_RECLAIM_ACCOUNT, -+ NULL); -+} -+ -+ -+/* Each rung of this ladder is a list of VMAs having a same scan ratio */ -+struct scan_rung { -+ //struct list_head scanned_list; -+ struct sradix_tree_root vma_root; -+ struct sradix_tree_root vma_root2; -+ -+ struct vma_slot *current_scan; -+ unsigned long current_offset; -+ -+ /* -+ * The initial value for current_offset, it should loop over -+ * [0~ step - 1] to let all slot have its chance to be scanned. -+ */ -+ unsigned long offset_init; -+ unsigned long step; /* dynamic step for current_offset */ -+ unsigned int flags; -+ unsigned long pages_to_scan; -+ //unsigned long fully_scanned_slots; -+ /* -+ * a little bit tricky - if cpu_time_ratio > 0, then the value is the -+ * the cpu time ratio it can spend in rung_i for every scan -+ * period. if < 0, then it is the cpu time ratio relative to the -+ * max cpu percentage user specified. Both in unit of -+ * 1/TIME_RATIO_SCALE -+ */ -+ int cpu_ratio; -+ -+ /* -+ * How long it will take for all slots in this rung to be fully -+ * scanned? If it's zero, we don't care about the cover time: -+ * it's fully scanned. -+ */ -+ unsigned int cover_msecs; -+ //unsigned long vma_num; -+ //unsigned long pages; /* Sum of all slot's pages in rung */ -+}; -+ -+/** -+ * node of either the stable or unstale rbtree -+ * -+ */ -+struct tree_node { -+ struct rb_node node; /* link in the main (un)stable rbtree */ -+ struct rb_root sub_root; /* rb_root for sublevel collision rbtree */ -+ u32 hash; -+ unsigned long count; /* TODO: merged with sub_root */ -+ struct list_head all_list; /* all tree nodes in stable/unstable tree */ -+}; -+ -+/** -+ * struct stable_node - node of the stable rbtree -+ * @node: rb node of this ksm page in the stable tree -+ * @hlist: hlist head of rmap_items using this ksm page -+ * @kpfn: page frame number of this ksm page -+ */ -+struct stable_node { -+ struct rb_node node; /* link in sub-rbtree */ -+ struct tree_node *tree_node; /* it's tree node root in stable tree, NULL if it's in hell list */ -+ struct hlist_head hlist; -+ unsigned long kpfn; -+ u32 hash_max; /* if ==0 then it's not been calculated yet */ -+ struct list_head all_list; /* in a list for all stable nodes */ -+}; -+ -+/** -+ * struct node_vma - group rmap_items linked in a same stable -+ * node together. -+ */ -+struct node_vma { -+ union { -+ struct vma_slot *slot; -+ unsigned long key; /* slot is used as key sorted on hlist */ -+ }; -+ struct hlist_node hlist; -+ struct hlist_head rmap_hlist; -+ struct stable_node *head; -+}; -+ -+/** -+ * struct rmap_item - reverse mapping item for virtual addresses -+ * @rmap_list: next rmap_item in mm_slot's singly-linked rmap_list -+ * @anon_vma: pointer to anon_vma for this mm,address, when in stable tree -+ * @mm: the memory structure this rmap_item is pointing into -+ * @address: the virtual address this rmap_item tracks (+ flags in low bits) -+ * @node: rb node of this rmap_item in the unstable tree -+ * @head: pointer to stable_node heading this list in the stable tree -+ * @hlist: link into hlist of rmap_items hanging off that stable_node -+ */ -+struct rmap_item { -+ struct vma_slot *slot; -+ struct page *page; -+ unsigned long address; /* + low bits used for flags below */ -+ unsigned long hash_round; -+ unsigned long entry_index; -+ union { -+ struct {/* when in unstable tree */ -+ struct rb_node node; -+ struct tree_node *tree_node; -+ u32 hash_max; -+ }; -+ struct { /* when in stable tree */ -+ struct node_vma *head; -+ struct hlist_node hlist; -+ struct anon_vma *anon_vma; -+ }; -+ }; -+} __aligned(4); -+ -+struct rmap_list_entry { -+ union { -+ struct rmap_item *item; -+ unsigned long addr; -+ }; -+ /* lowest bit is used for is_addr tag */ -+} __aligned(4); /* 4 aligned to fit in to pages*/ -+ -+ -+/* Basic data structure definition ends */ -+ -+ -+/* -+ * Flags for rmap_item to judge if it's listed in the stable/unstable tree. -+ * The flags use the low bits of rmap_item.address -+ */ -+#define UNSTABLE_FLAG 0x1 -+#define STABLE_FLAG 0x2 -+#define get_rmap_addr(x) ((x)->address & PAGE_MASK) -+ -+/* -+ * rmap_list_entry helpers -+ */ -+#define IS_ADDR_FLAG 1 -+#define is_addr(ptr) ((unsigned long)(ptr) & IS_ADDR_FLAG) -+#define set_is_addr(ptr) ((ptr) |= IS_ADDR_FLAG) -+#define get_clean_addr(ptr) (((ptr) & ~(__typeof__(ptr))IS_ADDR_FLAG)) -+ -+ -+/* -+ * High speed caches for frequently allocated and freed structs -+ */ -+static struct kmem_cache *rmap_item_cache; -+static struct kmem_cache *stable_node_cache; -+static struct kmem_cache *node_vma_cache; -+static struct kmem_cache *vma_slot_cache; -+static struct kmem_cache *tree_node_cache; -+#define UKSM_KMEM_CACHE(__struct, __flags) kmem_cache_create("uksm_"#__struct,\ -+ sizeof(struct __struct), __alignof__(struct __struct),\ -+ (__flags), NULL) -+ -+/* Array of all scan_rung, uksm_scan_ladder[0] having the minimum scan ratio */ -+#define SCAN_LADDER_SIZE 4 -+static struct scan_rung uksm_scan_ladder[SCAN_LADDER_SIZE]; -+ -+/* The evaluation rounds uksmd has finished */ -+static unsigned long long uksm_eval_round = 1; -+ -+/* -+ * we add 1 to this var when we consider we should rebuild the whole -+ * unstable tree. -+ */ -+static unsigned long uksm_hash_round = 1; -+ -+/* -+ * How many times the whole memory is scanned. -+ */ -+static unsigned long long fully_scanned_round = 1; -+ -+/* The total number of virtual pages of all vma slots */ -+static u64 uksm_pages_total; -+ -+/* The number of pages has been scanned since the start up */ -+static u64 uksm_pages_scanned; -+ -+static u64 scanned_virtual_pages; -+ -+/* The number of pages has been scanned since last encode_benefit call */ -+static u64 uksm_pages_scanned_last; -+ -+/* If the scanned number is tooo large, we encode it here */ -+static u64 pages_scanned_stored; -+ -+static unsigned long pages_scanned_base; -+ -+/* The number of nodes in the stable tree */ -+static unsigned long uksm_pages_shared; -+ -+/* The number of page slots additionally sharing those nodes */ -+static unsigned long uksm_pages_sharing; -+ -+/* The number of nodes in the unstable tree */ -+static unsigned long uksm_pages_unshared; -+ -+/* -+ * Milliseconds ksmd should sleep between scans, -+ * >= 100ms to be consistent with -+ * scan_time_to_sleep_msec() -+ */ -+static unsigned int uksm_sleep_jiffies; -+ -+/* The real value for the uksmd next sleep */ -+static unsigned int uksm_sleep_real; -+ -+/* Saved value for user input uksm_sleep_jiffies when it's enlarged */ -+static unsigned int uksm_sleep_saved; -+ -+/* Max percentage of cpu utilization ksmd can take to scan in one batch */ -+static unsigned int uksm_max_cpu_percentage; -+ -+static int uksm_cpu_governor; -+ -+static char *uksm_cpu_governor_str[4] = { "full", "medium", "low", "quiet" }; -+ -+struct uksm_cpu_preset_s { -+ int cpu_ratio[SCAN_LADDER_SIZE]; -+ unsigned int cover_msecs[SCAN_LADDER_SIZE]; -+ unsigned int max_cpu; /* percentage */ -+}; -+ -+struct uksm_cpu_preset_s uksm_cpu_preset[4] = { -+ { {20, 40, -2500, -10000}, {1000, 500, 200, 50}, 95}, -+ { {20, 30, -2500, -10000}, {1000, 500, 400, 100}, 50}, -+ { {10, 20, -5000, -10000}, {1500, 1000, 1000, 250}, 20}, -+ { {10, 20, 40, 75}, {2000, 1000, 1000, 1000}, 1}, -+}; -+ -+/* The default value for uksm_ema_page_time if it's not initialized */ -+#define UKSM_PAGE_TIME_DEFAULT 500 -+ -+/*cost to scan one page by expotional moving average in nsecs */ -+static unsigned long uksm_ema_page_time = UKSM_PAGE_TIME_DEFAULT; -+ -+/* The expotional moving average alpha weight, in percentage. */ -+#define EMA_ALPHA 20 -+ -+/* -+ * The threshold used to filter out thrashing areas, -+ * If it == 0, filtering is disabled, otherwise it's the percentage up-bound -+ * of the thrashing ratio of all areas. Any area with a bigger thrashing ratio -+ * will be considered as having a zero duplication ratio. -+ */ -+static unsigned int uksm_thrash_threshold = 50; -+ -+/* How much dedup ratio is considered to be abundant*/ -+static unsigned int uksm_abundant_threshold = 10; -+ -+/* All slots having merged pages in this eval round. */ -+struct list_head vma_slot_dedup = LIST_HEAD_INIT(vma_slot_dedup); -+ -+/* How many times the ksmd has slept since startup */ -+static unsigned long long uksm_sleep_times; -+ -+#define UKSM_RUN_STOP 0 -+#define UKSM_RUN_MERGE 1 -+static unsigned int uksm_run = 1; -+ -+static DECLARE_WAIT_QUEUE_HEAD(uksm_thread_wait); -+static DEFINE_MUTEX(uksm_thread_mutex); -+ -+/* -+ * List vma_slot_new is for newly created vma_slot waiting to be added by -+ * ksmd. If one cannot be added(e.g. due to it's too small), it's moved to -+ * vma_slot_noadd. vma_slot_del is the list for vma_slot whose corresponding -+ * VMA has been removed/freed. -+ */ -+struct list_head vma_slot_new = LIST_HEAD_INIT(vma_slot_new); -+struct list_head vma_slot_noadd = LIST_HEAD_INIT(vma_slot_noadd); -+struct list_head vma_slot_del = LIST_HEAD_INIT(vma_slot_del); -+static DEFINE_SPINLOCK(vma_slot_list_lock); -+ -+/* The unstable tree heads */ -+static struct rb_root root_unstable_tree = RB_ROOT; -+ -+/* -+ * All tree_nodes are in a list to be freed at once when unstable tree is -+ * freed after each scan round. -+ */ -+static struct list_head unstable_tree_node_list = -+ LIST_HEAD_INIT(unstable_tree_node_list); -+ -+/* List contains all stable nodes */ -+static struct list_head stable_node_list = LIST_HEAD_INIT(stable_node_list); -+ -+/* -+ * When the hash strength is changed, the stable tree must be delta_hashed and -+ * re-structured. We use two set of below structs to speed up the -+ * re-structuring of stable tree. -+ */ -+static struct list_head -+stable_tree_node_list[2] = {LIST_HEAD_INIT(stable_tree_node_list[0]), -+ LIST_HEAD_INIT(stable_tree_node_list[1])}; -+ -+static struct list_head *stable_tree_node_listp = &stable_tree_node_list[0]; -+static struct rb_root root_stable_tree[2] = {RB_ROOT, RB_ROOT}; -+static struct rb_root *root_stable_treep = &root_stable_tree[0]; -+static unsigned long stable_tree_index; -+ -+/* The hash strength needed to hash a full page */ -+#define HASH_STRENGTH_FULL (PAGE_SIZE / sizeof(u32)) -+ -+/* The hash strength needed for loop-back hashing */ -+#define HASH_STRENGTH_MAX (HASH_STRENGTH_FULL + 10) -+ -+/* The random offsets in a page */ -+static u32 *random_nums; -+ -+/* The hash strength */ -+static unsigned long hash_strength = HASH_STRENGTH_FULL >> 4; -+ -+/* The delta value each time the hash strength increases or decreases */ -+static unsigned long hash_strength_delta; -+#define HASH_STRENGTH_DELTA_MAX 5 -+ -+/* The time we have saved due to random_sample_hash */ -+static u64 rshash_pos; -+ -+/* The time we have wasted due to hash collision */ -+static u64 rshash_neg; -+ -+struct uksm_benefit { -+ u64 pos; -+ u64 neg; -+ u64 scanned; -+ unsigned long base; -+} benefit; -+ -+/* -+ * The relative cost of memcmp, compared to 1 time unit of random sample -+ * hash, this value is tested when ksm module is initialized -+ */ -+static unsigned long memcmp_cost; -+ -+static unsigned long rshash_neg_cont_zero; -+static unsigned long rshash_cont_obscure; -+ -+/* The possible states of hash strength adjustment heuristic */ -+enum rshash_states { -+ RSHASH_STILL, -+ RSHASH_TRYUP, -+ RSHASH_TRYDOWN, -+ RSHASH_NEW, -+ RSHASH_PRE_STILL, -+}; -+ -+/* The possible direction we are about to adjust hash strength */ -+enum rshash_direct { -+ GO_UP, -+ GO_DOWN, -+ OBSCURE, -+ STILL, -+}; -+ -+/* random sampling hash state machine */ -+static struct { -+ enum rshash_states state; -+ enum rshash_direct pre_direct; -+ u8 below_count; -+ /* Keep a lookup window of size 5, iff above_count/below_count > 3 -+ * in this window we stop trying. -+ */ -+ u8 lookup_window_index; -+ u64 stable_benefit; -+ unsigned long turn_point_down; -+ unsigned long turn_benefit_down; -+ unsigned long turn_point_up; -+ unsigned long turn_benefit_up; -+ unsigned long stable_point; -+} rshash_state; -+ -+/*zero page hash table, hash_strength [0 ~ HASH_STRENGTH_MAX]*/ -+static u32 *zero_hash_table; -+ -+static inline struct node_vma *alloc_node_vma(void) -+{ -+ struct node_vma *node_vma; -+ -+ node_vma = kmem_cache_zalloc(node_vma_cache, GFP_KERNEL | -+ __GFP_NORETRY | __GFP_NOWARN); -+ if (node_vma) { -+ INIT_HLIST_HEAD(&node_vma->rmap_hlist); -+ INIT_HLIST_NODE(&node_vma->hlist); -+ } -+ return node_vma; -+} -+ -+static inline void free_node_vma(struct node_vma *node_vma) -+{ -+ kmem_cache_free(node_vma_cache, node_vma); -+} -+ -+ -+static inline struct vma_slot *alloc_vma_slot(void) -+{ -+ struct vma_slot *slot; -+ -+ /* -+ * In case ksm is not initialized by now. -+ * Oops, we need to consider the call site of uksm_init() in the future. -+ */ -+ if (!vma_slot_cache) -+ return NULL; -+ -+ slot = kmem_cache_zalloc(vma_slot_cache, GFP_KERNEL | -+ __GFP_NORETRY | __GFP_NOWARN); -+ if (slot) { -+ INIT_LIST_HEAD(&slot->slot_list); -+ INIT_LIST_HEAD(&slot->dedup_list); -+ slot->flags |= UKSM_SLOT_NEED_RERAND; -+ } -+ return slot; -+} -+ -+static inline void free_vma_slot(struct vma_slot *vma_slot) -+{ -+ kmem_cache_free(vma_slot_cache, vma_slot); -+} -+ -+ -+ -+static inline struct rmap_item *alloc_rmap_item(void) -+{ -+ struct rmap_item *rmap_item; -+ -+ rmap_item = kmem_cache_zalloc(rmap_item_cache, GFP_KERNEL | -+ __GFP_NORETRY | __GFP_NOWARN); -+ if (rmap_item) { -+ /* bug on lowest bit is not clear for flag use */ -+ BUG_ON(is_addr(rmap_item)); -+ } -+ return rmap_item; -+} -+ -+static inline void free_rmap_item(struct rmap_item *rmap_item) -+{ -+ rmap_item->slot = NULL; /* debug safety */ -+ kmem_cache_free(rmap_item_cache, rmap_item); -+} -+ -+static inline struct stable_node *alloc_stable_node(void) -+{ -+ struct stable_node *node; -+ -+ node = kmem_cache_alloc(stable_node_cache, GFP_KERNEL | -+ __GFP_NORETRY | __GFP_NOWARN); -+ if (!node) -+ return NULL; -+ -+ INIT_HLIST_HEAD(&node->hlist); -+ list_add(&node->all_list, &stable_node_list); -+ return node; -+} -+ -+static inline void free_stable_node(struct stable_node *stable_node) -+{ -+ list_del(&stable_node->all_list); -+ kmem_cache_free(stable_node_cache, stable_node); -+} -+ -+static inline struct tree_node *alloc_tree_node(struct list_head *list) -+{ -+ struct tree_node *node; -+ -+ node = kmem_cache_zalloc(tree_node_cache, GFP_KERNEL | -+ __GFP_NORETRY | __GFP_NOWARN); -+ if (!node) -+ return NULL; -+ -+ list_add(&node->all_list, list); -+ return node; -+} -+ -+static inline void free_tree_node(struct tree_node *node) -+{ -+ list_del(&node->all_list); -+ kmem_cache_free(tree_node_cache, node); -+} -+ -+static void uksm_drop_anon_vma(struct rmap_item *rmap_item) -+{ -+ struct anon_vma *anon_vma = rmap_item->anon_vma; -+ -+ put_anon_vma(anon_vma); -+} -+ -+ -+/** -+ * Remove a stable node from stable_tree, may unlink from its tree_node and -+ * may remove its parent tree_node if no other stable node is pending. -+ * -+ * @stable_node The node need to be removed -+ * @unlink_rb Will this node be unlinked from the rbtree? -+ * @remove_tree_ node Will its tree_node be removed if empty? -+ */ -+static void remove_node_from_stable_tree(struct stable_node *stable_node, -+ int unlink_rb, int remove_tree_node) -+{ -+ struct node_vma *node_vma; -+ struct rmap_item *rmap_item; -+ struct hlist_node *n; -+ -+ if (!hlist_empty(&stable_node->hlist)) { -+ hlist_for_each_entry_safe(node_vma, n, -+ &stable_node->hlist, hlist) { -+ hlist_for_each_entry(rmap_item, &node_vma->rmap_hlist, hlist) { -+ uksm_pages_sharing--; -+ -+ uksm_drop_anon_vma(rmap_item); -+ rmap_item->address &= PAGE_MASK; -+ } -+ free_node_vma(node_vma); -+ cond_resched(); -+ } -+ -+ /* the last one is counted as shared */ -+ uksm_pages_shared--; -+ uksm_pages_sharing++; -+ } -+ -+ if (stable_node->tree_node && unlink_rb) { -+ rb_erase(&stable_node->node, -+ &stable_node->tree_node->sub_root); -+ -+ if (RB_EMPTY_ROOT(&stable_node->tree_node->sub_root) && -+ remove_tree_node) { -+ rb_erase(&stable_node->tree_node->node, -+ root_stable_treep); -+ free_tree_node(stable_node->tree_node); -+ } else { -+ stable_node->tree_node->count--; -+ } -+ } -+ -+ free_stable_node(stable_node); -+} -+ -+ -+/* -+ * get_uksm_page: checks if the page indicated by the stable node -+ * is still its ksm page, despite having held no reference to it. -+ * In which case we can trust the content of the page, and it -+ * returns the gotten page; but if the page has now been zapped, -+ * remove the stale node from the stable tree and return NULL. -+ * -+ * You would expect the stable_node to hold a reference to the ksm page. -+ * But if it increments the page's count, swapping out has to wait for -+ * ksmd to come around again before it can free the page, which may take -+ * seconds or even minutes: much too unresponsive. So instead we use a -+ * "keyhole reference": access to the ksm page from the stable node peeps -+ * out through its keyhole to see if that page still holds the right key, -+ * pointing back to this stable node. This relies on freeing a PageAnon -+ * page to reset its page->mapping to NULL, and relies on no other use of -+ * a page to put something that might look like our key in page->mapping. -+ * -+ * include/linux/pagemap.h page_cache_get_speculative() is a good reference, -+ * but this is different - made simpler by uksm_thread_mutex being held, but -+ * interesting for assuming that no other use of the struct page could ever -+ * put our expected_mapping into page->mapping (or a field of the union which -+ * coincides with page->mapping). The RCU calls are not for KSM at all, but -+ * to keep the page_count protocol described with page_cache_get_speculative. -+ * -+ * Note: it is possible that get_uksm_page() will return NULL one moment, -+ * then page the next, if the page is in between page_freeze_refs() and -+ * page_unfreeze_refs(): this shouldn't be a problem anywhere, the page -+ * is on its way to being freed; but it is an anomaly to bear in mind. -+ * -+ * @unlink_rb: if the removal of this node will firstly unlink from -+ * its rbtree. stable_node_reinsert will prevent this when restructuring the -+ * node from its old tree. -+ * -+ * @remove_tree_node: if this is the last one of its tree_node, will the -+ * tree_node be freed ? If we are inserting stable node, this tree_node may -+ * be reused, so don't free it. -+ */ -+static struct page *get_uksm_page(struct stable_node *stable_node, -+ int unlink_rb, int remove_tree_node) -+{ -+ struct page *page; -+ void *expected_mapping; -+ unsigned long kpfn; -+ -+ expected_mapping = (void *)((unsigned long)stable_node | -+ PAGE_MAPPING_KSM); -+again: -+ kpfn = READ_ONCE(stable_node->kpfn); -+ page = pfn_to_page(kpfn); -+ -+ /* -+ * page is computed from kpfn, so on most architectures reading -+ * page->mapping is naturally ordered after reading node->kpfn, -+ * but on Alpha we need to be more careful. -+ */ -+ smp_read_barrier_depends(); -+ -+ if (READ_ONCE(page->mapping) != expected_mapping) -+ goto stale; -+ -+ /* -+ * We cannot do anything with the page while its refcount is 0. -+ * Usually 0 means free, or tail of a higher-order page: in which -+ * case this node is no longer referenced, and should be freed; -+ * however, it might mean that the page is under page_freeze_refs(). -+ * The __remove_mapping() case is easy, again the node is now stale; -+ * but if page is swapcache in migrate_page_move_mapping(), it might -+ * still be our page, in which case it's essential to keep the node. -+ */ -+ while (!get_page_unless_zero(page)) { -+ /* -+ * Another check for page->mapping != expected_mapping would -+ * work here too. We have chosen the !PageSwapCache test to -+ * optimize the common case, when the page is or is about to -+ * be freed: PageSwapCache is cleared (under spin_lock_irq) -+ * in the freeze_refs section of __remove_mapping(); but Anon -+ * page->mapping reset to NULL later, in free_pages_prepare(). -+ */ -+ if (!PageSwapCache(page)) -+ goto stale; -+ cpu_relax(); -+ } -+ -+ if (READ_ONCE(page->mapping) != expected_mapping) { -+ put_page(page); -+ goto stale; -+ } -+ -+ lock_page(page); -+ if (READ_ONCE(page->mapping) != expected_mapping) { -+ unlock_page(page); -+ put_page(page); -+ goto stale; -+ } -+ unlock_page(page); -+ return page; -+stale: -+ /* -+ * We come here from above when page->mapping or !PageSwapCache -+ * suggests that the node is stale; but it might be under migration. -+ * We need smp_rmb(), matching the smp_wmb() in ksm_migrate_page(), -+ * before checking whether node->kpfn has been changed. -+ */ -+ smp_rmb(); -+ if (stable_node->kpfn != kpfn) -+ goto again; -+ -+ remove_node_from_stable_tree(stable_node, unlink_rb, remove_tree_node); -+ -+ return NULL; -+} -+ -+/* -+ * Removing rmap_item from stable or unstable tree. -+ * This function will clean the information from the stable/unstable tree. -+ */ -+static inline void remove_rmap_item_from_tree(struct rmap_item *rmap_item) -+{ -+ if (rmap_item->address & STABLE_FLAG) { -+ struct stable_node *stable_node; -+ struct node_vma *node_vma; -+ struct page *page; -+ -+ node_vma = rmap_item->head; -+ stable_node = node_vma->head; -+ page = get_uksm_page(stable_node, 1, 1); -+ if (!page) -+ goto out; -+ -+ /* -+ * page lock is needed because it's racing with -+ * try_to_unmap_ksm(), etc. -+ */ -+ lock_page(page); -+ hlist_del(&rmap_item->hlist); -+ -+ if (hlist_empty(&node_vma->rmap_hlist)) { -+ hlist_del(&node_vma->hlist); -+ free_node_vma(node_vma); -+ } -+ unlock_page(page); -+ -+ put_page(page); -+ if (hlist_empty(&stable_node->hlist)) { -+ /* do NOT call remove_node_from_stable_tree() here, -+ * it's possible for a forked rmap_item not in -+ * stable tree while the in-tree rmap_items were -+ * deleted. -+ */ -+ uksm_pages_shared--; -+ } else -+ uksm_pages_sharing--; -+ -+ -+ uksm_drop_anon_vma(rmap_item); -+ } else if (rmap_item->address & UNSTABLE_FLAG) { -+ if (rmap_item->hash_round == uksm_hash_round) { -+ -+ rb_erase(&rmap_item->node, -+ &rmap_item->tree_node->sub_root); -+ if (RB_EMPTY_ROOT(&rmap_item->tree_node->sub_root)) { -+ rb_erase(&rmap_item->tree_node->node, -+ &root_unstable_tree); -+ -+ free_tree_node(rmap_item->tree_node); -+ } else -+ rmap_item->tree_node->count--; -+ } -+ uksm_pages_unshared--; -+ } -+ -+ rmap_item->address &= PAGE_MASK; -+ rmap_item->hash_max = 0; -+ -+out: -+ cond_resched(); /* we're called from many long loops */ -+} -+ -+static inline int slot_in_uksm(struct vma_slot *slot) -+{ -+ return list_empty(&slot->slot_list); -+} -+ -+/* -+ * Test if the mm is exiting -+ */ -+static inline bool uksm_test_exit(struct mm_struct *mm) -+{ -+ return atomic_read(&mm->mm_users) == 0; -+} -+ -+static inline unsigned long vma_pool_size(struct vma_slot *slot) -+{ -+ return round_up(sizeof(struct rmap_list_entry) * slot->pages, -+ PAGE_SIZE) >> PAGE_SHIFT; -+} -+ -+#define CAN_OVERFLOW_U64(x, delta) (U64_MAX - (x) < (delta)) -+ -+/* must be done with sem locked */ -+static int slot_pool_alloc(struct vma_slot *slot) -+{ -+ unsigned long pool_size; -+ -+ if (slot->rmap_list_pool) -+ return 0; -+ -+ pool_size = vma_pool_size(slot); -+ slot->rmap_list_pool = kcalloc(pool_size, sizeof(struct page *), -+ GFP_KERNEL); -+ if (!slot->rmap_list_pool) -+ return -ENOMEM; -+ -+ slot->pool_counts = kcalloc(pool_size, sizeof(unsigned int), -+ GFP_KERNEL); -+ if (!slot->pool_counts) { -+ kfree(slot->rmap_list_pool); -+ return -ENOMEM; -+ } -+ -+ slot->pool_size = pool_size; -+ BUG_ON(CAN_OVERFLOW_U64(uksm_pages_total, slot->pages)); -+ slot->flags |= UKSM_SLOT_IN_UKSM; -+ uksm_pages_total += slot->pages; -+ -+ return 0; -+} -+ -+/* -+ * Called after vma is unlinked from its mm -+ */ -+void uksm_remove_vma(struct vm_area_struct *vma) -+{ -+ struct vma_slot *slot; -+ -+ if (!vma->uksm_vma_slot) -+ return; -+ -+ spin_lock(&vma_slot_list_lock); -+ slot = vma->uksm_vma_slot; -+ if (!slot) -+ goto out; -+ -+ if (slot_in_uksm(slot)) { -+ /** -+ * This slot has been added by ksmd, so move to the del list -+ * waiting ksmd to free it. -+ */ -+ list_add_tail(&slot->slot_list, &vma_slot_del); -+ } else { -+ /** -+ * It's still on new list. It's ok to free slot directly. -+ */ -+ list_del(&slot->slot_list); -+ free_vma_slot(slot); -+ } -+out: -+ vma->uksm_vma_slot = NULL; -+ spin_unlock(&vma_slot_list_lock); -+} -+ -+/** -+ * Need to do two things: -+ * 1. check if slot was moved to del list -+ * 2. make sure the mmap_sem is manipulated under valid vma. -+ * -+ * My concern here is that in some cases, this may make -+ * vma_slot_list_lock() waiters to serialized further by some -+ * sem->wait_lock, can this really be expensive? -+ * -+ * -+ * @return -+ * 0: if successfully locked mmap_sem -+ * -ENOENT: this slot was moved to del list -+ * -EBUSY: vma lock failed -+ */ -+static int try_down_read_slot_mmap_sem(struct vma_slot *slot) -+{ -+ struct vm_area_struct *vma; -+ struct mm_struct *mm; -+ struct rw_semaphore *sem; -+ -+ spin_lock(&vma_slot_list_lock); -+ -+ /* the slot_list was removed and inited from new list, when it enters -+ * uksm_list. If now it's not empty, then it must be moved to del list -+ */ -+ if (!slot_in_uksm(slot)) { -+ spin_unlock(&vma_slot_list_lock); -+ return -ENOENT; -+ } -+ -+ BUG_ON(slot->pages != vma_pages(slot->vma)); -+ /* Ok, vma still valid */ -+ vma = slot->vma; -+ mm = vma->vm_mm; -+ sem = &mm->mmap_sem; -+ -+ if (uksm_test_exit(mm)) { -+ spin_unlock(&vma_slot_list_lock); -+ return -ENOENT; -+ } -+ -+ if (down_read_trylock(sem)) { -+ spin_unlock(&vma_slot_list_lock); -+ if (slot_pool_alloc(slot)) { -+ uksm_remove_vma(vma); -+ up_read(sem); -+ return -ENOENT; -+ } -+ return 0; -+ } -+ -+ spin_unlock(&vma_slot_list_lock); -+ return -EBUSY; -+} -+ -+static inline unsigned long -+vma_page_address(struct page *page, struct vm_area_struct *vma) -+{ -+ pgoff_t pgoff = page->index; -+ unsigned long address; -+ -+ address = vma->vm_start + ((pgoff - vma->vm_pgoff) << PAGE_SHIFT); -+ if (unlikely(address < vma->vm_start || address >= vma->vm_end)) { -+ /* page should be within @vma mapping range */ -+ return -EFAULT; -+ } -+ return address; -+} -+ -+ -+/* return 0 on success with the item's mmap_sem locked */ -+static inline int get_mergeable_page_lock_mmap(struct rmap_item *item) -+{ -+ struct mm_struct *mm; -+ struct vma_slot *slot = item->slot; -+ int err = -EINVAL; -+ -+ struct page *page; -+ -+ /* -+ * try_down_read_slot_mmap_sem() returns non-zero if the slot -+ * has been removed by uksm_remove_vma(). -+ */ -+ if (try_down_read_slot_mmap_sem(slot)) -+ return -EBUSY; -+ -+ mm = slot->vma->vm_mm; -+ -+ if (uksm_test_exit(mm)) -+ goto failout_up; -+ -+ page = item->page; -+ rcu_read_lock(); -+ if (!get_page_unless_zero(page)) { -+ rcu_read_unlock(); -+ goto failout_up; -+ } -+ -+ /* No need to consider huge page here. */ -+ if (item->slot->vma->anon_vma != page_anon_vma(page) || -+ vma_page_address(page, item->slot->vma) != get_rmap_addr(item)) { -+ /* -+ * TODO: -+ * should we release this item becase of its stale page -+ * mapping? -+ */ -+ put_page(page); -+ rcu_read_unlock(); -+ goto failout_up; -+ } -+ rcu_read_unlock(); -+ return 0; -+ -+failout_up: -+ up_read(&mm->mmap_sem); -+ return err; -+} -+ -+/* -+ * What kind of VMA is considered ? -+ */ -+static inline int vma_can_enter(struct vm_area_struct *vma) -+{ -+ return uksm_flags_can_scan(vma->vm_flags); -+} -+ -+/* -+ * Called whenever a fresh new vma is created A new vma_slot. -+ * is created and inserted into a global list Must be called. -+ * after vma is inserted to its mm. -+ */ -+void uksm_vma_add_new(struct vm_area_struct *vma) -+{ -+ struct vma_slot *slot; -+ -+ if (!vma_can_enter(vma)) { -+ vma->uksm_vma_slot = NULL; -+ return; -+ } -+ -+ slot = alloc_vma_slot(); -+ if (!slot) { -+ vma->uksm_vma_slot = NULL; -+ return; -+ } -+ -+ vma->uksm_vma_slot = slot; -+ vma->vm_flags |= VM_MERGEABLE; -+ slot->vma = vma; -+ slot->mm = vma->vm_mm; -+ slot->ctime_j = jiffies; -+ slot->pages = vma_pages(vma); -+ spin_lock(&vma_slot_list_lock); -+ list_add_tail(&slot->slot_list, &vma_slot_new); -+ spin_unlock(&vma_slot_list_lock); -+} -+ -+/* 32/3 < they < 32/2 */ -+#define shiftl 8 -+#define shiftr 12 -+ -+#define HASH_FROM_TO(from, to) \ -+for (index = from; index < to; index++) { \ -+ pos = random_nums[index]; \ -+ hash += key[pos]; \ -+ hash += (hash << shiftl); \ -+ hash ^= (hash >> shiftr); \ -+} -+ -+ -+#define HASH_FROM_DOWN_TO(from, to) \ -+for (index = from - 1; index >= to; index--) { \ -+ hash ^= (hash >> shiftr); \ -+ hash ^= (hash >> (shiftr*2)); \ -+ hash -= (hash << shiftl); \ -+ hash += (hash << (shiftl*2)); \ -+ pos = random_nums[index]; \ -+ hash -= key[pos]; \ -+} -+ -+/* -+ * The main random sample hash function. -+ */ -+static u32 random_sample_hash(void *addr, u32 hash_strength) -+{ -+ u32 hash = 0xdeadbeef; -+ int index, pos, loop = hash_strength; -+ u32 *key = (u32 *)addr; -+ -+ if (loop > HASH_STRENGTH_FULL) -+ loop = HASH_STRENGTH_FULL; -+ -+ HASH_FROM_TO(0, loop); -+ -+ if (hash_strength > HASH_STRENGTH_FULL) { -+ loop = hash_strength - HASH_STRENGTH_FULL; -+ HASH_FROM_TO(0, loop); -+ } -+ -+ return hash; -+} -+ -+ -+/** -+ * It's used when hash strength is adjusted -+ * -+ * @addr The page's virtual address -+ * @from The original hash strength -+ * @to The hash strength changed to -+ * @hash The hash value generated with "from" hash value -+ * -+ * return the hash value -+ */ -+static u32 delta_hash(void *addr, int from, int to, u32 hash) -+{ -+ u32 *key = (u32 *)addr; -+ int index, pos; /* make sure they are int type */ -+ -+ if (to > from) { -+ if (from >= HASH_STRENGTH_FULL) { -+ from -= HASH_STRENGTH_FULL; -+ to -= HASH_STRENGTH_FULL; -+ HASH_FROM_TO(from, to); -+ } else if (to <= HASH_STRENGTH_FULL) { -+ HASH_FROM_TO(from, to); -+ } else { -+ HASH_FROM_TO(from, HASH_STRENGTH_FULL); -+ HASH_FROM_TO(0, to - HASH_STRENGTH_FULL); -+ } -+ } else { -+ if (from <= HASH_STRENGTH_FULL) { -+ HASH_FROM_DOWN_TO(from, to); -+ } else if (to >= HASH_STRENGTH_FULL) { -+ from -= HASH_STRENGTH_FULL; -+ to -= HASH_STRENGTH_FULL; -+ HASH_FROM_DOWN_TO(from, to); -+ } else { -+ HASH_FROM_DOWN_TO(from - HASH_STRENGTH_FULL, 0); -+ HASH_FROM_DOWN_TO(HASH_STRENGTH_FULL, to); -+ } -+ } -+ -+ return hash; -+} -+ -+/** -+ * -+ * Called when: rshash_pos or rshash_neg is about to overflow or a scan round -+ * has finished. -+ * -+ * return 0 if no page has been scanned since last call, 1 otherwise. -+ */ -+static inline int encode_benefit(void) -+{ -+ u64 scanned_delta, pos_delta, neg_delta; -+ unsigned long base = benefit.base; -+ -+ scanned_delta = uksm_pages_scanned - uksm_pages_scanned_last; -+ -+ if (!scanned_delta) -+ return 0; -+ -+ scanned_delta >>= base; -+ pos_delta = rshash_pos >> base; -+ neg_delta = rshash_neg >> base; -+ -+ if (CAN_OVERFLOW_U64(benefit.pos, pos_delta) || -+ CAN_OVERFLOW_U64(benefit.neg, neg_delta) || -+ CAN_OVERFLOW_U64(benefit.scanned, scanned_delta)) { -+ benefit.scanned >>= 1; -+ benefit.neg >>= 1; -+ benefit.pos >>= 1; -+ benefit.base++; -+ scanned_delta >>= 1; -+ pos_delta >>= 1; -+ neg_delta >>= 1; -+ } -+ -+ benefit.pos += pos_delta; -+ benefit.neg += neg_delta; -+ benefit.scanned += scanned_delta; -+ -+ BUG_ON(!benefit.scanned); -+ -+ rshash_pos = rshash_neg = 0; -+ uksm_pages_scanned_last = uksm_pages_scanned; -+ -+ return 1; -+} -+ -+static inline void reset_benefit(void) -+{ -+ benefit.pos = 0; -+ benefit.neg = 0; -+ benefit.base = 0; -+ benefit.scanned = 0; -+} -+ -+static inline void inc_rshash_pos(unsigned long delta) -+{ -+ if (CAN_OVERFLOW_U64(rshash_pos, delta)) -+ encode_benefit(); -+ -+ rshash_pos += delta; -+} -+ -+static inline void inc_rshash_neg(unsigned long delta) -+{ -+ if (CAN_OVERFLOW_U64(rshash_neg, delta)) -+ encode_benefit(); -+ -+ rshash_neg += delta; -+} -+ -+ -+static inline u32 page_hash(struct page *page, unsigned long hash_strength, -+ int cost_accounting) -+{ -+ u32 val; -+ unsigned long delta; -+ -+ void *addr = kmap_atomic(page); -+ -+ val = random_sample_hash(addr, hash_strength); -+ kunmap_atomic(addr); -+ -+ if (cost_accounting) { -+ if (hash_strength < HASH_STRENGTH_FULL) -+ delta = HASH_STRENGTH_FULL - hash_strength; -+ else -+ delta = 0; -+ -+ inc_rshash_pos(delta); -+ } -+ -+ return val; -+} -+ -+static int memcmp_pages(struct page *page1, struct page *page2, -+ int cost_accounting) -+{ -+ char *addr1, *addr2; -+ int ret; -+ -+ addr1 = kmap_atomic(page1); -+ addr2 = kmap_atomic(page2); -+ ret = memcmp(addr1, addr2, PAGE_SIZE); -+ kunmap_atomic(addr2); -+ kunmap_atomic(addr1); -+ -+ if (cost_accounting) -+ inc_rshash_neg(memcmp_cost); -+ -+ return ret; -+} -+ -+static inline int pages_identical(struct page *page1, struct page *page2) -+{ -+ return !memcmp_pages(page1, page2, 0); -+} -+ -+static inline int is_page_full_zero(struct page *page) -+{ -+ char *addr; -+ int ret; -+ -+ addr = kmap_atomic(page); -+ ret = is_full_zero(addr, PAGE_SIZE); -+ kunmap_atomic(addr); -+ -+ return ret; -+} -+ -+static int write_protect_page(struct vm_area_struct *vma, struct page *page, -+ pte_t *orig_pte, pte_t *old_pte) -+{ -+ struct mm_struct *mm = vma->vm_mm; -+ struct page_vma_mapped_walk pvmw = { -+ .page = page, -+ .vma = vma, -+ }; -+ int swapped; -+ int err = -EFAULT; -+ unsigned long mmun_start; /* For mmu_notifiers */ -+ unsigned long mmun_end; /* For mmu_notifiers */ -+ -+ pvmw.address = page_address_in_vma(page, vma); -+ if (pvmw.address == -EFAULT) -+ goto out; -+ -+ BUG_ON(PageTransCompound(page)); -+ -+ mmun_start = pvmw.address; -+ mmun_end = pvmw.address + PAGE_SIZE; -+ mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); -+ -+ if (!page_vma_mapped_walk(&pvmw)) -+ goto out_mn; -+ if (WARN_ONCE(!pvmw.pte, "Unexpected PMD mapping?")) -+ goto out_unlock; -+ -+ if (old_pte) -+ *old_pte = *pvmw.pte; -+ -+ if (pte_write(*pvmw.pte) || pte_dirty(*pvmw.pte) || -+ (pte_protnone(*pvmw.pte) && pte_savedwrite(*pvmw.pte)) || mm_tlb_flush_pending(mm)) { -+ pte_t entry; -+ -+ swapped = PageSwapCache(page); -+ flush_cache_page(vma, pvmw.address, page_to_pfn(page)); -+ /* -+ * Ok this is tricky, when get_user_pages_fast() run it doesn't -+ * take any lock, therefore the check that we are going to make -+ * with the pagecount against the mapcount is racey and -+ * O_DIRECT can happen right after the check. -+ * So we clear the pte and flush the tlb before the check -+ * this assure us that no O_DIRECT can happen after the check -+ * or in the middle of the check. -+ */ -+ entry = ptep_clear_flush_notify(vma, pvmw.address, pvmw.pte); -+ /* -+ * Check that no O_DIRECT or similar I/O is in progress on the -+ * page -+ */ -+ if (page_mapcount(page) + 1 + swapped != page_count(page)) { -+ set_pte_at(mm, pvmw.address, pvmw.pte, entry); -+ goto out_unlock; -+ } -+ if (pte_dirty(entry)) -+ set_page_dirty(page); -+ -+ if (pte_protnone(entry)) -+ entry = pte_mkclean(pte_clear_savedwrite(entry)); -+ else -+ entry = pte_mkclean(pte_wrprotect(entry)); -+ -+ set_pte_at_notify(mm, pvmw.address, pvmw.pte, entry); -+ } -+ *orig_pte = *pvmw.pte; -+ err = 0; -+ -+out_unlock: -+ page_vma_mapped_walk_done(&pvmw); -+out_mn: -+ mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); -+out: -+ return err; -+} -+ -+#define MERGE_ERR_PGERR 1 /* the page is invalid cannot continue */ -+#define MERGE_ERR_COLLI 2 /* there is a collision */ -+#define MERGE_ERR_COLLI_MAX 3 /* collision at the max hash strength */ -+#define MERGE_ERR_CHANGED 4 /* the page has changed since last hash */ -+ -+ -+/** -+ * replace_page - replace page in vma by new ksm page -+ * @vma: vma that holds the pte pointing to page -+ * @page: the page we are replacing by kpage -+ * @kpage: the ksm page we replace page by -+ * @orig_pte: the original value of the pte -+ * -+ * Returns 0 on success, MERGE_ERR_PGERR on failure. -+ */ -+static int replace_page(struct vm_area_struct *vma, struct page *page, -+ struct page *kpage, pte_t orig_pte) -+{ -+ struct mm_struct *mm = vma->vm_mm; -+ pgd_t *pgd; -+ p4d_t *p4d; -+ pud_t *pud; -+ pmd_t *pmd; -+ pte_t *ptep; -+ spinlock_t *ptl; -+ pte_t entry; -+ -+ unsigned long addr; -+ int err = MERGE_ERR_PGERR; -+ unsigned long mmun_start; /* For mmu_notifiers */ -+ unsigned long mmun_end; /* For mmu_notifiers */ -+ -+ addr = page_address_in_vma(page, vma); -+ if (addr == -EFAULT) -+ goto out; -+ -+ pgd = pgd_offset(mm, addr); -+ if (!pgd_present(*pgd)) -+ goto out; -+ -+ p4d = p4d_offset(pgd, addr); -+ pud = pud_offset(p4d, addr); -+ if (!pud_present(*pud)) -+ goto out; -+ -+ pmd = pmd_offset(pud, addr); -+ BUG_ON(pmd_trans_huge(*pmd)); -+ if (!pmd_present(*pmd)) -+ goto out; -+ -+ mmun_start = addr; -+ mmun_end = addr + PAGE_SIZE; -+ mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); -+ -+ ptep = pte_offset_map_lock(mm, pmd, addr, &ptl); -+ if (!pte_same(*ptep, orig_pte)) { -+ pte_unmap_unlock(ptep, ptl); -+ goto out_mn; -+ } -+ -+ flush_cache_page(vma, addr, pte_pfn(*ptep)); -+ ptep_clear_flush_notify(vma, addr, ptep); -+ entry = mk_pte(kpage, vma->vm_page_prot); -+ -+ /* special treatment is needed for zero_page */ -+ if ((page_to_pfn(kpage) == uksm_zero_pfn) || -+ (page_to_pfn(kpage) == zero_pfn)) { -+ entry = pte_mkspecial(entry); -+ dec_mm_counter(mm, MM_ANONPAGES); -+ inc_zone_page_state(page, NR_UKSM_ZERO_PAGES); -+ } else { -+ get_page(kpage); -+ page_add_anon_rmap(kpage, vma, addr, false); -+ } -+ -+ set_pte_at_notify(mm, addr, ptep, entry); -+ -+ page_remove_rmap(page, false); -+ if (!page_mapped(page)) -+ try_to_free_swap(page); -+ put_page(page); -+ -+ pte_unmap_unlock(ptep, ptl); -+ err = 0; -+out_mn: -+ mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); -+out: -+ return err; -+} -+ -+ -+/** -+ * Fully hash a page with HASH_STRENGTH_MAX return a non-zero hash value. The -+ * zero hash value at HASH_STRENGTH_MAX is used to indicated that its -+ * hash_max member has not been calculated. -+ * -+ * @page The page needs to be hashed -+ * @hash_old The hash value calculated with current hash strength -+ * -+ * return the new hash value calculated at HASH_STRENGTH_MAX -+ */ -+static inline u32 page_hash_max(struct page *page, u32 hash_old) -+{ -+ u32 hash_max = 0; -+ void *addr; -+ -+ addr = kmap_atomic(page); -+ hash_max = delta_hash(addr, hash_strength, -+ HASH_STRENGTH_MAX, hash_old); -+ -+ kunmap_atomic(addr); -+ -+ if (!hash_max) -+ hash_max = 1; -+ -+ inc_rshash_neg(HASH_STRENGTH_MAX - hash_strength); -+ return hash_max; -+} -+ -+/* -+ * We compare the hash again, to ensure that it is really a hash collision -+ * instead of being caused by page write. -+ */ -+static inline int check_collision(struct rmap_item *rmap_item, -+ u32 hash) -+{ -+ int err; -+ struct page *page = rmap_item->page; -+ -+ /* if this rmap_item has already been hash_maxed, then the collision -+ * must appears in the second-level rbtree search. In this case we check -+ * if its hash_max value has been changed. Otherwise, the collision -+ * happens in the first-level rbtree search, so we check against it's -+ * current hash value. -+ */ -+ if (rmap_item->hash_max) { -+ inc_rshash_neg(memcmp_cost); -+ inc_rshash_neg(HASH_STRENGTH_MAX - hash_strength); -+ -+ if (rmap_item->hash_max == page_hash_max(page, hash)) -+ err = MERGE_ERR_COLLI; -+ else -+ err = MERGE_ERR_CHANGED; -+ } else { -+ inc_rshash_neg(memcmp_cost + hash_strength); -+ -+ if (page_hash(page, hash_strength, 0) == hash) -+ err = MERGE_ERR_COLLI; -+ else -+ err = MERGE_ERR_CHANGED; -+ } -+ -+ return err; -+} -+ -+/** -+ * Try to merge a rmap_item.page with a kpage in stable node. kpage must -+ * already be a ksm page. -+ * -+ * @return 0 if the pages were merged, -EFAULT otherwise. -+ */ -+static int try_to_merge_with_uksm_page(struct rmap_item *rmap_item, -+ struct page *kpage, u32 hash) -+{ -+ struct vm_area_struct *vma = rmap_item->slot->vma; -+ struct mm_struct *mm = vma->vm_mm; -+ pte_t orig_pte = __pte(0); -+ int err = MERGE_ERR_PGERR; -+ struct page *page; -+ -+ if (uksm_test_exit(mm)) -+ goto out; -+ -+ page = rmap_item->page; -+ -+ if (page == kpage) { /* ksm page forked */ -+ err = 0; -+ goto out; -+ } -+ -+ /* -+ * We need the page lock to read a stable PageSwapCache in -+ * write_protect_page(). We use trylock_page() instead of -+ * lock_page() because we don't want to wait here - we -+ * prefer to continue scanning and merging different pages, -+ * then come back to this page when it is unlocked. -+ */ -+ if (!trylock_page(page)) -+ goto out; -+ -+ if (!PageAnon(page) || !PageKsm(kpage)) -+ goto out_unlock; -+ -+ if (PageTransCompound(page)) { -+ err = split_huge_page(page); -+ if (err) -+ goto out_unlock; -+ } -+ -+ /* -+ * If this anonymous page is mapped only here, its pte may need -+ * to be write-protected. If it's mapped elsewhere, all of its -+ * ptes are necessarily already write-protected. But in either -+ * case, we need to lock and check page_count is not raised. -+ */ -+ if (write_protect_page(vma, page, &orig_pte, NULL) == 0) { -+ if (pages_identical(page, kpage)) -+ err = replace_page(vma, page, kpage, orig_pte); -+ else -+ err = check_collision(rmap_item, hash); -+ } -+ -+ if ((vma->vm_flags & VM_LOCKED) && kpage && !err) { -+ munlock_vma_page(page); -+ if (!PageMlocked(kpage)) { -+ unlock_page(page); -+ lock_page(kpage); -+ mlock_vma_page(kpage); -+ page = kpage; /* for final unlock */ -+ } -+ } -+ -+out_unlock: -+ unlock_page(page); -+out: -+ return err; -+} -+ -+ -+ -+/** -+ * If two pages fail to merge in try_to_merge_two_pages, then we have a chance -+ * to restore a page mapping that has been changed in try_to_merge_two_pages. -+ * -+ * @return 0 on success. -+ */ -+static int restore_uksm_page_pte(struct vm_area_struct *vma, unsigned long addr, -+ pte_t orig_pte, pte_t wprt_pte) -+{ -+ struct mm_struct *mm = vma->vm_mm; -+ pgd_t *pgd; -+ p4d_t *p4d; -+ pud_t *pud; -+ pmd_t *pmd; -+ pte_t *ptep; -+ spinlock_t *ptl; -+ -+ int err = -EFAULT; -+ -+ pgd = pgd_offset(mm, addr); -+ if (!pgd_present(*pgd)) -+ goto out; -+ -+ p4d = p4d_offset(pgd, addr); -+ pud = pud_offset(p4d, addr); -+ if (!pud_present(*pud)) -+ goto out; -+ -+ pmd = pmd_offset(pud, addr); -+ if (!pmd_present(*pmd)) -+ goto out; -+ -+ ptep = pte_offset_map_lock(mm, pmd, addr, &ptl); -+ if (!pte_same(*ptep, wprt_pte)) { -+ /* already copied, let it be */ -+ pte_unmap_unlock(ptep, ptl); -+ goto out; -+ } -+ -+ /* -+ * Good boy, still here. When we still get the ksm page, it does not -+ * return to the free page pool, there is no way that a pte was changed -+ * to other page and gets back to this page. And remind that ksm page -+ * do not reuse in do_wp_page(). So it's safe to restore the original -+ * pte. -+ */ -+ flush_cache_page(vma, addr, pte_pfn(*ptep)); -+ ptep_clear_flush_notify(vma, addr, ptep); -+ set_pte_at_notify(mm, addr, ptep, orig_pte); -+ -+ pte_unmap_unlock(ptep, ptl); -+ err = 0; -+out: -+ return err; -+} -+ -+/** -+ * try_to_merge_two_pages() - take two identical pages and prepare -+ * them to be merged into one page(rmap_item->page) -+ * -+ * @return 0 if we successfully merged two identical pages into -+ * one ksm page. MERGE_ERR_COLLI if it's only a hash collision -+ * search in rbtree. MERGE_ERR_CHANGED if rmap_item has been -+ * changed since it's hashed. MERGE_ERR_PGERR otherwise. -+ * -+ */ -+static int try_to_merge_two_pages(struct rmap_item *rmap_item, -+ struct rmap_item *tree_rmap_item, -+ u32 hash) -+{ -+ pte_t orig_pte1 = __pte(0), orig_pte2 = __pte(0); -+ pte_t wprt_pte1 = __pte(0), wprt_pte2 = __pte(0); -+ struct vm_area_struct *vma1 = rmap_item->slot->vma; -+ struct vm_area_struct *vma2 = tree_rmap_item->slot->vma; -+ struct page *page = rmap_item->page; -+ struct page *tree_page = tree_rmap_item->page; -+ int err = MERGE_ERR_PGERR; -+ struct address_space *saved_mapping; -+ -+ -+ if (rmap_item->page == tree_rmap_item->page) -+ goto out; -+ -+ if (!trylock_page(page)) -+ goto out; -+ -+ if (!PageAnon(page)) -+ goto out_unlock; -+ -+ if (PageTransCompound(page)) { -+ err = split_huge_page(page); -+ if (err) -+ goto out_unlock; -+ } -+ -+ if (write_protect_page(vma1, page, &wprt_pte1, &orig_pte1) != 0) { -+ unlock_page(page); -+ goto out; -+ } -+ -+ /* -+ * While we hold page lock, upgrade page from -+ * PageAnon+anon_vma to PageKsm+NULL stable_node: -+ * stable_tree_insert() will update stable_node. -+ */ -+ saved_mapping = page->mapping; -+ set_page_stable_node(page, NULL); -+ mark_page_accessed(page); -+ if (!PageDirty(page)) -+ SetPageDirty(page); -+ -+ unlock_page(page); -+ -+ if (!trylock_page(tree_page)) -+ goto restore_out; -+ -+ if (!PageAnon(tree_page)) { -+ unlock_page(tree_page); -+ goto restore_out; -+ } -+ -+ if (PageTransCompound(tree_page)) { -+ err = split_huge_page(tree_page); -+ if (err) { -+ unlock_page(tree_page); -+ goto restore_out; -+ } -+ } -+ -+ if (write_protect_page(vma2, tree_page, &wprt_pte2, &orig_pte2) != 0) { -+ unlock_page(tree_page); -+ goto restore_out; -+ } -+ -+ if (pages_identical(page, tree_page)) { -+ err = replace_page(vma2, tree_page, page, wprt_pte2); -+ if (err) { -+ unlock_page(tree_page); -+ goto restore_out; -+ } -+ -+ if ((vma2->vm_flags & VM_LOCKED)) { -+ munlock_vma_page(tree_page); -+ if (!PageMlocked(page)) { -+ unlock_page(tree_page); -+ lock_page(page); -+ mlock_vma_page(page); -+ tree_page = page; /* for final unlock */ -+ } -+ } -+ -+ unlock_page(tree_page); -+ -+ goto out; /* success */ -+ -+ } else { -+ if (tree_rmap_item->hash_max && -+ tree_rmap_item->hash_max == rmap_item->hash_max) { -+ err = MERGE_ERR_COLLI_MAX; -+ } else if (page_hash(page, hash_strength, 0) == -+ page_hash(tree_page, hash_strength, 0)) { -+ inc_rshash_neg(memcmp_cost + hash_strength * 2); -+ err = MERGE_ERR_COLLI; -+ } else { -+ err = MERGE_ERR_CHANGED; -+ } -+ -+ unlock_page(tree_page); -+ } -+ -+restore_out: -+ lock_page(page); -+ if (!restore_uksm_page_pte(vma1, get_rmap_addr(rmap_item), -+ orig_pte1, wprt_pte1)) -+ page->mapping = saved_mapping; -+ -+out_unlock: -+ unlock_page(page); -+out: -+ return err; -+} -+ -+static inline int hash_cmp(u32 new_val, u32 node_val) -+{ -+ if (new_val > node_val) -+ return 1; -+ else if (new_val < node_val) -+ return -1; -+ else -+ return 0; -+} -+ -+static inline u32 rmap_item_hash_max(struct rmap_item *item, u32 hash) -+{ -+ u32 hash_max = item->hash_max; -+ -+ if (!hash_max) { -+ hash_max = page_hash_max(item->page, hash); -+ -+ item->hash_max = hash_max; -+ } -+ -+ return hash_max; -+} -+ -+ -+ -+/** -+ * stable_tree_search() - search the stable tree for a page -+ * -+ * @item: the rmap_item we are comparing with -+ * @hash: the hash value of this item->page already calculated -+ * -+ * @return the page we have found, NULL otherwise. The page returned has -+ * been gotten. -+ */ -+static struct page *stable_tree_search(struct rmap_item *item, u32 hash) -+{ -+ struct rb_node *node = root_stable_treep->rb_node; -+ struct tree_node *tree_node; -+ unsigned long hash_max; -+ struct page *page = item->page; -+ struct stable_node *stable_node; -+ -+ stable_node = page_stable_node(page); -+ if (stable_node) { -+ /* ksm page forked, that is -+ * if (PageKsm(page) && !in_stable_tree(rmap_item)) -+ * it's actually gotten once outside. -+ */ -+ get_page(page); -+ return page; -+ } -+ -+ while (node) { -+ int cmp; -+ -+ tree_node = rb_entry(node, struct tree_node, node); -+ -+ cmp = hash_cmp(hash, tree_node->hash); -+ -+ if (cmp < 0) -+ node = node->rb_left; -+ else if (cmp > 0) -+ node = node->rb_right; -+ else -+ break; -+ } -+ -+ if (!node) -+ return NULL; -+ -+ if (tree_node->count == 1) { -+ stable_node = rb_entry(tree_node->sub_root.rb_node, -+ struct stable_node, node); -+ BUG_ON(!stable_node); -+ -+ goto get_page_out; -+ } -+ -+ /* -+ * ok, we have to search the second -+ * level subtree, hash the page to a -+ * full strength. -+ */ -+ node = tree_node->sub_root.rb_node; -+ BUG_ON(!node); -+ hash_max = rmap_item_hash_max(item, hash); -+ -+ while (node) { -+ int cmp; -+ -+ stable_node = rb_entry(node, struct stable_node, node); -+ -+ cmp = hash_cmp(hash_max, stable_node->hash_max); -+ -+ if (cmp < 0) -+ node = node->rb_left; -+ else if (cmp > 0) -+ node = node->rb_right; -+ else -+ goto get_page_out; -+ } -+ -+ return NULL; -+ -+get_page_out: -+ page = get_uksm_page(stable_node, 1, 1); -+ return page; -+} -+ -+static int try_merge_rmap_item(struct rmap_item *item, -+ struct page *kpage, -+ struct page *tree_page) -+{ -+ struct vm_area_struct *vma = item->slot->vma; -+ struct page_vma_mapped_walk pvmw = { -+ .page = kpage, -+ .vma = vma, -+ }; -+ -+ pvmw.address = get_rmap_addr(item); -+ if (!page_vma_mapped_walk(&pvmw)) -+ return 0; -+ -+ if (pte_write(*pvmw.pte)) { -+ /* has changed, abort! */ -+ page_vma_mapped_walk_done(&pvmw); -+ return 0; -+ } -+ -+ get_page(tree_page); -+ page_add_anon_rmap(tree_page, vma, pvmw.address, false); -+ -+ flush_cache_page(vma, pvmw.address, page_to_pfn(kpage)); -+ ptep_clear_flush_notify(vma, pvmw.address, pvmw.pte); -+ set_pte_at_notify(vma->vm_mm, pvmw.address, pvmw.pte, -+ mk_pte(tree_page, vma->vm_page_prot)); -+ -+ page_remove_rmap(kpage, false); -+ put_page(kpage); -+ -+ page_vma_mapped_walk_done(&pvmw); -+ -+ return 1; -+} -+ -+/** -+ * try_to_merge_with_stable_page() - when two rmap_items need to be inserted -+ * into stable tree, the page was found to be identical to a stable ksm page, -+ * this is the last chance we can merge them into one. -+ * -+ * @item1: the rmap_item holding the page which we wanted to insert -+ * into stable tree. -+ * @item2: the other rmap_item we found when unstable tree search -+ * @oldpage: the page currently mapped by the two rmap_items -+ * @tree_page: the page we found identical in stable tree node -+ * @success1: return if item1 is successfully merged -+ * @success2: return if item2 is successfully merged -+ */ -+static void try_merge_with_stable(struct rmap_item *item1, -+ struct rmap_item *item2, -+ struct page **kpage, -+ struct page *tree_page, -+ int *success1, int *success2) -+{ -+ struct vm_area_struct *vma1 = item1->slot->vma; -+ struct vm_area_struct *vma2 = item2->slot->vma; -+ *success1 = 0; -+ *success2 = 0; -+ -+ if (unlikely(*kpage == tree_page)) { -+ /* I don't think this can really happen */ -+ pr_warn("UKSM: unexpected condition detected in " -+ "%s -- *kpage == tree_page !\n", __func__); -+ *success1 = 1; -+ *success2 = 1; -+ return; -+ } -+ -+ if (!PageAnon(*kpage) || !PageKsm(*kpage)) -+ goto failed; -+ -+ if (!trylock_page(tree_page)) -+ goto failed; -+ -+ /* If the oldpage is still ksm and still pointed -+ * to in the right place, and still write protected, -+ * we are confident it's not changed, no need to -+ * memcmp anymore. -+ * be ware, we cannot take nested pte locks, -+ * deadlock risk. -+ */ -+ if (!try_merge_rmap_item(item1, *kpage, tree_page)) -+ goto unlock_failed; -+ -+ /* ok, then vma2, remind that pte1 already set */ -+ if (!try_merge_rmap_item(item2, *kpage, tree_page)) -+ goto success_1; -+ -+ *success2 = 1; -+success_1: -+ *success1 = 1; -+ -+ -+ if ((*success1 && vma1->vm_flags & VM_LOCKED) || -+ (*success2 && vma2->vm_flags & VM_LOCKED)) { -+ munlock_vma_page(*kpage); -+ if (!PageMlocked(tree_page)) -+ mlock_vma_page(tree_page); -+ } -+ -+ /* -+ * We do not need oldpage any more in the caller, so can break the lock -+ * now. -+ */ -+ unlock_page(*kpage); -+ *kpage = tree_page; /* Get unlocked outside. */ -+ return; -+ -+unlock_failed: -+ unlock_page(tree_page); -+failed: -+ return; -+} -+ -+static inline void stable_node_hash_max(struct stable_node *node, -+ struct page *page, u32 hash) -+{ -+ u32 hash_max = node->hash_max; -+ -+ if (!hash_max) { -+ hash_max = page_hash_max(page, hash); -+ node->hash_max = hash_max; -+ } -+} -+ -+static inline -+struct stable_node *new_stable_node(struct tree_node *tree_node, -+ struct page *kpage, u32 hash_max) -+{ -+ struct stable_node *new_stable_node; -+ -+ new_stable_node = alloc_stable_node(); -+ if (!new_stable_node) -+ return NULL; -+ -+ new_stable_node->kpfn = page_to_pfn(kpage); -+ new_stable_node->hash_max = hash_max; -+ new_stable_node->tree_node = tree_node; -+ set_page_stable_node(kpage, new_stable_node); -+ -+ return new_stable_node; -+} -+ -+static inline -+struct stable_node *first_level_insert(struct tree_node *tree_node, -+ struct rmap_item *rmap_item, -+ struct rmap_item *tree_rmap_item, -+ struct page **kpage, u32 hash, -+ int *success1, int *success2) -+{ -+ int cmp; -+ struct page *tree_page; -+ u32 hash_max = 0; -+ struct stable_node *stable_node, *new_snode; -+ struct rb_node *parent = NULL, **new; -+ -+ /* this tree node contains no sub-tree yet */ -+ stable_node = rb_entry(tree_node->sub_root.rb_node, -+ struct stable_node, node); -+ -+ tree_page = get_uksm_page(stable_node, 1, 0); -+ if (tree_page) { -+ cmp = memcmp_pages(*kpage, tree_page, 1); -+ if (!cmp) { -+ try_merge_with_stable(rmap_item, tree_rmap_item, kpage, -+ tree_page, success1, success2); -+ put_page(tree_page); -+ if (!*success1 && !*success2) -+ goto failed; -+ -+ return stable_node; -+ -+ } else { -+ /* -+ * collision in first level try to create a subtree. -+ * A new node need to be created. -+ */ -+ put_page(tree_page); -+ -+ stable_node_hash_max(stable_node, tree_page, -+ tree_node->hash); -+ hash_max = rmap_item_hash_max(rmap_item, hash); -+ cmp = hash_cmp(hash_max, stable_node->hash_max); -+ -+ parent = &stable_node->node; -+ if (cmp < 0) -+ new = &parent->rb_left; -+ else if (cmp > 0) -+ new = &parent->rb_right; -+ else -+ goto failed; -+ } -+ -+ } else { -+ /* the only stable_node deleted, we reuse its tree_node. -+ */ -+ parent = NULL; -+ new = &tree_node->sub_root.rb_node; -+ } -+ -+ new_snode = new_stable_node(tree_node, *kpage, hash_max); -+ if (!new_snode) -+ goto failed; -+ -+ rb_link_node(&new_snode->node, parent, new); -+ rb_insert_color(&new_snode->node, &tree_node->sub_root); -+ tree_node->count++; -+ *success1 = *success2 = 1; -+ -+ return new_snode; -+ -+failed: -+ return NULL; -+} -+ -+static inline -+struct stable_node *stable_subtree_insert(struct tree_node *tree_node, -+ struct rmap_item *rmap_item, -+ struct rmap_item *tree_rmap_item, -+ struct page **kpage, u32 hash, -+ int *success1, int *success2) -+{ -+ struct page *tree_page; -+ u32 hash_max; -+ struct stable_node *stable_node, *new_snode; -+ struct rb_node *parent, **new; -+ -+research: -+ parent = NULL; -+ new = &tree_node->sub_root.rb_node; -+ BUG_ON(!*new); -+ hash_max = rmap_item_hash_max(rmap_item, hash); -+ while (*new) { -+ int cmp; -+ -+ stable_node = rb_entry(*new, struct stable_node, node); -+ -+ cmp = hash_cmp(hash_max, stable_node->hash_max); -+ -+ if (cmp < 0) { -+ parent = *new; -+ new = &parent->rb_left; -+ } else if (cmp > 0) { -+ parent = *new; -+ new = &parent->rb_right; -+ } else { -+ tree_page = get_uksm_page(stable_node, 1, 0); -+ if (tree_page) { -+ cmp = memcmp_pages(*kpage, tree_page, 1); -+ if (!cmp) { -+ try_merge_with_stable(rmap_item, -+ tree_rmap_item, kpage, -+ tree_page, success1, success2); -+ -+ put_page(tree_page); -+ if (!*success1 && !*success2) -+ goto failed; -+ /* -+ * successfully merged with a stable -+ * node -+ */ -+ return stable_node; -+ } else { -+ put_page(tree_page); -+ goto failed; -+ } -+ } else { -+ /* -+ * stable node may be deleted, -+ * and subtree maybe -+ * restructed, cannot -+ * continue, research it. -+ */ -+ if (tree_node->count) { -+ goto research; -+ } else { -+ /* reuse the tree node*/ -+ parent = NULL; -+ new = &tree_node->sub_root.rb_node; -+ } -+ } -+ } -+ } -+ -+ new_snode = new_stable_node(tree_node, *kpage, hash_max); -+ if (!new_snode) -+ goto failed; -+ -+ rb_link_node(&new_snode->node, parent, new); -+ rb_insert_color(&new_snode->node, &tree_node->sub_root); -+ tree_node->count++; -+ *success1 = *success2 = 1; -+ -+ return new_snode; -+ -+failed: -+ return NULL; -+} -+ -+ -+/** -+ * stable_tree_insert() - try to insert a merged page in unstable tree to -+ * the stable tree -+ * -+ * @kpage: the page need to be inserted -+ * @hash: the current hash of this page -+ * @rmap_item: the rmap_item being scanned -+ * @tree_rmap_item: the rmap_item found on unstable tree -+ * @success1: return if rmap_item is merged -+ * @success2: return if tree_rmap_item is merged -+ * -+ * @return the stable_node on stable tree if at least one -+ * rmap_item is inserted into stable tree, NULL -+ * otherwise. -+ */ -+static struct stable_node * -+stable_tree_insert(struct page **kpage, u32 hash, -+ struct rmap_item *rmap_item, -+ struct rmap_item *tree_rmap_item, -+ int *success1, int *success2) -+{ -+ struct rb_node **new = &root_stable_treep->rb_node; -+ struct rb_node *parent = NULL; -+ struct stable_node *stable_node; -+ struct tree_node *tree_node; -+ u32 hash_max = 0; -+ -+ *success1 = *success2 = 0; -+ -+ while (*new) { -+ int cmp; -+ -+ tree_node = rb_entry(*new, struct tree_node, node); -+ -+ cmp = hash_cmp(hash, tree_node->hash); -+ -+ if (cmp < 0) { -+ parent = *new; -+ new = &parent->rb_left; -+ } else if (cmp > 0) { -+ parent = *new; -+ new = &parent->rb_right; -+ } else -+ break; -+ } -+ -+ if (*new) { -+ if (tree_node->count == 1) { -+ stable_node = first_level_insert(tree_node, rmap_item, -+ tree_rmap_item, kpage, -+ hash, success1, success2); -+ } else { -+ stable_node = stable_subtree_insert(tree_node, -+ rmap_item, tree_rmap_item, kpage, -+ hash, success1, success2); -+ } -+ } else { -+ -+ /* no tree node found */ -+ tree_node = alloc_tree_node(stable_tree_node_listp); -+ if (!tree_node) { -+ stable_node = NULL; -+ goto out; -+ } -+ -+ stable_node = new_stable_node(tree_node, *kpage, hash_max); -+ if (!stable_node) { -+ free_tree_node(tree_node); -+ goto out; -+ } -+ -+ tree_node->hash = hash; -+ rb_link_node(&tree_node->node, parent, new); -+ rb_insert_color(&tree_node->node, root_stable_treep); -+ parent = NULL; -+ new = &tree_node->sub_root.rb_node; -+ -+ rb_link_node(&stable_node->node, parent, new); -+ rb_insert_color(&stable_node->node, &tree_node->sub_root); -+ tree_node->count++; -+ *success1 = *success2 = 1; -+ } -+ -+out: -+ return stable_node; -+} -+ -+ -+/** -+ * get_tree_rmap_item_page() - try to get the page and lock the mmap_sem -+ * -+ * @return 0 on success, -EBUSY if unable to lock the mmap_sem, -+ * -EINVAL if the page mapping has been changed. -+ */ -+static inline int get_tree_rmap_item_page(struct rmap_item *tree_rmap_item) -+{ -+ int err; -+ -+ err = get_mergeable_page_lock_mmap(tree_rmap_item); -+ -+ if (err == -EINVAL) { -+ /* its page map has been changed, remove it */ -+ remove_rmap_item_from_tree(tree_rmap_item); -+ } -+ -+ /* The page is gotten and mmap_sem is locked now. */ -+ return err; -+} -+ -+ -+/** -+ * unstable_tree_search_insert() - search an unstable tree rmap_item with the -+ * same hash value. Get its page and trylock the mmap_sem -+ */ -+static inline -+struct rmap_item *unstable_tree_search_insert(struct rmap_item *rmap_item, -+ u32 hash) -+ -+{ -+ struct rb_node **new = &root_unstable_tree.rb_node; -+ struct rb_node *parent = NULL; -+ struct tree_node *tree_node; -+ u32 hash_max; -+ struct rmap_item *tree_rmap_item; -+ -+ while (*new) { -+ int cmp; -+ -+ tree_node = rb_entry(*new, struct tree_node, node); -+ -+ cmp = hash_cmp(hash, tree_node->hash); -+ -+ if (cmp < 0) { -+ parent = *new; -+ new = &parent->rb_left; -+ } else if (cmp > 0) { -+ parent = *new; -+ new = &parent->rb_right; -+ } else -+ break; -+ } -+ -+ if (*new) { -+ /* got the tree_node */ -+ if (tree_node->count == 1) { -+ tree_rmap_item = rb_entry(tree_node->sub_root.rb_node, -+ struct rmap_item, node); -+ BUG_ON(!tree_rmap_item); -+ -+ goto get_page_out; -+ } -+ -+ /* well, search the collision subtree */ -+ new = &tree_node->sub_root.rb_node; -+ BUG_ON(!*new); -+ hash_max = rmap_item_hash_max(rmap_item, hash); -+ -+ while (*new) { -+ int cmp; -+ -+ tree_rmap_item = rb_entry(*new, struct rmap_item, -+ node); -+ -+ cmp = hash_cmp(hash_max, tree_rmap_item->hash_max); -+ parent = *new; -+ if (cmp < 0) -+ new = &parent->rb_left; -+ else if (cmp > 0) -+ new = &parent->rb_right; -+ else -+ goto get_page_out; -+ } -+ } else { -+ /* alloc a new tree_node */ -+ tree_node = alloc_tree_node(&unstable_tree_node_list); -+ if (!tree_node) -+ return NULL; -+ -+ tree_node->hash = hash; -+ rb_link_node(&tree_node->node, parent, new); -+ rb_insert_color(&tree_node->node, &root_unstable_tree); -+ parent = NULL; -+ new = &tree_node->sub_root.rb_node; -+ } -+ -+ /* did not found even in sub-tree */ -+ rmap_item->tree_node = tree_node; -+ rmap_item->address |= UNSTABLE_FLAG; -+ rmap_item->hash_round = uksm_hash_round; -+ rb_link_node(&rmap_item->node, parent, new); -+ rb_insert_color(&rmap_item->node, &tree_node->sub_root); -+ -+ uksm_pages_unshared++; -+ return NULL; -+ -+get_page_out: -+ if (tree_rmap_item->page == rmap_item->page) -+ return NULL; -+ -+ if (get_tree_rmap_item_page(tree_rmap_item)) -+ return NULL; -+ -+ return tree_rmap_item; -+} -+ -+static void hold_anon_vma(struct rmap_item *rmap_item, -+ struct anon_vma *anon_vma) -+{ -+ rmap_item->anon_vma = anon_vma; -+ get_anon_vma(anon_vma); -+} -+ -+ -+/** -+ * stable_tree_append() - append a rmap_item to a stable node. Deduplication -+ * ratio statistics is done in this function. -+ * -+ */ -+static void stable_tree_append(struct rmap_item *rmap_item, -+ struct stable_node *stable_node, int logdedup) -+{ -+ struct node_vma *node_vma = NULL, *new_node_vma, *node_vma_cont = NULL; -+ unsigned long key = (unsigned long)rmap_item->slot; -+ unsigned long factor = rmap_item->slot->rung->step; -+ -+ BUG_ON(!stable_node); -+ rmap_item->address |= STABLE_FLAG; -+ -+ if (hlist_empty(&stable_node->hlist)) { -+ uksm_pages_shared++; -+ goto node_vma_new; -+ } else { -+ uksm_pages_sharing++; -+ } -+ -+ hlist_for_each_entry(node_vma, &stable_node->hlist, hlist) { -+ if (node_vma->key >= key) -+ break; -+ -+ if (logdedup) { -+ node_vma->slot->pages_bemerged += factor; -+ if (list_empty(&node_vma->slot->dedup_list)) -+ list_add(&node_vma->slot->dedup_list, -+ &vma_slot_dedup); -+ } -+ } -+ -+ if (node_vma) { -+ if (node_vma->key == key) { -+ node_vma_cont = hlist_entry_safe(node_vma->hlist.next, struct node_vma, hlist); -+ goto node_vma_ok; -+ } else if (node_vma->key > key) { -+ node_vma_cont = node_vma; -+ } -+ } -+ -+node_vma_new: -+ /* no same vma already in node, alloc a new node_vma */ -+ new_node_vma = alloc_node_vma(); -+ BUG_ON(!new_node_vma); -+ new_node_vma->head = stable_node; -+ new_node_vma->slot = rmap_item->slot; -+ -+ if (!node_vma) { -+ hlist_add_head(&new_node_vma->hlist, &stable_node->hlist); -+ } else if (node_vma->key != key) { -+ if (node_vma->key < key) -+ hlist_add_behind(&new_node_vma->hlist, &node_vma->hlist); -+ else { -+ hlist_add_before(&new_node_vma->hlist, -+ &node_vma->hlist); -+ } -+ -+ } -+ node_vma = new_node_vma; -+ -+node_vma_ok: /* ok, ready to add to the list */ -+ rmap_item->head = node_vma; -+ hlist_add_head(&rmap_item->hlist, &node_vma->rmap_hlist); -+ hold_anon_vma(rmap_item, rmap_item->slot->vma->anon_vma); -+ if (logdedup) { -+ rmap_item->slot->pages_merged++; -+ if (node_vma_cont) { -+ node_vma = node_vma_cont; -+ hlist_for_each_entry_continue(node_vma, hlist) { -+ node_vma->slot->pages_bemerged += factor; -+ if (list_empty(&node_vma->slot->dedup_list)) -+ list_add(&node_vma->slot->dedup_list, -+ &vma_slot_dedup); -+ } -+ } -+ } -+} -+ -+/* -+ * We use break_ksm to break COW on a ksm page: it's a stripped down -+ * -+ * if (get_user_pages(addr, 1, 1, 1, &page, NULL) == 1) -+ * put_page(page); -+ * -+ * but taking great care only to touch a ksm page, in a VM_MERGEABLE vma, -+ * in case the application has unmapped and remapped mm,addr meanwhile. -+ * Could a ksm page appear anywhere else? Actually yes, in a VM_PFNMAP -+ * mmap of /dev/mem or /dev/kmem, where we would not want to touch it. -+ */ -+static int break_ksm(struct vm_area_struct *vma, unsigned long addr) -+{ -+ struct page *page; -+ int ret = 0; -+ -+ do { -+ cond_resched(); -+ page = follow_page(vma, addr, FOLL_GET | FOLL_MIGRATION | FOLL_REMOTE); -+ if (IS_ERR_OR_NULL(page)) -+ break; -+ if (PageKsm(page)) { -+ ret = handle_mm_fault(vma, addr, -+ FAULT_FLAG_WRITE | FAULT_FLAG_REMOTE); -+ } else -+ ret = VM_FAULT_WRITE; -+ put_page(page); -+ } while (!(ret & (VM_FAULT_WRITE | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | VM_FAULT_OOM))); -+ /* -+ * We must loop because handle_mm_fault() may back out if there's -+ * any difficulty e.g. if pte accessed bit gets updated concurrently. -+ * -+ * VM_FAULT_WRITE is what we have been hoping for: it indicates that -+ * COW has been broken, even if the vma does not permit VM_WRITE; -+ * but note that a concurrent fault might break PageKsm for us. -+ * -+ * VM_FAULT_SIGBUS could occur if we race with truncation of the -+ * backing file, which also invalidates anonymous pages: that's -+ * okay, that truncation will have unmapped the PageKsm for us. -+ * -+ * VM_FAULT_OOM: at the time of writing (late July 2009), setting -+ * aside mem_cgroup limits, VM_FAULT_OOM would only be set if the -+ * current task has TIF_MEMDIE set, and will be OOM killed on return -+ * to user; and ksmd, having no mm, would never be chosen for that. -+ * -+ * But if the mm is in a limited mem_cgroup, then the fault may fail -+ * with VM_FAULT_OOM even if the current task is not TIF_MEMDIE; and -+ * even ksmd can fail in this way - though it's usually breaking ksm -+ * just to undo a merge it made a moment before, so unlikely to oom. -+ * -+ * That's a pity: we might therefore have more kernel pages allocated -+ * than we're counting as nodes in the stable tree; but uksm_do_scan -+ * will retry to break_cow on each pass, so should recover the page -+ * in due course. The important thing is to not let VM_MERGEABLE -+ * be cleared while any such pages might remain in the area. -+ */ -+ return (ret & VM_FAULT_OOM) ? -ENOMEM : 0; -+} -+ -+static void break_cow(struct rmap_item *rmap_item) -+{ -+ struct vm_area_struct *vma = rmap_item->slot->vma; -+ struct mm_struct *mm = vma->vm_mm; -+ unsigned long addr = get_rmap_addr(rmap_item); -+ -+ if (uksm_test_exit(mm)) -+ goto out; -+ -+ break_ksm(vma, addr); -+out: -+ return; -+} -+ -+/* -+ * Though it's very tempting to unmerge in_stable_tree(rmap_item)s rather -+ * than check every pte of a given vma, the locking doesn't quite work for -+ * that - an rmap_item is assigned to the stable tree after inserting ksm -+ * page and upping mmap_sem. Nor does it fit with the way we skip dup'ing -+ * rmap_items from parent to child at fork time (so as not to waste time -+ * if exit comes before the next scan reaches it). -+ * -+ * Similarly, although we'd like to remove rmap_items (so updating counts -+ * and freeing memory) when unmerging an area, it's easier to leave that -+ * to the next pass of ksmd - consider, for example, how ksmd might be -+ * in cmp_and_merge_page on one of the rmap_items we would be removing. -+ */ -+inline int unmerge_uksm_pages(struct vm_area_struct *vma, -+ unsigned long start, unsigned long end) -+{ -+ unsigned long addr; -+ int err = 0; -+ -+ for (addr = start; addr < end && !err; addr += PAGE_SIZE) { -+ if (uksm_test_exit(vma->vm_mm)) -+ break; -+ if (signal_pending(current)) -+ err = -ERESTARTSYS; -+ else -+ err = break_ksm(vma, addr); -+ } -+ return err; -+} -+ -+static inline void inc_uksm_pages_scanned(void) -+{ -+ u64 delta; -+ -+ -+ if (uksm_pages_scanned == U64_MAX) { -+ encode_benefit(); -+ -+ delta = uksm_pages_scanned >> pages_scanned_base; -+ -+ if (CAN_OVERFLOW_U64(pages_scanned_stored, delta)) { -+ pages_scanned_stored >>= 1; -+ delta >>= 1; -+ pages_scanned_base++; -+ } -+ -+ pages_scanned_stored += delta; -+ -+ uksm_pages_scanned = uksm_pages_scanned_last = 0; -+ } -+ -+ uksm_pages_scanned++; -+} -+ -+static inline int find_zero_page_hash(int strength, u32 hash) -+{ -+ return (zero_hash_table[strength] == hash); -+} -+ -+static -+int cmp_and_merge_zero_page(struct vm_area_struct *vma, struct page *page) -+{ -+ struct page *zero_page = empty_uksm_zero_page; -+ struct mm_struct *mm = vma->vm_mm; -+ pte_t orig_pte = __pte(0); -+ int err = -EFAULT; -+ -+ if (uksm_test_exit(mm)) -+ goto out; -+ -+ if (!trylock_page(page)) -+ goto out; -+ -+ if (!PageAnon(page)) -+ goto out_unlock; -+ -+ if (PageTransCompound(page)) { -+ err = split_huge_page(page); -+ if (err) -+ goto out_unlock; -+ } -+ -+ if (write_protect_page(vma, page, &orig_pte, 0) == 0) { -+ if (is_page_full_zero(page)) -+ err = replace_page(vma, page, zero_page, orig_pte); -+ } -+ -+out_unlock: -+ unlock_page(page); -+out: -+ return err; -+} -+ -+/* -+ * cmp_and_merge_page() - first see if page can be merged into the stable -+ * tree; if not, compare hash to previous and if it's the same, see if page -+ * can be inserted into the unstable tree, or merged with a page already there -+ * and both transferred to the stable tree. -+ * -+ * @page: the page that we are searching identical page to. -+ * @rmap_item: the reverse mapping into the virtual address of this page -+ */ -+static void cmp_and_merge_page(struct rmap_item *rmap_item, u32 hash) -+{ -+ struct rmap_item *tree_rmap_item; -+ struct page *page; -+ struct page *kpage = NULL; -+ u32 hash_max; -+ int err; -+ unsigned int success1, success2; -+ struct stable_node *snode; -+ int cmp; -+ struct rb_node *parent = NULL, **new; -+ -+ remove_rmap_item_from_tree(rmap_item); -+ page = rmap_item->page; -+ -+ /* We first start with searching the page inside the stable tree */ -+ kpage = stable_tree_search(rmap_item, hash); -+ if (kpage) { -+ err = try_to_merge_with_uksm_page(rmap_item, kpage, -+ hash); -+ if (!err) { -+ /* -+ * The page was successfully merged, add -+ * its rmap_item to the stable tree. -+ * page lock is needed because it's -+ * racing with try_to_unmap_ksm(), etc. -+ */ -+ lock_page(kpage); -+ snode = page_stable_node(kpage); -+ stable_tree_append(rmap_item, snode, 1); -+ unlock_page(kpage); -+ put_page(kpage); -+ return; /* success */ -+ } -+ put_page(kpage); -+ -+ /* -+ * if it's a collision and it has been search in sub-rbtree -+ * (hash_max != 0), we want to abort, because if it is -+ * successfully merged in unstable tree, the collision trends to -+ * happen again. -+ */ -+ if (err == MERGE_ERR_COLLI && rmap_item->hash_max) -+ return; -+ } -+ -+ tree_rmap_item = -+ unstable_tree_search_insert(rmap_item, hash); -+ if (tree_rmap_item) { -+ err = try_to_merge_two_pages(rmap_item, tree_rmap_item, hash); -+ /* -+ * As soon as we merge this page, we want to remove the -+ * rmap_item of the page we have merged with from the unstable -+ * tree, and insert it instead as new node in the stable tree. -+ */ -+ if (!err) { -+ kpage = page; -+ remove_rmap_item_from_tree(tree_rmap_item); -+ lock_page(kpage); -+ snode = stable_tree_insert(&kpage, hash, -+ rmap_item, tree_rmap_item, -+ &success1, &success2); -+ -+ /* -+ * Do not log dedup for tree item, it's not counted as -+ * scanned in this round. -+ */ -+ if (success2) -+ stable_tree_append(tree_rmap_item, snode, 0); -+ -+ /* -+ * The order of these two stable append is important: -+ * we are scanning rmap_item. -+ */ -+ if (success1) -+ stable_tree_append(rmap_item, snode, 1); -+ -+ /* -+ * The original kpage may be unlocked inside -+ * stable_tree_insert() already. This page -+ * should be unlocked before doing -+ * break_cow(). -+ */ -+ unlock_page(kpage); -+ -+ if (!success1) -+ break_cow(rmap_item); -+ -+ if (!success2) -+ break_cow(tree_rmap_item); -+ -+ } else if (err == MERGE_ERR_COLLI) { -+ BUG_ON(tree_rmap_item->tree_node->count > 1); -+ -+ rmap_item_hash_max(tree_rmap_item, -+ tree_rmap_item->tree_node->hash); -+ -+ hash_max = rmap_item_hash_max(rmap_item, hash); -+ cmp = hash_cmp(hash_max, tree_rmap_item->hash_max); -+ parent = &tree_rmap_item->node; -+ if (cmp < 0) -+ new = &parent->rb_left; -+ else if (cmp > 0) -+ new = &parent->rb_right; -+ else -+ goto put_up_out; -+ -+ rmap_item->tree_node = tree_rmap_item->tree_node; -+ rmap_item->address |= UNSTABLE_FLAG; -+ rmap_item->hash_round = uksm_hash_round; -+ rb_link_node(&rmap_item->node, parent, new); -+ rb_insert_color(&rmap_item->node, -+ &tree_rmap_item->tree_node->sub_root); -+ rmap_item->tree_node->count++; -+ } else { -+ /* -+ * either one of the page has changed or they collide -+ * at the max hash, we consider them as ill items. -+ */ -+ remove_rmap_item_from_tree(tree_rmap_item); -+ } -+put_up_out: -+ put_page(tree_rmap_item->page); -+ up_read(&tree_rmap_item->slot->vma->vm_mm->mmap_sem); -+ } -+} -+ -+ -+ -+ -+static inline unsigned long get_pool_index(struct vma_slot *slot, -+ unsigned long index) -+{ -+ unsigned long pool_index; -+ -+ pool_index = (sizeof(struct rmap_list_entry *) * index) >> PAGE_SHIFT; -+ if (pool_index >= slot->pool_size) -+ BUG(); -+ return pool_index; -+} -+ -+static inline unsigned long index_page_offset(unsigned long index) -+{ -+ return offset_in_page(sizeof(struct rmap_list_entry *) * index); -+} -+ -+static inline -+struct rmap_list_entry *get_rmap_list_entry(struct vma_slot *slot, -+ unsigned long index, int need_alloc) -+{ -+ unsigned long pool_index; -+ struct page *page; -+ void *addr; -+ -+ -+ pool_index = get_pool_index(slot, index); -+ if (!slot->rmap_list_pool[pool_index]) { -+ if (!need_alloc) -+ return NULL; -+ -+ page = alloc_page(GFP_KERNEL | __GFP_ZERO | __GFP_NOWARN); -+ if (!page) -+ return NULL; -+ -+ slot->rmap_list_pool[pool_index] = page; -+ } -+ -+ addr = kmap(slot->rmap_list_pool[pool_index]); -+ addr += index_page_offset(index); -+ -+ return addr; -+} -+ -+static inline void put_rmap_list_entry(struct vma_slot *slot, -+ unsigned long index) -+{ -+ unsigned long pool_index; -+ -+ pool_index = get_pool_index(slot, index); -+ BUG_ON(!slot->rmap_list_pool[pool_index]); -+ kunmap(slot->rmap_list_pool[pool_index]); -+} -+ -+static inline int entry_is_new(struct rmap_list_entry *entry) -+{ -+ return !entry->item; -+} -+ -+static inline unsigned long get_index_orig_addr(struct vma_slot *slot, -+ unsigned long index) -+{ -+ return slot->vma->vm_start + (index << PAGE_SHIFT); -+} -+ -+static inline unsigned long get_entry_address(struct rmap_list_entry *entry) -+{ -+ unsigned long addr; -+ -+ if (is_addr(entry->addr)) -+ addr = get_clean_addr(entry->addr); -+ else if (entry->item) -+ addr = get_rmap_addr(entry->item); -+ else -+ BUG(); -+ -+ return addr; -+} -+ -+static inline struct rmap_item *get_entry_item(struct rmap_list_entry *entry) -+{ -+ if (is_addr(entry->addr)) -+ return NULL; -+ -+ return entry->item; -+} -+ -+static inline void inc_rmap_list_pool_count(struct vma_slot *slot, -+ unsigned long index) -+{ -+ unsigned long pool_index; -+ -+ pool_index = get_pool_index(slot, index); -+ BUG_ON(!slot->rmap_list_pool[pool_index]); -+ slot->pool_counts[pool_index]++; -+} -+ -+static inline void dec_rmap_list_pool_count(struct vma_slot *slot, -+ unsigned long index) -+{ -+ unsigned long pool_index; -+ -+ pool_index = get_pool_index(slot, index); -+ BUG_ON(!slot->rmap_list_pool[pool_index]); -+ BUG_ON(!slot->pool_counts[pool_index]); -+ slot->pool_counts[pool_index]--; -+} -+ -+static inline int entry_has_rmap(struct rmap_list_entry *entry) -+{ -+ return !is_addr(entry->addr) && entry->item; -+} -+ -+static inline void swap_entries(struct rmap_list_entry *entry1, -+ unsigned long index1, -+ struct rmap_list_entry *entry2, -+ unsigned long index2) -+{ -+ struct rmap_list_entry tmp; -+ -+ /* swapping two new entries is meaningless */ -+ BUG_ON(entry_is_new(entry1) && entry_is_new(entry2)); -+ -+ tmp = *entry1; -+ *entry1 = *entry2; -+ *entry2 = tmp; -+ -+ if (entry_has_rmap(entry1)) -+ entry1->item->entry_index = index1; -+ -+ if (entry_has_rmap(entry2)) -+ entry2->item->entry_index = index2; -+ -+ if (entry_has_rmap(entry1) && !entry_has_rmap(entry2)) { -+ inc_rmap_list_pool_count(entry1->item->slot, index1); -+ dec_rmap_list_pool_count(entry1->item->slot, index2); -+ } else if (!entry_has_rmap(entry1) && entry_has_rmap(entry2)) { -+ inc_rmap_list_pool_count(entry2->item->slot, index2); -+ dec_rmap_list_pool_count(entry2->item->slot, index1); -+ } -+} -+ -+static inline void free_entry_item(struct rmap_list_entry *entry) -+{ -+ unsigned long index; -+ struct rmap_item *item; -+ -+ if (!is_addr(entry->addr)) { -+ BUG_ON(!entry->item); -+ item = entry->item; -+ entry->addr = get_rmap_addr(item); -+ set_is_addr(entry->addr); -+ index = item->entry_index; -+ remove_rmap_item_from_tree(item); -+ dec_rmap_list_pool_count(item->slot, index); -+ free_rmap_item(item); -+ } -+} -+ -+static inline int pool_entry_boundary(unsigned long index) -+{ -+ unsigned long linear_addr; -+ -+ linear_addr = sizeof(struct rmap_list_entry *) * index; -+ return index && !offset_in_page(linear_addr); -+} -+ -+static inline void try_free_last_pool(struct vma_slot *slot, -+ unsigned long index) -+{ -+ unsigned long pool_index; -+ -+ pool_index = get_pool_index(slot, index); -+ if (slot->rmap_list_pool[pool_index] && -+ !slot->pool_counts[pool_index]) { -+ __free_page(slot->rmap_list_pool[pool_index]); -+ slot->rmap_list_pool[pool_index] = NULL; -+ slot->flags |= UKSM_SLOT_NEED_SORT; -+ } -+ -+} -+ -+static inline unsigned long vma_item_index(struct vm_area_struct *vma, -+ struct rmap_item *item) -+{ -+ return (get_rmap_addr(item) - vma->vm_start) >> PAGE_SHIFT; -+} -+ -+static int within_same_pool(struct vma_slot *slot, -+ unsigned long i, unsigned long j) -+{ -+ unsigned long pool_i, pool_j; -+ -+ pool_i = get_pool_index(slot, i); -+ pool_j = get_pool_index(slot, j); -+ -+ return (pool_i == pool_j); -+} -+ -+static void sort_rmap_entry_list(struct vma_slot *slot) -+{ -+ unsigned long i, j; -+ struct rmap_list_entry *entry, *swap_entry; -+ -+ entry = get_rmap_list_entry(slot, 0, 0); -+ for (i = 0; i < slot->pages; ) { -+ -+ if (!entry) -+ goto skip_whole_pool; -+ -+ if (entry_is_new(entry)) -+ goto next_entry; -+ -+ if (is_addr(entry->addr)) { -+ entry->addr = 0; -+ goto next_entry; -+ } -+ -+ j = vma_item_index(slot->vma, entry->item); -+ if (j == i) -+ goto next_entry; -+ -+ if (within_same_pool(slot, i, j)) -+ swap_entry = entry + j - i; -+ else -+ swap_entry = get_rmap_list_entry(slot, j, 1); -+ -+ swap_entries(entry, i, swap_entry, j); -+ if (!within_same_pool(slot, i, j)) -+ put_rmap_list_entry(slot, j); -+ continue; -+ -+skip_whole_pool: -+ i += PAGE_SIZE / sizeof(*entry); -+ if (i < slot->pages) -+ entry = get_rmap_list_entry(slot, i, 0); -+ continue; -+ -+next_entry: -+ if (i >= slot->pages - 1 || -+ !within_same_pool(slot, i, i + 1)) { -+ put_rmap_list_entry(slot, i); -+ if (i + 1 < slot->pages) -+ entry = get_rmap_list_entry(slot, i + 1, 0); -+ } else -+ entry++; -+ i++; -+ continue; -+ } -+ -+ /* free empty pool entries which contain no rmap_item */ -+ /* CAN be simplied to based on only pool_counts when bug freed !!!!! */ -+ for (i = 0; i < slot->pool_size; i++) { -+ unsigned char has_rmap; -+ void *addr; -+ -+ if (!slot->rmap_list_pool[i]) -+ continue; -+ -+ has_rmap = 0; -+ addr = kmap(slot->rmap_list_pool[i]); -+ BUG_ON(!addr); -+ for (j = 0; j < PAGE_SIZE / sizeof(*entry); j++) { -+ entry = (struct rmap_list_entry *)addr + j; -+ if (is_addr(entry->addr)) -+ continue; -+ if (!entry->item) -+ continue; -+ has_rmap = 1; -+ } -+ kunmap(slot->rmap_list_pool[i]); -+ if (!has_rmap) { -+ BUG_ON(slot->pool_counts[i]); -+ __free_page(slot->rmap_list_pool[i]); -+ slot->rmap_list_pool[i] = NULL; -+ } -+ } -+ -+ slot->flags &= ~UKSM_SLOT_NEED_SORT; -+} -+ -+/* -+ * vma_fully_scanned() - if all the pages in this slot have been scanned. -+ */ -+static inline int vma_fully_scanned(struct vma_slot *slot) -+{ -+ return slot->pages_scanned == slot->pages; -+} -+ -+/** -+ * get_next_rmap_item() - Get the next rmap_item in a vma_slot according to -+ * its random permutation. This function is embedded with the random -+ * permutation index management code. -+ */ -+static struct rmap_item *get_next_rmap_item(struct vma_slot *slot, u32 *hash) -+{ -+ unsigned long rand_range, addr, swap_index, scan_index; -+ struct rmap_item *item = NULL; -+ struct rmap_list_entry *scan_entry, *swap_entry = NULL; -+ struct page *page; -+ -+ scan_index = swap_index = slot->pages_scanned % slot->pages; -+ -+ if (pool_entry_boundary(scan_index)) -+ try_free_last_pool(slot, scan_index - 1); -+ -+ if (vma_fully_scanned(slot)) { -+ if (slot->flags & UKSM_SLOT_NEED_SORT) -+ slot->flags |= UKSM_SLOT_NEED_RERAND; -+ else -+ slot->flags &= ~UKSM_SLOT_NEED_RERAND; -+ if (slot->flags & UKSM_SLOT_NEED_SORT) -+ sort_rmap_entry_list(slot); -+ } -+ -+ scan_entry = get_rmap_list_entry(slot, scan_index, 1); -+ if (!scan_entry) -+ return NULL; -+ -+ if (entry_is_new(scan_entry)) { -+ scan_entry->addr = get_index_orig_addr(slot, scan_index); -+ set_is_addr(scan_entry->addr); -+ } -+ -+ if (slot->flags & UKSM_SLOT_NEED_RERAND) { -+ rand_range = slot->pages - scan_index; -+ BUG_ON(!rand_range); -+ swap_index = scan_index + (prandom_u32() % rand_range); -+ } -+ -+ if (swap_index != scan_index) { -+ swap_entry = get_rmap_list_entry(slot, swap_index, 1); -+ if (entry_is_new(swap_entry)) { -+ swap_entry->addr = get_index_orig_addr(slot, -+ swap_index); -+ set_is_addr(swap_entry->addr); -+ } -+ swap_entries(scan_entry, scan_index, swap_entry, swap_index); -+ } -+ -+ addr = get_entry_address(scan_entry); -+ item = get_entry_item(scan_entry); -+ BUG_ON(addr > slot->vma->vm_end || addr < slot->vma->vm_start); -+ -+ page = follow_page(slot->vma, addr, FOLL_GET); -+ if (IS_ERR_OR_NULL(page)) -+ goto nopage; -+ -+ if (!PageAnon(page)) -+ goto putpage; -+ -+ /*check is zero_page pfn or uksm_zero_page*/ -+ if ((page_to_pfn(page) == zero_pfn) -+ || (page_to_pfn(page) == uksm_zero_pfn)) -+ goto putpage; -+ -+ flush_anon_page(slot->vma, page, addr); -+ flush_dcache_page(page); -+ -+ -+ *hash = page_hash(page, hash_strength, 1); -+ inc_uksm_pages_scanned(); -+ /*if the page content all zero, re-map to zero-page*/ -+ if (find_zero_page_hash(hash_strength, *hash)) { -+ if (!cmp_and_merge_zero_page(slot->vma, page)) { -+ slot->pages_merged++; -+ -+ /* For full-zero pages, no need to create rmap item */ -+ goto putpage; -+ } else { -+ inc_rshash_neg(memcmp_cost / 2); -+ } -+ } -+ -+ if (!item) { -+ item = alloc_rmap_item(); -+ if (item) { -+ /* It has already been zeroed */ -+ item->slot = slot; -+ item->address = addr; -+ item->entry_index = scan_index; -+ scan_entry->item = item; -+ inc_rmap_list_pool_count(slot, scan_index); -+ } else -+ goto putpage; -+ } -+ -+ BUG_ON(item->slot != slot); -+ /* the page may have changed */ -+ item->page = page; -+ put_rmap_list_entry(slot, scan_index); -+ if (swap_entry) -+ put_rmap_list_entry(slot, swap_index); -+ return item; -+ -+putpage: -+ put_page(page); -+ page = NULL; -+nopage: -+ /* no page, store addr back and free rmap_item if possible */ -+ free_entry_item(scan_entry); -+ put_rmap_list_entry(slot, scan_index); -+ if (swap_entry) -+ put_rmap_list_entry(slot, swap_index); -+ return NULL; -+} -+ -+static inline int in_stable_tree(struct rmap_item *rmap_item) -+{ -+ return rmap_item->address & STABLE_FLAG; -+} -+ -+/** -+ * scan_vma_one_page() - scan the next page in a vma_slot. Called with -+ * mmap_sem locked. -+ */ -+static noinline void scan_vma_one_page(struct vma_slot *slot) -+{ -+ u32 hash; -+ struct mm_struct *mm; -+ struct rmap_item *rmap_item = NULL; -+ struct vm_area_struct *vma = slot->vma; -+ -+ mm = vma->vm_mm; -+ BUG_ON(!mm); -+ BUG_ON(!slot); -+ -+ rmap_item = get_next_rmap_item(slot, &hash); -+ if (!rmap_item) -+ goto out1; -+ -+ if (PageKsm(rmap_item->page) && in_stable_tree(rmap_item)) -+ goto out2; -+ -+ cmp_and_merge_page(rmap_item, hash); -+out2: -+ put_page(rmap_item->page); -+out1: -+ slot->pages_scanned++; -+ slot->this_sampled++; -+ if (slot->fully_scanned_round != fully_scanned_round) -+ scanned_virtual_pages++; -+ -+ if (vma_fully_scanned(slot)) -+ slot->fully_scanned_round = fully_scanned_round; -+} -+ -+static inline unsigned long rung_get_pages(struct scan_rung *rung) -+{ -+ struct slot_tree_node *node; -+ -+ if (!rung->vma_root.rnode) -+ return 0; -+ -+ node = container_of(rung->vma_root.rnode, struct slot_tree_node, snode); -+ -+ return node->size; -+} -+ -+#define RUNG_SAMPLED_MIN 3 -+ -+static inline -+void uksm_calc_rung_step(struct scan_rung *rung, -+ unsigned long page_time, unsigned long ratio) -+{ -+ unsigned long sampled, pages; -+ -+ /* will be fully scanned ? */ -+ if (!rung->cover_msecs) { -+ rung->step = 1; -+ return; -+ } -+ -+ sampled = rung->cover_msecs * (NSEC_PER_MSEC / TIME_RATIO_SCALE) -+ * ratio / page_time; -+ -+ /* -+ * Before we finsish a scan round and expensive per-round jobs, -+ * we need to have a chance to estimate the per page time. So -+ * the sampled number can not be too small. -+ */ -+ if (sampled < RUNG_SAMPLED_MIN) -+ sampled = RUNG_SAMPLED_MIN; -+ -+ pages = rung_get_pages(rung); -+ if (likely(pages > sampled)) -+ rung->step = pages / sampled; -+ else -+ rung->step = 1; -+} -+ -+static inline int step_need_recalc(struct scan_rung *rung) -+{ -+ unsigned long pages, stepmax; -+ -+ pages = rung_get_pages(rung); -+ stepmax = pages / RUNG_SAMPLED_MIN; -+ -+ return pages && (rung->step > pages || -+ (stepmax && rung->step > stepmax)); -+} -+ -+static inline -+void reset_current_scan(struct scan_rung *rung, int finished, int step_recalc) -+{ -+ struct vma_slot *slot; -+ -+ if (finished) -+ rung->flags |= UKSM_RUNG_ROUND_FINISHED; -+ -+ if (step_recalc || step_need_recalc(rung)) { -+ uksm_calc_rung_step(rung, uksm_ema_page_time, rung->cpu_ratio); -+ BUG_ON(step_need_recalc(rung)); -+ } -+ -+ slot_iter_index = prandom_u32() % rung->step; -+ BUG_ON(!rung->vma_root.rnode); -+ slot = sradix_tree_next(&rung->vma_root, NULL, 0, slot_iter); -+ BUG_ON(!slot); -+ -+ rung->current_scan = slot; -+ rung->current_offset = slot_iter_index; -+} -+ -+static inline struct sradix_tree_root *slot_get_root(struct vma_slot *slot) -+{ -+ return &slot->rung->vma_root; -+} -+ -+/* -+ * return if resetted. -+ */ -+static int advance_current_scan(struct scan_rung *rung) -+{ -+ unsigned short n; -+ struct vma_slot *slot, *next = NULL; -+ -+ BUG_ON(!rung->vma_root.num); -+ -+ slot = rung->current_scan; -+ n = (slot->pages - rung->current_offset) % rung->step; -+ slot_iter_index = rung->step - n; -+ next = sradix_tree_next(&rung->vma_root, slot->snode, -+ slot->sindex, slot_iter); -+ -+ if (next) { -+ rung->current_offset = slot_iter_index; -+ rung->current_scan = next; -+ return 0; -+ } else { -+ reset_current_scan(rung, 1, 0); -+ return 1; -+ } -+} -+ -+static inline void rung_rm_slot(struct vma_slot *slot) -+{ -+ struct scan_rung *rung = slot->rung; -+ struct sradix_tree_root *root; -+ -+ if (rung->current_scan == slot) -+ advance_current_scan(rung); -+ -+ root = slot_get_root(slot); -+ sradix_tree_delete_from_leaf(root, slot->snode, slot->sindex); -+ slot->snode = NULL; -+ if (step_need_recalc(rung)) { -+ uksm_calc_rung_step(rung, uksm_ema_page_time, rung->cpu_ratio); -+ BUG_ON(step_need_recalc(rung)); -+ } -+ -+ /* In case advance_current_scan loop back to this slot again */ -+ if (rung->vma_root.num && rung->current_scan == slot) -+ reset_current_scan(slot->rung, 1, 0); -+} -+ -+static inline void rung_add_new_slots(struct scan_rung *rung, -+ struct vma_slot **slots, unsigned long num) -+{ -+ int err; -+ struct vma_slot *slot; -+ unsigned long i; -+ struct sradix_tree_root *root = &rung->vma_root; -+ -+ err = sradix_tree_enter(root, (void **)slots, num); -+ BUG_ON(err); -+ -+ for (i = 0; i < num; i++) { -+ slot = slots[i]; -+ slot->rung = rung; -+ BUG_ON(vma_fully_scanned(slot)); -+ } -+ -+ if (rung->vma_root.num == num) -+ reset_current_scan(rung, 0, 1); -+} -+ -+static inline int rung_add_one_slot(struct scan_rung *rung, -+ struct vma_slot *slot) -+{ -+ int err; -+ -+ err = sradix_tree_enter(&rung->vma_root, (void **)&slot, 1); -+ if (err) -+ return err; -+ -+ slot->rung = rung; -+ if (rung->vma_root.num == 1) -+ reset_current_scan(rung, 0, 1); -+ -+ return 0; -+} -+ -+/* -+ * Return true if the slot is deleted from its rung. -+ */ -+static inline int vma_rung_enter(struct vma_slot *slot, struct scan_rung *rung) -+{ -+ struct scan_rung *old_rung = slot->rung; -+ int err; -+ -+ if (old_rung == rung) -+ return 0; -+ -+ rung_rm_slot(slot); -+ err = rung_add_one_slot(rung, slot); -+ if (err) { -+ err = rung_add_one_slot(old_rung, slot); -+ WARN_ON(err); /* OOPS, badly OOM, we lost this slot */ -+ } -+ -+ return 1; -+} -+ -+static inline int vma_rung_up(struct vma_slot *slot) -+{ -+ struct scan_rung *rung; -+ -+ rung = slot->rung; -+ if (slot->rung != &uksm_scan_ladder[SCAN_LADDER_SIZE-1]) -+ rung++; -+ -+ return vma_rung_enter(slot, rung); -+} -+ -+static inline int vma_rung_down(struct vma_slot *slot) -+{ -+ struct scan_rung *rung; -+ -+ rung = slot->rung; -+ if (slot->rung != &uksm_scan_ladder[0]) -+ rung--; -+ -+ return vma_rung_enter(slot, rung); -+} -+ -+/** -+ * cal_dedup_ratio() - Calculate the deduplication ratio for this slot. -+ */ -+static unsigned long cal_dedup_ratio(struct vma_slot *slot) -+{ -+ unsigned long ret; -+ unsigned long pages; -+ -+ pages = slot->this_sampled; -+ if (!pages) -+ return 0; -+ -+ BUG_ON(slot->pages_scanned == slot->last_scanned); -+ -+ ret = slot->pages_merged; -+ -+ /* Thrashing area filtering */ -+ if (ret && uksm_thrash_threshold) { -+ if (slot->pages_cowed * 100 / slot->pages_merged -+ > uksm_thrash_threshold) { -+ ret = 0; -+ } else { -+ ret = slot->pages_merged - slot->pages_cowed; -+ } -+ } -+ -+ return ret * 100 / pages; -+} -+ -+/** -+ * cal_dedup_ratio() - Calculate the deduplication ratio for this slot. -+ */ -+static unsigned long cal_dedup_ratio_old(struct vma_slot *slot) -+{ -+ unsigned long ret; -+ unsigned long pages; -+ -+ pages = slot->pages; -+ if (!pages) -+ return 0; -+ -+ ret = slot->pages_bemerged; -+ -+ /* Thrashing area filtering */ -+ if (ret && uksm_thrash_threshold) { -+ if (slot->pages_cowed * 100 / slot->pages_bemerged -+ > uksm_thrash_threshold) { -+ ret = 0; -+ } else { -+ ret = slot->pages_bemerged - slot->pages_cowed; -+ } -+ } -+ -+ return ret * 100 / pages; -+} -+ -+/** -+ * stable_node_reinsert() - When the hash_strength has been adjusted, the -+ * stable tree need to be restructured, this is the function re-inserting the -+ * stable node. -+ */ -+static inline void stable_node_reinsert(struct stable_node *new_node, -+ struct page *page, -+ struct rb_root *root_treep, -+ struct list_head *tree_node_listp, -+ u32 hash) -+{ -+ struct rb_node **new = &root_treep->rb_node; -+ struct rb_node *parent = NULL; -+ struct stable_node *stable_node; -+ struct tree_node *tree_node; -+ struct page *tree_page; -+ int cmp; -+ -+ while (*new) { -+ int cmp; -+ -+ tree_node = rb_entry(*new, struct tree_node, node); -+ -+ cmp = hash_cmp(hash, tree_node->hash); -+ -+ if (cmp < 0) { -+ parent = *new; -+ new = &parent->rb_left; -+ } else if (cmp > 0) { -+ parent = *new; -+ new = &parent->rb_right; -+ } else -+ break; -+ } -+ -+ if (*new) { -+ /* find a stable tree node with same first level hash value */ -+ stable_node_hash_max(new_node, page, hash); -+ if (tree_node->count == 1) { -+ stable_node = rb_entry(tree_node->sub_root.rb_node, -+ struct stable_node, node); -+ tree_page = get_uksm_page(stable_node, 1, 0); -+ if (tree_page) { -+ stable_node_hash_max(stable_node, -+ tree_page, hash); -+ put_page(tree_page); -+ -+ /* prepare for stable node insertion */ -+ -+ cmp = hash_cmp(new_node->hash_max, -+ stable_node->hash_max); -+ parent = &stable_node->node; -+ if (cmp < 0) -+ new = &parent->rb_left; -+ else if (cmp > 0) -+ new = &parent->rb_right; -+ else -+ goto failed; -+ -+ goto add_node; -+ } else { -+ /* the only stable_node deleted, the tree node -+ * was not deleted. -+ */ -+ goto tree_node_reuse; -+ } -+ } -+ -+ /* well, search the collision subtree */ -+ new = &tree_node->sub_root.rb_node; -+ parent = NULL; -+ BUG_ON(!*new); -+ while (*new) { -+ int cmp; -+ -+ stable_node = rb_entry(*new, struct stable_node, node); -+ -+ cmp = hash_cmp(new_node->hash_max, -+ stable_node->hash_max); -+ -+ if (cmp < 0) { -+ parent = *new; -+ new = &parent->rb_left; -+ } else if (cmp > 0) { -+ parent = *new; -+ new = &parent->rb_right; -+ } else { -+ /* oh, no, still a collision */ -+ goto failed; -+ } -+ } -+ -+ goto add_node; -+ } -+ -+ /* no tree node found */ -+ tree_node = alloc_tree_node(tree_node_listp); -+ if (!tree_node) { -+ pr_err("UKSM: memory allocation error!\n"); -+ goto failed; -+ } else { -+ tree_node->hash = hash; -+ rb_link_node(&tree_node->node, parent, new); -+ rb_insert_color(&tree_node->node, root_treep); -+ -+tree_node_reuse: -+ /* prepare for stable node insertion */ -+ parent = NULL; -+ new = &tree_node->sub_root.rb_node; -+ } -+ -+add_node: -+ rb_link_node(&new_node->node, parent, new); -+ rb_insert_color(&new_node->node, &tree_node->sub_root); -+ new_node->tree_node = tree_node; -+ tree_node->count++; -+ return; -+ -+failed: -+ /* This can only happen when two nodes have collided -+ * in two levels. -+ */ -+ new_node->tree_node = NULL; -+ return; -+} -+ -+static inline void free_all_tree_nodes(struct list_head *list) -+{ -+ struct tree_node *node, *tmp; -+ -+ list_for_each_entry_safe(node, tmp, list, all_list) { -+ free_tree_node(node); -+ } -+} -+ -+/** -+ * stable_tree_delta_hash() - Delta hash the stable tree from previous hash -+ * strength to the current hash_strength. It re-structures the hole tree. -+ */ -+static inline void stable_tree_delta_hash(u32 prev_hash_strength) -+{ -+ struct stable_node *node, *tmp; -+ struct rb_root *root_new_treep; -+ struct list_head *new_tree_node_listp; -+ -+ stable_tree_index = (stable_tree_index + 1) % 2; -+ root_new_treep = &root_stable_tree[stable_tree_index]; -+ new_tree_node_listp = &stable_tree_node_list[stable_tree_index]; -+ *root_new_treep = RB_ROOT; -+ BUG_ON(!list_empty(new_tree_node_listp)); -+ -+ /* -+ * we need to be safe, the node could be removed by get_uksm_page() -+ */ -+ list_for_each_entry_safe(node, tmp, &stable_node_list, all_list) { -+ void *addr; -+ struct page *node_page; -+ u32 hash; -+ -+ /* -+ * We are completely re-structuring the stable nodes to a new -+ * stable tree. We don't want to touch the old tree unlinks and -+ * old tree_nodes. The old tree_nodes will be freed at once. -+ */ -+ node_page = get_uksm_page(node, 0, 0); -+ if (!node_page) -+ continue; -+ -+ if (node->tree_node) { -+ hash = node->tree_node->hash; -+ -+ addr = kmap_atomic(node_page); -+ -+ hash = delta_hash(addr, prev_hash_strength, -+ hash_strength, hash); -+ kunmap_atomic(addr); -+ } else { -+ /* -+ *it was not inserted to rbtree due to collision in last -+ *round scan. -+ */ -+ hash = page_hash(node_page, hash_strength, 0); -+ } -+ -+ stable_node_reinsert(node, node_page, root_new_treep, -+ new_tree_node_listp, hash); -+ put_page(node_page); -+ } -+ -+ root_stable_treep = root_new_treep; -+ free_all_tree_nodes(stable_tree_node_listp); -+ BUG_ON(!list_empty(stable_tree_node_listp)); -+ stable_tree_node_listp = new_tree_node_listp; -+} -+ -+static inline void inc_hash_strength(unsigned long delta) -+{ -+ hash_strength += 1 << delta; -+ if (hash_strength > HASH_STRENGTH_MAX) -+ hash_strength = HASH_STRENGTH_MAX; -+} -+ -+static inline void dec_hash_strength(unsigned long delta) -+{ -+ unsigned long change = 1 << delta; -+ -+ if (hash_strength <= change + 1) -+ hash_strength = 1; -+ else -+ hash_strength -= change; -+} -+ -+static inline void inc_hash_strength_delta(void) -+{ -+ hash_strength_delta++; -+ if (hash_strength_delta > HASH_STRENGTH_DELTA_MAX) -+ hash_strength_delta = HASH_STRENGTH_DELTA_MAX; -+} -+ -+static inline unsigned long get_current_neg_ratio(void) -+{ -+ u64 pos = benefit.pos; -+ u64 neg = benefit.neg; -+ -+ if (!neg) -+ return 0; -+ -+ if (!pos || neg > pos) -+ return 100; -+ -+ if (neg > div64_u64(U64_MAX, 100)) -+ pos = div64_u64(pos, 100); -+ else -+ neg *= 100; -+ -+ return div64_u64(neg, pos); -+} -+ -+static inline unsigned long get_current_benefit(void) -+{ -+ u64 pos = benefit.pos; -+ u64 neg = benefit.neg; -+ u64 scanned = benefit.scanned; -+ -+ if (neg > pos) -+ return 0; -+ -+ return div64_u64((pos - neg), scanned); -+} -+ -+static inline int judge_rshash_direction(void) -+{ -+ u64 current_neg_ratio, stable_benefit; -+ u64 current_benefit, delta = 0; -+ int ret = STILL; -+ -+ /* -+ * Try to probe a value after the boot, and in case the system -+ * are still for a long time. -+ */ -+ if ((fully_scanned_round & 0xFFULL) == 10) { -+ ret = OBSCURE; -+ goto out; -+ } -+ -+ current_neg_ratio = get_current_neg_ratio(); -+ -+ if (current_neg_ratio == 0) { -+ rshash_neg_cont_zero++; -+ if (rshash_neg_cont_zero > 2) -+ return GO_DOWN; -+ else -+ return STILL; -+ } -+ rshash_neg_cont_zero = 0; -+ -+ if (current_neg_ratio > 90) { -+ ret = GO_UP; -+ goto out; -+ } -+ -+ current_benefit = get_current_benefit(); -+ stable_benefit = rshash_state.stable_benefit; -+ -+ if (!stable_benefit) { -+ ret = OBSCURE; -+ goto out; -+ } -+ -+ if (current_benefit > stable_benefit) -+ delta = current_benefit - stable_benefit; -+ else if (current_benefit < stable_benefit) -+ delta = stable_benefit - current_benefit; -+ -+ delta = div64_u64(100 * delta, stable_benefit); -+ -+ if (delta > 50) { -+ rshash_cont_obscure++; -+ if (rshash_cont_obscure > 2) -+ return OBSCURE; -+ else -+ return STILL; -+ } -+ -+out: -+ rshash_cont_obscure = 0; -+ return ret; -+} -+ -+/** -+ * rshash_adjust() - The main function to control the random sampling state -+ * machine for hash strength adapting. -+ * -+ * return true if hash_strength has changed. -+ */ -+static inline int rshash_adjust(void) -+{ -+ unsigned long prev_hash_strength = hash_strength; -+ -+ if (!encode_benefit()) -+ return 0; -+ -+ switch (rshash_state.state) { -+ case RSHASH_STILL: -+ switch (judge_rshash_direction()) { -+ case GO_UP: -+ if (rshash_state.pre_direct == GO_DOWN) -+ hash_strength_delta = 0; -+ -+ inc_hash_strength(hash_strength_delta); -+ inc_hash_strength_delta(); -+ rshash_state.stable_benefit = get_current_benefit(); -+ rshash_state.pre_direct = GO_UP; -+ break; -+ -+ case GO_DOWN: -+ if (rshash_state.pre_direct == GO_UP) -+ hash_strength_delta = 0; -+ -+ dec_hash_strength(hash_strength_delta); -+ inc_hash_strength_delta(); -+ rshash_state.stable_benefit = get_current_benefit(); -+ rshash_state.pre_direct = GO_DOWN; -+ break; -+ -+ case OBSCURE: -+ rshash_state.stable_point = hash_strength; -+ rshash_state.turn_point_down = hash_strength; -+ rshash_state.turn_point_up = hash_strength; -+ rshash_state.turn_benefit_down = get_current_benefit(); -+ rshash_state.turn_benefit_up = get_current_benefit(); -+ rshash_state.lookup_window_index = 0; -+ rshash_state.state = RSHASH_TRYDOWN; -+ dec_hash_strength(hash_strength_delta); -+ inc_hash_strength_delta(); -+ break; -+ -+ case STILL: -+ break; -+ default: -+ BUG(); -+ } -+ break; -+ -+ case RSHASH_TRYDOWN: -+ if (rshash_state.lookup_window_index++ % 5 == 0) -+ rshash_state.below_count = 0; -+ -+ if (get_current_benefit() < rshash_state.stable_benefit) -+ rshash_state.below_count++; -+ else if (get_current_benefit() > -+ rshash_state.turn_benefit_down) { -+ rshash_state.turn_point_down = hash_strength; -+ rshash_state.turn_benefit_down = get_current_benefit(); -+ } -+ -+ if (rshash_state.below_count >= 3 || -+ judge_rshash_direction() == GO_UP || -+ hash_strength == 1) { -+ hash_strength = rshash_state.stable_point; -+ hash_strength_delta = 0; -+ inc_hash_strength(hash_strength_delta); -+ inc_hash_strength_delta(); -+ rshash_state.lookup_window_index = 0; -+ rshash_state.state = RSHASH_TRYUP; -+ hash_strength_delta = 0; -+ } else { -+ dec_hash_strength(hash_strength_delta); -+ inc_hash_strength_delta(); -+ } -+ break; -+ -+ case RSHASH_TRYUP: -+ if (rshash_state.lookup_window_index++ % 5 == 0) -+ rshash_state.below_count = 0; -+ -+ if (get_current_benefit() < rshash_state.turn_benefit_down) -+ rshash_state.below_count++; -+ else if (get_current_benefit() > rshash_state.turn_benefit_up) { -+ rshash_state.turn_point_up = hash_strength; -+ rshash_state.turn_benefit_up = get_current_benefit(); -+ } -+ -+ if (rshash_state.below_count >= 3 || -+ judge_rshash_direction() == GO_DOWN || -+ hash_strength == HASH_STRENGTH_MAX) { -+ hash_strength = rshash_state.turn_benefit_up > -+ rshash_state.turn_benefit_down ? -+ rshash_state.turn_point_up : -+ rshash_state.turn_point_down; -+ -+ rshash_state.state = RSHASH_PRE_STILL; -+ } else { -+ inc_hash_strength(hash_strength_delta); -+ inc_hash_strength_delta(); -+ } -+ -+ break; -+ -+ case RSHASH_NEW: -+ case RSHASH_PRE_STILL: -+ rshash_state.stable_benefit = get_current_benefit(); -+ rshash_state.state = RSHASH_STILL; -+ hash_strength_delta = 0; -+ break; -+ default: -+ BUG(); -+ } -+ -+ /* rshash_neg = rshash_pos = 0; */ -+ reset_benefit(); -+ -+ if (prev_hash_strength != hash_strength) -+ stable_tree_delta_hash(prev_hash_strength); -+ -+ return prev_hash_strength != hash_strength; -+} -+ -+/** -+ * round_update_ladder() - The main function to do update of all the -+ * adjustments whenever a scan round is finished. -+ */ -+static noinline void round_update_ladder(void) -+{ -+ int i; -+ unsigned long dedup; -+ struct vma_slot *slot, *tmp_slot; -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) -+ uksm_scan_ladder[i].flags &= ~UKSM_RUNG_ROUND_FINISHED; -+ -+ list_for_each_entry_safe(slot, tmp_slot, &vma_slot_dedup, dedup_list) { -+ -+ /* slot may be rung_rm_slot() when mm exits */ -+ if (slot->snode) { -+ dedup = cal_dedup_ratio_old(slot); -+ if (dedup && dedup >= uksm_abundant_threshold) -+ vma_rung_up(slot); -+ } -+ -+ slot->pages_bemerged = 0; -+ slot->pages_cowed = 0; -+ -+ list_del_init(&slot->dedup_list); -+ } -+} -+ -+static void uksm_del_vma_slot(struct vma_slot *slot) -+{ -+ int i, j; -+ struct rmap_list_entry *entry; -+ -+ if (slot->snode) { -+ /* -+ * In case it just failed when entering the rung, it's not -+ * necessary. -+ */ -+ rung_rm_slot(slot); -+ } -+ -+ if (!list_empty(&slot->dedup_list)) -+ list_del(&slot->dedup_list); -+ -+ if (!slot->rmap_list_pool || !slot->pool_counts) { -+ /* In case it OOMed in uksm_vma_enter() */ -+ goto out; -+ } -+ -+ for (i = 0; i < slot->pool_size; i++) { -+ void *addr; -+ -+ if (!slot->rmap_list_pool[i]) -+ continue; -+ -+ addr = kmap(slot->rmap_list_pool[i]); -+ for (j = 0; j < PAGE_SIZE / sizeof(*entry); j++) { -+ entry = (struct rmap_list_entry *)addr + j; -+ if (is_addr(entry->addr)) -+ continue; -+ if (!entry->item) -+ continue; -+ -+ remove_rmap_item_from_tree(entry->item); -+ free_rmap_item(entry->item); -+ slot->pool_counts[i]--; -+ } -+ BUG_ON(slot->pool_counts[i]); -+ kunmap(slot->rmap_list_pool[i]); -+ __free_page(slot->rmap_list_pool[i]); -+ } -+ kfree(slot->rmap_list_pool); -+ kfree(slot->pool_counts); -+ -+out: -+ slot->rung = NULL; -+ if (slot->flags & UKSM_SLOT_IN_UKSM) { -+ BUG_ON(uksm_pages_total < slot->pages); -+ uksm_pages_total -= slot->pages; -+ } -+ -+ if (slot->fully_scanned_round == fully_scanned_round) -+ scanned_virtual_pages -= slot->pages; -+ else -+ scanned_virtual_pages -= slot->pages_scanned; -+ free_vma_slot(slot); -+} -+ -+ -+#define SPIN_LOCK_PERIOD 32 -+static struct vma_slot *cleanup_slots[SPIN_LOCK_PERIOD]; -+static inline void cleanup_vma_slots(void) -+{ -+ struct vma_slot *slot; -+ int i; -+ -+ i = 0; -+ spin_lock(&vma_slot_list_lock); -+ while (!list_empty(&vma_slot_del)) { -+ slot = list_entry(vma_slot_del.next, -+ struct vma_slot, slot_list); -+ list_del(&slot->slot_list); -+ cleanup_slots[i++] = slot; -+ if (i == SPIN_LOCK_PERIOD) { -+ spin_unlock(&vma_slot_list_lock); -+ while (--i >= 0) -+ uksm_del_vma_slot(cleanup_slots[i]); -+ i = 0; -+ spin_lock(&vma_slot_list_lock); -+ } -+ } -+ spin_unlock(&vma_slot_list_lock); -+ -+ while (--i >= 0) -+ uksm_del_vma_slot(cleanup_slots[i]); -+} -+ -+/* -+ * Expotional moving average formula -+ */ -+static inline unsigned long ema(unsigned long curr, unsigned long last_ema) -+{ -+ /* -+ * For a very high burst, even the ema cannot work well, a false very -+ * high per-page time estimation can result in feedback in very high -+ * overhead of context switch and rung update -- this will then lead -+ * to higher per-paper time, this may not converge. -+ * -+ * Instead, we try to approach this value in a binary manner. -+ */ -+ if (curr > last_ema * 10) -+ return last_ema * 2; -+ -+ return (EMA_ALPHA * curr + (100 - EMA_ALPHA) * last_ema) / 100; -+} -+ -+/* -+ * convert cpu ratio in 1/TIME_RATIO_SCALE configured by user to -+ * nanoseconds based on current uksm_sleep_jiffies. -+ */ -+static inline unsigned long cpu_ratio_to_nsec(unsigned int ratio) -+{ -+ return NSEC_PER_USEC * jiffies_to_usecs(uksm_sleep_jiffies) / -+ (TIME_RATIO_SCALE - ratio) * ratio; -+} -+ -+ -+static inline unsigned long rung_real_ratio(int cpu_time_ratio) -+{ -+ unsigned long ret; -+ -+ BUG_ON(!cpu_time_ratio); -+ -+ if (cpu_time_ratio > 0) -+ ret = cpu_time_ratio; -+ else -+ ret = (unsigned long)(-cpu_time_ratio) * -+ uksm_max_cpu_percentage / 100UL; -+ -+ return ret ? ret : 1; -+} -+ -+static noinline void uksm_calc_scan_pages(void) -+{ -+ struct scan_rung *ladder = uksm_scan_ladder; -+ unsigned long sleep_usecs, nsecs; -+ unsigned long ratio; -+ int i; -+ unsigned long per_page; -+ -+ if (uksm_ema_page_time > 100000 || -+ (((unsigned long) uksm_eval_round & (256UL - 1)) == 0UL)) -+ uksm_ema_page_time = UKSM_PAGE_TIME_DEFAULT; -+ -+ per_page = uksm_ema_page_time; -+ BUG_ON(!per_page); -+ -+ /* -+ * For every 8 eval round, we try to probe a uksm_sleep_jiffies value -+ * based on saved user input. -+ */ -+ if (((unsigned long) uksm_eval_round & (8UL - 1)) == 0UL) -+ uksm_sleep_jiffies = uksm_sleep_saved; -+ -+ /* We require a rung scan at least 1 page in a period. */ -+ nsecs = per_page; -+ ratio = rung_real_ratio(ladder[0].cpu_ratio); -+ if (cpu_ratio_to_nsec(ratio) < nsecs) { -+ sleep_usecs = nsecs * (TIME_RATIO_SCALE - ratio) / ratio -+ / NSEC_PER_USEC; -+ uksm_sleep_jiffies = usecs_to_jiffies(sleep_usecs) + 1; -+ } -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ ratio = rung_real_ratio(ladder[i].cpu_ratio); -+ ladder[i].pages_to_scan = cpu_ratio_to_nsec(ratio) / -+ per_page; -+ BUG_ON(!ladder[i].pages_to_scan); -+ uksm_calc_rung_step(&ladder[i], per_page, ratio); -+ } -+} -+ -+/* -+ * From the scan time of this round (ns) to next expected min sleep time -+ * (ms), be careful of the possible overflows. ratio is taken from -+ * rung_real_ratio() -+ */ -+static inline -+unsigned int scan_time_to_sleep(unsigned long long scan_time, unsigned long ratio) -+{ -+ scan_time >>= 20; /* to msec level now */ -+ BUG_ON(scan_time > (ULONG_MAX / TIME_RATIO_SCALE)); -+ -+ return (unsigned int) ((unsigned long) scan_time * -+ (TIME_RATIO_SCALE - ratio) / ratio); -+} -+ -+#define __round_mask(x, y) ((__typeof__(x))((y)-1)) -+#define round_up(x, y) ((((x)-1) | __round_mask(x, y))+1) -+ -+static void uksm_vma_enter(struct vma_slot **slots, unsigned long num) -+{ -+ struct scan_rung *rung; -+ -+ rung = &uksm_scan_ladder[0]; -+ rung_add_new_slots(rung, slots, num); -+} -+ -+static struct vma_slot *batch_slots[SLOT_TREE_NODE_STORE_SIZE]; -+ -+static void uksm_enter_all_slots(void) -+{ -+ struct vma_slot *slot; -+ unsigned long index; -+ struct list_head empty_vma_list; -+ int i; -+ -+ i = 0; -+ index = 0; -+ INIT_LIST_HEAD(&empty_vma_list); -+ -+ spin_lock(&vma_slot_list_lock); -+ while (!list_empty(&vma_slot_new)) { -+ slot = list_entry(vma_slot_new.next, -+ struct vma_slot, slot_list); -+ -+ if (!slot->vma->anon_vma) { -+ list_move(&slot->slot_list, &empty_vma_list); -+ } else if (vma_can_enter(slot->vma)) { -+ batch_slots[index++] = slot; -+ list_del_init(&slot->slot_list); -+ } else { -+ list_move(&slot->slot_list, &vma_slot_noadd); -+ } -+ -+ if (++i == SPIN_LOCK_PERIOD || -+ (index && !(index % SLOT_TREE_NODE_STORE_SIZE))) { -+ spin_unlock(&vma_slot_list_lock); -+ -+ if (index && !(index % SLOT_TREE_NODE_STORE_SIZE)) { -+ uksm_vma_enter(batch_slots, index); -+ index = 0; -+ } -+ i = 0; -+ cond_resched(); -+ spin_lock(&vma_slot_list_lock); -+ } -+ } -+ -+ list_splice(&empty_vma_list, &vma_slot_new); -+ -+ spin_unlock(&vma_slot_list_lock); -+ -+ if (index) -+ uksm_vma_enter(batch_slots, index); -+ -+} -+ -+static inline int rung_round_finished(struct scan_rung *rung) -+{ -+ return rung->flags & UKSM_RUNG_ROUND_FINISHED; -+} -+ -+static inline void judge_slot(struct vma_slot *slot) -+{ -+ struct scan_rung *rung = slot->rung; -+ unsigned long dedup; -+ int deleted; -+ -+ dedup = cal_dedup_ratio(slot); -+ if (vma_fully_scanned(slot) && uksm_thrash_threshold) -+ deleted = vma_rung_enter(slot, &uksm_scan_ladder[0]); -+ else if (dedup && dedup >= uksm_abundant_threshold) -+ deleted = vma_rung_up(slot); -+ else -+ deleted = vma_rung_down(slot); -+ -+ slot->pages_merged = 0; -+ slot->pages_cowed = 0; -+ slot->this_sampled = 0; -+ -+ if (vma_fully_scanned(slot)) -+ slot->pages_scanned = 0; -+ -+ slot->last_scanned = slot->pages_scanned; -+ -+ /* If its deleted in above, then rung was already advanced. */ -+ if (!deleted) -+ advance_current_scan(rung); -+} -+ -+ -+static inline int hash_round_finished(void) -+{ -+ if (scanned_virtual_pages > (uksm_pages_total >> 2)) { -+ scanned_virtual_pages = 0; -+ if (uksm_pages_scanned) -+ fully_scanned_round++; -+ -+ return 1; -+ } else { -+ return 0; -+ } -+} -+ -+#define UKSM_MMSEM_BATCH 5 -+#define BUSY_RETRY 100 -+ -+/** -+ * uksm_do_scan() - the main worker function. -+ */ -+static noinline void uksm_do_scan(void) -+{ -+ struct vma_slot *slot, *iter; -+ struct mm_struct *busy_mm; -+ unsigned char round_finished, all_rungs_emtpy; -+ int i, err, mmsem_batch; -+ unsigned long pcost; -+ long long delta_exec; -+ unsigned long vpages, max_cpu_ratio; -+ unsigned long long start_time, end_time, scan_time; -+ unsigned int expected_jiffies; -+ -+ might_sleep(); -+ -+ vpages = 0; -+ -+ start_time = task_sched_runtime(current); -+ max_cpu_ratio = 0; -+ mmsem_batch = 0; -+ -+ for (i = 0; i < SCAN_LADDER_SIZE;) { -+ struct scan_rung *rung = &uksm_scan_ladder[i]; -+ unsigned long ratio; -+ int busy_retry; -+ -+ if (!rung->pages_to_scan) { -+ i++; -+ continue; -+ } -+ -+ if (!rung->vma_root.num) { -+ rung->pages_to_scan = 0; -+ i++; -+ continue; -+ } -+ -+ ratio = rung_real_ratio(rung->cpu_ratio); -+ if (ratio > max_cpu_ratio) -+ max_cpu_ratio = ratio; -+ -+ busy_retry = BUSY_RETRY; -+ /* -+ * Do not consider rung_round_finished() here, just used up the -+ * rung->pages_to_scan quota. -+ */ -+ while (rung->pages_to_scan && rung->vma_root.num && -+ likely(!freezing(current))) { -+ int reset = 0; -+ -+ slot = rung->current_scan; -+ -+ BUG_ON(vma_fully_scanned(slot)); -+ -+ if (mmsem_batch) -+ err = 0; -+ else -+ err = try_down_read_slot_mmap_sem(slot); -+ -+ if (err == -ENOENT) { -+rm_slot: -+ rung_rm_slot(slot); -+ continue; -+ } -+ -+ busy_mm = slot->mm; -+ -+ if (err == -EBUSY) { -+ /* skip other vmas on the same mm */ -+ do { -+ reset = advance_current_scan(rung); -+ iter = rung->current_scan; -+ busy_retry--; -+ if (iter->vma->vm_mm != busy_mm || -+ !busy_retry || reset) -+ break; -+ } while (1); -+ -+ if (iter->vma->vm_mm != busy_mm) { -+ continue; -+ } else { -+ /* scan round finsished */ -+ break; -+ } -+ } -+ -+ BUG_ON(!vma_can_enter(slot->vma)); -+ if (uksm_test_exit(slot->vma->vm_mm)) { -+ mmsem_batch = 0; -+ up_read(&slot->vma->vm_mm->mmap_sem); -+ goto rm_slot; -+ } -+ -+ if (mmsem_batch) -+ mmsem_batch--; -+ else -+ mmsem_batch = UKSM_MMSEM_BATCH; -+ -+ /* Ok, we have take the mmap_sem, ready to scan */ -+ scan_vma_one_page(slot); -+ rung->pages_to_scan--; -+ vpages++; -+ -+ if (rung->current_offset + rung->step > slot->pages - 1 -+ || vma_fully_scanned(slot)) { -+ up_read(&slot->vma->vm_mm->mmap_sem); -+ judge_slot(slot); -+ mmsem_batch = 0; -+ } else { -+ rung->current_offset += rung->step; -+ if (!mmsem_batch) -+ up_read(&slot->vma->vm_mm->mmap_sem); -+ } -+ -+ busy_retry = BUSY_RETRY; -+ cond_resched(); -+ } -+ -+ if (mmsem_batch) { -+ up_read(&slot->vma->vm_mm->mmap_sem); -+ mmsem_batch = 0; -+ } -+ -+ if (freezing(current)) -+ break; -+ -+ cond_resched(); -+ } -+ end_time = task_sched_runtime(current); -+ delta_exec = end_time - start_time; -+ -+ if (freezing(current)) -+ return; -+ -+ cleanup_vma_slots(); -+ uksm_enter_all_slots(); -+ -+ round_finished = 1; -+ all_rungs_emtpy = 1; -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ struct scan_rung *rung = &uksm_scan_ladder[i]; -+ -+ if (rung->vma_root.num) { -+ all_rungs_emtpy = 0; -+ if (!rung_round_finished(rung)) -+ round_finished = 0; -+ } -+ } -+ -+ if (all_rungs_emtpy) -+ round_finished = 0; -+ -+ if (round_finished) { -+ round_update_ladder(); -+ uksm_eval_round++; -+ -+ if (hash_round_finished() && rshash_adjust()) { -+ /* Reset the unstable root iff hash strength changed */ -+ uksm_hash_round++; -+ root_unstable_tree = RB_ROOT; -+ free_all_tree_nodes(&unstable_tree_node_list); -+ } -+ -+ /* -+ * A number of pages can hang around indefinitely on per-cpu -+ * pagevecs, raised page count preventing write_protect_page -+ * from merging them. Though it doesn't really matter much, -+ * it is puzzling to see some stuck in pages_volatile until -+ * other activity jostles them out, and they also prevented -+ * LTP's KSM test from succeeding deterministically; so drain -+ * them here (here rather than on entry to uksm_do_scan(), -+ * so we don't IPI too often when pages_to_scan is set low). -+ */ -+ lru_add_drain_all(); -+ } -+ -+ -+ if (vpages && delta_exec > 0) { -+ pcost = (unsigned long) delta_exec / vpages; -+ if (likely(uksm_ema_page_time)) -+ uksm_ema_page_time = ema(pcost, uksm_ema_page_time); -+ else -+ uksm_ema_page_time = pcost; -+ } -+ -+ uksm_calc_scan_pages(); -+ uksm_sleep_real = uksm_sleep_jiffies; -+ /* in case of radical cpu bursts, apply the upper bound */ -+ end_time = task_sched_runtime(current); -+ if (max_cpu_ratio && end_time > start_time) { -+ scan_time = end_time - start_time; -+ expected_jiffies = msecs_to_jiffies( -+ scan_time_to_sleep(scan_time, max_cpu_ratio)); -+ -+ if (expected_jiffies > uksm_sleep_real) -+ uksm_sleep_real = expected_jiffies; -+ -+ /* We have a 1 second up bound for responsiveness. */ -+ if (jiffies_to_msecs(uksm_sleep_real) > MSEC_PER_SEC) -+ uksm_sleep_real = msecs_to_jiffies(1000); -+ } -+ -+ return; -+} -+ -+static int ksmd_should_run(void) -+{ -+ return uksm_run & UKSM_RUN_MERGE; -+} -+ -+static int uksm_scan_thread(void *nothing) -+{ -+ set_freezable(); -+ set_user_nice(current, 5); -+ -+ while (!kthread_should_stop()) { -+ mutex_lock(&uksm_thread_mutex); -+ if (ksmd_should_run()) -+ uksm_do_scan(); -+ mutex_unlock(&uksm_thread_mutex); -+ -+ try_to_freeze(); -+ -+ if (ksmd_should_run()) { -+ schedule_timeout_interruptible(uksm_sleep_real); -+ uksm_sleep_times++; -+ } else { -+ wait_event_freezable(uksm_thread_wait, -+ ksmd_should_run() || kthread_should_stop()); -+ } -+ } -+ return 0; -+} -+ -+void rmap_walk_ksm(struct page *page, struct rmap_walk_control *rwc) -+{ -+ struct stable_node *stable_node; -+ struct node_vma *node_vma; -+ struct rmap_item *rmap_item; -+ int search_new_forks = 0; -+ unsigned long address; -+ -+ VM_BUG_ON_PAGE(!PageKsm(page), page); -+ VM_BUG_ON_PAGE(!PageLocked(page), page); -+ -+ stable_node = page_stable_node(page); -+ if (!stable_node) -+ return; -+again: -+ hlist_for_each_entry(node_vma, &stable_node->hlist, hlist) { -+ hlist_for_each_entry(rmap_item, &node_vma->rmap_hlist, hlist) { -+ struct anon_vma *anon_vma = rmap_item->anon_vma; -+ struct anon_vma_chain *vmac; -+ struct vm_area_struct *vma; -+ -+ cond_resched(); -+ anon_vma_lock_read(anon_vma); -+ anon_vma_interval_tree_foreach(vmac, &anon_vma->rb_root, -+ 0, ULONG_MAX) { -+ cond_resched(); -+ vma = vmac->vma; -+ address = get_rmap_addr(rmap_item); -+ -+ if (address < vma->vm_start || -+ address >= vma->vm_end) -+ continue; -+ -+ if ((rmap_item->slot->vma == vma) == -+ search_new_forks) -+ continue; -+ -+ if (rwc->invalid_vma && rwc->invalid_vma(vma, rwc->arg)) -+ continue; -+ -+ if (!rwc->rmap_one(page, vma, address, rwc->arg)) { -+ anon_vma_unlock_read(anon_vma); -+ return; -+ } -+ -+ if (rwc->done && rwc->done(page)) { -+ anon_vma_unlock_read(anon_vma); -+ return; -+ } -+ } -+ anon_vma_unlock_read(anon_vma); -+ } -+ } -+ if (!search_new_forks++) -+ goto again; -+} -+ -+#ifdef CONFIG_MIGRATION -+/* Common ksm interface but may be specific to uksm */ -+void ksm_migrate_page(struct page *newpage, struct page *oldpage) -+{ -+ struct stable_node *stable_node; -+ -+ VM_BUG_ON_PAGE(!PageLocked(oldpage), oldpage); -+ VM_BUG_ON_PAGE(!PageLocked(newpage), newpage); -+ VM_BUG_ON(newpage->mapping != oldpage->mapping); -+ -+ stable_node = page_stable_node(newpage); -+ if (stable_node) { -+ VM_BUG_ON(stable_node->kpfn != page_to_pfn(oldpage)); -+ stable_node->kpfn = page_to_pfn(newpage); -+ /* -+ * newpage->mapping was set in advance; now we need smp_wmb() -+ * to make sure that the new stable_node->kpfn is visible -+ * to get_ksm_page() before it can see that oldpage->mapping -+ * has gone stale (or that PageSwapCache has been cleared). -+ */ -+ smp_wmb(); -+ set_page_stable_node(oldpage, NULL); -+ } -+} -+#endif /* CONFIG_MIGRATION */ -+ -+#ifdef CONFIG_MEMORY_HOTREMOVE -+static struct stable_node *uksm_check_stable_tree(unsigned long start_pfn, -+ unsigned long end_pfn) -+{ -+ struct rb_node *node; -+ -+ for (node = rb_first(root_stable_treep); node; node = rb_next(node)) { -+ struct stable_node *stable_node; -+ -+ stable_node = rb_entry(node, struct stable_node, node); -+ if (stable_node->kpfn >= start_pfn && -+ stable_node->kpfn < end_pfn) -+ return stable_node; -+ } -+ return NULL; -+} -+ -+static int uksm_memory_callback(struct notifier_block *self, -+ unsigned long action, void *arg) -+{ -+ struct memory_notify *mn = arg; -+ struct stable_node *stable_node; -+ -+ switch (action) { -+ case MEM_GOING_OFFLINE: -+ /* -+ * Keep it very simple for now: just lock out ksmd and -+ * MADV_UNMERGEABLE while any memory is going offline. -+ * mutex_lock_nested() is necessary because lockdep was alarmed -+ * that here we take uksm_thread_mutex inside notifier chain -+ * mutex, and later take notifier chain mutex inside -+ * uksm_thread_mutex to unlock it. But that's safe because both -+ * are inside mem_hotplug_mutex. -+ */ -+ mutex_lock_nested(&uksm_thread_mutex, SINGLE_DEPTH_NESTING); -+ break; -+ -+ case MEM_OFFLINE: -+ /* -+ * Most of the work is done by page migration; but there might -+ * be a few stable_nodes left over, still pointing to struct -+ * pages which have been offlined: prune those from the tree. -+ */ -+ while ((stable_node = uksm_check_stable_tree(mn->start_pfn, -+ mn->start_pfn + mn->nr_pages)) != NULL) -+ remove_node_from_stable_tree(stable_node, 1, 1); -+ /* fallthrough */ -+ -+ case MEM_CANCEL_OFFLINE: -+ mutex_unlock(&uksm_thread_mutex); -+ break; -+ } -+ return NOTIFY_OK; -+} -+#endif /* CONFIG_MEMORY_HOTREMOVE */ -+ -+#ifdef CONFIG_SYSFS -+/* -+ * This all compiles without CONFIG_SYSFS, but is a waste of space. -+ */ -+ -+#define UKSM_ATTR_RO(_name) \ -+ static struct kobj_attribute _name##_attr = __ATTR_RO(_name) -+#define UKSM_ATTR(_name) \ -+ static struct kobj_attribute _name##_attr = \ -+ __ATTR(_name, 0644, _name##_show, _name##_store) -+ -+static ssize_t max_cpu_percentage_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%u\n", uksm_max_cpu_percentage); -+} -+ -+static ssize_t max_cpu_percentage_store(struct kobject *kobj, -+ struct kobj_attribute *attr, -+ const char *buf, size_t count) -+{ -+ unsigned long max_cpu_percentage; -+ int err; -+ -+ err = kstrtoul(buf, 10, &max_cpu_percentage); -+ if (err || max_cpu_percentage > 100) -+ return -EINVAL; -+ -+ if (max_cpu_percentage == 100) -+ max_cpu_percentage = 99; -+ else if (max_cpu_percentage < 10) -+ max_cpu_percentage = 10; -+ -+ uksm_max_cpu_percentage = max_cpu_percentage; -+ -+ return count; -+} -+UKSM_ATTR(max_cpu_percentage); -+ -+static ssize_t sleep_millisecs_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%u\n", jiffies_to_msecs(uksm_sleep_jiffies)); -+} -+ -+static ssize_t sleep_millisecs_store(struct kobject *kobj, -+ struct kobj_attribute *attr, -+ const char *buf, size_t count) -+{ -+ unsigned long msecs; -+ int err; -+ -+ err = kstrtoul(buf, 10, &msecs); -+ if (err || msecs > MSEC_PER_SEC) -+ return -EINVAL; -+ -+ uksm_sleep_jiffies = msecs_to_jiffies(msecs); -+ uksm_sleep_saved = uksm_sleep_jiffies; -+ -+ return count; -+} -+UKSM_ATTR(sleep_millisecs); -+ -+ -+static ssize_t cpu_governor_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ int n = sizeof(uksm_cpu_governor_str) / sizeof(char *); -+ int i; -+ -+ buf[0] = '\0'; -+ for (i = 0; i < n ; i++) { -+ if (uksm_cpu_governor == i) -+ strcat(buf, "["); -+ -+ strcat(buf, uksm_cpu_governor_str[i]); -+ -+ if (uksm_cpu_governor == i) -+ strcat(buf, "]"); -+ -+ strcat(buf, " "); -+ } -+ strcat(buf, "\n"); -+ -+ return strlen(buf); -+} -+ -+static inline void init_performance_values(void) -+{ -+ int i; -+ struct scan_rung *rung; -+ struct uksm_cpu_preset_s *preset = uksm_cpu_preset + uksm_cpu_governor; -+ -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ rung = uksm_scan_ladder + i; -+ rung->cpu_ratio = preset->cpu_ratio[i]; -+ rung->cover_msecs = preset->cover_msecs[i]; -+ } -+ -+ uksm_max_cpu_percentage = preset->max_cpu; -+} -+ -+static ssize_t cpu_governor_store(struct kobject *kobj, -+ struct kobj_attribute *attr, -+ const char *buf, size_t count) -+{ -+ int n = sizeof(uksm_cpu_governor_str) / sizeof(char *); -+ -+ for (n--; n >= 0 ; n--) { -+ if (!strncmp(buf, uksm_cpu_governor_str[n], -+ strlen(uksm_cpu_governor_str[n]))) -+ break; -+ } -+ -+ if (n < 0) -+ return -EINVAL; -+ else -+ uksm_cpu_governor = n; -+ -+ init_performance_values(); -+ -+ return count; -+} -+UKSM_ATTR(cpu_governor); -+ -+static ssize_t run_show(struct kobject *kobj, struct kobj_attribute *attr, -+ char *buf) -+{ -+ return sprintf(buf, "%u\n", uksm_run); -+} -+ -+static ssize_t run_store(struct kobject *kobj, struct kobj_attribute *attr, -+ const char *buf, size_t count) -+{ -+ int err; -+ unsigned long flags; -+ -+ err = kstrtoul(buf, 10, &flags); -+ if (err || flags > UINT_MAX) -+ return -EINVAL; -+ if (flags > UKSM_RUN_MERGE) -+ return -EINVAL; -+ -+ mutex_lock(&uksm_thread_mutex); -+ if (uksm_run != flags) -+ uksm_run = flags; -+ mutex_unlock(&uksm_thread_mutex); -+ -+ if (flags & UKSM_RUN_MERGE) -+ wake_up_interruptible(&uksm_thread_wait); -+ -+ return count; -+} -+UKSM_ATTR(run); -+ -+static ssize_t abundant_threshold_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%u\n", uksm_abundant_threshold); -+} -+ -+static ssize_t abundant_threshold_store(struct kobject *kobj, -+ struct kobj_attribute *attr, -+ const char *buf, size_t count) -+{ -+ int err; -+ unsigned long flags; -+ -+ err = kstrtoul(buf, 10, &flags); -+ if (err || flags > 99) -+ return -EINVAL; -+ -+ uksm_abundant_threshold = flags; -+ -+ return count; -+} -+UKSM_ATTR(abundant_threshold); -+ -+static ssize_t thrash_threshold_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%u\n", uksm_thrash_threshold); -+} -+ -+static ssize_t thrash_threshold_store(struct kobject *kobj, -+ struct kobj_attribute *attr, -+ const char *buf, size_t count) -+{ -+ int err; -+ unsigned long flags; -+ -+ err = kstrtoul(buf, 10, &flags); -+ if (err || flags > 99) -+ return -EINVAL; -+ -+ uksm_thrash_threshold = flags; -+ -+ return count; -+} -+UKSM_ATTR(thrash_threshold); -+ -+static ssize_t cpu_ratios_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ int i, size; -+ struct scan_rung *rung; -+ char *p = buf; -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ rung = &uksm_scan_ladder[i]; -+ -+ if (rung->cpu_ratio > 0) -+ size = sprintf(p, "%d ", rung->cpu_ratio); -+ else -+ size = sprintf(p, "MAX/%d ", -+ TIME_RATIO_SCALE / -rung->cpu_ratio); -+ -+ p += size; -+ } -+ -+ *p++ = '\n'; -+ *p = '\0'; -+ -+ return p - buf; -+} -+ -+static ssize_t cpu_ratios_store(struct kobject *kobj, -+ struct kobj_attribute *attr, -+ const char *buf, size_t count) -+{ -+ int i, cpuratios[SCAN_LADDER_SIZE], err; -+ unsigned long value; -+ struct scan_rung *rung; -+ char *p, *end = NULL; -+ -+ p = kzalloc(count, GFP_KERNEL); -+ if (!p) -+ return -ENOMEM; -+ -+ memcpy(p, buf, count); -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ if (i != SCAN_LADDER_SIZE - 1) { -+ end = strchr(p, ' '); -+ if (!end) -+ return -EINVAL; -+ -+ *end = '\0'; -+ } -+ -+ if (strstr(p, "MAX/")) { -+ p = strchr(p, '/') + 1; -+ err = kstrtoul(p, 10, &value); -+ if (err || value > TIME_RATIO_SCALE || !value) -+ return -EINVAL; -+ -+ cpuratios[i] = -(int) (TIME_RATIO_SCALE / value); -+ } else { -+ err = kstrtoul(p, 10, &value); -+ if (err || value > TIME_RATIO_SCALE || !value) -+ return -EINVAL; -+ -+ cpuratios[i] = value; -+ } -+ -+ p = end + 1; -+ } -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ rung = &uksm_scan_ladder[i]; -+ -+ rung->cpu_ratio = cpuratios[i]; -+ } -+ -+ return count; -+} -+UKSM_ATTR(cpu_ratios); -+ -+static ssize_t eval_intervals_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ int i, size; -+ struct scan_rung *rung; -+ char *p = buf; -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ rung = &uksm_scan_ladder[i]; -+ size = sprintf(p, "%u ", rung->cover_msecs); -+ p += size; -+ } -+ -+ *p++ = '\n'; -+ *p = '\0'; -+ -+ return p - buf; -+} -+ -+static ssize_t eval_intervals_store(struct kobject *kobj, -+ struct kobj_attribute *attr, -+ const char *buf, size_t count) -+{ -+ int i, err; -+ unsigned long values[SCAN_LADDER_SIZE]; -+ struct scan_rung *rung; -+ char *p, *end = NULL; -+ ssize_t ret = count; -+ -+ p = kzalloc(count + 2, GFP_KERNEL); -+ if (!p) -+ return -ENOMEM; -+ -+ memcpy(p, buf, count); -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ if (i != SCAN_LADDER_SIZE - 1) { -+ end = strchr(p, ' '); -+ if (!end) { -+ ret = -EINVAL; -+ goto out; -+ } -+ -+ *end = '\0'; -+ } -+ -+ err = kstrtoul(p, 10, &values[i]); -+ if (err) { -+ ret = -EINVAL; -+ goto out; -+ } -+ -+ p = end + 1; -+ } -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ rung = &uksm_scan_ladder[i]; -+ -+ rung->cover_msecs = values[i]; -+ } -+ -+out: -+ kfree(p); -+ return ret; -+} -+UKSM_ATTR(eval_intervals); -+ -+static ssize_t ema_per_page_time_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%lu\n", uksm_ema_page_time); -+} -+UKSM_ATTR_RO(ema_per_page_time); -+ -+static ssize_t pages_shared_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%lu\n", uksm_pages_shared); -+} -+UKSM_ATTR_RO(pages_shared); -+ -+static ssize_t pages_sharing_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%lu\n", uksm_pages_sharing); -+} -+UKSM_ATTR_RO(pages_sharing); -+ -+static ssize_t pages_unshared_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%lu\n", uksm_pages_unshared); -+} -+UKSM_ATTR_RO(pages_unshared); -+ -+static ssize_t full_scans_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%llu\n", fully_scanned_round); -+} -+UKSM_ATTR_RO(full_scans); -+ -+static ssize_t pages_scanned_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ unsigned long base = 0; -+ u64 delta, ret; -+ -+ if (pages_scanned_stored) { -+ base = pages_scanned_base; -+ ret = pages_scanned_stored; -+ delta = uksm_pages_scanned >> base; -+ if (CAN_OVERFLOW_U64(ret, delta)) { -+ ret >>= 1; -+ delta >>= 1; -+ base++; -+ ret += delta; -+ } -+ } else { -+ ret = uksm_pages_scanned; -+ } -+ -+ while (ret > ULONG_MAX) { -+ ret >>= 1; -+ base++; -+ } -+ -+ if (base) -+ return sprintf(buf, "%lu * 2^%lu\n", (unsigned long)ret, base); -+ else -+ return sprintf(buf, "%lu\n", (unsigned long)ret); -+} -+UKSM_ATTR_RO(pages_scanned); -+ -+static ssize_t hash_strength_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%lu\n", hash_strength); -+} -+UKSM_ATTR_RO(hash_strength); -+ -+static ssize_t sleep_times_show(struct kobject *kobj, -+ struct kobj_attribute *attr, char *buf) -+{ -+ return sprintf(buf, "%llu\n", uksm_sleep_times); -+} -+UKSM_ATTR_RO(sleep_times); -+ -+ -+static struct attribute *uksm_attrs[] = { -+ &max_cpu_percentage_attr.attr, -+ &sleep_millisecs_attr.attr, -+ &cpu_governor_attr.attr, -+ &run_attr.attr, -+ &ema_per_page_time_attr.attr, -+ &pages_shared_attr.attr, -+ &pages_sharing_attr.attr, -+ &pages_unshared_attr.attr, -+ &full_scans_attr.attr, -+ &pages_scanned_attr.attr, -+ &hash_strength_attr.attr, -+ &sleep_times_attr.attr, -+ &thrash_threshold_attr.attr, -+ &abundant_threshold_attr.attr, -+ &cpu_ratios_attr.attr, -+ &eval_intervals_attr.attr, -+ NULL, -+}; -+ -+static struct attribute_group uksm_attr_group = { -+ .attrs = uksm_attrs, -+ .name = "uksm", -+}; -+#endif /* CONFIG_SYSFS */ -+ -+static inline void init_scan_ladder(void) -+{ -+ int i; -+ struct scan_rung *rung; -+ -+ for (i = 0; i < SCAN_LADDER_SIZE; i++) { -+ rung = uksm_scan_ladder + i; -+ slot_tree_init_root(&rung->vma_root); -+ } -+ -+ init_performance_values(); -+ uksm_calc_scan_pages(); -+} -+ -+static inline int cal_positive_negative_costs(void) -+{ -+ struct page *p1, *p2; -+ unsigned char *addr1, *addr2; -+ unsigned long i, time_start, hash_cost; -+ unsigned long loopnum = 0; -+ -+ /*IMPORTANT: volatile is needed to prevent over-optimization by gcc. */ -+ volatile u32 hash; -+ volatile int ret; -+ -+ p1 = alloc_page(GFP_KERNEL); -+ if (!p1) -+ return -ENOMEM; -+ -+ p2 = alloc_page(GFP_KERNEL); -+ if (!p2) -+ return -ENOMEM; -+ -+ addr1 = kmap_atomic(p1); -+ addr2 = kmap_atomic(p2); -+ memset(addr1, prandom_u32(), PAGE_SIZE); -+ memcpy(addr2, addr1, PAGE_SIZE); -+ -+ /* make sure that the two pages differ in last byte */ -+ addr2[PAGE_SIZE-1] = ~addr2[PAGE_SIZE-1]; -+ kunmap_atomic(addr2); -+ kunmap_atomic(addr1); -+ -+ time_start = jiffies; -+ while (jiffies - time_start < 100) { -+ for (i = 0; i < 100; i++) -+ hash = page_hash(p1, HASH_STRENGTH_FULL, 0); -+ loopnum += 100; -+ } -+ hash_cost = (jiffies - time_start); -+ -+ time_start = jiffies; -+ for (i = 0; i < loopnum; i++) -+ ret = pages_identical(p1, p2); -+ memcmp_cost = HASH_STRENGTH_FULL * (jiffies - time_start); -+ memcmp_cost /= hash_cost; -+ pr_info("UKSM: relative memcmp_cost = %lu " -+ "hash=%u cmp_ret=%d.\n", -+ memcmp_cost, hash, ret); -+ -+ __free_page(p1); -+ __free_page(p2); -+ return 0; -+} -+ -+static int init_zeropage_hash_table(void) -+{ -+ struct page *page; -+ char *addr; -+ int i; -+ -+ page = alloc_page(GFP_KERNEL); -+ if (!page) -+ return -ENOMEM; -+ -+ addr = kmap_atomic(page); -+ memset(addr, 0, PAGE_SIZE); -+ kunmap_atomic(addr); -+ -+ zero_hash_table = kmalloc_array(HASH_STRENGTH_MAX, sizeof(u32), -+ GFP_KERNEL); -+ if (!zero_hash_table) -+ return -ENOMEM; -+ -+ for (i = 0; i < HASH_STRENGTH_MAX; i++) -+ zero_hash_table[i] = page_hash(page, i, 0); -+ -+ __free_page(page); -+ -+ return 0; -+} -+ -+static inline int init_random_sampling(void) -+{ -+ unsigned long i; -+ -+ random_nums = kmalloc(PAGE_SIZE, GFP_KERNEL); -+ if (!random_nums) -+ return -ENOMEM; -+ -+ for (i = 0; i < HASH_STRENGTH_FULL; i++) -+ random_nums[i] = i; -+ -+ for (i = 0; i < HASH_STRENGTH_FULL; i++) { -+ unsigned long rand_range, swap_index, tmp; -+ -+ rand_range = HASH_STRENGTH_FULL - i; -+ swap_index = i + prandom_u32() % rand_range; -+ tmp = random_nums[i]; -+ random_nums[i] = random_nums[swap_index]; -+ random_nums[swap_index] = tmp; -+ } -+ -+ rshash_state.state = RSHASH_NEW; -+ rshash_state.below_count = 0; -+ rshash_state.lookup_window_index = 0; -+ -+ return cal_positive_negative_costs(); -+} -+ -+static int __init uksm_slab_init(void) -+{ -+ rmap_item_cache = UKSM_KMEM_CACHE(rmap_item, 0); -+ if (!rmap_item_cache) -+ goto out; -+ -+ stable_node_cache = UKSM_KMEM_CACHE(stable_node, 0); -+ if (!stable_node_cache) -+ goto out_free1; -+ -+ node_vma_cache = UKSM_KMEM_CACHE(node_vma, 0); -+ if (!node_vma_cache) -+ goto out_free2; -+ -+ vma_slot_cache = UKSM_KMEM_CACHE(vma_slot, 0); -+ if (!vma_slot_cache) -+ goto out_free3; -+ -+ tree_node_cache = UKSM_KMEM_CACHE(tree_node, 0); -+ if (!tree_node_cache) -+ goto out_free4; -+ -+ return 0; -+ -+out_free4: -+ kmem_cache_destroy(vma_slot_cache); -+out_free3: -+ kmem_cache_destroy(node_vma_cache); -+out_free2: -+ kmem_cache_destroy(stable_node_cache); -+out_free1: -+ kmem_cache_destroy(rmap_item_cache); -+out: -+ return -ENOMEM; -+} -+ -+static void __init uksm_slab_free(void) -+{ -+ kmem_cache_destroy(stable_node_cache); -+ kmem_cache_destroy(rmap_item_cache); -+ kmem_cache_destroy(node_vma_cache); -+ kmem_cache_destroy(vma_slot_cache); -+ kmem_cache_destroy(tree_node_cache); -+} -+ -+/* Common interface to ksm, different to it. */ -+int ksm_madvise(struct vm_area_struct *vma, unsigned long start, -+ unsigned long end, int advice, unsigned long *vm_flags) -+{ -+ int err; -+ -+ switch (advice) { -+ case MADV_MERGEABLE: -+ return 0; /* just ignore the advice */ -+ -+ case MADV_UNMERGEABLE: -+ if (!(*vm_flags & VM_MERGEABLE) || !uksm_flags_can_scan(*vm_flags)) -+ return 0; /* just ignore the advice */ -+ -+ if (vma->anon_vma) { -+ err = unmerge_uksm_pages(vma, start, end); -+ if (err) -+ return err; -+ } -+ -+ uksm_remove_vma(vma); -+ *vm_flags &= ~VM_MERGEABLE; -+ break; -+ } -+ -+ return 0; -+} -+ -+/* Common interface to ksm, actually the same. */ -+struct page *ksm_might_need_to_copy(struct page *page, -+ struct vm_area_struct *vma, unsigned long address) -+{ -+ struct anon_vma *anon_vma = page_anon_vma(page); -+ struct page *new_page; -+ -+ if (PageKsm(page)) { -+ if (page_stable_node(page)) -+ return page; /* no need to copy it */ -+ } else if (!anon_vma) { -+ return page; /* no need to copy it */ -+ } else if (anon_vma->root == vma->anon_vma->root && -+ page->index == linear_page_index(vma, address)) { -+ return page; /* still no need to copy it */ -+ } -+ if (!PageUptodate(page)) -+ return page; /* let do_swap_page report the error */ -+ -+ new_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address); -+ if (new_page) { -+ copy_user_highpage(new_page, page, address, vma); -+ -+ SetPageDirty(new_page); -+ __SetPageUptodate(new_page); -+ __SetPageLocked(new_page); -+ } -+ -+ return new_page; -+} -+ -+static int __init uksm_init(void) -+{ -+ struct task_struct *uksm_thread; -+ int err; -+ -+ uksm_sleep_jiffies = msecs_to_jiffies(100); -+ uksm_sleep_saved = uksm_sleep_jiffies; -+ -+ slot_tree_init(); -+ init_scan_ladder(); -+ -+ -+ err = init_random_sampling(); -+ if (err) -+ goto out_free2; -+ -+ err = uksm_slab_init(); -+ if (err) -+ goto out_free1; -+ -+ err = init_zeropage_hash_table(); -+ if (err) -+ goto out_free0; -+ -+ uksm_thread = kthread_run(uksm_scan_thread, NULL, "uksmd"); -+ if (IS_ERR(uksm_thread)) { -+ pr_err("uksm: creating kthread failed\n"); -+ err = PTR_ERR(uksm_thread); -+ goto out_free; -+ } -+ -+#ifdef CONFIG_SYSFS -+ err = sysfs_create_group(mm_kobj, &uksm_attr_group); -+ if (err) { -+ pr_err("uksm: register sysfs failed\n"); -+ kthread_stop(uksm_thread); -+ goto out_free; -+ } -+#else -+ uksm_run = UKSM_RUN_MERGE; /* no way for user to start it */ -+ -+#endif /* CONFIG_SYSFS */ -+ -+#ifdef CONFIG_MEMORY_HOTREMOVE -+ /* -+ * Choose a high priority since the callback takes uksm_thread_mutex: -+ * later callbacks could only be taking locks which nest within that. -+ */ -+ hotplug_memory_notifier(uksm_memory_callback, 100); -+#endif -+ return 0; -+ -+out_free: -+ kfree(zero_hash_table); -+out_free0: -+ uksm_slab_free(); -+out_free1: -+ kfree(random_nums); -+out_free2: -+ kfree(uksm_scan_ladder); -+ return err; -+} -+ -+#ifdef MODULE -+subsys_initcall(ksm_init); -+#else -+late_initcall(uksm_init); -+#endif -+ -diff -Nur a/mm/vmstat.c b/mm/vmstat.c ---- a/mm/vmstat.c 2018-05-25 15:18:02.000000000 +0100 -+++ b/mm/vmstat.c 2018-05-26 19:30:55.791140570 +0100 -@@ -1091,6 +1091,9 @@ - "nr_dirtied", - "nr_written", - -+#ifdef CONFIG_UKSM -+ "nr_uksm_zero_pages", -+#endif - /* enum writeback_stat_item counters */ - "nr_dirty_threshold", - "nr_dirty_background_threshold", diff --git a/sys-kernel/linux-sources-redcore-lts/files/uksm-linux-hardened.patch b/sys-kernel/linux-sources-redcore-lts/files/uksm-linux-hardened.patch new file mode 100644 index 00000000..f0596117 --- /dev/null +++ b/sys-kernel/linux-sources-redcore-lts/files/uksm-linux-hardened.patch @@ -0,0 +1,6919 @@ +diff -Nur a/Documentation/vm/00-INDEX b/Documentation/vm/00-INDEX +--- a/Documentation/vm/00-INDEX 2018-05-25 15:18:02.000000000 +0100 ++++ b/Documentation/vm/00-INDEX 2018-05-26 19:30:55.783140311 +0100 +@@ -20,6 +20,8 @@ + - description of the idle page tracking feature. + ksm.txt + - how to use the Kernel Samepage Merging feature. ++uksm.txt ++ - Introduction to Ultra KSM + numa + - information about NUMA specific code in the Linux vm. + numa_memory_policy.txt +diff -Nur a/Documentation/vm/uksm.txt b/Documentation/vm/uksm.txt +--- a/Documentation/vm/uksm.txt 1970-01-01 01:00:00.000000000 +0100 ++++ b/Documentation/vm/uksm.txt 2018-05-26 19:30:55.783140311 +0100 +@@ -0,0 +1,61 @@ ++The Ultra Kernel Samepage Merging feature ++---------------------------------------------- ++/* ++ * Ultra KSM. Copyright (C) 2011-2012 Nai Xia ++ * ++ * This is an improvement upon KSM. Some basic data structures and routines ++ * are borrowed from ksm.c . ++ * ++ * Its new features: ++ * 1. Full system scan: ++ * It automatically scans all user processes' anonymous VMAs. Kernel-user ++ * interaction to submit a memory area to KSM is no longer needed. ++ * ++ * 2. Rich area detection: ++ * It automatically detects rich areas containing abundant duplicated ++ * pages based. Rich areas are given a full scan speed. Poor areas are ++ * sampled at a reasonable speed with very low CPU consumption. ++ * ++ * 3. Ultra Per-page scan speed improvement: ++ * A new hash algorithm is proposed. As a result, on a machine with ++ * Core(TM)2 Quad Q9300 CPU in 32-bit mode and 800MHZ DDR2 main memory, it ++ * can scan memory areas that does not contain duplicated pages at speed of ++ * 627MB/sec ~ 2445MB/sec and can merge duplicated areas at speed of ++ * 477MB/sec ~ 923MB/sec. ++ * ++ * 4. Thrashing area avoidance: ++ * Thrashing area(an VMA that has frequent Ksm page break-out) can be ++ * filtered out. My benchmark shows it's more efficient than KSM's per-page ++ * hash value based volatile page detection. ++ * ++ * ++ * 5. Misc changes upon KSM: ++ * * It has a fully x86-opitmized memcmp dedicated for 4-byte-aligned page ++ * comparison. It's much faster than default C version on x86. ++ * * rmap_item now has an struct *page member to loosely cache a ++ * address-->page mapping, which reduces too much time-costly ++ * follow_page(). ++ * * The VMA creation/exit procedures are hooked to let the Ultra KSM know. ++ * * try_to_merge_two_pages() now can revert a pte if it fails. No break_ ++ * ksm is needed for this case. ++ * ++ * 6. Full Zero Page consideration(contributed by Figo Zhang) ++ * Now uksmd consider full zero pages as special pages and merge them to an ++ * special unswappable uksm zero page. ++ */ ++ ++ChangeLog: ++ ++2012-05-05 The creation of this Doc ++2012-05-08 UKSM 0.1.1.1 libc crash bug fix, api clean up, doc clean up. ++2012-05-28 UKSM 0.1.1.2 bug fix release ++2012-06-26 UKSM 0.1.2-beta1 first beta release for 0.1.2 ++2012-07-2 UKSM 0.1.2-beta2 ++2012-07-10 UKSM 0.1.2-beta3 ++2012-07-26 UKSM 0.1.2 Fine grained speed control, more scan optimization. ++2012-10-13 UKSM 0.1.2.1 Bug fixes. ++2012-12-31 UKSM 0.1.2.2 Minor bug fixes. ++2014-07-02 UKSM 0.1.2.3 Fix a " __this_cpu_read() in preemptible bug". ++2015-04-22 UKSM 0.1.2.4 Fix a race condition that can sometimes trigger anonying warnings. ++2016-09-10 UKSM 0.1.2.5 Fix a bug in dedup ratio calculation. ++2017-02-26 UKSM 0.1.2.6 Fix a bug in hugetlbpage handling and a race bug with page migration. +diff -Nur a/fs/exec.c b/fs/exec.c +--- a/fs/exec.c 2018-05-26 19:24:34.831782903 +0100 ++++ b/fs/exec.c 2018-05-26 19:31:18.404873956 +0100 +@@ -63,6 +63,7 @@ + #include + #include + #include ++#include + + #include + #include +@@ -1377,6 +1378,7 @@ + /* An exec changes our domain. We are no longer part of the thread + group */ + current->self_exec_id++; ++ + flush_signal_handlers(current, 0); + } + EXPORT_SYMBOL(setup_new_exec); +diff -Nur a/fs/proc/meminfo.c b/fs/proc/meminfo.c +--- a/fs/proc/meminfo.c 2018-05-25 15:18:02.000000000 +0100 ++++ b/fs/proc/meminfo.c 2018-05-26 19:30:55.784140344 +0100 +@@ -118,6 +118,10 @@ + global_zone_page_state(NR_KERNEL_STACK_KB)); + show_val_kb(m, "PageTables: ", + global_zone_page_state(NR_PAGETABLE)); ++#ifdef CONFIG_UKSM ++ show_val_kb(m, "KsmZeroPages: ", ++ global_zone_page_state(NR_UKSM_ZERO_PAGES)); ++#endif + #ifdef CONFIG_QUICKLIST + show_val_kb(m, "Quicklists: ", quicklist_total_size()); + #endif +diff -Nur a/include/asm-generic/pgtable.h b/include/asm-generic/pgtable.h +--- a/include/asm-generic/pgtable.h 2018-05-25 15:18:02.000000000 +0100 ++++ b/include/asm-generic/pgtable.h 2018-05-26 19:30:55.784140344 +0100 +@@ -789,12 +789,25 @@ + extern void untrack_pfn_moved(struct vm_area_struct *vma); + #endif + ++#ifdef CONFIG_UKSM ++static inline int is_uksm_zero_pfn(unsigned long pfn) ++{ ++ extern unsigned long uksm_zero_pfn; ++ return pfn == uksm_zero_pfn; ++} ++#else ++static inline int is_uksm_zero_pfn(unsigned long pfn) ++{ ++ return 0; ++} ++#endif ++ + #ifdef __HAVE_COLOR_ZERO_PAGE + static inline int is_zero_pfn(unsigned long pfn) + { + extern unsigned long zero_pfn; + unsigned long offset_from_zero_pfn = pfn - zero_pfn; +- return offset_from_zero_pfn <= (zero_page_mask >> PAGE_SHIFT); ++ return offset_from_zero_pfn <= (zero_page_mask >> PAGE_SHIFT) || is_uksm_zero_pfn(pfn); + } + + #define my_zero_pfn(addr) page_to_pfn(ZERO_PAGE(addr)) +@@ -803,7 +816,7 @@ + static inline int is_zero_pfn(unsigned long pfn) + { + extern unsigned long zero_pfn; +- return pfn == zero_pfn; ++ return (pfn == zero_pfn) || (is_uksm_zero_pfn(pfn)); + } + + static inline unsigned long my_zero_pfn(unsigned long addr) +diff -Nur a/include/linux/ksm.h b/include/linux/ksm.h +--- a/include/linux/ksm.h 2018-05-25 15:18:02.000000000 +0100 ++++ b/include/linux/ksm.h 2018-05-26 19:30:55.784140344 +0100 +@@ -21,21 +21,6 @@ + #ifdef CONFIG_KSM + int ksm_madvise(struct vm_area_struct *vma, unsigned long start, + unsigned long end, int advice, unsigned long *vm_flags); +-int __ksm_enter(struct mm_struct *mm); +-void __ksm_exit(struct mm_struct *mm); +- +-static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm) +-{ +- if (test_bit(MMF_VM_MERGEABLE, &oldmm->flags)) +- return __ksm_enter(mm); +- return 0; +-} +- +-static inline void ksm_exit(struct mm_struct *mm) +-{ +- if (test_bit(MMF_VM_MERGEABLE, &mm->flags)) +- __ksm_exit(mm); +-} + + static inline struct stable_node *page_stable_node(struct page *page) + { +@@ -65,6 +50,33 @@ + void rmap_walk_ksm(struct page *page, struct rmap_walk_control *rwc); + void ksm_migrate_page(struct page *newpage, struct page *oldpage); + ++#ifdef CONFIG_KSM_LEGACY ++int __ksm_enter(struct mm_struct *mm); ++void __ksm_exit(struct mm_struct *mm); ++static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm) ++{ ++ if (test_bit(MMF_VM_MERGEABLE, &oldmm->flags)) ++ return __ksm_enter(mm); ++ return 0; ++} ++ ++static inline void ksm_exit(struct mm_struct *mm) ++{ ++ if (test_bit(MMF_VM_MERGEABLE, &mm->flags)) ++ __ksm_exit(mm); ++} ++ ++#elif defined(CONFIG_UKSM) ++static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm) ++{ ++ return 0; ++} ++ ++static inline void ksm_exit(struct mm_struct *mm) ++{ ++} ++#endif /* !CONFIG_UKSM */ ++ + #else /* !CONFIG_KSM */ + + static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm) +@@ -106,4 +118,6 @@ + #endif /* CONFIG_MMU */ + #endif /* !CONFIG_KSM */ + ++#include ++ + #endif /* __LINUX_KSM_H */ +diff -Nur a/include/linux/mm_types.h b/include/linux/mm_types.h +--- a/include/linux/mm_types.h 2018-05-25 15:18:02.000000000 +0100 ++++ b/include/linux/mm_types.h 2018-05-26 19:30:55.784140344 +0100 +@@ -337,6 +337,9 @@ + struct mempolicy *vm_policy; /* NUMA policy for the VMA */ + #endif + struct vm_userfaultfd_ctx vm_userfaultfd_ctx; ++#ifdef CONFIG_UKSM ++ struct vma_slot *uksm_vma_slot; ++#endif + } __randomize_layout; + + struct core_thread { +diff -Nur a/include/linux/mmzone.h b/include/linux/mmzone.h +--- a/include/linux/mmzone.h 2018-05-25 15:18:02.000000000 +0100 ++++ b/include/linux/mmzone.h 2018-05-26 19:30:55.785140376 +0100 +@@ -148,6 +148,9 @@ + NR_ZSPAGES, /* allocated in zsmalloc */ + #endif + NR_FREE_CMA_PAGES, ++#ifdef CONFIG_UKSM ++ NR_UKSM_ZERO_PAGES, ++#endif + NR_VM_ZONE_STAT_ITEMS }; + + enum node_stat_item { +@@ -872,7 +875,7 @@ + } + + /** +- * is_highmem - helper function to quickly check if a struct zone is a ++ * is_highmem - helper function to quickly check if a struct zone is a + * highmem zone or not. This is an attempt to keep references + * to ZONE_{DMA/NORMAL/HIGHMEM/etc} in general code to a minimum. + * @zone - pointer to struct zone variable +diff -Nur a/include/linux/sradix-tree.h b/include/linux/sradix-tree.h +--- a/include/linux/sradix-tree.h 1970-01-01 01:00:00.000000000 +0100 ++++ b/include/linux/sradix-tree.h 2018-05-26 19:30:55.785140376 +0100 +@@ -0,0 +1,77 @@ ++#ifndef _LINUX_SRADIX_TREE_H ++#define _LINUX_SRADIX_TREE_H ++ ++ ++#define INIT_SRADIX_TREE(root, mask) \ ++do { \ ++ (root)->height = 0; \ ++ (root)->gfp_mask = (mask); \ ++ (root)->rnode = NULL; \ ++} while (0) ++ ++#define ULONG_BITS (sizeof(unsigned long) * 8) ++#define SRADIX_TREE_INDEX_BITS (8 /* CHAR_BIT */ * sizeof(unsigned long)) ++//#define SRADIX_TREE_MAP_SHIFT 6 ++//#define SRADIX_TREE_MAP_SIZE (1UL << SRADIX_TREE_MAP_SHIFT) ++//#define SRADIX_TREE_MAP_MASK (SRADIX_TREE_MAP_SIZE-1) ++ ++struct sradix_tree_node { ++ unsigned int height; /* Height from the bottom */ ++ unsigned int count; ++ unsigned int fulls; /* Number of full sublevel trees */ ++ struct sradix_tree_node *parent; ++ void *stores[0]; ++}; ++ ++/* A simple radix tree implementation */ ++struct sradix_tree_root { ++ unsigned int height; ++ struct sradix_tree_node *rnode; ++ ++ /* Where found to have available empty stores in its sublevels */ ++ struct sradix_tree_node *enter_node; ++ unsigned int shift; ++ unsigned int stores_size; ++ unsigned int mask; ++ unsigned long min; /* The first hole index */ ++ unsigned long num; ++ //unsigned long *height_to_maxindex; ++ ++ /* How the node is allocated and freed. */ ++ struct sradix_tree_node *(*alloc)(void); ++ void (*free)(struct sradix_tree_node *node); ++ ++ /* When a new node is added and removed */ ++ void (*extend)(struct sradix_tree_node *parent, struct sradix_tree_node *child); ++ void (*assign)(struct sradix_tree_node *node, unsigned int index, void *item); ++ void (*rm)(struct sradix_tree_node *node, unsigned int offset); ++}; ++ ++struct sradix_tree_path { ++ struct sradix_tree_node *node; ++ int offset; ++}; ++ ++static inline ++void init_sradix_tree_root(struct sradix_tree_root *root, unsigned long shift) ++{ ++ root->height = 0; ++ root->rnode = NULL; ++ root->shift = shift; ++ root->stores_size = 1UL << shift; ++ root->mask = root->stores_size - 1; ++} ++ ++ ++extern void *sradix_tree_next(struct sradix_tree_root *root, ++ struct sradix_tree_node *node, unsigned long index, ++ int (*iter)(void *, unsigned long)); ++ ++extern int sradix_tree_enter(struct sradix_tree_root *root, void **item, int num); ++ ++extern void sradix_tree_delete_from_leaf(struct sradix_tree_root *root, ++ struct sradix_tree_node *node, unsigned long index); ++ ++extern void *sradix_tree_lookup(struct sradix_tree_root *root, unsigned long index); ++ ++#endif /* _LINUX_SRADIX_TREE_H */ +diff -Nur a/include/linux/uksm.h b/include/linux/uksm.h +--- a/include/linux/uksm.h 1970-01-01 01:00:00.000000000 +0100 ++++ b/include/linux/uksm.h 2018-05-26 19:30:55.785140376 +0100 +@@ -0,0 +1,149 @@ ++#ifndef __LINUX_UKSM_H ++#define __LINUX_UKSM_H ++/* ++ * Memory merging support. ++ * ++ * This code enables dynamic sharing of identical pages found in different ++ * memory areas, even if they are not shared by fork(). ++ */ ++ ++/* if !CONFIG_UKSM this file should not be compiled at all. */ ++#ifdef CONFIG_UKSM ++ ++#include ++#include ++#include ++#include ++#include ++ ++extern unsigned long zero_pfn __read_mostly; ++extern unsigned long uksm_zero_pfn __read_mostly; ++extern struct page *empty_uksm_zero_page; ++ ++/* must be done before linked to mm */ ++extern void uksm_vma_add_new(struct vm_area_struct *vma); ++extern void uksm_remove_vma(struct vm_area_struct *vma); ++ ++#define UKSM_SLOT_NEED_SORT (1 << 0) ++#define UKSM_SLOT_NEED_RERAND (1 << 1) ++#define UKSM_SLOT_SCANNED (1 << 2) /* It's scanned in this round */ ++#define UKSM_SLOT_FUL_SCANNED (1 << 3) ++#define UKSM_SLOT_IN_UKSM (1 << 4) ++ ++struct vma_slot { ++ struct sradix_tree_node *snode; ++ unsigned long sindex; ++ ++ struct list_head slot_list; ++ unsigned long fully_scanned_round; ++ unsigned long dedup_num; ++ unsigned long pages_scanned; ++ unsigned long this_sampled; ++ unsigned long last_scanned; ++ unsigned long pages_to_scan; ++ struct scan_rung *rung; ++ struct page **rmap_list_pool; ++ unsigned int *pool_counts; ++ unsigned long pool_size; ++ struct vm_area_struct *vma; ++ struct mm_struct *mm; ++ unsigned long ctime_j; ++ unsigned long pages; ++ unsigned long flags; ++ unsigned long pages_cowed; /* pages cowed this round */ ++ unsigned long pages_merged; /* pages merged this round */ ++ unsigned long pages_bemerged; ++ ++ /* when it has page merged in this eval round */ ++ struct list_head dedup_list; ++}; ++ ++static inline void uksm_unmap_zero_page(pte_t pte) ++{ ++ if (pte_pfn(pte) == uksm_zero_pfn) ++ __dec_zone_page_state(empty_uksm_zero_page, NR_UKSM_ZERO_PAGES); ++} ++ ++static inline void uksm_map_zero_page(pte_t pte) ++{ ++ if (pte_pfn(pte) == uksm_zero_pfn) ++ __inc_zone_page_state(empty_uksm_zero_page, NR_UKSM_ZERO_PAGES); ++} ++ ++static inline void uksm_cow_page(struct vm_area_struct *vma, struct page *page) ++{ ++ if (vma->uksm_vma_slot && PageKsm(page)) ++ vma->uksm_vma_slot->pages_cowed++; ++} ++ ++static inline void uksm_cow_pte(struct vm_area_struct *vma, pte_t pte) ++{ ++ if (vma->uksm_vma_slot && pte_pfn(pte) == uksm_zero_pfn) ++ vma->uksm_vma_slot->pages_cowed++; ++} ++ ++static inline int uksm_flags_can_scan(unsigned long vm_flags) ++{ ++#ifdef VM_SAO ++ if (vm_flags & VM_SAO) ++ return 0; ++#endif ++ ++ return !(vm_flags & (VM_PFNMAP | VM_IO | VM_DONTEXPAND | ++ VM_HUGETLB | VM_MIXEDMAP | VM_SHARED ++ | VM_MAYSHARE | VM_GROWSUP | VM_GROWSDOWN)); ++} ++ ++static inline void uksm_vm_flags_mod(unsigned long *vm_flags_p) ++{ ++ if (uksm_flags_can_scan(*vm_flags_p)) ++ *vm_flags_p |= VM_MERGEABLE; ++} ++ ++/* ++ * Just a wrapper for BUG_ON for where ksm_zeropage must not be. TODO: it will ++ * be removed when uksm zero page patch is stable enough. ++ */ ++static inline void uksm_bugon_zeropage(pte_t pte) ++{ ++ BUG_ON(pte_pfn(pte) == uksm_zero_pfn); ++} ++#else ++static inline void uksm_vma_add_new(struct vm_area_struct *vma) ++{ ++} ++ ++static inline void uksm_remove_vma(struct vm_area_struct *vma) ++{ ++} ++ ++static inline void uksm_unmap_zero_page(pte_t pte) ++{ ++} ++ ++static inline void uksm_map_zero_page(pte_t pte) ++{ ++} ++ ++static inline void uksm_cow_page(struct vm_area_struct *vma, struct page *page) ++{ ++} ++ ++static inline void uksm_cow_pte(struct vm_area_struct *vma, pte_t pte) ++{ ++} ++ ++static inline int uksm_flags_can_scan(unsigned long vm_flags) ++{ ++ return 0; ++} ++ ++static inline void uksm_vm_flags_mod(unsigned long *vm_flags_p) ++{ ++} ++ ++static inline void uksm_bugon_zeropage(pte_t pte) ++{ ++} ++#endif /* !CONFIG_UKSM */ ++#endif /* __LINUX_UKSM_H */ +diff -Nur a/kernel/fork.c b/kernel/fork.c +--- a/kernel/fork.c 2018-05-26 19:24:34.840783196 +0100 ++++ b/kernel/fork.c 2018-05-26 19:30:55.785140376 +0100 +@@ -655,7 +655,7 @@ + goto fail_nomem; + charge = len; + } +- tmp = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL); ++ tmp = kmem_cache_zalloc(vm_area_cachep, GFP_KERNEL); + if (!tmp) + goto fail_nomem; + *tmp = *mpnt; +@@ -714,7 +714,7 @@ + __vma_link_rb(mm, tmp, rb_link, rb_parent); + rb_link = &tmp->vm_rb.rb_right; + rb_parent = &tmp->vm_rb; +- ++ uksm_vma_add_new(tmp); + mm->map_count++; + if (!(tmp->vm_flags & VM_WIPEONFORK)) + retval = copy_page_range(mm, oldmm, mpnt); +diff -Nur a/lib/Makefile b/lib/Makefile +--- a/lib/Makefile 2018-05-25 15:18:02.000000000 +0100 ++++ b/lib/Makefile 2018-05-26 19:30:55.786140408 +0100 +@@ -18,7 +18,7 @@ + KCOV_INSTRUMENT_dynamic_debug.o := n + + lib-y := ctype.o string.o vsprintf.o cmdline.o \ +- rbtree.o radix-tree.o dump_stack.o timerqueue.o\ ++ rbtree.o radix-tree.o sradix-tree.o dump_stack.o timerqueue.o\ + idr.o int_sqrt.o extable.o \ + sha1.o chacha20.o irq_regs.o argv_split.o \ + flex_proportions.o ratelimit.o show_mem.o \ +diff -Nur a/lib/sradix-tree.c b/lib/sradix-tree.c +--- a/lib/sradix-tree.c 1970-01-01 01:00:00.000000000 +0100 ++++ b/lib/sradix-tree.c 2018-05-26 19:30:55.786140408 +0100 +@@ -0,0 +1,476 @@ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++static inline int sradix_node_full(struct sradix_tree_root *root, struct sradix_tree_node *node) ++{ ++ return node->fulls == root->stores_size || ++ (node->height == 1 && node->count == root->stores_size); ++} ++ ++/* ++ * Extend a sradix tree so it can store key @index. ++ */ ++static int sradix_tree_extend(struct sradix_tree_root *root, unsigned long index) ++{ ++ struct sradix_tree_node *node; ++ unsigned int height; ++ ++ if (unlikely(root->rnode == NULL)) { ++ if (!(node = root->alloc())) ++ return -ENOMEM; ++ ++ node->height = 1; ++ root->rnode = node; ++ root->height = 1; ++ } ++ ++ /* Figure out what the height should be. */ ++ height = root->height; ++ index >>= root->shift * height; ++ ++ while (index) { ++ index >>= root->shift; ++ height++; ++ } ++ ++ while (height > root->height) { ++ unsigned int newheight; ++ ++ if (!(node = root->alloc())) ++ return -ENOMEM; ++ ++ /* Increase the height. */ ++ node->stores[0] = root->rnode; ++ root->rnode->parent = node; ++ if (root->extend) ++ root->extend(node, root->rnode); ++ ++ newheight = root->height + 1; ++ node->height = newheight; ++ node->count = 1; ++ if (sradix_node_full(root, root->rnode)) ++ node->fulls = 1; ++ ++ root->rnode = node; ++ root->height = newheight; ++ } ++ ++ return 0; ++} ++ ++/* ++ * Search the next item from the current node, that is not NULL ++ * and can satify root->iter(). ++ */ ++void *sradix_tree_next(struct sradix_tree_root *root, ++ struct sradix_tree_node *node, unsigned long index, ++ int (*iter)(void *item, unsigned long height)) ++{ ++ unsigned long offset; ++ void *item; ++ ++ if (unlikely(node == NULL)) { ++ node = root->rnode; ++ for (offset = 0; offset < root->stores_size; offset++) { ++ item = node->stores[offset]; ++ if (item && (!iter || iter(item, node->height))) ++ break; ++ } ++ ++ if (unlikely(offset >= root->stores_size)) ++ return NULL; ++ ++ if (node->height == 1) ++ return item; ++ else ++ goto go_down; ++ } ++ ++ while (node) { ++ offset = (index & root->mask) + 1; ++ for (; offset < root->stores_size; offset++) { ++ item = node->stores[offset]; ++ if (item && (!iter || iter(item, node->height))) ++ break; ++ } ++ ++ if (offset < root->stores_size) ++ break; ++ ++ node = node->parent; ++ index >>= root->shift; ++ } ++ ++ if (!node) ++ return NULL; ++ ++ while (node->height > 1) { ++go_down: ++ node = item; ++ for (offset = 0; offset < root->stores_size; offset++) { ++ item = node->stores[offset]; ++ if (item && (!iter || iter(item, node->height))) ++ break; ++ } ++ ++ if (unlikely(offset >= root->stores_size)) ++ return NULL; ++ } ++ ++ BUG_ON(offset > root->stores_size); ++ ++ return item; ++} ++ ++/* ++ * Blindly insert the item to the tree. Typically, we reuse the ++ * first empty store item. ++ */ ++int sradix_tree_enter(struct sradix_tree_root *root, void **item, int num) ++{ ++ unsigned long index; ++ unsigned int height; ++ struct sradix_tree_node *node, *tmp = NULL; ++ int offset, offset_saved; ++ void **store = NULL; ++ int error, i, j, shift; ++ ++go_on: ++ index = root->min; ++ ++ if (root->enter_node && !sradix_node_full(root, root->enter_node)) { ++ node = root->enter_node; ++ BUG_ON((index >> (root->shift * root->height))); ++ } else { ++ node = root->rnode; ++ if (node == NULL || (index >> (root->shift * root->height)) ++ || sradix_node_full(root, node)) { ++ error = sradix_tree_extend(root, index); ++ if (error) ++ return error; ++ ++ node = root->rnode; ++ } ++ } ++ ++ ++ height = node->height; ++ shift = (height - 1) * root->shift; ++ offset = (index >> shift) & root->mask; ++ while (shift > 0) { ++ offset_saved = offset; ++ for (; offset < root->stores_size; offset++) { ++ store = &node->stores[offset]; ++ tmp = *store; ++ ++ if (!tmp || !sradix_node_full(root, tmp)) ++ break; ++ } ++ BUG_ON(offset >= root->stores_size); ++ ++ if (offset != offset_saved) { ++ index += (offset - offset_saved) << shift; ++ index &= ~((1UL << shift) - 1); ++ } ++ ++ if (!tmp) { ++ if (!(tmp = root->alloc())) ++ return -ENOMEM; ++ ++ tmp->height = shift / root->shift; ++ *store = tmp; ++ tmp->parent = node; ++ node->count++; ++// if (root->extend) ++// root->extend(node, tmp); ++ } ++ ++ node = tmp; ++ shift -= root->shift; ++ offset = (index >> shift) & root->mask; ++ } ++ ++ BUG_ON(node->height != 1); ++ ++ ++ store = &node->stores[offset]; ++ for (i = 0, j = 0; ++ j < root->stores_size - node->count && ++ i < root->stores_size - offset && j < num; i++) { ++ if (!store[i]) { ++ store[i] = item[j]; ++ if (root->assign) ++ root->assign(node, index + i, item[j]); ++ j++; ++ } ++ } ++ ++ node->count += j; ++ root->num += j; ++ num -= j; ++ ++ while (sradix_node_full(root, node)) { ++ node = node->parent; ++ if (!node) ++ break; ++ ++ node->fulls++; ++ } ++ ++ if (unlikely(!node)) { ++ /* All nodes are full */ ++ root->min = 1 << (root->height * root->shift); ++ root->enter_node = NULL; ++ } else { ++ root->min = index + i - 1; ++ root->min |= (1UL << (node->height - 1)) - 1; ++ root->min++; ++ root->enter_node = node; ++ } ++ ++ if (num) { ++ item += j; ++ goto go_on; ++ } ++ ++ return 0; ++} ++ ++ ++/** ++ * sradix_tree_shrink - shrink height of a sradix tree to minimal ++ * @root sradix tree root ++ * ++ */ ++static inline void sradix_tree_shrink(struct sradix_tree_root *root) ++{ ++ /* try to shrink tree height */ ++ while (root->height > 1) { ++ struct sradix_tree_node *to_free = root->rnode; ++ ++ /* ++ * The candidate node has more than one child, or its child ++ * is not at the leftmost store, we cannot shrink. ++ */ ++ if (to_free->count != 1 || !to_free->stores[0]) ++ break; ++ ++ root->rnode = to_free->stores[0]; ++ root->rnode->parent = NULL; ++ root->height--; ++ if (unlikely(root->enter_node == to_free)) ++ root->enter_node = NULL; ++ root->free(to_free); ++ } ++} ++ ++/* ++ * Del the item on the known leaf node and index ++ */ ++void sradix_tree_delete_from_leaf(struct sradix_tree_root *root, ++ struct sradix_tree_node *node, unsigned long index) ++{ ++ unsigned int offset; ++ struct sradix_tree_node *start, *end; ++ ++ BUG_ON(node->height != 1); ++ ++ start = node; ++ while (node && !(--node->count)) ++ node = node->parent; ++ ++ end = node; ++ if (!node) { ++ root->rnode = NULL; ++ root->height = 0; ++ root->min = 0; ++ root->num = 0; ++ root->enter_node = NULL; ++ } else { ++ offset = (index >> (root->shift * (node->height - 1))) & root->mask; ++ if (root->rm) ++ root->rm(node, offset); ++ node->stores[offset] = NULL; ++ root->num--; ++ if (root->min > index) { ++ root->min = index; ++ root->enter_node = node; ++ } ++ } ++ ++ if (start != end) { ++ do { ++ node = start; ++ start = start->parent; ++ if (unlikely(root->enter_node == node)) ++ root->enter_node = end; ++ root->free(node); ++ } while (start != end); ++ ++ /* ++ * Note that shrink may free "end", so enter_node still need to ++ * be checked inside. ++ */ ++ sradix_tree_shrink(root); ++ } else if (node->count == root->stores_size - 1) { ++ /* It WAS a full leaf node. Update the ancestors */ ++ node = node->parent; ++ while (node) { ++ node->fulls--; ++ if (node->fulls != root->stores_size - 1) ++ break; ++ ++ node = node->parent; ++ } ++ } ++} ++ ++void *sradix_tree_lookup(struct sradix_tree_root *root, unsigned long index) ++{ ++ unsigned int height, offset; ++ struct sradix_tree_node *node; ++ int shift; ++ ++ node = root->rnode; ++ if (node == NULL || (index >> (root->shift * root->height))) ++ return NULL; ++ ++ height = root->height; ++ shift = (height - 1) * root->shift; ++ ++ do { ++ offset = (index >> shift) & root->mask; ++ node = node->stores[offset]; ++ if (!node) ++ return NULL; ++ ++ shift -= root->shift; ++ } while (shift >= 0); ++ ++ return node; ++} ++ ++/* ++ * Return the item if it exists, otherwise create it in place ++ * and return the created item. ++ */ ++void *sradix_tree_lookup_create(struct sradix_tree_root *root, ++ unsigned long index, void *(*item_alloc)(void)) ++{ ++ unsigned int height, offset; ++ struct sradix_tree_node *node, *tmp; ++ void *item; ++ int shift, error; ++ ++ if (root->rnode == NULL || (index >> (root->shift * root->height))) { ++ if (item_alloc) { ++ error = sradix_tree_extend(root, index); ++ if (error) ++ return NULL; ++ } else { ++ return NULL; ++ } ++ } ++ ++ node = root->rnode; ++ height = root->height; ++ shift = (height - 1) * root->shift; ++ ++ do { ++ offset = (index >> shift) & root->mask; ++ if (!node->stores[offset]) { ++ if (!(tmp = root->alloc())) ++ return NULL; ++ ++ tmp->height = shift / root->shift; ++ node->stores[offset] = tmp; ++ tmp->parent = node; ++ node->count++; ++ node = tmp; ++ } else { ++ node = node->stores[offset]; ++ } ++ ++ shift -= root->shift; ++ } while (shift > 0); ++ ++ BUG_ON(node->height != 1); ++ offset = index & root->mask; ++ if (node->stores[offset]) { ++ return node->stores[offset]; ++ } else if (item_alloc) { ++ if (!(item = item_alloc())) ++ return NULL; ++ ++ node->stores[offset] = item; ++ ++ /* ++ * NOTE: we do NOT call root->assign here, since this item is ++ * newly created by us having no meaning. Caller can call this ++ * if it's necessary to do so. ++ */ ++ ++ node->count++; ++ root->num++; ++ ++ while (sradix_node_full(root, node)) { ++ node = node->parent; ++ if (!node) ++ break; ++ ++ node->fulls++; ++ } ++ ++ if (unlikely(!node)) { ++ /* All nodes are full */ ++ root->min = 1 << (root->height * root->shift); ++ } else { ++ if (root->min == index) { ++ root->min |= (1UL << (node->height - 1)) - 1; ++ root->min++; ++ root->enter_node = node; ++ } ++ } ++ ++ return item; ++ } else { ++ return NULL; ++ } ++ ++} ++ ++int sradix_tree_delete(struct sradix_tree_root *root, unsigned long index) ++{ ++ unsigned int height, offset; ++ struct sradix_tree_node *node; ++ int shift; ++ ++ node = root->rnode; ++ if (node == NULL || (index >> (root->shift * root->height))) ++ return -ENOENT; ++ ++ height = root->height; ++ shift = (height - 1) * root->shift; ++ ++ do { ++ offset = (index >> shift) & root->mask; ++ node = node->stores[offset]; ++ if (!node) ++ return -ENOENT; ++ ++ shift -= root->shift; ++ } while (shift > 0); ++ ++ offset = index & root->mask; ++ if (!node->stores[offset]) ++ return -ENOENT; ++ ++ sradix_tree_delete_from_leaf(root, node, index); ++ ++ return 0; ++} +diff -Nur a/mm/Kconfig b/mm/Kconfig +--- a/mm/Kconfig 2018-05-26 19:24:34.846783391 +0100 ++++ b/mm/Kconfig 2018-05-26 19:30:55.786140408 +0100 +@@ -315,6 +315,32 @@ + See Documentation/vm/ksm.txt for more information: KSM is inactive + until a program has madvised that an area is MADV_MERGEABLE, and + root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set). ++choice ++ prompt "Choose UKSM/KSM strategy" ++ default UKSM ++ depends on KSM ++ help ++ This option allows to select a UKSM/KSM stragety. ++ ++config UKSM ++ bool "Ultra-KSM for page merging" ++ depends on KSM ++ help ++ UKSM is inspired by the Linux kernel project \u2014 KSM(Kernel Same ++ page Merging), but with a fundamentally rewritten core algorithm. With ++ an advanced algorithm, UKSM now can transparently scans all anonymously ++ mapped user space applications with an significantly improved scan speed ++ and CPU efficiency. Since KVM is friendly to KSM, KVM can also benefit from ++ UKSM. Now UKSM has its first stable release and first real world enterprise user. ++ For more information, please goto its project page. ++ (www.kerneldedup.org) ++ ++config KSM_LEGACY ++ bool "Legacy KSM implementation" ++ depends on KSM ++ help ++ The legacy KSM implementation from Red Hat. ++endchoice + + config DEFAULT_MMAP_MIN_ADDR + int "Low address space to protect from user allocation" +diff -Nur a/mm/Makefile b/mm/Makefile +--- a/mm/Makefile 2018-05-25 15:18:02.000000000 +0100 ++++ b/mm/Makefile 2018-05-26 19:30:55.786140408 +0100 +@@ -65,7 +65,8 @@ + obj-$(CONFIG_SPARSEMEM_VMEMMAP) += sparse-vmemmap.o + obj-$(CONFIG_SLOB) += slob.o + obj-$(CONFIG_MMU_NOTIFIER) += mmu_notifier.o +-obj-$(CONFIG_KSM) += ksm.o ++obj-$(CONFIG_KSM_LEGACY) += ksm.o ++obj-$(CONFIG_UKSM) += uksm.o + obj-$(CONFIG_PAGE_POISONING) += page_poison.o + obj-$(CONFIG_SLAB) += slab.o + obj-$(CONFIG_SLUB) += slub.o +diff -Nur a/mm/memory.c b/mm/memory.c +--- a/mm/memory.c 2018-05-25 15:18:02.000000000 +0100 ++++ b/mm/memory.c 2018-05-26 19:30:55.787140441 +0100 +@@ -129,6 +129,25 @@ + + unsigned long highest_memmap_pfn __read_mostly; + ++#ifdef CONFIG_UKSM ++unsigned long uksm_zero_pfn __read_mostly; ++EXPORT_SYMBOL_GPL(uksm_zero_pfn); ++struct page *empty_uksm_zero_page; ++ ++static int __init setup_uksm_zero_page(void) ++{ ++ empty_uksm_zero_page = alloc_pages(__GFP_ZERO & ~__GFP_MOVABLE, 0); ++ if (!empty_uksm_zero_page) ++ panic("Oh boy, that early out of memory?"); ++ ++ SetPageReserved(empty_uksm_zero_page); ++ uksm_zero_pfn = page_to_pfn(empty_uksm_zero_page); ++ ++ return 0; ++} ++core_initcall(setup_uksm_zero_page); ++#endif ++ + /* + * CONFIG_MMU architectures set up ZERO_PAGE in their paging_init() + */ +@@ -140,6 +159,7 @@ + core_initcall(init_zero_pfn); + + ++ + #if defined(SPLIT_RSS_COUNTING) + + void sync_mm_rss(struct mm_struct *mm) +@@ -1035,6 +1055,9 @@ + get_page(page); + page_dup_rmap(page, false); + rss[mm_counter(page)]++; ++ ++ /* Should return NULL in vm_normal_page() */ ++ uksm_bugon_zeropage(pte); + } else if (pte_devmap(pte)) { + page = pte_page(pte); + +@@ -1048,6 +1071,8 @@ + page_dup_rmap(page, false); + rss[mm_counter(page)]++; + } ++ } else { ++ uksm_map_zero_page(pte); + } + + out_set_pte: +@@ -1317,8 +1342,10 @@ + ptent = ptep_get_and_clear_full(mm, addr, pte, + tlb->fullmm); + tlb_remove_tlb_entry(tlb, pte, addr); +- if (unlikely(!page)) ++ if (unlikely(!page)) { ++ uksm_unmap_zero_page(ptent); + continue; ++ } + + if (!PageAnon(page)) { + if (pte_dirty(ptent)) { +@@ -2318,8 +2345,10 @@ + clear_page(kaddr); + kunmap_atomic(kaddr); + flush_dcache_page(dst); +- } else ++ } else { + copy_user_highpage(dst, src, va, vma); ++ uksm_cow_page(vma, src); ++ } + } + + static gfp_t __get_fault_gfp_mask(struct vm_area_struct *vma) +@@ -2468,6 +2497,7 @@ + vmf->address); + if (!new_page) + goto oom; ++ uksm_cow_pte(vma, vmf->orig_pte); + } else { + new_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, + vmf->address); +@@ -2494,7 +2524,9 @@ + mm_counter_file(old_page)); + inc_mm_counter_fast(mm, MM_ANONPAGES); + } ++ uksm_bugon_zeropage(vmf->orig_pte); + } else { ++ uksm_unmap_zero_page(vmf->orig_pte); + inc_mm_counter_fast(mm, MM_ANONPAGES); + } + flush_cache_page(vma, vmf->address, pte_pfn(vmf->orig_pte)); +diff -Nur a/mm/mmap.c b/mm/mmap.c +--- a/mm/mmap.c 2018-05-26 19:24:34.847783423 +0100 ++++ b/mm/mmap.c 2018-05-26 19:30:55.788140473 +0100 +@@ -45,6 +45,7 @@ + #include + #include + #include ++#include + + #include + #include +@@ -173,6 +174,7 @@ + if (vma->vm_file) + fput(vma->vm_file); + mpol_put(vma_policy(vma)); ++ uksm_remove_vma(vma); + kmem_cache_free(vm_area_cachep, vma); + return next; + } +@@ -699,9 +701,16 @@ + long adjust_next = 0; + int remove_next = 0; + ++/* ++ * to avoid deadlock, ksm_remove_vma must be done before any spin_lock is ++ * acquired ++ */ ++ uksm_remove_vma(vma); ++ + if (next && !insert) { + struct vm_area_struct *exporter = NULL, *importer = NULL; + ++ uksm_remove_vma(next); + if (end >= next->vm_end) { + /* + * vma expands, overlapping all the next, and +@@ -834,6 +843,7 @@ + end_changed = true; + } + vma->vm_pgoff = pgoff; ++ + if (adjust_next) { + next->vm_start += adjust_next << PAGE_SHIFT; + next->vm_pgoff += adjust_next; +@@ -939,6 +949,7 @@ + if (remove_next == 2) { + remove_next = 1; + end = next->vm_end; ++ uksm_remove_vma(next); + goto again; + } + else if (next) +@@ -965,10 +976,14 @@ + */ + VM_WARN_ON(mm->highest_vm_end != vm_end_gap(vma)); + } ++ } else { ++ if (next && !insert) ++ uksm_vma_add_new(next); + } + if (insert && file) + uprobe_mmap(insert); + ++ uksm_vma_add_new(vma); + validate_mm(mm); + + return 0; +@@ -1385,6 +1400,9 @@ + vm_flags |= calc_vm_prot_bits(prot, pkey) | calc_vm_flag_bits(flags) | + mm->def_flags | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC; + ++ /* If uksm is enabled, we add VM_MERGEABLE to new VMAs. */ ++ uksm_vm_flags_mod(&vm_flags); ++ + if (flags & MAP_LOCKED) + if (!can_do_mlock()) + return -EPERM; +@@ -1724,6 +1742,7 @@ + allow_write_access(file); + } + file = vma->vm_file; ++ uksm_vma_add_new(vma); + out: + perf_event_mmap(vma); + +@@ -1765,6 +1784,7 @@ + if (vm_flags & VM_DENYWRITE) + allow_write_access(file); + free_vma: ++ uksm_remove_vma(vma); + kmem_cache_free(vm_area_cachep, vma); + unacct_error: + if (charged) +@@ -2589,6 +2609,8 @@ + else + err = vma_adjust(vma, vma->vm_start, addr, vma->vm_pgoff, new); + ++ uksm_vma_add_new(new); ++ + /* Success. */ + if (!err) + return 0; +@@ -2881,6 +2903,7 @@ + if ((flags & (~VM_EXEC)) != 0) + return -EINVAL; + flags |= VM_DATA_DEFAULT_FLAGS | VM_ACCOUNT | mm->def_flags; ++ uksm_vm_flags_mod(&flags); + + error = get_unmapped_area(NULL, addr, len, 0, MAP_FIXED); + if (offset_in_page(error)) +@@ -2938,6 +2961,7 @@ + vma->vm_flags = flags; + vma->vm_page_prot = vm_get_page_prot(flags); + vma_link(mm, vma, prev, rb_link, rb_parent); ++ uksm_vma_add_new(vma); + out: + perf_event_mmap(vma); + mm->total_vm += len >> PAGE_SHIFT; +@@ -3015,6 +3039,12 @@ + up_write(&mm->mmap_sem); + } + ++ /* ++ * Taking write lock on mmap_sem does not harm others, ++ * but it's crucial for uksm to avoid races. ++ */ ++ down_write(&mm->mmap_sem); ++ + if (mm->locked_vm) { + vma = mm->mmap; + while (vma) { +@@ -3049,6 +3079,11 @@ + vma = remove_vma(vma); + } + vm_unacct_memory(nr_accounted); ++ ++ mm->mmap = NULL; ++ mm->mm_rb = RB_ROOT; ++ vmacache_invalidate(mm); ++ up_write(&mm->mmap_sem); + } + + /* Insert vm structure into process list sorted by address +@@ -3158,6 +3193,7 @@ + new_vma->vm_ops->open(new_vma); + vma_link(mm, new_vma, prev, rb_link, rb_parent); + *need_rmap_locks = false; ++ uksm_vma_add_new(new_vma); + } + return new_vma; + +@@ -3308,6 +3344,7 @@ + vm_stat_account(mm, vma->vm_flags, len >> PAGE_SHIFT); + + perf_event_mmap(vma); ++ uksm_vma_add_new(vma); + + return vma; + +diff -Nur a/mm/rmap.c b/mm/rmap.c +--- a/mm/rmap.c 2018-05-25 15:18:02.000000000 +0100 ++++ b/mm/rmap.c 2018-05-26 19:30:55.788140473 +0100 +@@ -1013,9 +1013,9 @@ + + /** + * __page_set_anon_rmap - set up new anonymous rmap +- * @page: Page to add to rmap ++ * @page: Page to add to rmap + * @vma: VM area to add page to. +- * @address: User virtual address of the mapping ++ * @address: User virtual address of the mapping + * @exclusive: the page is exclusively owned by the current process + */ + static void __page_set_anon_rmap(struct page *page, +diff -Nur a/mm/uksm.c b/mm/uksm.c +--- a/mm/uksm.c 1970-01-01 01:00:00.000000000 +0100 ++++ b/mm/uksm.c 2018-05-26 19:30:55.791140570 +0100 +@@ -0,0 +1,5584 @@ ++/* ++ * Ultra KSM. Copyright (C) 2011-2012 Nai Xia ++ * ++ * This is an improvement upon KSM. Some basic data structures and routines ++ * are borrowed from ksm.c . ++ * ++ * Its new features: ++ * 1. Full system scan: ++ * It automatically scans all user processes' anonymous VMAs. Kernel-user ++ * interaction to submit a memory area to KSM is no longer needed. ++ * ++ * 2. Rich area detection: ++ * It automatically detects rich areas containing abundant duplicated ++ * pages based. Rich areas are given a full scan speed. Poor areas are ++ * sampled at a reasonable speed with very low CPU consumption. ++ * ++ * 3. Ultra Per-page scan speed improvement: ++ * A new hash algorithm is proposed. As a result, on a machine with ++ * Core(TM)2 Quad Q9300 CPU in 32-bit mode and 800MHZ DDR2 main memory, it ++ * can scan memory areas that does not contain duplicated pages at speed of ++ * 627MB/sec ~ 2445MB/sec and can merge duplicated areas at speed of ++ * 477MB/sec ~ 923MB/sec. ++ * ++ * 4. Thrashing area avoidance: ++ * Thrashing area(an VMA that has frequent Ksm page break-out) can be ++ * filtered out. My benchmark shows it's more efficient than KSM's per-page ++ * hash value based volatile page detection. ++ * ++ * ++ * 5. Misc changes upon KSM: ++ * * It has a fully x86-opitmized memcmp dedicated for 4-byte-aligned page ++ * comparison. It's much faster than default C version on x86. ++ * * rmap_item now has an struct *page member to loosely cache a ++ * address-->page mapping, which reduces too much time-costly ++ * follow_page(). ++ * * The VMA creation/exit procedures are hooked to let the Ultra KSM know. ++ * * try_to_merge_two_pages() now can revert a pte if it fails. No break_ ++ * ksm is needed for this case. ++ * ++ * 6. Full Zero Page consideration(contributed by Figo Zhang) ++ * Now uksmd consider full zero pages as special pages and merge them to an ++ * special unswappable uksm zero page. ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++#include "internal.h" ++ ++#ifdef CONFIG_X86 ++#undef memcmp ++ ++#ifdef CONFIG_X86_32 ++#define memcmp memcmpx86_32 ++/* ++ * Compare 4-byte-aligned address s1 and s2, with length n ++ */ ++int memcmpx86_32(void *s1, void *s2, size_t n) ++{ ++ size_t num = n / 4; ++ register int res; ++ ++ __asm__ __volatile__ ++ ( ++ "testl %3,%3\n\t" ++ "repe; cmpsd\n\t" ++ "je 1f\n\t" ++ "sbbl %0,%0\n\t" ++ "orl $1,%0\n" ++ "1:" ++ : "=&a" (res), "+&S" (s1), "+&D" (s2), "+&c" (num) ++ : "0" (0) ++ : "cc"); ++ ++ return res; ++} ++ ++/* ++ * Check the page is all zero ? ++ */ ++static int is_full_zero(const void *s1, size_t len) ++{ ++ unsigned char same; ++ ++ len /= 4; ++ ++ __asm__ __volatile__ ++ ("repe; scasl;" ++ "sete %0" ++ : "=qm" (same), "+D" (s1), "+c" (len) ++ : "a" (0) ++ : "cc"); ++ ++ return same; ++} ++ ++ ++#elif defined(CONFIG_X86_64) ++#define memcmp memcmpx86_64 ++/* ++ * Compare 8-byte-aligned address s1 and s2, with length n ++ */ ++int memcmpx86_64(void *s1, void *s2, size_t n) ++{ ++ size_t num = n / 8; ++ register int res; ++ ++ __asm__ __volatile__ ++ ( ++ "testq %q3,%q3\n\t" ++ "repe; cmpsq\n\t" ++ "je 1f\n\t" ++ "sbbq %q0,%q0\n\t" ++ "orq $1,%q0\n" ++ "1:" ++ : "=&a" (res), "+&S" (s1), "+&D" (s2), "+&c" (num) ++ : "0" (0) ++ : "cc"); ++ ++ return res; ++} ++ ++static int is_full_zero(const void *s1, size_t len) ++{ ++ unsigned char same; ++ ++ len /= 8; ++ ++ __asm__ __volatile__ ++ ("repe; scasq;" ++ "sete %0" ++ : "=qm" (same), "+D" (s1), "+c" (len) ++ : "a" (0) ++ : "cc"); ++ ++ return same; ++} ++ ++#endif ++#else ++static int is_full_zero(const void *s1, size_t len) ++{ ++ unsigned long *src = s1; ++ int i; ++ ++ len /= sizeof(*src); ++ ++ for (i = 0; i < len; i++) { ++ if (src[i]) ++ return 0; ++ } ++ ++ return 1; ++} ++#endif ++ ++#define UKSM_RUNG_ROUND_FINISHED (1 << 0) ++#define TIME_RATIO_SCALE 10000 ++ ++#define SLOT_TREE_NODE_SHIFT 8 ++#define SLOT_TREE_NODE_STORE_SIZE (1UL << SLOT_TREE_NODE_SHIFT) ++struct slot_tree_node { ++ unsigned long size; ++ struct sradix_tree_node snode; ++ void *stores[SLOT_TREE_NODE_STORE_SIZE]; ++}; ++ ++static struct kmem_cache *slot_tree_node_cachep; ++ ++static struct sradix_tree_node *slot_tree_node_alloc(void) ++{ ++ struct slot_tree_node *p; ++ ++ p = kmem_cache_zalloc(slot_tree_node_cachep, GFP_KERNEL | ++ __GFP_NORETRY | __GFP_NOWARN); ++ if (!p) ++ return NULL; ++ ++ return &p->snode; ++} ++ ++static void slot_tree_node_free(struct sradix_tree_node *node) ++{ ++ struct slot_tree_node *p; ++ ++ p = container_of(node, struct slot_tree_node, snode); ++ kmem_cache_free(slot_tree_node_cachep, p); ++} ++ ++static void slot_tree_node_extend(struct sradix_tree_node *parent, ++ struct sradix_tree_node *child) ++{ ++ struct slot_tree_node *p, *c; ++ ++ p = container_of(parent, struct slot_tree_node, snode); ++ c = container_of(child, struct slot_tree_node, snode); ++ ++ p->size += c->size; ++} ++ ++void slot_tree_node_assign(struct sradix_tree_node *node, ++ unsigned int index, void *item) ++{ ++ struct vma_slot *slot = item; ++ struct slot_tree_node *cur; ++ ++ slot->snode = node; ++ slot->sindex = index; ++ ++ while (node) { ++ cur = container_of(node, struct slot_tree_node, snode); ++ cur->size += slot->pages; ++ node = node->parent; ++ } ++} ++ ++void slot_tree_node_rm(struct sradix_tree_node *node, unsigned int offset) ++{ ++ struct vma_slot *slot; ++ struct slot_tree_node *cur; ++ unsigned long pages; ++ ++ if (node->height == 1) { ++ slot = node->stores[offset]; ++ pages = slot->pages; ++ } else { ++ cur = container_of(node->stores[offset], ++ struct slot_tree_node, snode); ++ pages = cur->size; ++ } ++ ++ while (node) { ++ cur = container_of(node, struct slot_tree_node, snode); ++ cur->size -= pages; ++ node = node->parent; ++ } ++} ++ ++unsigned long slot_iter_index; ++int slot_iter(void *item, unsigned long height) ++{ ++ struct slot_tree_node *node; ++ struct vma_slot *slot; ++ ++ if (height == 1) { ++ slot = item; ++ if (slot_iter_index < slot->pages) { ++ /*in this one*/ ++ return 1; ++ } else { ++ slot_iter_index -= slot->pages; ++ return 0; ++ } ++ ++ } else { ++ node = container_of(item, struct slot_tree_node, snode); ++ if (slot_iter_index < node->size) { ++ /*in this one*/ ++ return 1; ++ } else { ++ slot_iter_index -= node->size; ++ return 0; ++ } ++ } ++} ++ ++ ++static inline void slot_tree_init_root(struct sradix_tree_root *root) ++{ ++ init_sradix_tree_root(root, SLOT_TREE_NODE_SHIFT); ++ root->alloc = slot_tree_node_alloc; ++ root->free = slot_tree_node_free; ++ root->extend = slot_tree_node_extend; ++ root->assign = slot_tree_node_assign; ++ root->rm = slot_tree_node_rm; ++} ++ ++void slot_tree_init(void) ++{ ++ slot_tree_node_cachep = kmem_cache_create("slot_tree_node", ++ sizeof(struct slot_tree_node), 0, ++ SLAB_PANIC | SLAB_RECLAIM_ACCOUNT, ++ NULL); ++} ++ ++ ++/* Each rung of this ladder is a list of VMAs having a same scan ratio */ ++struct scan_rung { ++ //struct list_head scanned_list; ++ struct sradix_tree_root vma_root; ++ struct sradix_tree_root vma_root2; ++ ++ struct vma_slot *current_scan; ++ unsigned long current_offset; ++ ++ /* ++ * The initial value for current_offset, it should loop over ++ * [0~ step - 1] to let all slot have its chance to be scanned. ++ */ ++ unsigned long offset_init; ++ unsigned long step; /* dynamic step for current_offset */ ++ unsigned int flags; ++ unsigned long pages_to_scan; ++ //unsigned long fully_scanned_slots; ++ /* ++ * a little bit tricky - if cpu_time_ratio > 0, then the value is the ++ * the cpu time ratio it can spend in rung_i for every scan ++ * period. if < 0, then it is the cpu time ratio relative to the ++ * max cpu percentage user specified. Both in unit of ++ * 1/TIME_RATIO_SCALE ++ */ ++ int cpu_ratio; ++ ++ /* ++ * How long it will take for all slots in this rung to be fully ++ * scanned? If it's zero, we don't care about the cover time: ++ * it's fully scanned. ++ */ ++ unsigned int cover_msecs; ++ //unsigned long vma_num; ++ //unsigned long pages; /* Sum of all slot's pages in rung */ ++}; ++ ++/** ++ * node of either the stable or unstale rbtree ++ * ++ */ ++struct tree_node { ++ struct rb_node node; /* link in the main (un)stable rbtree */ ++ struct rb_root sub_root; /* rb_root for sublevel collision rbtree */ ++ u32 hash; ++ unsigned long count; /* TODO: merged with sub_root */ ++ struct list_head all_list; /* all tree nodes in stable/unstable tree */ ++}; ++ ++/** ++ * struct stable_node - node of the stable rbtree ++ * @node: rb node of this ksm page in the stable tree ++ * @hlist: hlist head of rmap_items using this ksm page ++ * @kpfn: page frame number of this ksm page ++ */ ++struct stable_node { ++ struct rb_node node; /* link in sub-rbtree */ ++ struct tree_node *tree_node; /* it's tree node root in stable tree, NULL if it's in hell list */ ++ struct hlist_head hlist; ++ unsigned long kpfn; ++ u32 hash_max; /* if ==0 then it's not been calculated yet */ ++ struct list_head all_list; /* in a list for all stable nodes */ ++}; ++ ++/** ++ * struct node_vma - group rmap_items linked in a same stable ++ * node together. ++ */ ++struct node_vma { ++ union { ++ struct vma_slot *slot; ++ unsigned long key; /* slot is used as key sorted on hlist */ ++ }; ++ struct hlist_node hlist; ++ struct hlist_head rmap_hlist; ++ struct stable_node *head; ++}; ++ ++/** ++ * struct rmap_item - reverse mapping item for virtual addresses ++ * @rmap_list: next rmap_item in mm_slot's singly-linked rmap_list ++ * @anon_vma: pointer to anon_vma for this mm,address, when in stable tree ++ * @mm: the memory structure this rmap_item is pointing into ++ * @address: the virtual address this rmap_item tracks (+ flags in low bits) ++ * @node: rb node of this rmap_item in the unstable tree ++ * @head: pointer to stable_node heading this list in the stable tree ++ * @hlist: link into hlist of rmap_items hanging off that stable_node ++ */ ++struct rmap_item { ++ struct vma_slot *slot; ++ struct page *page; ++ unsigned long address; /* + low bits used for flags below */ ++ unsigned long hash_round; ++ unsigned long entry_index; ++ union { ++ struct {/* when in unstable tree */ ++ struct rb_node node; ++ struct tree_node *tree_node; ++ u32 hash_max; ++ }; ++ struct { /* when in stable tree */ ++ struct node_vma *head; ++ struct hlist_node hlist; ++ struct anon_vma *anon_vma; ++ }; ++ }; ++} __aligned(4); ++ ++struct rmap_list_entry { ++ union { ++ struct rmap_item *item; ++ unsigned long addr; ++ }; ++ /* lowest bit is used for is_addr tag */ ++} __aligned(4); /* 4 aligned to fit in to pages*/ ++ ++ ++/* Basic data structure definition ends */ ++ ++ ++/* ++ * Flags for rmap_item to judge if it's listed in the stable/unstable tree. ++ * The flags use the low bits of rmap_item.address ++ */ ++#define UNSTABLE_FLAG 0x1 ++#define STABLE_FLAG 0x2 ++#define get_rmap_addr(x) ((x)->address & PAGE_MASK) ++ ++/* ++ * rmap_list_entry helpers ++ */ ++#define IS_ADDR_FLAG 1 ++#define is_addr(ptr) ((unsigned long)(ptr) & IS_ADDR_FLAG) ++#define set_is_addr(ptr) ((ptr) |= IS_ADDR_FLAG) ++#define get_clean_addr(ptr) (((ptr) & ~(__typeof__(ptr))IS_ADDR_FLAG)) ++ ++ ++/* ++ * High speed caches for frequently allocated and freed structs ++ */ ++static struct kmem_cache *rmap_item_cache; ++static struct kmem_cache *stable_node_cache; ++static struct kmem_cache *node_vma_cache; ++static struct kmem_cache *vma_slot_cache; ++static struct kmem_cache *tree_node_cache; ++#define UKSM_KMEM_CACHE(__struct, __flags) kmem_cache_create("uksm_"#__struct,\ ++ sizeof(struct __struct), __alignof__(struct __struct),\ ++ (__flags), NULL) ++ ++/* Array of all scan_rung, uksm_scan_ladder[0] having the minimum scan ratio */ ++#define SCAN_LADDER_SIZE 4 ++static struct scan_rung uksm_scan_ladder[SCAN_LADDER_SIZE]; ++ ++/* The evaluation rounds uksmd has finished */ ++static unsigned long long uksm_eval_round = 1; ++ ++/* ++ * we add 1 to this var when we consider we should rebuild the whole ++ * unstable tree. ++ */ ++static unsigned long uksm_hash_round = 1; ++ ++/* ++ * How many times the whole memory is scanned. ++ */ ++static unsigned long long fully_scanned_round = 1; ++ ++/* The total number of virtual pages of all vma slots */ ++static u64 uksm_pages_total; ++ ++/* The number of pages has been scanned since the start up */ ++static u64 uksm_pages_scanned; ++ ++static u64 scanned_virtual_pages; ++ ++/* The number of pages has been scanned since last encode_benefit call */ ++static u64 uksm_pages_scanned_last; ++ ++/* If the scanned number is tooo large, we encode it here */ ++static u64 pages_scanned_stored; ++ ++static unsigned long pages_scanned_base; ++ ++/* The number of nodes in the stable tree */ ++static unsigned long uksm_pages_shared; ++ ++/* The number of page slots additionally sharing those nodes */ ++static unsigned long uksm_pages_sharing; ++ ++/* The number of nodes in the unstable tree */ ++static unsigned long uksm_pages_unshared; ++ ++/* ++ * Milliseconds ksmd should sleep between scans, ++ * >= 100ms to be consistent with ++ * scan_time_to_sleep_msec() ++ */ ++static unsigned int uksm_sleep_jiffies; ++ ++/* The real value for the uksmd next sleep */ ++static unsigned int uksm_sleep_real; ++ ++/* Saved value for user input uksm_sleep_jiffies when it's enlarged */ ++static unsigned int uksm_sleep_saved; ++ ++/* Max percentage of cpu utilization ksmd can take to scan in one batch */ ++static unsigned int uksm_max_cpu_percentage; ++ ++static int uksm_cpu_governor; ++ ++static char *uksm_cpu_governor_str[4] = { "full", "medium", "low", "quiet" }; ++ ++struct uksm_cpu_preset_s { ++ int cpu_ratio[SCAN_LADDER_SIZE]; ++ unsigned int cover_msecs[SCAN_LADDER_SIZE]; ++ unsigned int max_cpu; /* percentage */ ++}; ++ ++struct uksm_cpu_preset_s uksm_cpu_preset[4] = { ++ { {20, 40, -2500, -10000}, {1000, 500, 200, 50}, 95}, ++ { {20, 30, -2500, -10000}, {1000, 500, 400, 100}, 50}, ++ { {10, 20, -5000, -10000}, {1500, 1000, 1000, 250}, 20}, ++ { {10, 20, 40, 75}, {2000, 1000, 1000, 1000}, 1}, ++}; ++ ++/* The default value for uksm_ema_page_time if it's not initialized */ ++#define UKSM_PAGE_TIME_DEFAULT 500 ++ ++/*cost to scan one page by expotional moving average in nsecs */ ++static unsigned long uksm_ema_page_time = UKSM_PAGE_TIME_DEFAULT; ++ ++/* The expotional moving average alpha weight, in percentage. */ ++#define EMA_ALPHA 20 ++ ++/* ++ * The threshold used to filter out thrashing areas, ++ * If it == 0, filtering is disabled, otherwise it's the percentage up-bound ++ * of the thrashing ratio of all areas. Any area with a bigger thrashing ratio ++ * will be considered as having a zero duplication ratio. ++ */ ++static unsigned int uksm_thrash_threshold = 50; ++ ++/* How much dedup ratio is considered to be abundant*/ ++static unsigned int uksm_abundant_threshold = 10; ++ ++/* All slots having merged pages in this eval round. */ ++struct list_head vma_slot_dedup = LIST_HEAD_INIT(vma_slot_dedup); ++ ++/* How many times the ksmd has slept since startup */ ++static unsigned long long uksm_sleep_times; ++ ++#define UKSM_RUN_STOP 0 ++#define UKSM_RUN_MERGE 1 ++static unsigned int uksm_run = 1; ++ ++static DECLARE_WAIT_QUEUE_HEAD(uksm_thread_wait); ++static DEFINE_MUTEX(uksm_thread_mutex); ++ ++/* ++ * List vma_slot_new is for newly created vma_slot waiting to be added by ++ * ksmd. If one cannot be added(e.g. due to it's too small), it's moved to ++ * vma_slot_noadd. vma_slot_del is the list for vma_slot whose corresponding ++ * VMA has been removed/freed. ++ */ ++struct list_head vma_slot_new = LIST_HEAD_INIT(vma_slot_new); ++struct list_head vma_slot_noadd = LIST_HEAD_INIT(vma_slot_noadd); ++struct list_head vma_slot_del = LIST_HEAD_INIT(vma_slot_del); ++static DEFINE_SPINLOCK(vma_slot_list_lock); ++ ++/* The unstable tree heads */ ++static struct rb_root root_unstable_tree = RB_ROOT; ++ ++/* ++ * All tree_nodes are in a list to be freed at once when unstable tree is ++ * freed after each scan round. ++ */ ++static struct list_head unstable_tree_node_list = ++ LIST_HEAD_INIT(unstable_tree_node_list); ++ ++/* List contains all stable nodes */ ++static struct list_head stable_node_list = LIST_HEAD_INIT(stable_node_list); ++ ++/* ++ * When the hash strength is changed, the stable tree must be delta_hashed and ++ * re-structured. We use two set of below structs to speed up the ++ * re-structuring of stable tree. ++ */ ++static struct list_head ++stable_tree_node_list[2] = {LIST_HEAD_INIT(stable_tree_node_list[0]), ++ LIST_HEAD_INIT(stable_tree_node_list[1])}; ++ ++static struct list_head *stable_tree_node_listp = &stable_tree_node_list[0]; ++static struct rb_root root_stable_tree[2] = {RB_ROOT, RB_ROOT}; ++static struct rb_root *root_stable_treep = &root_stable_tree[0]; ++static unsigned long stable_tree_index; ++ ++/* The hash strength needed to hash a full page */ ++#define HASH_STRENGTH_FULL (PAGE_SIZE / sizeof(u32)) ++ ++/* The hash strength needed for loop-back hashing */ ++#define HASH_STRENGTH_MAX (HASH_STRENGTH_FULL + 10) ++ ++/* The random offsets in a page */ ++static u32 *random_nums; ++ ++/* The hash strength */ ++static unsigned long hash_strength = HASH_STRENGTH_FULL >> 4; ++ ++/* The delta value each time the hash strength increases or decreases */ ++static unsigned long hash_strength_delta; ++#define HASH_STRENGTH_DELTA_MAX 5 ++ ++/* The time we have saved due to random_sample_hash */ ++static u64 rshash_pos; ++ ++/* The time we have wasted due to hash collision */ ++static u64 rshash_neg; ++ ++struct uksm_benefit { ++ u64 pos; ++ u64 neg; ++ u64 scanned; ++ unsigned long base; ++} benefit; ++ ++/* ++ * The relative cost of memcmp, compared to 1 time unit of random sample ++ * hash, this value is tested when ksm module is initialized ++ */ ++static unsigned long memcmp_cost; ++ ++static unsigned long rshash_neg_cont_zero; ++static unsigned long rshash_cont_obscure; ++ ++/* The possible states of hash strength adjustment heuristic */ ++enum rshash_states { ++ RSHASH_STILL, ++ RSHASH_TRYUP, ++ RSHASH_TRYDOWN, ++ RSHASH_NEW, ++ RSHASH_PRE_STILL, ++}; ++ ++/* The possible direction we are about to adjust hash strength */ ++enum rshash_direct { ++ GO_UP, ++ GO_DOWN, ++ OBSCURE, ++ STILL, ++}; ++ ++/* random sampling hash state machine */ ++static struct { ++ enum rshash_states state; ++ enum rshash_direct pre_direct; ++ u8 below_count; ++ /* Keep a lookup window of size 5, iff above_count/below_count > 3 ++ * in this window we stop trying. ++ */ ++ u8 lookup_window_index; ++ u64 stable_benefit; ++ unsigned long turn_point_down; ++ unsigned long turn_benefit_down; ++ unsigned long turn_point_up; ++ unsigned long turn_benefit_up; ++ unsigned long stable_point; ++} rshash_state; ++ ++/*zero page hash table, hash_strength [0 ~ HASH_STRENGTH_MAX]*/ ++static u32 *zero_hash_table; ++ ++static inline struct node_vma *alloc_node_vma(void) ++{ ++ struct node_vma *node_vma; ++ ++ node_vma = kmem_cache_zalloc(node_vma_cache, GFP_KERNEL | ++ __GFP_NORETRY | __GFP_NOWARN); ++ if (node_vma) { ++ INIT_HLIST_HEAD(&node_vma->rmap_hlist); ++ INIT_HLIST_NODE(&node_vma->hlist); ++ } ++ return node_vma; ++} ++ ++static inline void free_node_vma(struct node_vma *node_vma) ++{ ++ kmem_cache_free(node_vma_cache, node_vma); ++} ++ ++ ++static inline struct vma_slot *alloc_vma_slot(void) ++{ ++ struct vma_slot *slot; ++ ++ /* ++ * In case ksm is not initialized by now. ++ * Oops, we need to consider the call site of uksm_init() in the future. ++ */ ++ if (!vma_slot_cache) ++ return NULL; ++ ++ slot = kmem_cache_zalloc(vma_slot_cache, GFP_KERNEL | ++ __GFP_NORETRY | __GFP_NOWARN); ++ if (slot) { ++ INIT_LIST_HEAD(&slot->slot_list); ++ INIT_LIST_HEAD(&slot->dedup_list); ++ slot->flags |= UKSM_SLOT_NEED_RERAND; ++ } ++ return slot; ++} ++ ++static inline void free_vma_slot(struct vma_slot *vma_slot) ++{ ++ kmem_cache_free(vma_slot_cache, vma_slot); ++} ++ ++ ++ ++static inline struct rmap_item *alloc_rmap_item(void) ++{ ++ struct rmap_item *rmap_item; ++ ++ rmap_item = kmem_cache_zalloc(rmap_item_cache, GFP_KERNEL | ++ __GFP_NORETRY | __GFP_NOWARN); ++ if (rmap_item) { ++ /* bug on lowest bit is not clear for flag use */ ++ BUG_ON(is_addr(rmap_item)); ++ } ++ return rmap_item; ++} ++ ++static inline void free_rmap_item(struct rmap_item *rmap_item) ++{ ++ rmap_item->slot = NULL; /* debug safety */ ++ kmem_cache_free(rmap_item_cache, rmap_item); ++} ++ ++static inline struct stable_node *alloc_stable_node(void) ++{ ++ struct stable_node *node; ++ ++ node = kmem_cache_alloc(stable_node_cache, GFP_KERNEL | ++ __GFP_NORETRY | __GFP_NOWARN); ++ if (!node) ++ return NULL; ++ ++ INIT_HLIST_HEAD(&node->hlist); ++ list_add(&node->all_list, &stable_node_list); ++ return node; ++} ++ ++static inline void free_stable_node(struct stable_node *stable_node) ++{ ++ list_del(&stable_node->all_list); ++ kmem_cache_free(stable_node_cache, stable_node); ++} ++ ++static inline struct tree_node *alloc_tree_node(struct list_head *list) ++{ ++ struct tree_node *node; ++ ++ node = kmem_cache_zalloc(tree_node_cache, GFP_KERNEL | ++ __GFP_NORETRY | __GFP_NOWARN); ++ if (!node) ++ return NULL; ++ ++ list_add(&node->all_list, list); ++ return node; ++} ++ ++static inline void free_tree_node(struct tree_node *node) ++{ ++ list_del(&node->all_list); ++ kmem_cache_free(tree_node_cache, node); ++} ++ ++static void uksm_drop_anon_vma(struct rmap_item *rmap_item) ++{ ++ struct anon_vma *anon_vma = rmap_item->anon_vma; ++ ++ put_anon_vma(anon_vma); ++} ++ ++ ++/** ++ * Remove a stable node from stable_tree, may unlink from its tree_node and ++ * may remove its parent tree_node if no other stable node is pending. ++ * ++ * @stable_node The node need to be removed ++ * @unlink_rb Will this node be unlinked from the rbtree? ++ * @remove_tree_ node Will its tree_node be removed if empty? ++ */ ++static void remove_node_from_stable_tree(struct stable_node *stable_node, ++ int unlink_rb, int remove_tree_node) ++{ ++ struct node_vma *node_vma; ++ struct rmap_item *rmap_item; ++ struct hlist_node *n; ++ ++ if (!hlist_empty(&stable_node->hlist)) { ++ hlist_for_each_entry_safe(node_vma, n, ++ &stable_node->hlist, hlist) { ++ hlist_for_each_entry(rmap_item, &node_vma->rmap_hlist, hlist) { ++ uksm_pages_sharing--; ++ ++ uksm_drop_anon_vma(rmap_item); ++ rmap_item->address &= PAGE_MASK; ++ } ++ free_node_vma(node_vma); ++ cond_resched(); ++ } ++ ++ /* the last one is counted as shared */ ++ uksm_pages_shared--; ++ uksm_pages_sharing++; ++ } ++ ++ if (stable_node->tree_node && unlink_rb) { ++ rb_erase(&stable_node->node, ++ &stable_node->tree_node->sub_root); ++ ++ if (RB_EMPTY_ROOT(&stable_node->tree_node->sub_root) && ++ remove_tree_node) { ++ rb_erase(&stable_node->tree_node->node, ++ root_stable_treep); ++ free_tree_node(stable_node->tree_node); ++ } else { ++ stable_node->tree_node->count--; ++ } ++ } ++ ++ free_stable_node(stable_node); ++} ++ ++ ++/* ++ * get_uksm_page: checks if the page indicated by the stable node ++ * is still its ksm page, despite having held no reference to it. ++ * In which case we can trust the content of the page, and it ++ * returns the gotten page; but if the page has now been zapped, ++ * remove the stale node from the stable tree and return NULL. ++ * ++ * You would expect the stable_node to hold a reference to the ksm page. ++ * But if it increments the page's count, swapping out has to wait for ++ * ksmd to come around again before it can free the page, which may take ++ * seconds or even minutes: much too unresponsive. So instead we use a ++ * "keyhole reference": access to the ksm page from the stable node peeps ++ * out through its keyhole to see if that page still holds the right key, ++ * pointing back to this stable node. This relies on freeing a PageAnon ++ * page to reset its page->mapping to NULL, and relies on no other use of ++ * a page to put something that might look like our key in page->mapping. ++ * ++ * include/linux/pagemap.h page_cache_get_speculative() is a good reference, ++ * but this is different - made simpler by uksm_thread_mutex being held, but ++ * interesting for assuming that no other use of the struct page could ever ++ * put our expected_mapping into page->mapping (or a field of the union which ++ * coincides with page->mapping). The RCU calls are not for KSM at all, but ++ * to keep the page_count protocol described with page_cache_get_speculative. ++ * ++ * Note: it is possible that get_uksm_page() will return NULL one moment, ++ * then page the next, if the page is in between page_freeze_refs() and ++ * page_unfreeze_refs(): this shouldn't be a problem anywhere, the page ++ * is on its way to being freed; but it is an anomaly to bear in mind. ++ * ++ * @unlink_rb: if the removal of this node will firstly unlink from ++ * its rbtree. stable_node_reinsert will prevent this when restructuring the ++ * node from its old tree. ++ * ++ * @remove_tree_node: if this is the last one of its tree_node, will the ++ * tree_node be freed ? If we are inserting stable node, this tree_node may ++ * be reused, so don't free it. ++ */ ++static struct page *get_uksm_page(struct stable_node *stable_node, ++ int unlink_rb, int remove_tree_node) ++{ ++ struct page *page; ++ void *expected_mapping; ++ unsigned long kpfn; ++ ++ expected_mapping = (void *)((unsigned long)stable_node | ++ PAGE_MAPPING_KSM); ++again: ++ kpfn = READ_ONCE(stable_node->kpfn); ++ page = pfn_to_page(kpfn); ++ ++ /* ++ * page is computed from kpfn, so on most architectures reading ++ * page->mapping is naturally ordered after reading node->kpfn, ++ * but on Alpha we need to be more careful. ++ */ ++ smp_read_barrier_depends(); ++ ++ if (READ_ONCE(page->mapping) != expected_mapping) ++ goto stale; ++ ++ /* ++ * We cannot do anything with the page while its refcount is 0. ++ * Usually 0 means free, or tail of a higher-order page: in which ++ * case this node is no longer referenced, and should be freed; ++ * however, it might mean that the page is under page_freeze_refs(). ++ * The __remove_mapping() case is easy, again the node is now stale; ++ * but if page is swapcache in migrate_page_move_mapping(), it might ++ * still be our page, in which case it's essential to keep the node. ++ */ ++ while (!get_page_unless_zero(page)) { ++ /* ++ * Another check for page->mapping != expected_mapping would ++ * work here too. We have chosen the !PageSwapCache test to ++ * optimize the common case, when the page is or is about to ++ * be freed: PageSwapCache is cleared (under spin_lock_irq) ++ * in the freeze_refs section of __remove_mapping(); but Anon ++ * page->mapping reset to NULL later, in free_pages_prepare(). ++ */ ++ if (!PageSwapCache(page)) ++ goto stale; ++ cpu_relax(); ++ } ++ ++ if (READ_ONCE(page->mapping) != expected_mapping) { ++ put_page(page); ++ goto stale; ++ } ++ ++ lock_page(page); ++ if (READ_ONCE(page->mapping) != expected_mapping) { ++ unlock_page(page); ++ put_page(page); ++ goto stale; ++ } ++ unlock_page(page); ++ return page; ++stale: ++ /* ++ * We come here from above when page->mapping or !PageSwapCache ++ * suggests that the node is stale; but it might be under migration. ++ * We need smp_rmb(), matching the smp_wmb() in ksm_migrate_page(), ++ * before checking whether node->kpfn has been changed. ++ */ ++ smp_rmb(); ++ if (stable_node->kpfn != kpfn) ++ goto again; ++ ++ remove_node_from_stable_tree(stable_node, unlink_rb, remove_tree_node); ++ ++ return NULL; ++} ++ ++/* ++ * Removing rmap_item from stable or unstable tree. ++ * This function will clean the information from the stable/unstable tree. ++ */ ++static inline void remove_rmap_item_from_tree(struct rmap_item *rmap_item) ++{ ++ if (rmap_item->address & STABLE_FLAG) { ++ struct stable_node *stable_node; ++ struct node_vma *node_vma; ++ struct page *page; ++ ++ node_vma = rmap_item->head; ++ stable_node = node_vma->head; ++ page = get_uksm_page(stable_node, 1, 1); ++ if (!page) ++ goto out; ++ ++ /* ++ * page lock is needed because it's racing with ++ * try_to_unmap_ksm(), etc. ++ */ ++ lock_page(page); ++ hlist_del(&rmap_item->hlist); ++ ++ if (hlist_empty(&node_vma->rmap_hlist)) { ++ hlist_del(&node_vma->hlist); ++ free_node_vma(node_vma); ++ } ++ unlock_page(page); ++ ++ put_page(page); ++ if (hlist_empty(&stable_node->hlist)) { ++ /* do NOT call remove_node_from_stable_tree() here, ++ * it's possible for a forked rmap_item not in ++ * stable tree while the in-tree rmap_items were ++ * deleted. ++ */ ++ uksm_pages_shared--; ++ } else ++ uksm_pages_sharing--; ++ ++ ++ uksm_drop_anon_vma(rmap_item); ++ } else if (rmap_item->address & UNSTABLE_FLAG) { ++ if (rmap_item->hash_round == uksm_hash_round) { ++ ++ rb_erase(&rmap_item->node, ++ &rmap_item->tree_node->sub_root); ++ if (RB_EMPTY_ROOT(&rmap_item->tree_node->sub_root)) { ++ rb_erase(&rmap_item->tree_node->node, ++ &root_unstable_tree); ++ ++ free_tree_node(rmap_item->tree_node); ++ } else ++ rmap_item->tree_node->count--; ++ } ++ uksm_pages_unshared--; ++ } ++ ++ rmap_item->address &= PAGE_MASK; ++ rmap_item->hash_max = 0; ++ ++out: ++ cond_resched(); /* we're called from many long loops */ ++} ++ ++static inline int slot_in_uksm(struct vma_slot *slot) ++{ ++ return list_empty(&slot->slot_list); ++} ++ ++/* ++ * Test if the mm is exiting ++ */ ++static inline bool uksm_test_exit(struct mm_struct *mm) ++{ ++ return atomic_read(&mm->mm_users) == 0; ++} ++ ++static inline unsigned long vma_pool_size(struct vma_slot *slot) ++{ ++ return round_up(sizeof(struct rmap_list_entry) * slot->pages, ++ PAGE_SIZE) >> PAGE_SHIFT; ++} ++ ++#define CAN_OVERFLOW_U64(x, delta) (U64_MAX - (x) < (delta)) ++ ++/* must be done with sem locked */ ++static int slot_pool_alloc(struct vma_slot *slot) ++{ ++ unsigned long pool_size; ++ ++ if (slot->rmap_list_pool) ++ return 0; ++ ++ pool_size = vma_pool_size(slot); ++ slot->rmap_list_pool = kcalloc(pool_size, sizeof(struct page *), ++ GFP_KERNEL); ++ if (!slot->rmap_list_pool) ++ return -ENOMEM; ++ ++ slot->pool_counts = kcalloc(pool_size, sizeof(unsigned int), ++ GFP_KERNEL); ++ if (!slot->pool_counts) { ++ kfree(slot->rmap_list_pool); ++ return -ENOMEM; ++ } ++ ++ slot->pool_size = pool_size; ++ BUG_ON(CAN_OVERFLOW_U64(uksm_pages_total, slot->pages)); ++ slot->flags |= UKSM_SLOT_IN_UKSM; ++ uksm_pages_total += slot->pages; ++ ++ return 0; ++} ++ ++/* ++ * Called after vma is unlinked from its mm ++ */ ++void uksm_remove_vma(struct vm_area_struct *vma) ++{ ++ struct vma_slot *slot; ++ ++ if (!vma->uksm_vma_slot) ++ return; ++ ++ spin_lock(&vma_slot_list_lock); ++ slot = vma->uksm_vma_slot; ++ if (!slot) ++ goto out; ++ ++ if (slot_in_uksm(slot)) { ++ /** ++ * This slot has been added by ksmd, so move to the del list ++ * waiting ksmd to free it. ++ */ ++ list_add_tail(&slot->slot_list, &vma_slot_del); ++ } else { ++ /** ++ * It's still on new list. It's ok to free slot directly. ++ */ ++ list_del(&slot->slot_list); ++ free_vma_slot(slot); ++ } ++out: ++ vma->uksm_vma_slot = NULL; ++ spin_unlock(&vma_slot_list_lock); ++} ++ ++/** ++ * Need to do two things: ++ * 1. check if slot was moved to del list ++ * 2. make sure the mmap_sem is manipulated under valid vma. ++ * ++ * My concern here is that in some cases, this may make ++ * vma_slot_list_lock() waiters to serialized further by some ++ * sem->wait_lock, can this really be expensive? ++ * ++ * ++ * @return ++ * 0: if successfully locked mmap_sem ++ * -ENOENT: this slot was moved to del list ++ * -EBUSY: vma lock failed ++ */ ++static int try_down_read_slot_mmap_sem(struct vma_slot *slot) ++{ ++ struct vm_area_struct *vma; ++ struct mm_struct *mm; ++ struct rw_semaphore *sem; ++ ++ spin_lock(&vma_slot_list_lock); ++ ++ /* the slot_list was removed and inited from new list, when it enters ++ * uksm_list. If now it's not empty, then it must be moved to del list ++ */ ++ if (!slot_in_uksm(slot)) { ++ spin_unlock(&vma_slot_list_lock); ++ return -ENOENT; ++ } ++ ++ BUG_ON(slot->pages != vma_pages(slot->vma)); ++ /* Ok, vma still valid */ ++ vma = slot->vma; ++ mm = vma->vm_mm; ++ sem = &mm->mmap_sem; ++ ++ if (uksm_test_exit(mm)) { ++ spin_unlock(&vma_slot_list_lock); ++ return -ENOENT; ++ } ++ ++ if (down_read_trylock(sem)) { ++ spin_unlock(&vma_slot_list_lock); ++ if (slot_pool_alloc(slot)) { ++ uksm_remove_vma(vma); ++ up_read(sem); ++ return -ENOENT; ++ } ++ return 0; ++ } ++ ++ spin_unlock(&vma_slot_list_lock); ++ return -EBUSY; ++} ++ ++static inline unsigned long ++vma_page_address(struct page *page, struct vm_area_struct *vma) ++{ ++ pgoff_t pgoff = page->index; ++ unsigned long address; ++ ++ address = vma->vm_start + ((pgoff - vma->vm_pgoff) << PAGE_SHIFT); ++ if (unlikely(address < vma->vm_start || address >= vma->vm_end)) { ++ /* page should be within @vma mapping range */ ++ return -EFAULT; ++ } ++ return address; ++} ++ ++ ++/* return 0 on success with the item's mmap_sem locked */ ++static inline int get_mergeable_page_lock_mmap(struct rmap_item *item) ++{ ++ struct mm_struct *mm; ++ struct vma_slot *slot = item->slot; ++ int err = -EINVAL; ++ ++ struct page *page; ++ ++ /* ++ * try_down_read_slot_mmap_sem() returns non-zero if the slot ++ * has been removed by uksm_remove_vma(). ++ */ ++ if (try_down_read_slot_mmap_sem(slot)) ++ return -EBUSY; ++ ++ mm = slot->vma->vm_mm; ++ ++ if (uksm_test_exit(mm)) ++ goto failout_up; ++ ++ page = item->page; ++ rcu_read_lock(); ++ if (!get_page_unless_zero(page)) { ++ rcu_read_unlock(); ++ goto failout_up; ++ } ++ ++ /* No need to consider huge page here. */ ++ if (item->slot->vma->anon_vma != page_anon_vma(page) || ++ vma_page_address(page, item->slot->vma) != get_rmap_addr(item)) { ++ /* ++ * TODO: ++ * should we release this item becase of its stale page ++ * mapping? ++ */ ++ put_page(page); ++ rcu_read_unlock(); ++ goto failout_up; ++ } ++ rcu_read_unlock(); ++ return 0; ++ ++failout_up: ++ up_read(&mm->mmap_sem); ++ return err; ++} ++ ++/* ++ * What kind of VMA is considered ? ++ */ ++static inline int vma_can_enter(struct vm_area_struct *vma) ++{ ++ return uksm_flags_can_scan(vma->vm_flags); ++} ++ ++/* ++ * Called whenever a fresh new vma is created A new vma_slot. ++ * is created and inserted into a global list Must be called. ++ * after vma is inserted to its mm. ++ */ ++void uksm_vma_add_new(struct vm_area_struct *vma) ++{ ++ struct vma_slot *slot; ++ ++ if (!vma_can_enter(vma)) { ++ vma->uksm_vma_slot = NULL; ++ return; ++ } ++ ++ slot = alloc_vma_slot(); ++ if (!slot) { ++ vma->uksm_vma_slot = NULL; ++ return; ++ } ++ ++ vma->uksm_vma_slot = slot; ++ vma->vm_flags |= VM_MERGEABLE; ++ slot->vma = vma; ++ slot->mm = vma->vm_mm; ++ slot->ctime_j = jiffies; ++ slot->pages = vma_pages(vma); ++ spin_lock(&vma_slot_list_lock); ++ list_add_tail(&slot->slot_list, &vma_slot_new); ++ spin_unlock(&vma_slot_list_lock); ++} ++ ++/* 32/3 < they < 32/2 */ ++#define shiftl 8 ++#define shiftr 12 ++ ++#define HASH_FROM_TO(from, to) \ ++for (index = from; index < to; index++) { \ ++ pos = random_nums[index]; \ ++ hash += key[pos]; \ ++ hash += (hash << shiftl); \ ++ hash ^= (hash >> shiftr); \ ++} ++ ++ ++#define HASH_FROM_DOWN_TO(from, to) \ ++for (index = from - 1; index >= to; index--) { \ ++ hash ^= (hash >> shiftr); \ ++ hash ^= (hash >> (shiftr*2)); \ ++ hash -= (hash << shiftl); \ ++ hash += (hash << (shiftl*2)); \ ++ pos = random_nums[index]; \ ++ hash -= key[pos]; \ ++} ++ ++/* ++ * The main random sample hash function. ++ */ ++static u32 random_sample_hash(void *addr, u32 hash_strength) ++{ ++ u32 hash = 0xdeadbeef; ++ int index, pos, loop = hash_strength; ++ u32 *key = (u32 *)addr; ++ ++ if (loop > HASH_STRENGTH_FULL) ++ loop = HASH_STRENGTH_FULL; ++ ++ HASH_FROM_TO(0, loop); ++ ++ if (hash_strength > HASH_STRENGTH_FULL) { ++ loop = hash_strength - HASH_STRENGTH_FULL; ++ HASH_FROM_TO(0, loop); ++ } ++ ++ return hash; ++} ++ ++ ++/** ++ * It's used when hash strength is adjusted ++ * ++ * @addr The page's virtual address ++ * @from The original hash strength ++ * @to The hash strength changed to ++ * @hash The hash value generated with "from" hash value ++ * ++ * return the hash value ++ */ ++static u32 delta_hash(void *addr, int from, int to, u32 hash) ++{ ++ u32 *key = (u32 *)addr; ++ int index, pos; /* make sure they are int type */ ++ ++ if (to > from) { ++ if (from >= HASH_STRENGTH_FULL) { ++ from -= HASH_STRENGTH_FULL; ++ to -= HASH_STRENGTH_FULL; ++ HASH_FROM_TO(from, to); ++ } else if (to <= HASH_STRENGTH_FULL) { ++ HASH_FROM_TO(from, to); ++ } else { ++ HASH_FROM_TO(from, HASH_STRENGTH_FULL); ++ HASH_FROM_TO(0, to - HASH_STRENGTH_FULL); ++ } ++ } else { ++ if (from <= HASH_STRENGTH_FULL) { ++ HASH_FROM_DOWN_TO(from, to); ++ } else if (to >= HASH_STRENGTH_FULL) { ++ from -= HASH_STRENGTH_FULL; ++ to -= HASH_STRENGTH_FULL; ++ HASH_FROM_DOWN_TO(from, to); ++ } else { ++ HASH_FROM_DOWN_TO(from - HASH_STRENGTH_FULL, 0); ++ HASH_FROM_DOWN_TO(HASH_STRENGTH_FULL, to); ++ } ++ } ++ ++ return hash; ++} ++ ++/** ++ * ++ * Called when: rshash_pos or rshash_neg is about to overflow or a scan round ++ * has finished. ++ * ++ * return 0 if no page has been scanned since last call, 1 otherwise. ++ */ ++static inline int encode_benefit(void) ++{ ++ u64 scanned_delta, pos_delta, neg_delta; ++ unsigned long base = benefit.base; ++ ++ scanned_delta = uksm_pages_scanned - uksm_pages_scanned_last; ++ ++ if (!scanned_delta) ++ return 0; ++ ++ scanned_delta >>= base; ++ pos_delta = rshash_pos >> base; ++ neg_delta = rshash_neg >> base; ++ ++ if (CAN_OVERFLOW_U64(benefit.pos, pos_delta) || ++ CAN_OVERFLOW_U64(benefit.neg, neg_delta) || ++ CAN_OVERFLOW_U64(benefit.scanned, scanned_delta)) { ++ benefit.scanned >>= 1; ++ benefit.neg >>= 1; ++ benefit.pos >>= 1; ++ benefit.base++; ++ scanned_delta >>= 1; ++ pos_delta >>= 1; ++ neg_delta >>= 1; ++ } ++ ++ benefit.pos += pos_delta; ++ benefit.neg += neg_delta; ++ benefit.scanned += scanned_delta; ++ ++ BUG_ON(!benefit.scanned); ++ ++ rshash_pos = rshash_neg = 0; ++ uksm_pages_scanned_last = uksm_pages_scanned; ++ ++ return 1; ++} ++ ++static inline void reset_benefit(void) ++{ ++ benefit.pos = 0; ++ benefit.neg = 0; ++ benefit.base = 0; ++ benefit.scanned = 0; ++} ++ ++static inline void inc_rshash_pos(unsigned long delta) ++{ ++ if (CAN_OVERFLOW_U64(rshash_pos, delta)) ++ encode_benefit(); ++ ++ rshash_pos += delta; ++} ++ ++static inline void inc_rshash_neg(unsigned long delta) ++{ ++ if (CAN_OVERFLOW_U64(rshash_neg, delta)) ++ encode_benefit(); ++ ++ rshash_neg += delta; ++} ++ ++ ++static inline u32 page_hash(struct page *page, unsigned long hash_strength, ++ int cost_accounting) ++{ ++ u32 val; ++ unsigned long delta; ++ ++ void *addr = kmap_atomic(page); ++ ++ val = random_sample_hash(addr, hash_strength); ++ kunmap_atomic(addr); ++ ++ if (cost_accounting) { ++ if (hash_strength < HASH_STRENGTH_FULL) ++ delta = HASH_STRENGTH_FULL - hash_strength; ++ else ++ delta = 0; ++ ++ inc_rshash_pos(delta); ++ } ++ ++ return val; ++} ++ ++static int memcmp_pages(struct page *page1, struct page *page2, ++ int cost_accounting) ++{ ++ char *addr1, *addr2; ++ int ret; ++ ++ addr1 = kmap_atomic(page1); ++ addr2 = kmap_atomic(page2); ++ ret = memcmp(addr1, addr2, PAGE_SIZE); ++ kunmap_atomic(addr2); ++ kunmap_atomic(addr1); ++ ++ if (cost_accounting) ++ inc_rshash_neg(memcmp_cost); ++ ++ return ret; ++} ++ ++static inline int pages_identical(struct page *page1, struct page *page2) ++{ ++ return !memcmp_pages(page1, page2, 0); ++} ++ ++static inline int is_page_full_zero(struct page *page) ++{ ++ char *addr; ++ int ret; ++ ++ addr = kmap_atomic(page); ++ ret = is_full_zero(addr, PAGE_SIZE); ++ kunmap_atomic(addr); ++ ++ return ret; ++} ++ ++static int write_protect_page(struct vm_area_struct *vma, struct page *page, ++ pte_t *orig_pte, pte_t *old_pte) ++{ ++ struct mm_struct *mm = vma->vm_mm; ++ struct page_vma_mapped_walk pvmw = { ++ .page = page, ++ .vma = vma, ++ }; ++ int swapped; ++ int err = -EFAULT; ++ unsigned long mmun_start; /* For mmu_notifiers */ ++ unsigned long mmun_end; /* For mmu_notifiers */ ++ ++ pvmw.address = page_address_in_vma(page, vma); ++ if (pvmw.address == -EFAULT) ++ goto out; ++ ++ BUG_ON(PageTransCompound(page)); ++ ++ mmun_start = pvmw.address; ++ mmun_end = pvmw.address + PAGE_SIZE; ++ mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); ++ ++ if (!page_vma_mapped_walk(&pvmw)) ++ goto out_mn; ++ if (WARN_ONCE(!pvmw.pte, "Unexpected PMD mapping?")) ++ goto out_unlock; ++ ++ if (old_pte) ++ *old_pte = *pvmw.pte; ++ ++ if (pte_write(*pvmw.pte) || pte_dirty(*pvmw.pte) || ++ (pte_protnone(*pvmw.pte) && pte_savedwrite(*pvmw.pte)) || mm_tlb_flush_pending(mm)) { ++ pte_t entry; ++ ++ swapped = PageSwapCache(page); ++ flush_cache_page(vma, pvmw.address, page_to_pfn(page)); ++ /* ++ * Ok this is tricky, when get_user_pages_fast() run it doesn't ++ * take any lock, therefore the check that we are going to make ++ * with the pagecount against the mapcount is racey and ++ * O_DIRECT can happen right after the check. ++ * So we clear the pte and flush the tlb before the check ++ * this assure us that no O_DIRECT can happen after the check ++ * or in the middle of the check. ++ */ ++ entry = ptep_clear_flush_notify(vma, pvmw.address, pvmw.pte); ++ /* ++ * Check that no O_DIRECT or similar I/O is in progress on the ++ * page ++ */ ++ if (page_mapcount(page) + 1 + swapped != page_count(page)) { ++ set_pte_at(mm, pvmw.address, pvmw.pte, entry); ++ goto out_unlock; ++ } ++ if (pte_dirty(entry)) ++ set_page_dirty(page); ++ ++ if (pte_protnone(entry)) ++ entry = pte_mkclean(pte_clear_savedwrite(entry)); ++ else ++ entry = pte_mkclean(pte_wrprotect(entry)); ++ ++ set_pte_at_notify(mm, pvmw.address, pvmw.pte, entry); ++ } ++ *orig_pte = *pvmw.pte; ++ err = 0; ++ ++out_unlock: ++ page_vma_mapped_walk_done(&pvmw); ++out_mn: ++ mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); ++out: ++ return err; ++} ++ ++#define MERGE_ERR_PGERR 1 /* the page is invalid cannot continue */ ++#define MERGE_ERR_COLLI 2 /* there is a collision */ ++#define MERGE_ERR_COLLI_MAX 3 /* collision at the max hash strength */ ++#define MERGE_ERR_CHANGED 4 /* the page has changed since last hash */ ++ ++ ++/** ++ * replace_page - replace page in vma by new ksm page ++ * @vma: vma that holds the pte pointing to page ++ * @page: the page we are replacing by kpage ++ * @kpage: the ksm page we replace page by ++ * @orig_pte: the original value of the pte ++ * ++ * Returns 0 on success, MERGE_ERR_PGERR on failure. ++ */ ++static int replace_page(struct vm_area_struct *vma, struct page *page, ++ struct page *kpage, pte_t orig_pte) ++{ ++ struct mm_struct *mm = vma->vm_mm; ++ pgd_t *pgd; ++ p4d_t *p4d; ++ pud_t *pud; ++ pmd_t *pmd; ++ pte_t *ptep; ++ spinlock_t *ptl; ++ pte_t entry; ++ ++ unsigned long addr; ++ int err = MERGE_ERR_PGERR; ++ unsigned long mmun_start; /* For mmu_notifiers */ ++ unsigned long mmun_end; /* For mmu_notifiers */ ++ ++ addr = page_address_in_vma(page, vma); ++ if (addr == -EFAULT) ++ goto out; ++ ++ pgd = pgd_offset(mm, addr); ++ if (!pgd_present(*pgd)) ++ goto out; ++ ++ p4d = p4d_offset(pgd, addr); ++ pud = pud_offset(p4d, addr); ++ if (!pud_present(*pud)) ++ goto out; ++ ++ pmd = pmd_offset(pud, addr); ++ BUG_ON(pmd_trans_huge(*pmd)); ++ if (!pmd_present(*pmd)) ++ goto out; ++ ++ mmun_start = addr; ++ mmun_end = addr + PAGE_SIZE; ++ mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); ++ ++ ptep = pte_offset_map_lock(mm, pmd, addr, &ptl); ++ if (!pte_same(*ptep, orig_pte)) { ++ pte_unmap_unlock(ptep, ptl); ++ goto out_mn; ++ } ++ ++ flush_cache_page(vma, addr, pte_pfn(*ptep)); ++ ptep_clear_flush_notify(vma, addr, ptep); ++ entry = mk_pte(kpage, vma->vm_page_prot); ++ ++ /* special treatment is needed for zero_page */ ++ if ((page_to_pfn(kpage) == uksm_zero_pfn) || ++ (page_to_pfn(kpage) == zero_pfn)) { ++ entry = pte_mkspecial(entry); ++ dec_mm_counter(mm, MM_ANONPAGES); ++ inc_zone_page_state(page, NR_UKSM_ZERO_PAGES); ++ } else { ++ get_page(kpage); ++ page_add_anon_rmap(kpage, vma, addr, false); ++ } ++ ++ set_pte_at_notify(mm, addr, ptep, entry); ++ ++ page_remove_rmap(page, false); ++ if (!page_mapped(page)) ++ try_to_free_swap(page); ++ put_page(page); ++ ++ pte_unmap_unlock(ptep, ptl); ++ err = 0; ++out_mn: ++ mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); ++out: ++ return err; ++} ++ ++ ++/** ++ * Fully hash a page with HASH_STRENGTH_MAX return a non-zero hash value. The ++ * zero hash value at HASH_STRENGTH_MAX is used to indicated that its ++ * hash_max member has not been calculated. ++ * ++ * @page The page needs to be hashed ++ * @hash_old The hash value calculated with current hash strength ++ * ++ * return the new hash value calculated at HASH_STRENGTH_MAX ++ */ ++static inline u32 page_hash_max(struct page *page, u32 hash_old) ++{ ++ u32 hash_max = 0; ++ void *addr; ++ ++ addr = kmap_atomic(page); ++ hash_max = delta_hash(addr, hash_strength, ++ HASH_STRENGTH_MAX, hash_old); ++ ++ kunmap_atomic(addr); ++ ++ if (!hash_max) ++ hash_max = 1; ++ ++ inc_rshash_neg(HASH_STRENGTH_MAX - hash_strength); ++ return hash_max; ++} ++ ++/* ++ * We compare the hash again, to ensure that it is really a hash collision ++ * instead of being caused by page write. ++ */ ++static inline int check_collision(struct rmap_item *rmap_item, ++ u32 hash) ++{ ++ int err; ++ struct page *page = rmap_item->page; ++ ++ /* if this rmap_item has already been hash_maxed, then the collision ++ * must appears in the second-level rbtree search. In this case we check ++ * if its hash_max value has been changed. Otherwise, the collision ++ * happens in the first-level rbtree search, so we check against it's ++ * current hash value. ++ */ ++ if (rmap_item->hash_max) { ++ inc_rshash_neg(memcmp_cost); ++ inc_rshash_neg(HASH_STRENGTH_MAX - hash_strength); ++ ++ if (rmap_item->hash_max == page_hash_max(page, hash)) ++ err = MERGE_ERR_COLLI; ++ else ++ err = MERGE_ERR_CHANGED; ++ } else { ++ inc_rshash_neg(memcmp_cost + hash_strength); ++ ++ if (page_hash(page, hash_strength, 0) == hash) ++ err = MERGE_ERR_COLLI; ++ else ++ err = MERGE_ERR_CHANGED; ++ } ++ ++ return err; ++} ++ ++/** ++ * Try to merge a rmap_item.page with a kpage in stable node. kpage must ++ * already be a ksm page. ++ * ++ * @return 0 if the pages were merged, -EFAULT otherwise. ++ */ ++static int try_to_merge_with_uksm_page(struct rmap_item *rmap_item, ++ struct page *kpage, u32 hash) ++{ ++ struct vm_area_struct *vma = rmap_item->slot->vma; ++ struct mm_struct *mm = vma->vm_mm; ++ pte_t orig_pte = __pte(0); ++ int err = MERGE_ERR_PGERR; ++ struct page *page; ++ ++ if (uksm_test_exit(mm)) ++ goto out; ++ ++ page = rmap_item->page; ++ ++ if (page == kpage) { /* ksm page forked */ ++ err = 0; ++ goto out; ++ } ++ ++ /* ++ * We need the page lock to read a stable PageSwapCache in ++ * write_protect_page(). We use trylock_page() instead of ++ * lock_page() because we don't want to wait here - we ++ * prefer to continue scanning and merging different pages, ++ * then come back to this page when it is unlocked. ++ */ ++ if (!trylock_page(page)) ++ goto out; ++ ++ if (!PageAnon(page) || !PageKsm(kpage)) ++ goto out_unlock; ++ ++ if (PageTransCompound(page)) { ++ err = split_huge_page(page); ++ if (err) ++ goto out_unlock; ++ } ++ ++ /* ++ * If this anonymous page is mapped only here, its pte may need ++ * to be write-protected. If it's mapped elsewhere, all of its ++ * ptes are necessarily already write-protected. But in either ++ * case, we need to lock and check page_count is not raised. ++ */ ++ if (write_protect_page(vma, page, &orig_pte, NULL) == 0) { ++ if (pages_identical(page, kpage)) ++ err = replace_page(vma, page, kpage, orig_pte); ++ else ++ err = check_collision(rmap_item, hash); ++ } ++ ++ if ((vma->vm_flags & VM_LOCKED) && kpage && !err) { ++ munlock_vma_page(page); ++ if (!PageMlocked(kpage)) { ++ unlock_page(page); ++ lock_page(kpage); ++ mlock_vma_page(kpage); ++ page = kpage; /* for final unlock */ ++ } ++ } ++ ++out_unlock: ++ unlock_page(page); ++out: ++ return err; ++} ++ ++ ++ ++/** ++ * If two pages fail to merge in try_to_merge_two_pages, then we have a chance ++ * to restore a page mapping that has been changed in try_to_merge_two_pages. ++ * ++ * @return 0 on success. ++ */ ++static int restore_uksm_page_pte(struct vm_area_struct *vma, unsigned long addr, ++ pte_t orig_pte, pte_t wprt_pte) ++{ ++ struct mm_struct *mm = vma->vm_mm; ++ pgd_t *pgd; ++ p4d_t *p4d; ++ pud_t *pud; ++ pmd_t *pmd; ++ pte_t *ptep; ++ spinlock_t *ptl; ++ ++ int err = -EFAULT; ++ ++ pgd = pgd_offset(mm, addr); ++ if (!pgd_present(*pgd)) ++ goto out; ++ ++ p4d = p4d_offset(pgd, addr); ++ pud = pud_offset(p4d, addr); ++ if (!pud_present(*pud)) ++ goto out; ++ ++ pmd = pmd_offset(pud, addr); ++ if (!pmd_present(*pmd)) ++ goto out; ++ ++ ptep = pte_offset_map_lock(mm, pmd, addr, &ptl); ++ if (!pte_same(*ptep, wprt_pte)) { ++ /* already copied, let it be */ ++ pte_unmap_unlock(ptep, ptl); ++ goto out; ++ } ++ ++ /* ++ * Good boy, still here. When we still get the ksm page, it does not ++ * return to the free page pool, there is no way that a pte was changed ++ * to other page and gets back to this page. And remind that ksm page ++ * do not reuse in do_wp_page(). So it's safe to restore the original ++ * pte. ++ */ ++ flush_cache_page(vma, addr, pte_pfn(*ptep)); ++ ptep_clear_flush_notify(vma, addr, ptep); ++ set_pte_at_notify(mm, addr, ptep, orig_pte); ++ ++ pte_unmap_unlock(ptep, ptl); ++ err = 0; ++out: ++ return err; ++} ++ ++/** ++ * try_to_merge_two_pages() - take two identical pages and prepare ++ * them to be merged into one page(rmap_item->page) ++ * ++ * @return 0 if we successfully merged two identical pages into ++ * one ksm page. MERGE_ERR_COLLI if it's only a hash collision ++ * search in rbtree. MERGE_ERR_CHANGED if rmap_item has been ++ * changed since it's hashed. MERGE_ERR_PGERR otherwise. ++ * ++ */ ++static int try_to_merge_two_pages(struct rmap_item *rmap_item, ++ struct rmap_item *tree_rmap_item, ++ u32 hash) ++{ ++ pte_t orig_pte1 = __pte(0), orig_pte2 = __pte(0); ++ pte_t wprt_pte1 = __pte(0), wprt_pte2 = __pte(0); ++ struct vm_area_struct *vma1 = rmap_item->slot->vma; ++ struct vm_area_struct *vma2 = tree_rmap_item->slot->vma; ++ struct page *page = rmap_item->page; ++ struct page *tree_page = tree_rmap_item->page; ++ int err = MERGE_ERR_PGERR; ++ struct address_space *saved_mapping; ++ ++ ++ if (rmap_item->page == tree_rmap_item->page) ++ goto out; ++ ++ if (!trylock_page(page)) ++ goto out; ++ ++ if (!PageAnon(page)) ++ goto out_unlock; ++ ++ if (PageTransCompound(page)) { ++ err = split_huge_page(page); ++ if (err) ++ goto out_unlock; ++ } ++ ++ if (write_protect_page(vma1, page, &wprt_pte1, &orig_pte1) != 0) { ++ unlock_page(page); ++ goto out; ++ } ++ ++ /* ++ * While we hold page lock, upgrade page from ++ * PageAnon+anon_vma to PageKsm+NULL stable_node: ++ * stable_tree_insert() will update stable_node. ++ */ ++ saved_mapping = page->mapping; ++ set_page_stable_node(page, NULL); ++ mark_page_accessed(page); ++ if (!PageDirty(page)) ++ SetPageDirty(page); ++ ++ unlock_page(page); ++ ++ if (!trylock_page(tree_page)) ++ goto restore_out; ++ ++ if (!PageAnon(tree_page)) { ++ unlock_page(tree_page); ++ goto restore_out; ++ } ++ ++ if (PageTransCompound(tree_page)) { ++ err = split_huge_page(tree_page); ++ if (err) { ++ unlock_page(tree_page); ++ goto restore_out; ++ } ++ } ++ ++ if (write_protect_page(vma2, tree_page, &wprt_pte2, &orig_pte2) != 0) { ++ unlock_page(tree_page); ++ goto restore_out; ++ } ++ ++ if (pages_identical(page, tree_page)) { ++ err = replace_page(vma2, tree_page, page, wprt_pte2); ++ if (err) { ++ unlock_page(tree_page); ++ goto restore_out; ++ } ++ ++ if ((vma2->vm_flags & VM_LOCKED)) { ++ munlock_vma_page(tree_page); ++ if (!PageMlocked(page)) { ++ unlock_page(tree_page); ++ lock_page(page); ++ mlock_vma_page(page); ++ tree_page = page; /* for final unlock */ ++ } ++ } ++ ++ unlock_page(tree_page); ++ ++ goto out; /* success */ ++ ++ } else { ++ if (tree_rmap_item->hash_max && ++ tree_rmap_item->hash_max == rmap_item->hash_max) { ++ err = MERGE_ERR_COLLI_MAX; ++ } else if (page_hash(page, hash_strength, 0) == ++ page_hash(tree_page, hash_strength, 0)) { ++ inc_rshash_neg(memcmp_cost + hash_strength * 2); ++ err = MERGE_ERR_COLLI; ++ } else { ++ err = MERGE_ERR_CHANGED; ++ } ++ ++ unlock_page(tree_page); ++ } ++ ++restore_out: ++ lock_page(page); ++ if (!restore_uksm_page_pte(vma1, get_rmap_addr(rmap_item), ++ orig_pte1, wprt_pte1)) ++ page->mapping = saved_mapping; ++ ++out_unlock: ++ unlock_page(page); ++out: ++ return err; ++} ++ ++static inline int hash_cmp(u32 new_val, u32 node_val) ++{ ++ if (new_val > node_val) ++ return 1; ++ else if (new_val < node_val) ++ return -1; ++ else ++ return 0; ++} ++ ++static inline u32 rmap_item_hash_max(struct rmap_item *item, u32 hash) ++{ ++ u32 hash_max = item->hash_max; ++ ++ if (!hash_max) { ++ hash_max = page_hash_max(item->page, hash); ++ ++ item->hash_max = hash_max; ++ } ++ ++ return hash_max; ++} ++ ++ ++ ++/** ++ * stable_tree_search() - search the stable tree for a page ++ * ++ * @item: the rmap_item we are comparing with ++ * @hash: the hash value of this item->page already calculated ++ * ++ * @return the page we have found, NULL otherwise. The page returned has ++ * been gotten. ++ */ ++static struct page *stable_tree_search(struct rmap_item *item, u32 hash) ++{ ++ struct rb_node *node = root_stable_treep->rb_node; ++ struct tree_node *tree_node; ++ unsigned long hash_max; ++ struct page *page = item->page; ++ struct stable_node *stable_node; ++ ++ stable_node = page_stable_node(page); ++ if (stable_node) { ++ /* ksm page forked, that is ++ * if (PageKsm(page) && !in_stable_tree(rmap_item)) ++ * it's actually gotten once outside. ++ */ ++ get_page(page); ++ return page; ++ } ++ ++ while (node) { ++ int cmp; ++ ++ tree_node = rb_entry(node, struct tree_node, node); ++ ++ cmp = hash_cmp(hash, tree_node->hash); ++ ++ if (cmp < 0) ++ node = node->rb_left; ++ else if (cmp > 0) ++ node = node->rb_right; ++ else ++ break; ++ } ++ ++ if (!node) ++ return NULL; ++ ++ if (tree_node->count == 1) { ++ stable_node = rb_entry(tree_node->sub_root.rb_node, ++ struct stable_node, node); ++ BUG_ON(!stable_node); ++ ++ goto get_page_out; ++ } ++ ++ /* ++ * ok, we have to search the second ++ * level subtree, hash the page to a ++ * full strength. ++ */ ++ node = tree_node->sub_root.rb_node; ++ BUG_ON(!node); ++ hash_max = rmap_item_hash_max(item, hash); ++ ++ while (node) { ++ int cmp; ++ ++ stable_node = rb_entry(node, struct stable_node, node); ++ ++ cmp = hash_cmp(hash_max, stable_node->hash_max); ++ ++ if (cmp < 0) ++ node = node->rb_left; ++ else if (cmp > 0) ++ node = node->rb_right; ++ else ++ goto get_page_out; ++ } ++ ++ return NULL; ++ ++get_page_out: ++ page = get_uksm_page(stable_node, 1, 1); ++ return page; ++} ++ ++static int try_merge_rmap_item(struct rmap_item *item, ++ struct page *kpage, ++ struct page *tree_page) ++{ ++ struct vm_area_struct *vma = item->slot->vma; ++ struct page_vma_mapped_walk pvmw = { ++ .page = kpage, ++ .vma = vma, ++ }; ++ ++ pvmw.address = get_rmap_addr(item); ++ if (!page_vma_mapped_walk(&pvmw)) ++ return 0; ++ ++ if (pte_write(*pvmw.pte)) { ++ /* has changed, abort! */ ++ page_vma_mapped_walk_done(&pvmw); ++ return 0; ++ } ++ ++ get_page(tree_page); ++ page_add_anon_rmap(tree_page, vma, pvmw.address, false); ++ ++ flush_cache_page(vma, pvmw.address, page_to_pfn(kpage)); ++ ptep_clear_flush_notify(vma, pvmw.address, pvmw.pte); ++ set_pte_at_notify(vma->vm_mm, pvmw.address, pvmw.pte, ++ mk_pte(tree_page, vma->vm_page_prot)); ++ ++ page_remove_rmap(kpage, false); ++ put_page(kpage); ++ ++ page_vma_mapped_walk_done(&pvmw); ++ ++ return 1; ++} ++ ++/** ++ * try_to_merge_with_stable_page() - when two rmap_items need to be inserted ++ * into stable tree, the page was found to be identical to a stable ksm page, ++ * this is the last chance we can merge them into one. ++ * ++ * @item1: the rmap_item holding the page which we wanted to insert ++ * into stable tree. ++ * @item2: the other rmap_item we found when unstable tree search ++ * @oldpage: the page currently mapped by the two rmap_items ++ * @tree_page: the page we found identical in stable tree node ++ * @success1: return if item1 is successfully merged ++ * @success2: return if item2 is successfully merged ++ */ ++static void try_merge_with_stable(struct rmap_item *item1, ++ struct rmap_item *item2, ++ struct page **kpage, ++ struct page *tree_page, ++ int *success1, int *success2) ++{ ++ struct vm_area_struct *vma1 = item1->slot->vma; ++ struct vm_area_struct *vma2 = item2->slot->vma; ++ *success1 = 0; ++ *success2 = 0; ++ ++ if (unlikely(*kpage == tree_page)) { ++ /* I don't think this can really happen */ ++ pr_warn("UKSM: unexpected condition detected in " ++ "%s -- *kpage == tree_page !\n", __func__); ++ *success1 = 1; ++ *success2 = 1; ++ return; ++ } ++ ++ if (!PageAnon(*kpage) || !PageKsm(*kpage)) ++ goto failed; ++ ++ if (!trylock_page(tree_page)) ++ goto failed; ++ ++ /* If the oldpage is still ksm and still pointed ++ * to in the right place, and still write protected, ++ * we are confident it's not changed, no need to ++ * memcmp anymore. ++ * be ware, we cannot take nested pte locks, ++ * deadlock risk. ++ */ ++ if (!try_merge_rmap_item(item1, *kpage, tree_page)) ++ goto unlock_failed; ++ ++ /* ok, then vma2, remind that pte1 already set */ ++ if (!try_merge_rmap_item(item2, *kpage, tree_page)) ++ goto success_1; ++ ++ *success2 = 1; ++success_1: ++ *success1 = 1; ++ ++ ++ if ((*success1 && vma1->vm_flags & VM_LOCKED) || ++ (*success2 && vma2->vm_flags & VM_LOCKED)) { ++ munlock_vma_page(*kpage); ++ if (!PageMlocked(tree_page)) ++ mlock_vma_page(tree_page); ++ } ++ ++ /* ++ * We do not need oldpage any more in the caller, so can break the lock ++ * now. ++ */ ++ unlock_page(*kpage); ++ *kpage = tree_page; /* Get unlocked outside. */ ++ return; ++ ++unlock_failed: ++ unlock_page(tree_page); ++failed: ++ return; ++} ++ ++static inline void stable_node_hash_max(struct stable_node *node, ++ struct page *page, u32 hash) ++{ ++ u32 hash_max = node->hash_max; ++ ++ if (!hash_max) { ++ hash_max = page_hash_max(page, hash); ++ node->hash_max = hash_max; ++ } ++} ++ ++static inline ++struct stable_node *new_stable_node(struct tree_node *tree_node, ++ struct page *kpage, u32 hash_max) ++{ ++ struct stable_node *new_stable_node; ++ ++ new_stable_node = alloc_stable_node(); ++ if (!new_stable_node) ++ return NULL; ++ ++ new_stable_node->kpfn = page_to_pfn(kpage); ++ new_stable_node->hash_max = hash_max; ++ new_stable_node->tree_node = tree_node; ++ set_page_stable_node(kpage, new_stable_node); ++ ++ return new_stable_node; ++} ++ ++static inline ++struct stable_node *first_level_insert(struct tree_node *tree_node, ++ struct rmap_item *rmap_item, ++ struct rmap_item *tree_rmap_item, ++ struct page **kpage, u32 hash, ++ int *success1, int *success2) ++{ ++ int cmp; ++ struct page *tree_page; ++ u32 hash_max = 0; ++ struct stable_node *stable_node, *new_snode; ++ struct rb_node *parent = NULL, **new; ++ ++ /* this tree node contains no sub-tree yet */ ++ stable_node = rb_entry(tree_node->sub_root.rb_node, ++ struct stable_node, node); ++ ++ tree_page = get_uksm_page(stable_node, 1, 0); ++ if (tree_page) { ++ cmp = memcmp_pages(*kpage, tree_page, 1); ++ if (!cmp) { ++ try_merge_with_stable(rmap_item, tree_rmap_item, kpage, ++ tree_page, success1, success2); ++ put_page(tree_page); ++ if (!*success1 && !*success2) ++ goto failed; ++ ++ return stable_node; ++ ++ } else { ++ /* ++ * collision in first level try to create a subtree. ++ * A new node need to be created. ++ */ ++ put_page(tree_page); ++ ++ stable_node_hash_max(stable_node, tree_page, ++ tree_node->hash); ++ hash_max = rmap_item_hash_max(rmap_item, hash); ++ cmp = hash_cmp(hash_max, stable_node->hash_max); ++ ++ parent = &stable_node->node; ++ if (cmp < 0) ++ new = &parent->rb_left; ++ else if (cmp > 0) ++ new = &parent->rb_right; ++ else ++ goto failed; ++ } ++ ++ } else { ++ /* the only stable_node deleted, we reuse its tree_node. ++ */ ++ parent = NULL; ++ new = &tree_node->sub_root.rb_node; ++ } ++ ++ new_snode = new_stable_node(tree_node, *kpage, hash_max); ++ if (!new_snode) ++ goto failed; ++ ++ rb_link_node(&new_snode->node, parent, new); ++ rb_insert_color(&new_snode->node, &tree_node->sub_root); ++ tree_node->count++; ++ *success1 = *success2 = 1; ++ ++ return new_snode; ++ ++failed: ++ return NULL; ++} ++ ++static inline ++struct stable_node *stable_subtree_insert(struct tree_node *tree_node, ++ struct rmap_item *rmap_item, ++ struct rmap_item *tree_rmap_item, ++ struct page **kpage, u32 hash, ++ int *success1, int *success2) ++{ ++ struct page *tree_page; ++ u32 hash_max; ++ struct stable_node *stable_node, *new_snode; ++ struct rb_node *parent, **new; ++ ++research: ++ parent = NULL; ++ new = &tree_node->sub_root.rb_node; ++ BUG_ON(!*new); ++ hash_max = rmap_item_hash_max(rmap_item, hash); ++ while (*new) { ++ int cmp; ++ ++ stable_node = rb_entry(*new, struct stable_node, node); ++ ++ cmp = hash_cmp(hash_max, stable_node->hash_max); ++ ++ if (cmp < 0) { ++ parent = *new; ++ new = &parent->rb_left; ++ } else if (cmp > 0) { ++ parent = *new; ++ new = &parent->rb_right; ++ } else { ++ tree_page = get_uksm_page(stable_node, 1, 0); ++ if (tree_page) { ++ cmp = memcmp_pages(*kpage, tree_page, 1); ++ if (!cmp) { ++ try_merge_with_stable(rmap_item, ++ tree_rmap_item, kpage, ++ tree_page, success1, success2); ++ ++ put_page(tree_page); ++ if (!*success1 && !*success2) ++ goto failed; ++ /* ++ * successfully merged with a stable ++ * node ++ */ ++ return stable_node; ++ } else { ++ put_page(tree_page); ++ goto failed; ++ } ++ } else { ++ /* ++ * stable node may be deleted, ++ * and subtree maybe ++ * restructed, cannot ++ * continue, research it. ++ */ ++ if (tree_node->count) { ++ goto research; ++ } else { ++ /* reuse the tree node*/ ++ parent = NULL; ++ new = &tree_node->sub_root.rb_node; ++ } ++ } ++ } ++ } ++ ++ new_snode = new_stable_node(tree_node, *kpage, hash_max); ++ if (!new_snode) ++ goto failed; ++ ++ rb_link_node(&new_snode->node, parent, new); ++ rb_insert_color(&new_snode->node, &tree_node->sub_root); ++ tree_node->count++; ++ *success1 = *success2 = 1; ++ ++ return new_snode; ++ ++failed: ++ return NULL; ++} ++ ++ ++/** ++ * stable_tree_insert() - try to insert a merged page in unstable tree to ++ * the stable tree ++ * ++ * @kpage: the page need to be inserted ++ * @hash: the current hash of this page ++ * @rmap_item: the rmap_item being scanned ++ * @tree_rmap_item: the rmap_item found on unstable tree ++ * @success1: return if rmap_item is merged ++ * @success2: return if tree_rmap_item is merged ++ * ++ * @return the stable_node on stable tree if at least one ++ * rmap_item is inserted into stable tree, NULL ++ * otherwise. ++ */ ++static struct stable_node * ++stable_tree_insert(struct page **kpage, u32 hash, ++ struct rmap_item *rmap_item, ++ struct rmap_item *tree_rmap_item, ++ int *success1, int *success2) ++{ ++ struct rb_node **new = &root_stable_treep->rb_node; ++ struct rb_node *parent = NULL; ++ struct stable_node *stable_node; ++ struct tree_node *tree_node; ++ u32 hash_max = 0; ++ ++ *success1 = *success2 = 0; ++ ++ while (*new) { ++ int cmp; ++ ++ tree_node = rb_entry(*new, struct tree_node, node); ++ ++ cmp = hash_cmp(hash, tree_node->hash); ++ ++ if (cmp < 0) { ++ parent = *new; ++ new = &parent->rb_left; ++ } else if (cmp > 0) { ++ parent = *new; ++ new = &parent->rb_right; ++ } else ++ break; ++ } ++ ++ if (*new) { ++ if (tree_node->count == 1) { ++ stable_node = first_level_insert(tree_node, rmap_item, ++ tree_rmap_item, kpage, ++ hash, success1, success2); ++ } else { ++ stable_node = stable_subtree_insert(tree_node, ++ rmap_item, tree_rmap_item, kpage, ++ hash, success1, success2); ++ } ++ } else { ++ ++ /* no tree node found */ ++ tree_node = alloc_tree_node(stable_tree_node_listp); ++ if (!tree_node) { ++ stable_node = NULL; ++ goto out; ++ } ++ ++ stable_node = new_stable_node(tree_node, *kpage, hash_max); ++ if (!stable_node) { ++ free_tree_node(tree_node); ++ goto out; ++ } ++ ++ tree_node->hash = hash; ++ rb_link_node(&tree_node->node, parent, new); ++ rb_insert_color(&tree_node->node, root_stable_treep); ++ parent = NULL; ++ new = &tree_node->sub_root.rb_node; ++ ++ rb_link_node(&stable_node->node, parent, new); ++ rb_insert_color(&stable_node->node, &tree_node->sub_root); ++ tree_node->count++; ++ *success1 = *success2 = 1; ++ } ++ ++out: ++ return stable_node; ++} ++ ++ ++/** ++ * get_tree_rmap_item_page() - try to get the page and lock the mmap_sem ++ * ++ * @return 0 on success, -EBUSY if unable to lock the mmap_sem, ++ * -EINVAL if the page mapping has been changed. ++ */ ++static inline int get_tree_rmap_item_page(struct rmap_item *tree_rmap_item) ++{ ++ int err; ++ ++ err = get_mergeable_page_lock_mmap(tree_rmap_item); ++ ++ if (err == -EINVAL) { ++ /* its page map has been changed, remove it */ ++ remove_rmap_item_from_tree(tree_rmap_item); ++ } ++ ++ /* The page is gotten and mmap_sem is locked now. */ ++ return err; ++} ++ ++ ++/** ++ * unstable_tree_search_insert() - search an unstable tree rmap_item with the ++ * same hash value. Get its page and trylock the mmap_sem ++ */ ++static inline ++struct rmap_item *unstable_tree_search_insert(struct rmap_item *rmap_item, ++ u32 hash) ++ ++{ ++ struct rb_node **new = &root_unstable_tree.rb_node; ++ struct rb_node *parent = NULL; ++ struct tree_node *tree_node; ++ u32 hash_max; ++ struct rmap_item *tree_rmap_item; ++ ++ while (*new) { ++ int cmp; ++ ++ tree_node = rb_entry(*new, struct tree_node, node); ++ ++ cmp = hash_cmp(hash, tree_node->hash); ++ ++ if (cmp < 0) { ++ parent = *new; ++ new = &parent->rb_left; ++ } else if (cmp > 0) { ++ parent = *new; ++ new = &parent->rb_right; ++ } else ++ break; ++ } ++ ++ if (*new) { ++ /* got the tree_node */ ++ if (tree_node->count == 1) { ++ tree_rmap_item = rb_entry(tree_node->sub_root.rb_node, ++ struct rmap_item, node); ++ BUG_ON(!tree_rmap_item); ++ ++ goto get_page_out; ++ } ++ ++ /* well, search the collision subtree */ ++ new = &tree_node->sub_root.rb_node; ++ BUG_ON(!*new); ++ hash_max = rmap_item_hash_max(rmap_item, hash); ++ ++ while (*new) { ++ int cmp; ++ ++ tree_rmap_item = rb_entry(*new, struct rmap_item, ++ node); ++ ++ cmp = hash_cmp(hash_max, tree_rmap_item->hash_max); ++ parent = *new; ++ if (cmp < 0) ++ new = &parent->rb_left; ++ else if (cmp > 0) ++ new = &parent->rb_right; ++ else ++ goto get_page_out; ++ } ++ } else { ++ /* alloc a new tree_node */ ++ tree_node = alloc_tree_node(&unstable_tree_node_list); ++ if (!tree_node) ++ return NULL; ++ ++ tree_node->hash = hash; ++ rb_link_node(&tree_node->node, parent, new); ++ rb_insert_color(&tree_node->node, &root_unstable_tree); ++ parent = NULL; ++ new = &tree_node->sub_root.rb_node; ++ } ++ ++ /* did not found even in sub-tree */ ++ rmap_item->tree_node = tree_node; ++ rmap_item->address |= UNSTABLE_FLAG; ++ rmap_item->hash_round = uksm_hash_round; ++ rb_link_node(&rmap_item->node, parent, new); ++ rb_insert_color(&rmap_item->node, &tree_node->sub_root); ++ ++ uksm_pages_unshared++; ++ return NULL; ++ ++get_page_out: ++ if (tree_rmap_item->page == rmap_item->page) ++ return NULL; ++ ++ if (get_tree_rmap_item_page(tree_rmap_item)) ++ return NULL; ++ ++ return tree_rmap_item; ++} ++ ++static void hold_anon_vma(struct rmap_item *rmap_item, ++ struct anon_vma *anon_vma) ++{ ++ rmap_item->anon_vma = anon_vma; ++ get_anon_vma(anon_vma); ++} ++ ++ ++/** ++ * stable_tree_append() - append a rmap_item to a stable node. Deduplication ++ * ratio statistics is done in this function. ++ * ++ */ ++static void stable_tree_append(struct rmap_item *rmap_item, ++ struct stable_node *stable_node, int logdedup) ++{ ++ struct node_vma *node_vma = NULL, *new_node_vma, *node_vma_cont = NULL; ++ unsigned long key = (unsigned long)rmap_item->slot; ++ unsigned long factor = rmap_item->slot->rung->step; ++ ++ BUG_ON(!stable_node); ++ rmap_item->address |= STABLE_FLAG; ++ ++ if (hlist_empty(&stable_node->hlist)) { ++ uksm_pages_shared++; ++ goto node_vma_new; ++ } else { ++ uksm_pages_sharing++; ++ } ++ ++ hlist_for_each_entry(node_vma, &stable_node->hlist, hlist) { ++ if (node_vma->key >= key) ++ break; ++ ++ if (logdedup) { ++ node_vma->slot->pages_bemerged += factor; ++ if (list_empty(&node_vma->slot->dedup_list)) ++ list_add(&node_vma->slot->dedup_list, ++ &vma_slot_dedup); ++ } ++ } ++ ++ if (node_vma) { ++ if (node_vma->key == key) { ++ node_vma_cont = hlist_entry_safe(node_vma->hlist.next, struct node_vma, hlist); ++ goto node_vma_ok; ++ } else if (node_vma->key > key) { ++ node_vma_cont = node_vma; ++ } ++ } ++ ++node_vma_new: ++ /* no same vma already in node, alloc a new node_vma */ ++ new_node_vma = alloc_node_vma(); ++ BUG_ON(!new_node_vma); ++ new_node_vma->head = stable_node; ++ new_node_vma->slot = rmap_item->slot; ++ ++ if (!node_vma) { ++ hlist_add_head(&new_node_vma->hlist, &stable_node->hlist); ++ } else if (node_vma->key != key) { ++ if (node_vma->key < key) ++ hlist_add_behind(&new_node_vma->hlist, &node_vma->hlist); ++ else { ++ hlist_add_before(&new_node_vma->hlist, ++ &node_vma->hlist); ++ } ++ ++ } ++ node_vma = new_node_vma; ++ ++node_vma_ok: /* ok, ready to add to the list */ ++ rmap_item->head = node_vma; ++ hlist_add_head(&rmap_item->hlist, &node_vma->rmap_hlist); ++ hold_anon_vma(rmap_item, rmap_item->slot->vma->anon_vma); ++ if (logdedup) { ++ rmap_item->slot->pages_merged++; ++ if (node_vma_cont) { ++ node_vma = node_vma_cont; ++ hlist_for_each_entry_continue(node_vma, hlist) { ++ node_vma->slot->pages_bemerged += factor; ++ if (list_empty(&node_vma->slot->dedup_list)) ++ list_add(&node_vma->slot->dedup_list, ++ &vma_slot_dedup); ++ } ++ } ++ } ++} ++ ++/* ++ * We use break_ksm to break COW on a ksm page: it's a stripped down ++ * ++ * if (get_user_pages(addr, 1, 1, 1, &page, NULL) == 1) ++ * put_page(page); ++ * ++ * but taking great care only to touch a ksm page, in a VM_MERGEABLE vma, ++ * in case the application has unmapped and remapped mm,addr meanwhile. ++ * Could a ksm page appear anywhere else? Actually yes, in a VM_PFNMAP ++ * mmap of /dev/mem or /dev/kmem, where we would not want to touch it. ++ */ ++static int break_ksm(struct vm_area_struct *vma, unsigned long addr) ++{ ++ struct page *page; ++ int ret = 0; ++ ++ do { ++ cond_resched(); ++ page = follow_page(vma, addr, FOLL_GET | FOLL_MIGRATION | FOLL_REMOTE); ++ if (IS_ERR_OR_NULL(page)) ++ break; ++ if (PageKsm(page)) { ++ ret = handle_mm_fault(vma, addr, ++ FAULT_FLAG_WRITE | FAULT_FLAG_REMOTE); ++ } else ++ ret = VM_FAULT_WRITE; ++ put_page(page); ++ } while (!(ret & (VM_FAULT_WRITE | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | VM_FAULT_OOM))); ++ /* ++ * We must loop because handle_mm_fault() may back out if there's ++ * any difficulty e.g. if pte accessed bit gets updated concurrently. ++ * ++ * VM_FAULT_WRITE is what we have been hoping for: it indicates that ++ * COW has been broken, even if the vma does not permit VM_WRITE; ++ * but note that a concurrent fault might break PageKsm for us. ++ * ++ * VM_FAULT_SIGBUS could occur if we race with truncation of the ++ * backing file, which also invalidates anonymous pages: that's ++ * okay, that truncation will have unmapped the PageKsm for us. ++ * ++ * VM_FAULT_OOM: at the time of writing (late July 2009), setting ++ * aside mem_cgroup limits, VM_FAULT_OOM would only be set if the ++ * current task has TIF_MEMDIE set, and will be OOM killed on return ++ * to user; and ksmd, having no mm, would never be chosen for that. ++ * ++ * But if the mm is in a limited mem_cgroup, then the fault may fail ++ * with VM_FAULT_OOM even if the current task is not TIF_MEMDIE; and ++ * even ksmd can fail in this way - though it's usually breaking ksm ++ * just to undo a merge it made a moment before, so unlikely to oom. ++ * ++ * That's a pity: we might therefore have more kernel pages allocated ++ * than we're counting as nodes in the stable tree; but uksm_do_scan ++ * will retry to break_cow on each pass, so should recover the page ++ * in due course. The important thing is to not let VM_MERGEABLE ++ * be cleared while any such pages might remain in the area. ++ */ ++ return (ret & VM_FAULT_OOM) ? -ENOMEM : 0; ++} ++ ++static void break_cow(struct rmap_item *rmap_item) ++{ ++ struct vm_area_struct *vma = rmap_item->slot->vma; ++ struct mm_struct *mm = vma->vm_mm; ++ unsigned long addr = get_rmap_addr(rmap_item); ++ ++ if (uksm_test_exit(mm)) ++ goto out; ++ ++ break_ksm(vma, addr); ++out: ++ return; ++} ++ ++/* ++ * Though it's very tempting to unmerge in_stable_tree(rmap_item)s rather ++ * than check every pte of a given vma, the locking doesn't quite work for ++ * that - an rmap_item is assigned to the stable tree after inserting ksm ++ * page and upping mmap_sem. Nor does it fit with the way we skip dup'ing ++ * rmap_items from parent to child at fork time (so as not to waste time ++ * if exit comes before the next scan reaches it). ++ * ++ * Similarly, although we'd like to remove rmap_items (so updating counts ++ * and freeing memory) when unmerging an area, it's easier to leave that ++ * to the next pass of ksmd - consider, for example, how ksmd might be ++ * in cmp_and_merge_page on one of the rmap_items we would be removing. ++ */ ++inline int unmerge_uksm_pages(struct vm_area_struct *vma, ++ unsigned long start, unsigned long end) ++{ ++ unsigned long addr; ++ int err = 0; ++ ++ for (addr = start; addr < end && !err; addr += PAGE_SIZE) { ++ if (uksm_test_exit(vma->vm_mm)) ++ break; ++ if (signal_pending(current)) ++ err = -ERESTARTSYS; ++ else ++ err = break_ksm(vma, addr); ++ } ++ return err; ++} ++ ++static inline void inc_uksm_pages_scanned(void) ++{ ++ u64 delta; ++ ++ ++ if (uksm_pages_scanned == U64_MAX) { ++ encode_benefit(); ++ ++ delta = uksm_pages_scanned >> pages_scanned_base; ++ ++ if (CAN_OVERFLOW_U64(pages_scanned_stored, delta)) { ++ pages_scanned_stored >>= 1; ++ delta >>= 1; ++ pages_scanned_base++; ++ } ++ ++ pages_scanned_stored += delta; ++ ++ uksm_pages_scanned = uksm_pages_scanned_last = 0; ++ } ++ ++ uksm_pages_scanned++; ++} ++ ++static inline int find_zero_page_hash(int strength, u32 hash) ++{ ++ return (zero_hash_table[strength] == hash); ++} ++ ++static ++int cmp_and_merge_zero_page(struct vm_area_struct *vma, struct page *page) ++{ ++ struct page *zero_page = empty_uksm_zero_page; ++ struct mm_struct *mm = vma->vm_mm; ++ pte_t orig_pte = __pte(0); ++ int err = -EFAULT; ++ ++ if (uksm_test_exit(mm)) ++ goto out; ++ ++ if (!trylock_page(page)) ++ goto out; ++ ++ if (!PageAnon(page)) ++ goto out_unlock; ++ ++ if (PageTransCompound(page)) { ++ err = split_huge_page(page); ++ if (err) ++ goto out_unlock; ++ } ++ ++ if (write_protect_page(vma, page, &orig_pte, 0) == 0) { ++ if (is_page_full_zero(page)) ++ err = replace_page(vma, page, zero_page, orig_pte); ++ } ++ ++out_unlock: ++ unlock_page(page); ++out: ++ return err; ++} ++ ++/* ++ * cmp_and_merge_page() - first see if page can be merged into the stable ++ * tree; if not, compare hash to previous and if it's the same, see if page ++ * can be inserted into the unstable tree, or merged with a page already there ++ * and both transferred to the stable tree. ++ * ++ * @page: the page that we are searching identical page to. ++ * @rmap_item: the reverse mapping into the virtual address of this page ++ */ ++static void cmp_and_merge_page(struct rmap_item *rmap_item, u32 hash) ++{ ++ struct rmap_item *tree_rmap_item; ++ struct page *page; ++ struct page *kpage = NULL; ++ u32 hash_max; ++ int err; ++ unsigned int success1, success2; ++ struct stable_node *snode; ++ int cmp; ++ struct rb_node *parent = NULL, **new; ++ ++ remove_rmap_item_from_tree(rmap_item); ++ page = rmap_item->page; ++ ++ /* We first start with searching the page inside the stable tree */ ++ kpage = stable_tree_search(rmap_item, hash); ++ if (kpage) { ++ err = try_to_merge_with_uksm_page(rmap_item, kpage, ++ hash); ++ if (!err) { ++ /* ++ * The page was successfully merged, add ++ * its rmap_item to the stable tree. ++ * page lock is needed because it's ++ * racing with try_to_unmap_ksm(), etc. ++ */ ++ lock_page(kpage); ++ snode = page_stable_node(kpage); ++ stable_tree_append(rmap_item, snode, 1); ++ unlock_page(kpage); ++ put_page(kpage); ++ return; /* success */ ++ } ++ put_page(kpage); ++ ++ /* ++ * if it's a collision and it has been search in sub-rbtree ++ * (hash_max != 0), we want to abort, because if it is ++ * successfully merged in unstable tree, the collision trends to ++ * happen again. ++ */ ++ if (err == MERGE_ERR_COLLI && rmap_item->hash_max) ++ return; ++ } ++ ++ tree_rmap_item = ++ unstable_tree_search_insert(rmap_item, hash); ++ if (tree_rmap_item) { ++ err = try_to_merge_two_pages(rmap_item, tree_rmap_item, hash); ++ /* ++ * As soon as we merge this page, we want to remove the ++ * rmap_item of the page we have merged with from the unstable ++ * tree, and insert it instead as new node in the stable tree. ++ */ ++ if (!err) { ++ kpage = page; ++ remove_rmap_item_from_tree(tree_rmap_item); ++ lock_page(kpage); ++ snode = stable_tree_insert(&kpage, hash, ++ rmap_item, tree_rmap_item, ++ &success1, &success2); ++ ++ /* ++ * Do not log dedup for tree item, it's not counted as ++ * scanned in this round. ++ */ ++ if (success2) ++ stable_tree_append(tree_rmap_item, snode, 0); ++ ++ /* ++ * The order of these two stable append is important: ++ * we are scanning rmap_item. ++ */ ++ if (success1) ++ stable_tree_append(rmap_item, snode, 1); ++ ++ /* ++ * The original kpage may be unlocked inside ++ * stable_tree_insert() already. This page ++ * should be unlocked before doing ++ * break_cow(). ++ */ ++ unlock_page(kpage); ++ ++ if (!success1) ++ break_cow(rmap_item); ++ ++ if (!success2) ++ break_cow(tree_rmap_item); ++ ++ } else if (err == MERGE_ERR_COLLI) { ++ BUG_ON(tree_rmap_item->tree_node->count > 1); ++ ++ rmap_item_hash_max(tree_rmap_item, ++ tree_rmap_item->tree_node->hash); ++ ++ hash_max = rmap_item_hash_max(rmap_item, hash); ++ cmp = hash_cmp(hash_max, tree_rmap_item->hash_max); ++ parent = &tree_rmap_item->node; ++ if (cmp < 0) ++ new = &parent->rb_left; ++ else if (cmp > 0) ++ new = &parent->rb_right; ++ else ++ goto put_up_out; ++ ++ rmap_item->tree_node = tree_rmap_item->tree_node; ++ rmap_item->address |= UNSTABLE_FLAG; ++ rmap_item->hash_round = uksm_hash_round; ++ rb_link_node(&rmap_item->node, parent, new); ++ rb_insert_color(&rmap_item->node, ++ &tree_rmap_item->tree_node->sub_root); ++ rmap_item->tree_node->count++; ++ } else { ++ /* ++ * either one of the page has changed or they collide ++ * at the max hash, we consider them as ill items. ++ */ ++ remove_rmap_item_from_tree(tree_rmap_item); ++ } ++put_up_out: ++ put_page(tree_rmap_item->page); ++ up_read(&tree_rmap_item->slot->vma->vm_mm->mmap_sem); ++ } ++} ++ ++ ++ ++ ++static inline unsigned long get_pool_index(struct vma_slot *slot, ++ unsigned long index) ++{ ++ unsigned long pool_index; ++ ++ pool_index = (sizeof(struct rmap_list_entry *) * index) >> PAGE_SHIFT; ++ if (pool_index >= slot->pool_size) ++ BUG(); ++ return pool_index; ++} ++ ++static inline unsigned long index_page_offset(unsigned long index) ++{ ++ return offset_in_page(sizeof(struct rmap_list_entry *) * index); ++} ++ ++static inline ++struct rmap_list_entry *get_rmap_list_entry(struct vma_slot *slot, ++ unsigned long index, int need_alloc) ++{ ++ unsigned long pool_index; ++ struct page *page; ++ void *addr; ++ ++ ++ pool_index = get_pool_index(slot, index); ++ if (!slot->rmap_list_pool[pool_index]) { ++ if (!need_alloc) ++ return NULL; ++ ++ page = alloc_page(GFP_KERNEL | __GFP_ZERO | __GFP_NOWARN); ++ if (!page) ++ return NULL; ++ ++ slot->rmap_list_pool[pool_index] = page; ++ } ++ ++ addr = kmap(slot->rmap_list_pool[pool_index]); ++ addr += index_page_offset(index); ++ ++ return addr; ++} ++ ++static inline void put_rmap_list_entry(struct vma_slot *slot, ++ unsigned long index) ++{ ++ unsigned long pool_index; ++ ++ pool_index = get_pool_index(slot, index); ++ BUG_ON(!slot->rmap_list_pool[pool_index]); ++ kunmap(slot->rmap_list_pool[pool_index]); ++} ++ ++static inline int entry_is_new(struct rmap_list_entry *entry) ++{ ++ return !entry->item; ++} ++ ++static inline unsigned long get_index_orig_addr(struct vma_slot *slot, ++ unsigned long index) ++{ ++ return slot->vma->vm_start + (index << PAGE_SHIFT); ++} ++ ++static inline unsigned long get_entry_address(struct rmap_list_entry *entry) ++{ ++ unsigned long addr; ++ ++ if (is_addr(entry->addr)) ++ addr = get_clean_addr(entry->addr); ++ else if (entry->item) ++ addr = get_rmap_addr(entry->item); ++ else ++ BUG(); ++ ++ return addr; ++} ++ ++static inline struct rmap_item *get_entry_item(struct rmap_list_entry *entry) ++{ ++ if (is_addr(entry->addr)) ++ return NULL; ++ ++ return entry->item; ++} ++ ++static inline void inc_rmap_list_pool_count(struct vma_slot *slot, ++ unsigned long index) ++{ ++ unsigned long pool_index; ++ ++ pool_index = get_pool_index(slot, index); ++ BUG_ON(!slot->rmap_list_pool[pool_index]); ++ slot->pool_counts[pool_index]++; ++} ++ ++static inline void dec_rmap_list_pool_count(struct vma_slot *slot, ++ unsigned long index) ++{ ++ unsigned long pool_index; ++ ++ pool_index = get_pool_index(slot, index); ++ BUG_ON(!slot->rmap_list_pool[pool_index]); ++ BUG_ON(!slot->pool_counts[pool_index]); ++ slot->pool_counts[pool_index]--; ++} ++ ++static inline int entry_has_rmap(struct rmap_list_entry *entry) ++{ ++ return !is_addr(entry->addr) && entry->item; ++} ++ ++static inline void swap_entries(struct rmap_list_entry *entry1, ++ unsigned long index1, ++ struct rmap_list_entry *entry2, ++ unsigned long index2) ++{ ++ struct rmap_list_entry tmp; ++ ++ /* swapping two new entries is meaningless */ ++ BUG_ON(entry_is_new(entry1) && entry_is_new(entry2)); ++ ++ tmp = *entry1; ++ *entry1 = *entry2; ++ *entry2 = tmp; ++ ++ if (entry_has_rmap(entry1)) ++ entry1->item->entry_index = index1; ++ ++ if (entry_has_rmap(entry2)) ++ entry2->item->entry_index = index2; ++ ++ if (entry_has_rmap(entry1) && !entry_has_rmap(entry2)) { ++ inc_rmap_list_pool_count(entry1->item->slot, index1); ++ dec_rmap_list_pool_count(entry1->item->slot, index2); ++ } else if (!entry_has_rmap(entry1) && entry_has_rmap(entry2)) { ++ inc_rmap_list_pool_count(entry2->item->slot, index2); ++ dec_rmap_list_pool_count(entry2->item->slot, index1); ++ } ++} ++ ++static inline void free_entry_item(struct rmap_list_entry *entry) ++{ ++ unsigned long index; ++ struct rmap_item *item; ++ ++ if (!is_addr(entry->addr)) { ++ BUG_ON(!entry->item); ++ item = entry->item; ++ entry->addr = get_rmap_addr(item); ++ set_is_addr(entry->addr); ++ index = item->entry_index; ++ remove_rmap_item_from_tree(item); ++ dec_rmap_list_pool_count(item->slot, index); ++ free_rmap_item(item); ++ } ++} ++ ++static inline int pool_entry_boundary(unsigned long index) ++{ ++ unsigned long linear_addr; ++ ++ linear_addr = sizeof(struct rmap_list_entry *) * index; ++ return index && !offset_in_page(linear_addr); ++} ++ ++static inline void try_free_last_pool(struct vma_slot *slot, ++ unsigned long index) ++{ ++ unsigned long pool_index; ++ ++ pool_index = get_pool_index(slot, index); ++ if (slot->rmap_list_pool[pool_index] && ++ !slot->pool_counts[pool_index]) { ++ __free_page(slot->rmap_list_pool[pool_index]); ++ slot->rmap_list_pool[pool_index] = NULL; ++ slot->flags |= UKSM_SLOT_NEED_SORT; ++ } ++ ++} ++ ++static inline unsigned long vma_item_index(struct vm_area_struct *vma, ++ struct rmap_item *item) ++{ ++ return (get_rmap_addr(item) - vma->vm_start) >> PAGE_SHIFT; ++} ++ ++static int within_same_pool(struct vma_slot *slot, ++ unsigned long i, unsigned long j) ++{ ++ unsigned long pool_i, pool_j; ++ ++ pool_i = get_pool_index(slot, i); ++ pool_j = get_pool_index(slot, j); ++ ++ return (pool_i == pool_j); ++} ++ ++static void sort_rmap_entry_list(struct vma_slot *slot) ++{ ++ unsigned long i, j; ++ struct rmap_list_entry *entry, *swap_entry; ++ ++ entry = get_rmap_list_entry(slot, 0, 0); ++ for (i = 0; i < slot->pages; ) { ++ ++ if (!entry) ++ goto skip_whole_pool; ++ ++ if (entry_is_new(entry)) ++ goto next_entry; ++ ++ if (is_addr(entry->addr)) { ++ entry->addr = 0; ++ goto next_entry; ++ } ++ ++ j = vma_item_index(slot->vma, entry->item); ++ if (j == i) ++ goto next_entry; ++ ++ if (within_same_pool(slot, i, j)) ++ swap_entry = entry + j - i; ++ else ++ swap_entry = get_rmap_list_entry(slot, j, 1); ++ ++ swap_entries(entry, i, swap_entry, j); ++ if (!within_same_pool(slot, i, j)) ++ put_rmap_list_entry(slot, j); ++ continue; ++ ++skip_whole_pool: ++ i += PAGE_SIZE / sizeof(*entry); ++ if (i < slot->pages) ++ entry = get_rmap_list_entry(slot, i, 0); ++ continue; ++ ++next_entry: ++ if (i >= slot->pages - 1 || ++ !within_same_pool(slot, i, i + 1)) { ++ put_rmap_list_entry(slot, i); ++ if (i + 1 < slot->pages) ++ entry = get_rmap_list_entry(slot, i + 1, 0); ++ } else ++ entry++; ++ i++; ++ continue; ++ } ++ ++ /* free empty pool entries which contain no rmap_item */ ++ /* CAN be simplied to based on only pool_counts when bug freed !!!!! */ ++ for (i = 0; i < slot->pool_size; i++) { ++ unsigned char has_rmap; ++ void *addr; ++ ++ if (!slot->rmap_list_pool[i]) ++ continue; ++ ++ has_rmap = 0; ++ addr = kmap(slot->rmap_list_pool[i]); ++ BUG_ON(!addr); ++ for (j = 0; j < PAGE_SIZE / sizeof(*entry); j++) { ++ entry = (struct rmap_list_entry *)addr + j; ++ if (is_addr(entry->addr)) ++ continue; ++ if (!entry->item) ++ continue; ++ has_rmap = 1; ++ } ++ kunmap(slot->rmap_list_pool[i]); ++ if (!has_rmap) { ++ BUG_ON(slot->pool_counts[i]); ++ __free_page(slot->rmap_list_pool[i]); ++ slot->rmap_list_pool[i] = NULL; ++ } ++ } ++ ++ slot->flags &= ~UKSM_SLOT_NEED_SORT; ++} ++ ++/* ++ * vma_fully_scanned() - if all the pages in this slot have been scanned. ++ */ ++static inline int vma_fully_scanned(struct vma_slot *slot) ++{ ++ return slot->pages_scanned == slot->pages; ++} ++ ++/** ++ * get_next_rmap_item() - Get the next rmap_item in a vma_slot according to ++ * its random permutation. This function is embedded with the random ++ * permutation index management code. ++ */ ++static struct rmap_item *get_next_rmap_item(struct vma_slot *slot, u32 *hash) ++{ ++ unsigned long rand_range, addr, swap_index, scan_index; ++ struct rmap_item *item = NULL; ++ struct rmap_list_entry *scan_entry, *swap_entry = NULL; ++ struct page *page; ++ ++ scan_index = swap_index = slot->pages_scanned % slot->pages; ++ ++ if (pool_entry_boundary(scan_index)) ++ try_free_last_pool(slot, scan_index - 1); ++ ++ if (vma_fully_scanned(slot)) { ++ if (slot->flags & UKSM_SLOT_NEED_SORT) ++ slot->flags |= UKSM_SLOT_NEED_RERAND; ++ else ++ slot->flags &= ~UKSM_SLOT_NEED_RERAND; ++ if (slot->flags & UKSM_SLOT_NEED_SORT) ++ sort_rmap_entry_list(slot); ++ } ++ ++ scan_entry = get_rmap_list_entry(slot, scan_index, 1); ++ if (!scan_entry) ++ return NULL; ++ ++ if (entry_is_new(scan_entry)) { ++ scan_entry->addr = get_index_orig_addr(slot, scan_index); ++ set_is_addr(scan_entry->addr); ++ } ++ ++ if (slot->flags & UKSM_SLOT_NEED_RERAND) { ++ rand_range = slot->pages - scan_index; ++ BUG_ON(!rand_range); ++ swap_index = scan_index + (prandom_u32() % rand_range); ++ } ++ ++ if (swap_index != scan_index) { ++ swap_entry = get_rmap_list_entry(slot, swap_index, 1); ++ if (entry_is_new(swap_entry)) { ++ swap_entry->addr = get_index_orig_addr(slot, ++ swap_index); ++ set_is_addr(swap_entry->addr); ++ } ++ swap_entries(scan_entry, scan_index, swap_entry, swap_index); ++ } ++ ++ addr = get_entry_address(scan_entry); ++ item = get_entry_item(scan_entry); ++ BUG_ON(addr > slot->vma->vm_end || addr < slot->vma->vm_start); ++ ++ page = follow_page(slot->vma, addr, FOLL_GET); ++ if (IS_ERR_OR_NULL(page)) ++ goto nopage; ++ ++ if (!PageAnon(page)) ++ goto putpage; ++ ++ /*check is zero_page pfn or uksm_zero_page*/ ++ if ((page_to_pfn(page) == zero_pfn) ++ || (page_to_pfn(page) == uksm_zero_pfn)) ++ goto putpage; ++ ++ flush_anon_page(slot->vma, page, addr); ++ flush_dcache_page(page); ++ ++ ++ *hash = page_hash(page, hash_strength, 1); ++ inc_uksm_pages_scanned(); ++ /*if the page content all zero, re-map to zero-page*/ ++ if (find_zero_page_hash(hash_strength, *hash)) { ++ if (!cmp_and_merge_zero_page(slot->vma, page)) { ++ slot->pages_merged++; ++ ++ /* For full-zero pages, no need to create rmap item */ ++ goto putpage; ++ } else { ++ inc_rshash_neg(memcmp_cost / 2); ++ } ++ } ++ ++ if (!item) { ++ item = alloc_rmap_item(); ++ if (item) { ++ /* It has already been zeroed */ ++ item->slot = slot; ++ item->address = addr; ++ item->entry_index = scan_index; ++ scan_entry->item = item; ++ inc_rmap_list_pool_count(slot, scan_index); ++ } else ++ goto putpage; ++ } ++ ++ BUG_ON(item->slot != slot); ++ /* the page may have changed */ ++ item->page = page; ++ put_rmap_list_entry(slot, scan_index); ++ if (swap_entry) ++ put_rmap_list_entry(slot, swap_index); ++ return item; ++ ++putpage: ++ put_page(page); ++ page = NULL; ++nopage: ++ /* no page, store addr back and free rmap_item if possible */ ++ free_entry_item(scan_entry); ++ put_rmap_list_entry(slot, scan_index); ++ if (swap_entry) ++ put_rmap_list_entry(slot, swap_index); ++ return NULL; ++} ++ ++static inline int in_stable_tree(struct rmap_item *rmap_item) ++{ ++ return rmap_item->address & STABLE_FLAG; ++} ++ ++/** ++ * scan_vma_one_page() - scan the next page in a vma_slot. Called with ++ * mmap_sem locked. ++ */ ++static noinline void scan_vma_one_page(struct vma_slot *slot) ++{ ++ u32 hash; ++ struct mm_struct *mm; ++ struct rmap_item *rmap_item = NULL; ++ struct vm_area_struct *vma = slot->vma; ++ ++ mm = vma->vm_mm; ++ BUG_ON(!mm); ++ BUG_ON(!slot); ++ ++ rmap_item = get_next_rmap_item(slot, &hash); ++ if (!rmap_item) ++ goto out1; ++ ++ if (PageKsm(rmap_item->page) && in_stable_tree(rmap_item)) ++ goto out2; ++ ++ cmp_and_merge_page(rmap_item, hash); ++out2: ++ put_page(rmap_item->page); ++out1: ++ slot->pages_scanned++; ++ slot->this_sampled++; ++ if (slot->fully_scanned_round != fully_scanned_round) ++ scanned_virtual_pages++; ++ ++ if (vma_fully_scanned(slot)) ++ slot->fully_scanned_round = fully_scanned_round; ++} ++ ++static inline unsigned long rung_get_pages(struct scan_rung *rung) ++{ ++ struct slot_tree_node *node; ++ ++ if (!rung->vma_root.rnode) ++ return 0; ++ ++ node = container_of(rung->vma_root.rnode, struct slot_tree_node, snode); ++ ++ return node->size; ++} ++ ++#define RUNG_SAMPLED_MIN 3 ++ ++static inline ++void uksm_calc_rung_step(struct scan_rung *rung, ++ unsigned long page_time, unsigned long ratio) ++{ ++ unsigned long sampled, pages; ++ ++ /* will be fully scanned ? */ ++ if (!rung->cover_msecs) { ++ rung->step = 1; ++ return; ++ } ++ ++ sampled = rung->cover_msecs * (NSEC_PER_MSEC / TIME_RATIO_SCALE) ++ * ratio / page_time; ++ ++ /* ++ * Before we finsish a scan round and expensive per-round jobs, ++ * we need to have a chance to estimate the per page time. So ++ * the sampled number can not be too small. ++ */ ++ if (sampled < RUNG_SAMPLED_MIN) ++ sampled = RUNG_SAMPLED_MIN; ++ ++ pages = rung_get_pages(rung); ++ if (likely(pages > sampled)) ++ rung->step = pages / sampled; ++ else ++ rung->step = 1; ++} ++ ++static inline int step_need_recalc(struct scan_rung *rung) ++{ ++ unsigned long pages, stepmax; ++ ++ pages = rung_get_pages(rung); ++ stepmax = pages / RUNG_SAMPLED_MIN; ++ ++ return pages && (rung->step > pages || ++ (stepmax && rung->step > stepmax)); ++} ++ ++static inline ++void reset_current_scan(struct scan_rung *rung, int finished, int step_recalc) ++{ ++ struct vma_slot *slot; ++ ++ if (finished) ++ rung->flags |= UKSM_RUNG_ROUND_FINISHED; ++ ++ if (step_recalc || step_need_recalc(rung)) { ++ uksm_calc_rung_step(rung, uksm_ema_page_time, rung->cpu_ratio); ++ BUG_ON(step_need_recalc(rung)); ++ } ++ ++ slot_iter_index = prandom_u32() % rung->step; ++ BUG_ON(!rung->vma_root.rnode); ++ slot = sradix_tree_next(&rung->vma_root, NULL, 0, slot_iter); ++ BUG_ON(!slot); ++ ++ rung->current_scan = slot; ++ rung->current_offset = slot_iter_index; ++} ++ ++static inline struct sradix_tree_root *slot_get_root(struct vma_slot *slot) ++{ ++ return &slot->rung->vma_root; ++} ++ ++/* ++ * return if resetted. ++ */ ++static int advance_current_scan(struct scan_rung *rung) ++{ ++ unsigned short n; ++ struct vma_slot *slot, *next = NULL; ++ ++ BUG_ON(!rung->vma_root.num); ++ ++ slot = rung->current_scan; ++ n = (slot->pages - rung->current_offset) % rung->step; ++ slot_iter_index = rung->step - n; ++ next = sradix_tree_next(&rung->vma_root, slot->snode, ++ slot->sindex, slot_iter); ++ ++ if (next) { ++ rung->current_offset = slot_iter_index; ++ rung->current_scan = next; ++ return 0; ++ } else { ++ reset_current_scan(rung, 1, 0); ++ return 1; ++ } ++} ++ ++static inline void rung_rm_slot(struct vma_slot *slot) ++{ ++ struct scan_rung *rung = slot->rung; ++ struct sradix_tree_root *root; ++ ++ if (rung->current_scan == slot) ++ advance_current_scan(rung); ++ ++ root = slot_get_root(slot); ++ sradix_tree_delete_from_leaf(root, slot->snode, slot->sindex); ++ slot->snode = NULL; ++ if (step_need_recalc(rung)) { ++ uksm_calc_rung_step(rung, uksm_ema_page_time, rung->cpu_ratio); ++ BUG_ON(step_need_recalc(rung)); ++ } ++ ++ /* In case advance_current_scan loop back to this slot again */ ++ if (rung->vma_root.num && rung->current_scan == slot) ++ reset_current_scan(slot->rung, 1, 0); ++} ++ ++static inline void rung_add_new_slots(struct scan_rung *rung, ++ struct vma_slot **slots, unsigned long num) ++{ ++ int err; ++ struct vma_slot *slot; ++ unsigned long i; ++ struct sradix_tree_root *root = &rung->vma_root; ++ ++ err = sradix_tree_enter(root, (void **)slots, num); ++ BUG_ON(err); ++ ++ for (i = 0; i < num; i++) { ++ slot = slots[i]; ++ slot->rung = rung; ++ BUG_ON(vma_fully_scanned(slot)); ++ } ++ ++ if (rung->vma_root.num == num) ++ reset_current_scan(rung, 0, 1); ++} ++ ++static inline int rung_add_one_slot(struct scan_rung *rung, ++ struct vma_slot *slot) ++{ ++ int err; ++ ++ err = sradix_tree_enter(&rung->vma_root, (void **)&slot, 1); ++ if (err) ++ return err; ++ ++ slot->rung = rung; ++ if (rung->vma_root.num == 1) ++ reset_current_scan(rung, 0, 1); ++ ++ return 0; ++} ++ ++/* ++ * Return true if the slot is deleted from its rung. ++ */ ++static inline int vma_rung_enter(struct vma_slot *slot, struct scan_rung *rung) ++{ ++ struct scan_rung *old_rung = slot->rung; ++ int err; ++ ++ if (old_rung == rung) ++ return 0; ++ ++ rung_rm_slot(slot); ++ err = rung_add_one_slot(rung, slot); ++ if (err) { ++ err = rung_add_one_slot(old_rung, slot); ++ WARN_ON(err); /* OOPS, badly OOM, we lost this slot */ ++ } ++ ++ return 1; ++} ++ ++static inline int vma_rung_up(struct vma_slot *slot) ++{ ++ struct scan_rung *rung; ++ ++ rung = slot->rung; ++ if (slot->rung != &uksm_scan_ladder[SCAN_LADDER_SIZE-1]) ++ rung++; ++ ++ return vma_rung_enter(slot, rung); ++} ++ ++static inline int vma_rung_down(struct vma_slot *slot) ++{ ++ struct scan_rung *rung; ++ ++ rung = slot->rung; ++ if (slot->rung != &uksm_scan_ladder[0]) ++ rung--; ++ ++ return vma_rung_enter(slot, rung); ++} ++ ++/** ++ * cal_dedup_ratio() - Calculate the deduplication ratio for this slot. ++ */ ++static unsigned long cal_dedup_ratio(struct vma_slot *slot) ++{ ++ unsigned long ret; ++ unsigned long pages; ++ ++ pages = slot->this_sampled; ++ if (!pages) ++ return 0; ++ ++ BUG_ON(slot->pages_scanned == slot->last_scanned); ++ ++ ret = slot->pages_merged; ++ ++ /* Thrashing area filtering */ ++ if (ret && uksm_thrash_threshold) { ++ if (slot->pages_cowed * 100 / slot->pages_merged ++ > uksm_thrash_threshold) { ++ ret = 0; ++ } else { ++ ret = slot->pages_merged - slot->pages_cowed; ++ } ++ } ++ ++ return ret * 100 / pages; ++} ++ ++/** ++ * cal_dedup_ratio() - Calculate the deduplication ratio for this slot. ++ */ ++static unsigned long cal_dedup_ratio_old(struct vma_slot *slot) ++{ ++ unsigned long ret; ++ unsigned long pages; ++ ++ pages = slot->pages; ++ if (!pages) ++ return 0; ++ ++ ret = slot->pages_bemerged; ++ ++ /* Thrashing area filtering */ ++ if (ret && uksm_thrash_threshold) { ++ if (slot->pages_cowed * 100 / slot->pages_bemerged ++ > uksm_thrash_threshold) { ++ ret = 0; ++ } else { ++ ret = slot->pages_bemerged - slot->pages_cowed; ++ } ++ } ++ ++ return ret * 100 / pages; ++} ++ ++/** ++ * stable_node_reinsert() - When the hash_strength has been adjusted, the ++ * stable tree need to be restructured, this is the function re-inserting the ++ * stable node. ++ */ ++static inline void stable_node_reinsert(struct stable_node *new_node, ++ struct page *page, ++ struct rb_root *root_treep, ++ struct list_head *tree_node_listp, ++ u32 hash) ++{ ++ struct rb_node **new = &root_treep->rb_node; ++ struct rb_node *parent = NULL; ++ struct stable_node *stable_node; ++ struct tree_node *tree_node; ++ struct page *tree_page; ++ int cmp; ++ ++ while (*new) { ++ int cmp; ++ ++ tree_node = rb_entry(*new, struct tree_node, node); ++ ++ cmp = hash_cmp(hash, tree_node->hash); ++ ++ if (cmp < 0) { ++ parent = *new; ++ new = &parent->rb_left; ++ } else if (cmp > 0) { ++ parent = *new; ++ new = &parent->rb_right; ++ } else ++ break; ++ } ++ ++ if (*new) { ++ /* find a stable tree node with same first level hash value */ ++ stable_node_hash_max(new_node, page, hash); ++ if (tree_node->count == 1) { ++ stable_node = rb_entry(tree_node->sub_root.rb_node, ++ struct stable_node, node); ++ tree_page = get_uksm_page(stable_node, 1, 0); ++ if (tree_page) { ++ stable_node_hash_max(stable_node, ++ tree_page, hash); ++ put_page(tree_page); ++ ++ /* prepare for stable node insertion */ ++ ++ cmp = hash_cmp(new_node->hash_max, ++ stable_node->hash_max); ++ parent = &stable_node->node; ++ if (cmp < 0) ++ new = &parent->rb_left; ++ else if (cmp > 0) ++ new = &parent->rb_right; ++ else ++ goto failed; ++ ++ goto add_node; ++ } else { ++ /* the only stable_node deleted, the tree node ++ * was not deleted. ++ */ ++ goto tree_node_reuse; ++ } ++ } ++ ++ /* well, search the collision subtree */ ++ new = &tree_node->sub_root.rb_node; ++ parent = NULL; ++ BUG_ON(!*new); ++ while (*new) { ++ int cmp; ++ ++ stable_node = rb_entry(*new, struct stable_node, node); ++ ++ cmp = hash_cmp(new_node->hash_max, ++ stable_node->hash_max); ++ ++ if (cmp < 0) { ++ parent = *new; ++ new = &parent->rb_left; ++ } else if (cmp > 0) { ++ parent = *new; ++ new = &parent->rb_right; ++ } else { ++ /* oh, no, still a collision */ ++ goto failed; ++ } ++ } ++ ++ goto add_node; ++ } ++ ++ /* no tree node found */ ++ tree_node = alloc_tree_node(tree_node_listp); ++ if (!tree_node) { ++ pr_err("UKSM: memory allocation error!\n"); ++ goto failed; ++ } else { ++ tree_node->hash = hash; ++ rb_link_node(&tree_node->node, parent, new); ++ rb_insert_color(&tree_node->node, root_treep); ++ ++tree_node_reuse: ++ /* prepare for stable node insertion */ ++ parent = NULL; ++ new = &tree_node->sub_root.rb_node; ++ } ++ ++add_node: ++ rb_link_node(&new_node->node, parent, new); ++ rb_insert_color(&new_node->node, &tree_node->sub_root); ++ new_node->tree_node = tree_node; ++ tree_node->count++; ++ return; ++ ++failed: ++ /* This can only happen when two nodes have collided ++ * in two levels. ++ */ ++ new_node->tree_node = NULL; ++ return; ++} ++ ++static inline void free_all_tree_nodes(struct list_head *list) ++{ ++ struct tree_node *node, *tmp; ++ ++ list_for_each_entry_safe(node, tmp, list, all_list) { ++ free_tree_node(node); ++ } ++} ++ ++/** ++ * stable_tree_delta_hash() - Delta hash the stable tree from previous hash ++ * strength to the current hash_strength. It re-structures the hole tree. ++ */ ++static inline void stable_tree_delta_hash(u32 prev_hash_strength) ++{ ++ struct stable_node *node, *tmp; ++ struct rb_root *root_new_treep; ++ struct list_head *new_tree_node_listp; ++ ++ stable_tree_index = (stable_tree_index + 1) % 2; ++ root_new_treep = &root_stable_tree[stable_tree_index]; ++ new_tree_node_listp = &stable_tree_node_list[stable_tree_index]; ++ *root_new_treep = RB_ROOT; ++ BUG_ON(!list_empty(new_tree_node_listp)); ++ ++ /* ++ * we need to be safe, the node could be removed by get_uksm_page() ++ */ ++ list_for_each_entry_safe(node, tmp, &stable_node_list, all_list) { ++ void *addr; ++ struct page *node_page; ++ u32 hash; ++ ++ /* ++ * We are completely re-structuring the stable nodes to a new ++ * stable tree. We don't want to touch the old tree unlinks and ++ * old tree_nodes. The old tree_nodes will be freed at once. ++ */ ++ node_page = get_uksm_page(node, 0, 0); ++ if (!node_page) ++ continue; ++ ++ if (node->tree_node) { ++ hash = node->tree_node->hash; ++ ++ addr = kmap_atomic(node_page); ++ ++ hash = delta_hash(addr, prev_hash_strength, ++ hash_strength, hash); ++ kunmap_atomic(addr); ++ } else { ++ /* ++ *it was not inserted to rbtree due to collision in last ++ *round scan. ++ */ ++ hash = page_hash(node_page, hash_strength, 0); ++ } ++ ++ stable_node_reinsert(node, node_page, root_new_treep, ++ new_tree_node_listp, hash); ++ put_page(node_page); ++ } ++ ++ root_stable_treep = root_new_treep; ++ free_all_tree_nodes(stable_tree_node_listp); ++ BUG_ON(!list_empty(stable_tree_node_listp)); ++ stable_tree_node_listp = new_tree_node_listp; ++} ++ ++static inline void inc_hash_strength(unsigned long delta) ++{ ++ hash_strength += 1 << delta; ++ if (hash_strength > HASH_STRENGTH_MAX) ++ hash_strength = HASH_STRENGTH_MAX; ++} ++ ++static inline void dec_hash_strength(unsigned long delta) ++{ ++ unsigned long change = 1 << delta; ++ ++ if (hash_strength <= change + 1) ++ hash_strength = 1; ++ else ++ hash_strength -= change; ++} ++ ++static inline void inc_hash_strength_delta(void) ++{ ++ hash_strength_delta++; ++ if (hash_strength_delta > HASH_STRENGTH_DELTA_MAX) ++ hash_strength_delta = HASH_STRENGTH_DELTA_MAX; ++} ++ ++static inline unsigned long get_current_neg_ratio(void) ++{ ++ u64 pos = benefit.pos; ++ u64 neg = benefit.neg; ++ ++ if (!neg) ++ return 0; ++ ++ if (!pos || neg > pos) ++ return 100; ++ ++ if (neg > div64_u64(U64_MAX, 100)) ++ pos = div64_u64(pos, 100); ++ else ++ neg *= 100; ++ ++ return div64_u64(neg, pos); ++} ++ ++static inline unsigned long get_current_benefit(void) ++{ ++ u64 pos = benefit.pos; ++ u64 neg = benefit.neg; ++ u64 scanned = benefit.scanned; ++ ++ if (neg > pos) ++ return 0; ++ ++ return div64_u64((pos - neg), scanned); ++} ++ ++static inline int judge_rshash_direction(void) ++{ ++ u64 current_neg_ratio, stable_benefit; ++ u64 current_benefit, delta = 0; ++ int ret = STILL; ++ ++ /* ++ * Try to probe a value after the boot, and in case the system ++ * are still for a long time. ++ */ ++ if ((fully_scanned_round & 0xFFULL) == 10) { ++ ret = OBSCURE; ++ goto out; ++ } ++ ++ current_neg_ratio = get_current_neg_ratio(); ++ ++ if (current_neg_ratio == 0) { ++ rshash_neg_cont_zero++; ++ if (rshash_neg_cont_zero > 2) ++ return GO_DOWN; ++ else ++ return STILL; ++ } ++ rshash_neg_cont_zero = 0; ++ ++ if (current_neg_ratio > 90) { ++ ret = GO_UP; ++ goto out; ++ } ++ ++ current_benefit = get_current_benefit(); ++ stable_benefit = rshash_state.stable_benefit; ++ ++ if (!stable_benefit) { ++ ret = OBSCURE; ++ goto out; ++ } ++ ++ if (current_benefit > stable_benefit) ++ delta = current_benefit - stable_benefit; ++ else if (current_benefit < stable_benefit) ++ delta = stable_benefit - current_benefit; ++ ++ delta = div64_u64(100 * delta, stable_benefit); ++ ++ if (delta > 50) { ++ rshash_cont_obscure++; ++ if (rshash_cont_obscure > 2) ++ return OBSCURE; ++ else ++ return STILL; ++ } ++ ++out: ++ rshash_cont_obscure = 0; ++ return ret; ++} ++ ++/** ++ * rshash_adjust() - The main function to control the random sampling state ++ * machine for hash strength adapting. ++ * ++ * return true if hash_strength has changed. ++ */ ++static inline int rshash_adjust(void) ++{ ++ unsigned long prev_hash_strength = hash_strength; ++ ++ if (!encode_benefit()) ++ return 0; ++ ++ switch (rshash_state.state) { ++ case RSHASH_STILL: ++ switch (judge_rshash_direction()) { ++ case GO_UP: ++ if (rshash_state.pre_direct == GO_DOWN) ++ hash_strength_delta = 0; ++ ++ inc_hash_strength(hash_strength_delta); ++ inc_hash_strength_delta(); ++ rshash_state.stable_benefit = get_current_benefit(); ++ rshash_state.pre_direct = GO_UP; ++ break; ++ ++ case GO_DOWN: ++ if (rshash_state.pre_direct == GO_UP) ++ hash_strength_delta = 0; ++ ++ dec_hash_strength(hash_strength_delta); ++ inc_hash_strength_delta(); ++ rshash_state.stable_benefit = get_current_benefit(); ++ rshash_state.pre_direct = GO_DOWN; ++ break; ++ ++ case OBSCURE: ++ rshash_state.stable_point = hash_strength; ++ rshash_state.turn_point_down = hash_strength; ++ rshash_state.turn_point_up = hash_strength; ++ rshash_state.turn_benefit_down = get_current_benefit(); ++ rshash_state.turn_benefit_up = get_current_benefit(); ++ rshash_state.lookup_window_index = 0; ++ rshash_state.state = RSHASH_TRYDOWN; ++ dec_hash_strength(hash_strength_delta); ++ inc_hash_strength_delta(); ++ break; ++ ++ case STILL: ++ break; ++ default: ++ BUG(); ++ } ++ break; ++ ++ case RSHASH_TRYDOWN: ++ if (rshash_state.lookup_window_index++ % 5 == 0) ++ rshash_state.below_count = 0; ++ ++ if (get_current_benefit() < rshash_state.stable_benefit) ++ rshash_state.below_count++; ++ else if (get_current_benefit() > ++ rshash_state.turn_benefit_down) { ++ rshash_state.turn_point_down = hash_strength; ++ rshash_state.turn_benefit_down = get_current_benefit(); ++ } ++ ++ if (rshash_state.below_count >= 3 || ++ judge_rshash_direction() == GO_UP || ++ hash_strength == 1) { ++ hash_strength = rshash_state.stable_point; ++ hash_strength_delta = 0; ++ inc_hash_strength(hash_strength_delta); ++ inc_hash_strength_delta(); ++ rshash_state.lookup_window_index = 0; ++ rshash_state.state = RSHASH_TRYUP; ++ hash_strength_delta = 0; ++ } else { ++ dec_hash_strength(hash_strength_delta); ++ inc_hash_strength_delta(); ++ } ++ break; ++ ++ case RSHASH_TRYUP: ++ if (rshash_state.lookup_window_index++ % 5 == 0) ++ rshash_state.below_count = 0; ++ ++ if (get_current_benefit() < rshash_state.turn_benefit_down) ++ rshash_state.below_count++; ++ else if (get_current_benefit() > rshash_state.turn_benefit_up) { ++ rshash_state.turn_point_up = hash_strength; ++ rshash_state.turn_benefit_up = get_current_benefit(); ++ } ++ ++ if (rshash_state.below_count >= 3 || ++ judge_rshash_direction() == GO_DOWN || ++ hash_strength == HASH_STRENGTH_MAX) { ++ hash_strength = rshash_state.turn_benefit_up > ++ rshash_state.turn_benefit_down ? ++ rshash_state.turn_point_up : ++ rshash_state.turn_point_down; ++ ++ rshash_state.state = RSHASH_PRE_STILL; ++ } else { ++ inc_hash_strength(hash_strength_delta); ++ inc_hash_strength_delta(); ++ } ++ ++ break; ++ ++ case RSHASH_NEW: ++ case RSHASH_PRE_STILL: ++ rshash_state.stable_benefit = get_current_benefit(); ++ rshash_state.state = RSHASH_STILL; ++ hash_strength_delta = 0; ++ break; ++ default: ++ BUG(); ++ } ++ ++ /* rshash_neg = rshash_pos = 0; */ ++ reset_benefit(); ++ ++ if (prev_hash_strength != hash_strength) ++ stable_tree_delta_hash(prev_hash_strength); ++ ++ return prev_hash_strength != hash_strength; ++} ++ ++/** ++ * round_update_ladder() - The main function to do update of all the ++ * adjustments whenever a scan round is finished. ++ */ ++static noinline void round_update_ladder(void) ++{ ++ int i; ++ unsigned long dedup; ++ struct vma_slot *slot, *tmp_slot; ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) ++ uksm_scan_ladder[i].flags &= ~UKSM_RUNG_ROUND_FINISHED; ++ ++ list_for_each_entry_safe(slot, tmp_slot, &vma_slot_dedup, dedup_list) { ++ ++ /* slot may be rung_rm_slot() when mm exits */ ++ if (slot->snode) { ++ dedup = cal_dedup_ratio_old(slot); ++ if (dedup && dedup >= uksm_abundant_threshold) ++ vma_rung_up(slot); ++ } ++ ++ slot->pages_bemerged = 0; ++ slot->pages_cowed = 0; ++ ++ list_del_init(&slot->dedup_list); ++ } ++} ++ ++static void uksm_del_vma_slot(struct vma_slot *slot) ++{ ++ int i, j; ++ struct rmap_list_entry *entry; ++ ++ if (slot->snode) { ++ /* ++ * In case it just failed when entering the rung, it's not ++ * necessary. ++ */ ++ rung_rm_slot(slot); ++ } ++ ++ if (!list_empty(&slot->dedup_list)) ++ list_del(&slot->dedup_list); ++ ++ if (!slot->rmap_list_pool || !slot->pool_counts) { ++ /* In case it OOMed in uksm_vma_enter() */ ++ goto out; ++ } ++ ++ for (i = 0; i < slot->pool_size; i++) { ++ void *addr; ++ ++ if (!slot->rmap_list_pool[i]) ++ continue; ++ ++ addr = kmap(slot->rmap_list_pool[i]); ++ for (j = 0; j < PAGE_SIZE / sizeof(*entry); j++) { ++ entry = (struct rmap_list_entry *)addr + j; ++ if (is_addr(entry->addr)) ++ continue; ++ if (!entry->item) ++ continue; ++ ++ remove_rmap_item_from_tree(entry->item); ++ free_rmap_item(entry->item); ++ slot->pool_counts[i]--; ++ } ++ BUG_ON(slot->pool_counts[i]); ++ kunmap(slot->rmap_list_pool[i]); ++ __free_page(slot->rmap_list_pool[i]); ++ } ++ kfree(slot->rmap_list_pool); ++ kfree(slot->pool_counts); ++ ++out: ++ slot->rung = NULL; ++ if (slot->flags & UKSM_SLOT_IN_UKSM) { ++ BUG_ON(uksm_pages_total < slot->pages); ++ uksm_pages_total -= slot->pages; ++ } ++ ++ if (slot->fully_scanned_round == fully_scanned_round) ++ scanned_virtual_pages -= slot->pages; ++ else ++ scanned_virtual_pages -= slot->pages_scanned; ++ free_vma_slot(slot); ++} ++ ++ ++#define SPIN_LOCK_PERIOD 32 ++static struct vma_slot *cleanup_slots[SPIN_LOCK_PERIOD]; ++static inline void cleanup_vma_slots(void) ++{ ++ struct vma_slot *slot; ++ int i; ++ ++ i = 0; ++ spin_lock(&vma_slot_list_lock); ++ while (!list_empty(&vma_slot_del)) { ++ slot = list_entry(vma_slot_del.next, ++ struct vma_slot, slot_list); ++ list_del(&slot->slot_list); ++ cleanup_slots[i++] = slot; ++ if (i == SPIN_LOCK_PERIOD) { ++ spin_unlock(&vma_slot_list_lock); ++ while (--i >= 0) ++ uksm_del_vma_slot(cleanup_slots[i]); ++ i = 0; ++ spin_lock(&vma_slot_list_lock); ++ } ++ } ++ spin_unlock(&vma_slot_list_lock); ++ ++ while (--i >= 0) ++ uksm_del_vma_slot(cleanup_slots[i]); ++} ++ ++/* ++ * Expotional moving average formula ++ */ ++static inline unsigned long ema(unsigned long curr, unsigned long last_ema) ++{ ++ /* ++ * For a very high burst, even the ema cannot work well, a false very ++ * high per-page time estimation can result in feedback in very high ++ * overhead of context switch and rung update -- this will then lead ++ * to higher per-paper time, this may not converge. ++ * ++ * Instead, we try to approach this value in a binary manner. ++ */ ++ if (curr > last_ema * 10) ++ return last_ema * 2; ++ ++ return (EMA_ALPHA * curr + (100 - EMA_ALPHA) * last_ema) / 100; ++} ++ ++/* ++ * convert cpu ratio in 1/TIME_RATIO_SCALE configured by user to ++ * nanoseconds based on current uksm_sleep_jiffies. ++ */ ++static inline unsigned long cpu_ratio_to_nsec(unsigned int ratio) ++{ ++ return NSEC_PER_USEC * jiffies_to_usecs(uksm_sleep_jiffies) / ++ (TIME_RATIO_SCALE - ratio) * ratio; ++} ++ ++ ++static inline unsigned long rung_real_ratio(int cpu_time_ratio) ++{ ++ unsigned long ret; ++ ++ BUG_ON(!cpu_time_ratio); ++ ++ if (cpu_time_ratio > 0) ++ ret = cpu_time_ratio; ++ else ++ ret = (unsigned long)(-cpu_time_ratio) * ++ uksm_max_cpu_percentage / 100UL; ++ ++ return ret ? ret : 1; ++} ++ ++static noinline void uksm_calc_scan_pages(void) ++{ ++ struct scan_rung *ladder = uksm_scan_ladder; ++ unsigned long sleep_usecs, nsecs; ++ unsigned long ratio; ++ int i; ++ unsigned long per_page; ++ ++ if (uksm_ema_page_time > 100000 || ++ (((unsigned long) uksm_eval_round & (256UL - 1)) == 0UL)) ++ uksm_ema_page_time = UKSM_PAGE_TIME_DEFAULT; ++ ++ per_page = uksm_ema_page_time; ++ BUG_ON(!per_page); ++ ++ /* ++ * For every 8 eval round, we try to probe a uksm_sleep_jiffies value ++ * based on saved user input. ++ */ ++ if (((unsigned long) uksm_eval_round & (8UL - 1)) == 0UL) ++ uksm_sleep_jiffies = uksm_sleep_saved; ++ ++ /* We require a rung scan at least 1 page in a period. */ ++ nsecs = per_page; ++ ratio = rung_real_ratio(ladder[0].cpu_ratio); ++ if (cpu_ratio_to_nsec(ratio) < nsecs) { ++ sleep_usecs = nsecs * (TIME_RATIO_SCALE - ratio) / ratio ++ / NSEC_PER_USEC; ++ uksm_sleep_jiffies = usecs_to_jiffies(sleep_usecs) + 1; ++ } ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ ratio = rung_real_ratio(ladder[i].cpu_ratio); ++ ladder[i].pages_to_scan = cpu_ratio_to_nsec(ratio) / ++ per_page; ++ BUG_ON(!ladder[i].pages_to_scan); ++ uksm_calc_rung_step(&ladder[i], per_page, ratio); ++ } ++} ++ ++/* ++ * From the scan time of this round (ns) to next expected min sleep time ++ * (ms), be careful of the possible overflows. ratio is taken from ++ * rung_real_ratio() ++ */ ++static inline ++unsigned int scan_time_to_sleep(unsigned long long scan_time, unsigned long ratio) ++{ ++ scan_time >>= 20; /* to msec level now */ ++ BUG_ON(scan_time > (ULONG_MAX / TIME_RATIO_SCALE)); ++ ++ return (unsigned int) ((unsigned long) scan_time * ++ (TIME_RATIO_SCALE - ratio) / ratio); ++} ++ ++#define __round_mask(x, y) ((__typeof__(x))((y)-1)) ++#define round_up(x, y) ((((x)-1) | __round_mask(x, y))+1) ++ ++static void uksm_vma_enter(struct vma_slot **slots, unsigned long num) ++{ ++ struct scan_rung *rung; ++ ++ rung = &uksm_scan_ladder[0]; ++ rung_add_new_slots(rung, slots, num); ++} ++ ++static struct vma_slot *batch_slots[SLOT_TREE_NODE_STORE_SIZE]; ++ ++static void uksm_enter_all_slots(void) ++{ ++ struct vma_slot *slot; ++ unsigned long index; ++ struct list_head empty_vma_list; ++ int i; ++ ++ i = 0; ++ index = 0; ++ INIT_LIST_HEAD(&empty_vma_list); ++ ++ spin_lock(&vma_slot_list_lock); ++ while (!list_empty(&vma_slot_new)) { ++ slot = list_entry(vma_slot_new.next, ++ struct vma_slot, slot_list); ++ ++ if (!slot->vma->anon_vma) { ++ list_move(&slot->slot_list, &empty_vma_list); ++ } else if (vma_can_enter(slot->vma)) { ++ batch_slots[index++] = slot; ++ list_del_init(&slot->slot_list); ++ } else { ++ list_move(&slot->slot_list, &vma_slot_noadd); ++ } ++ ++ if (++i == SPIN_LOCK_PERIOD || ++ (index && !(index % SLOT_TREE_NODE_STORE_SIZE))) { ++ spin_unlock(&vma_slot_list_lock); ++ ++ if (index && !(index % SLOT_TREE_NODE_STORE_SIZE)) { ++ uksm_vma_enter(batch_slots, index); ++ index = 0; ++ } ++ i = 0; ++ cond_resched(); ++ spin_lock(&vma_slot_list_lock); ++ } ++ } ++ ++ list_splice(&empty_vma_list, &vma_slot_new); ++ ++ spin_unlock(&vma_slot_list_lock); ++ ++ if (index) ++ uksm_vma_enter(batch_slots, index); ++ ++} ++ ++static inline int rung_round_finished(struct scan_rung *rung) ++{ ++ return rung->flags & UKSM_RUNG_ROUND_FINISHED; ++} ++ ++static inline void judge_slot(struct vma_slot *slot) ++{ ++ struct scan_rung *rung = slot->rung; ++ unsigned long dedup; ++ int deleted; ++ ++ dedup = cal_dedup_ratio(slot); ++ if (vma_fully_scanned(slot) && uksm_thrash_threshold) ++ deleted = vma_rung_enter(slot, &uksm_scan_ladder[0]); ++ else if (dedup && dedup >= uksm_abundant_threshold) ++ deleted = vma_rung_up(slot); ++ else ++ deleted = vma_rung_down(slot); ++ ++ slot->pages_merged = 0; ++ slot->pages_cowed = 0; ++ slot->this_sampled = 0; ++ ++ if (vma_fully_scanned(slot)) ++ slot->pages_scanned = 0; ++ ++ slot->last_scanned = slot->pages_scanned; ++ ++ /* If its deleted in above, then rung was already advanced. */ ++ if (!deleted) ++ advance_current_scan(rung); ++} ++ ++ ++static inline int hash_round_finished(void) ++{ ++ if (scanned_virtual_pages > (uksm_pages_total >> 2)) { ++ scanned_virtual_pages = 0; ++ if (uksm_pages_scanned) ++ fully_scanned_round++; ++ ++ return 1; ++ } else { ++ return 0; ++ } ++} ++ ++#define UKSM_MMSEM_BATCH 5 ++#define BUSY_RETRY 100 ++ ++/** ++ * uksm_do_scan() - the main worker function. ++ */ ++static noinline void uksm_do_scan(void) ++{ ++ struct vma_slot *slot, *iter; ++ struct mm_struct *busy_mm; ++ unsigned char round_finished, all_rungs_emtpy; ++ int i, err, mmsem_batch; ++ unsigned long pcost; ++ long long delta_exec; ++ unsigned long vpages, max_cpu_ratio; ++ unsigned long long start_time, end_time, scan_time; ++ unsigned int expected_jiffies; ++ ++ might_sleep(); ++ ++ vpages = 0; ++ ++ start_time = task_sched_runtime(current); ++ max_cpu_ratio = 0; ++ mmsem_batch = 0; ++ ++ for (i = 0; i < SCAN_LADDER_SIZE;) { ++ struct scan_rung *rung = &uksm_scan_ladder[i]; ++ unsigned long ratio; ++ int busy_retry; ++ ++ if (!rung->pages_to_scan) { ++ i++; ++ continue; ++ } ++ ++ if (!rung->vma_root.num) { ++ rung->pages_to_scan = 0; ++ i++; ++ continue; ++ } ++ ++ ratio = rung_real_ratio(rung->cpu_ratio); ++ if (ratio > max_cpu_ratio) ++ max_cpu_ratio = ratio; ++ ++ busy_retry = BUSY_RETRY; ++ /* ++ * Do not consider rung_round_finished() here, just used up the ++ * rung->pages_to_scan quota. ++ */ ++ while (rung->pages_to_scan && rung->vma_root.num && ++ likely(!freezing(current))) { ++ int reset = 0; ++ ++ slot = rung->current_scan; ++ ++ BUG_ON(vma_fully_scanned(slot)); ++ ++ if (mmsem_batch) ++ err = 0; ++ else ++ err = try_down_read_slot_mmap_sem(slot); ++ ++ if (err == -ENOENT) { ++rm_slot: ++ rung_rm_slot(slot); ++ continue; ++ } ++ ++ busy_mm = slot->mm; ++ ++ if (err == -EBUSY) { ++ /* skip other vmas on the same mm */ ++ do { ++ reset = advance_current_scan(rung); ++ iter = rung->current_scan; ++ busy_retry--; ++ if (iter->vma->vm_mm != busy_mm || ++ !busy_retry || reset) ++ break; ++ } while (1); ++ ++ if (iter->vma->vm_mm != busy_mm) { ++ continue; ++ } else { ++ /* scan round finsished */ ++ break; ++ } ++ } ++ ++ BUG_ON(!vma_can_enter(slot->vma)); ++ if (uksm_test_exit(slot->vma->vm_mm)) { ++ mmsem_batch = 0; ++ up_read(&slot->vma->vm_mm->mmap_sem); ++ goto rm_slot; ++ } ++ ++ if (mmsem_batch) ++ mmsem_batch--; ++ else ++ mmsem_batch = UKSM_MMSEM_BATCH; ++ ++ /* Ok, we have take the mmap_sem, ready to scan */ ++ scan_vma_one_page(slot); ++ rung->pages_to_scan--; ++ vpages++; ++ ++ if (rung->current_offset + rung->step > slot->pages - 1 ++ || vma_fully_scanned(slot)) { ++ up_read(&slot->vma->vm_mm->mmap_sem); ++ judge_slot(slot); ++ mmsem_batch = 0; ++ } else { ++ rung->current_offset += rung->step; ++ if (!mmsem_batch) ++ up_read(&slot->vma->vm_mm->mmap_sem); ++ } ++ ++ busy_retry = BUSY_RETRY; ++ cond_resched(); ++ } ++ ++ if (mmsem_batch) { ++ up_read(&slot->vma->vm_mm->mmap_sem); ++ mmsem_batch = 0; ++ } ++ ++ if (freezing(current)) ++ break; ++ ++ cond_resched(); ++ } ++ end_time = task_sched_runtime(current); ++ delta_exec = end_time - start_time; ++ ++ if (freezing(current)) ++ return; ++ ++ cleanup_vma_slots(); ++ uksm_enter_all_slots(); ++ ++ round_finished = 1; ++ all_rungs_emtpy = 1; ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ struct scan_rung *rung = &uksm_scan_ladder[i]; ++ ++ if (rung->vma_root.num) { ++ all_rungs_emtpy = 0; ++ if (!rung_round_finished(rung)) ++ round_finished = 0; ++ } ++ } ++ ++ if (all_rungs_emtpy) ++ round_finished = 0; ++ ++ if (round_finished) { ++ round_update_ladder(); ++ uksm_eval_round++; ++ ++ if (hash_round_finished() && rshash_adjust()) { ++ /* Reset the unstable root iff hash strength changed */ ++ uksm_hash_round++; ++ root_unstable_tree = RB_ROOT; ++ free_all_tree_nodes(&unstable_tree_node_list); ++ } ++ ++ /* ++ * A number of pages can hang around indefinitely on per-cpu ++ * pagevecs, raised page count preventing write_protect_page ++ * from merging them. Though it doesn't really matter much, ++ * it is puzzling to see some stuck in pages_volatile until ++ * other activity jostles them out, and they also prevented ++ * LTP's KSM test from succeeding deterministically; so drain ++ * them here (here rather than on entry to uksm_do_scan(), ++ * so we don't IPI too often when pages_to_scan is set low). ++ */ ++ lru_add_drain_all(); ++ } ++ ++ ++ if (vpages && delta_exec > 0) { ++ pcost = (unsigned long) delta_exec / vpages; ++ if (likely(uksm_ema_page_time)) ++ uksm_ema_page_time = ema(pcost, uksm_ema_page_time); ++ else ++ uksm_ema_page_time = pcost; ++ } ++ ++ uksm_calc_scan_pages(); ++ uksm_sleep_real = uksm_sleep_jiffies; ++ /* in case of radical cpu bursts, apply the upper bound */ ++ end_time = task_sched_runtime(current); ++ if (max_cpu_ratio && end_time > start_time) { ++ scan_time = end_time - start_time; ++ expected_jiffies = msecs_to_jiffies( ++ scan_time_to_sleep(scan_time, max_cpu_ratio)); ++ ++ if (expected_jiffies > uksm_sleep_real) ++ uksm_sleep_real = expected_jiffies; ++ ++ /* We have a 1 second up bound for responsiveness. */ ++ if (jiffies_to_msecs(uksm_sleep_real) > MSEC_PER_SEC) ++ uksm_sleep_real = msecs_to_jiffies(1000); ++ } ++ ++ return; ++} ++ ++static int ksmd_should_run(void) ++{ ++ return uksm_run & UKSM_RUN_MERGE; ++} ++ ++static int uksm_scan_thread(void *nothing) ++{ ++ set_freezable(); ++ set_user_nice(current, 5); ++ ++ while (!kthread_should_stop()) { ++ mutex_lock(&uksm_thread_mutex); ++ if (ksmd_should_run()) ++ uksm_do_scan(); ++ mutex_unlock(&uksm_thread_mutex); ++ ++ try_to_freeze(); ++ ++ if (ksmd_should_run()) { ++ schedule_timeout_interruptible(uksm_sleep_real); ++ uksm_sleep_times++; ++ } else { ++ wait_event_freezable(uksm_thread_wait, ++ ksmd_should_run() || kthread_should_stop()); ++ } ++ } ++ return 0; ++} ++ ++void rmap_walk_ksm(struct page *page, struct rmap_walk_control *rwc) ++{ ++ struct stable_node *stable_node; ++ struct node_vma *node_vma; ++ struct rmap_item *rmap_item; ++ int search_new_forks = 0; ++ unsigned long address; ++ ++ VM_BUG_ON_PAGE(!PageKsm(page), page); ++ VM_BUG_ON_PAGE(!PageLocked(page), page); ++ ++ stable_node = page_stable_node(page); ++ if (!stable_node) ++ return; ++again: ++ hlist_for_each_entry(node_vma, &stable_node->hlist, hlist) { ++ hlist_for_each_entry(rmap_item, &node_vma->rmap_hlist, hlist) { ++ struct anon_vma *anon_vma = rmap_item->anon_vma; ++ struct anon_vma_chain *vmac; ++ struct vm_area_struct *vma; ++ ++ cond_resched(); ++ anon_vma_lock_read(anon_vma); ++ anon_vma_interval_tree_foreach(vmac, &anon_vma->rb_root, ++ 0, ULONG_MAX) { ++ cond_resched(); ++ vma = vmac->vma; ++ address = get_rmap_addr(rmap_item); ++ ++ if (address < vma->vm_start || ++ address >= vma->vm_end) ++ continue; ++ ++ if ((rmap_item->slot->vma == vma) == ++ search_new_forks) ++ continue; ++ ++ if (rwc->invalid_vma && rwc->invalid_vma(vma, rwc->arg)) ++ continue; ++ ++ if (!rwc->rmap_one(page, vma, address, rwc->arg)) { ++ anon_vma_unlock_read(anon_vma); ++ return; ++ } ++ ++ if (rwc->done && rwc->done(page)) { ++ anon_vma_unlock_read(anon_vma); ++ return; ++ } ++ } ++ anon_vma_unlock_read(anon_vma); ++ } ++ } ++ if (!search_new_forks++) ++ goto again; ++} ++ ++#ifdef CONFIG_MIGRATION ++/* Common ksm interface but may be specific to uksm */ ++void ksm_migrate_page(struct page *newpage, struct page *oldpage) ++{ ++ struct stable_node *stable_node; ++ ++ VM_BUG_ON_PAGE(!PageLocked(oldpage), oldpage); ++ VM_BUG_ON_PAGE(!PageLocked(newpage), newpage); ++ VM_BUG_ON(newpage->mapping != oldpage->mapping); ++ ++ stable_node = page_stable_node(newpage); ++ if (stable_node) { ++ VM_BUG_ON(stable_node->kpfn != page_to_pfn(oldpage)); ++ stable_node->kpfn = page_to_pfn(newpage); ++ /* ++ * newpage->mapping was set in advance; now we need smp_wmb() ++ * to make sure that the new stable_node->kpfn is visible ++ * to get_ksm_page() before it can see that oldpage->mapping ++ * has gone stale (or that PageSwapCache has been cleared). ++ */ ++ smp_wmb(); ++ set_page_stable_node(oldpage, NULL); ++ } ++} ++#endif /* CONFIG_MIGRATION */ ++ ++#ifdef CONFIG_MEMORY_HOTREMOVE ++static struct stable_node *uksm_check_stable_tree(unsigned long start_pfn, ++ unsigned long end_pfn) ++{ ++ struct rb_node *node; ++ ++ for (node = rb_first(root_stable_treep); node; node = rb_next(node)) { ++ struct stable_node *stable_node; ++ ++ stable_node = rb_entry(node, struct stable_node, node); ++ if (stable_node->kpfn >= start_pfn && ++ stable_node->kpfn < end_pfn) ++ return stable_node; ++ } ++ return NULL; ++} ++ ++static int uksm_memory_callback(struct notifier_block *self, ++ unsigned long action, void *arg) ++{ ++ struct memory_notify *mn = arg; ++ struct stable_node *stable_node; ++ ++ switch (action) { ++ case MEM_GOING_OFFLINE: ++ /* ++ * Keep it very simple for now: just lock out ksmd and ++ * MADV_UNMERGEABLE while any memory is going offline. ++ * mutex_lock_nested() is necessary because lockdep was alarmed ++ * that here we take uksm_thread_mutex inside notifier chain ++ * mutex, and later take notifier chain mutex inside ++ * uksm_thread_mutex to unlock it. But that's safe because both ++ * are inside mem_hotplug_mutex. ++ */ ++ mutex_lock_nested(&uksm_thread_mutex, SINGLE_DEPTH_NESTING); ++ break; ++ ++ case MEM_OFFLINE: ++ /* ++ * Most of the work is done by page migration; but there might ++ * be a few stable_nodes left over, still pointing to struct ++ * pages which have been offlined: prune those from the tree. ++ */ ++ while ((stable_node = uksm_check_stable_tree(mn->start_pfn, ++ mn->start_pfn + mn->nr_pages)) != NULL) ++ remove_node_from_stable_tree(stable_node, 1, 1); ++ /* fallthrough */ ++ ++ case MEM_CANCEL_OFFLINE: ++ mutex_unlock(&uksm_thread_mutex); ++ break; ++ } ++ return NOTIFY_OK; ++} ++#endif /* CONFIG_MEMORY_HOTREMOVE */ ++ ++#ifdef CONFIG_SYSFS ++/* ++ * This all compiles without CONFIG_SYSFS, but is a waste of space. ++ */ ++ ++#define UKSM_ATTR_RO(_name) \ ++ static struct kobj_attribute _name##_attr = __ATTR_RO(_name) ++#define UKSM_ATTR(_name) \ ++ static struct kobj_attribute _name##_attr = \ ++ __ATTR(_name, 0644, _name##_show, _name##_store) ++ ++static ssize_t max_cpu_percentage_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%u\n", uksm_max_cpu_percentage); ++} ++ ++static ssize_t max_cpu_percentage_store(struct kobject *kobj, ++ struct kobj_attribute *attr, ++ const char *buf, size_t count) ++{ ++ unsigned long max_cpu_percentage; ++ int err; ++ ++ err = kstrtoul(buf, 10, &max_cpu_percentage); ++ if (err || max_cpu_percentage > 100) ++ return -EINVAL; ++ ++ if (max_cpu_percentage == 100) ++ max_cpu_percentage = 99; ++ else if (max_cpu_percentage < 10) ++ max_cpu_percentage = 10; ++ ++ uksm_max_cpu_percentage = max_cpu_percentage; ++ ++ return count; ++} ++UKSM_ATTR(max_cpu_percentage); ++ ++static ssize_t sleep_millisecs_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%u\n", jiffies_to_msecs(uksm_sleep_jiffies)); ++} ++ ++static ssize_t sleep_millisecs_store(struct kobject *kobj, ++ struct kobj_attribute *attr, ++ const char *buf, size_t count) ++{ ++ unsigned long msecs; ++ int err; ++ ++ err = kstrtoul(buf, 10, &msecs); ++ if (err || msecs > MSEC_PER_SEC) ++ return -EINVAL; ++ ++ uksm_sleep_jiffies = msecs_to_jiffies(msecs); ++ uksm_sleep_saved = uksm_sleep_jiffies; ++ ++ return count; ++} ++UKSM_ATTR(sleep_millisecs); ++ ++ ++static ssize_t cpu_governor_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ int n = sizeof(uksm_cpu_governor_str) / sizeof(char *); ++ int i; ++ ++ buf[0] = '\0'; ++ for (i = 0; i < n ; i++) { ++ if (uksm_cpu_governor == i) ++ strcat(buf, "["); ++ ++ strcat(buf, uksm_cpu_governor_str[i]); ++ ++ if (uksm_cpu_governor == i) ++ strcat(buf, "]"); ++ ++ strcat(buf, " "); ++ } ++ strcat(buf, "\n"); ++ ++ return strlen(buf); ++} ++ ++static inline void init_performance_values(void) ++{ ++ int i; ++ struct scan_rung *rung; ++ struct uksm_cpu_preset_s *preset = uksm_cpu_preset + uksm_cpu_governor; ++ ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ rung = uksm_scan_ladder + i; ++ rung->cpu_ratio = preset->cpu_ratio[i]; ++ rung->cover_msecs = preset->cover_msecs[i]; ++ } ++ ++ uksm_max_cpu_percentage = preset->max_cpu; ++} ++ ++static ssize_t cpu_governor_store(struct kobject *kobj, ++ struct kobj_attribute *attr, ++ const char *buf, size_t count) ++{ ++ int n = sizeof(uksm_cpu_governor_str) / sizeof(char *); ++ ++ for (n--; n >= 0 ; n--) { ++ if (!strncmp(buf, uksm_cpu_governor_str[n], ++ strlen(uksm_cpu_governor_str[n]))) ++ break; ++ } ++ ++ if (n < 0) ++ return -EINVAL; ++ else ++ uksm_cpu_governor = n; ++ ++ init_performance_values(); ++ ++ return count; ++} ++UKSM_ATTR(cpu_governor); ++ ++static ssize_t run_show(struct kobject *kobj, struct kobj_attribute *attr, ++ char *buf) ++{ ++ return sprintf(buf, "%u\n", uksm_run); ++} ++ ++static ssize_t run_store(struct kobject *kobj, struct kobj_attribute *attr, ++ const char *buf, size_t count) ++{ ++ int err; ++ unsigned long flags; ++ ++ err = kstrtoul(buf, 10, &flags); ++ if (err || flags > UINT_MAX) ++ return -EINVAL; ++ if (flags > UKSM_RUN_MERGE) ++ return -EINVAL; ++ ++ mutex_lock(&uksm_thread_mutex); ++ if (uksm_run != flags) ++ uksm_run = flags; ++ mutex_unlock(&uksm_thread_mutex); ++ ++ if (flags & UKSM_RUN_MERGE) ++ wake_up_interruptible(&uksm_thread_wait); ++ ++ return count; ++} ++UKSM_ATTR(run); ++ ++static ssize_t abundant_threshold_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%u\n", uksm_abundant_threshold); ++} ++ ++static ssize_t abundant_threshold_store(struct kobject *kobj, ++ struct kobj_attribute *attr, ++ const char *buf, size_t count) ++{ ++ int err; ++ unsigned long flags; ++ ++ err = kstrtoul(buf, 10, &flags); ++ if (err || flags > 99) ++ return -EINVAL; ++ ++ uksm_abundant_threshold = flags; ++ ++ return count; ++} ++UKSM_ATTR(abundant_threshold); ++ ++static ssize_t thrash_threshold_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%u\n", uksm_thrash_threshold); ++} ++ ++static ssize_t thrash_threshold_store(struct kobject *kobj, ++ struct kobj_attribute *attr, ++ const char *buf, size_t count) ++{ ++ int err; ++ unsigned long flags; ++ ++ err = kstrtoul(buf, 10, &flags); ++ if (err || flags > 99) ++ return -EINVAL; ++ ++ uksm_thrash_threshold = flags; ++ ++ return count; ++} ++UKSM_ATTR(thrash_threshold); ++ ++static ssize_t cpu_ratios_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ int i, size; ++ struct scan_rung *rung; ++ char *p = buf; ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ rung = &uksm_scan_ladder[i]; ++ ++ if (rung->cpu_ratio > 0) ++ size = sprintf(p, "%d ", rung->cpu_ratio); ++ else ++ size = sprintf(p, "MAX/%d ", ++ TIME_RATIO_SCALE / -rung->cpu_ratio); ++ ++ p += size; ++ } ++ ++ *p++ = '\n'; ++ *p = '\0'; ++ ++ return p - buf; ++} ++ ++static ssize_t cpu_ratios_store(struct kobject *kobj, ++ struct kobj_attribute *attr, ++ const char *buf, size_t count) ++{ ++ int i, cpuratios[SCAN_LADDER_SIZE], err; ++ unsigned long value; ++ struct scan_rung *rung; ++ char *p, *end = NULL; ++ ++ p = kzalloc(count, GFP_KERNEL); ++ if (!p) ++ return -ENOMEM; ++ ++ memcpy(p, buf, count); ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ if (i != SCAN_LADDER_SIZE - 1) { ++ end = strchr(p, ' '); ++ if (!end) ++ return -EINVAL; ++ ++ *end = '\0'; ++ } ++ ++ if (strstr(p, "MAX/")) { ++ p = strchr(p, '/') + 1; ++ err = kstrtoul(p, 10, &value); ++ if (err || value > TIME_RATIO_SCALE || !value) ++ return -EINVAL; ++ ++ cpuratios[i] = -(int) (TIME_RATIO_SCALE / value); ++ } else { ++ err = kstrtoul(p, 10, &value); ++ if (err || value > TIME_RATIO_SCALE || !value) ++ return -EINVAL; ++ ++ cpuratios[i] = value; ++ } ++ ++ p = end + 1; ++ } ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ rung = &uksm_scan_ladder[i]; ++ ++ rung->cpu_ratio = cpuratios[i]; ++ } ++ ++ return count; ++} ++UKSM_ATTR(cpu_ratios); ++ ++static ssize_t eval_intervals_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ int i, size; ++ struct scan_rung *rung; ++ char *p = buf; ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ rung = &uksm_scan_ladder[i]; ++ size = sprintf(p, "%u ", rung->cover_msecs); ++ p += size; ++ } ++ ++ *p++ = '\n'; ++ *p = '\0'; ++ ++ return p - buf; ++} ++ ++static ssize_t eval_intervals_store(struct kobject *kobj, ++ struct kobj_attribute *attr, ++ const char *buf, size_t count) ++{ ++ int i, err; ++ unsigned long values[SCAN_LADDER_SIZE]; ++ struct scan_rung *rung; ++ char *p, *end = NULL; ++ ssize_t ret = count; ++ ++ p = kzalloc(count + 2, GFP_KERNEL); ++ if (!p) ++ return -ENOMEM; ++ ++ memcpy(p, buf, count); ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ if (i != SCAN_LADDER_SIZE - 1) { ++ end = strchr(p, ' '); ++ if (!end) { ++ ret = -EINVAL; ++ goto out; ++ } ++ ++ *end = '\0'; ++ } ++ ++ err = kstrtoul(p, 10, &values[i]); ++ if (err) { ++ ret = -EINVAL; ++ goto out; ++ } ++ ++ p = end + 1; ++ } ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ rung = &uksm_scan_ladder[i]; ++ ++ rung->cover_msecs = values[i]; ++ } ++ ++out: ++ kfree(p); ++ return ret; ++} ++UKSM_ATTR(eval_intervals); ++ ++static ssize_t ema_per_page_time_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%lu\n", uksm_ema_page_time); ++} ++UKSM_ATTR_RO(ema_per_page_time); ++ ++static ssize_t pages_shared_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%lu\n", uksm_pages_shared); ++} ++UKSM_ATTR_RO(pages_shared); ++ ++static ssize_t pages_sharing_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%lu\n", uksm_pages_sharing); ++} ++UKSM_ATTR_RO(pages_sharing); ++ ++static ssize_t pages_unshared_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%lu\n", uksm_pages_unshared); ++} ++UKSM_ATTR_RO(pages_unshared); ++ ++static ssize_t full_scans_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%llu\n", fully_scanned_round); ++} ++UKSM_ATTR_RO(full_scans); ++ ++static ssize_t pages_scanned_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ unsigned long base = 0; ++ u64 delta, ret; ++ ++ if (pages_scanned_stored) { ++ base = pages_scanned_base; ++ ret = pages_scanned_stored; ++ delta = uksm_pages_scanned >> base; ++ if (CAN_OVERFLOW_U64(ret, delta)) { ++ ret >>= 1; ++ delta >>= 1; ++ base++; ++ ret += delta; ++ } ++ } else { ++ ret = uksm_pages_scanned; ++ } ++ ++ while (ret > ULONG_MAX) { ++ ret >>= 1; ++ base++; ++ } ++ ++ if (base) ++ return sprintf(buf, "%lu * 2^%lu\n", (unsigned long)ret, base); ++ else ++ return sprintf(buf, "%lu\n", (unsigned long)ret); ++} ++UKSM_ATTR_RO(pages_scanned); ++ ++static ssize_t hash_strength_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%lu\n", hash_strength); ++} ++UKSM_ATTR_RO(hash_strength); ++ ++static ssize_t sleep_times_show(struct kobject *kobj, ++ struct kobj_attribute *attr, char *buf) ++{ ++ return sprintf(buf, "%llu\n", uksm_sleep_times); ++} ++UKSM_ATTR_RO(sleep_times); ++ ++ ++static struct attribute *uksm_attrs[] = { ++ &max_cpu_percentage_attr.attr, ++ &sleep_millisecs_attr.attr, ++ &cpu_governor_attr.attr, ++ &run_attr.attr, ++ &ema_per_page_time_attr.attr, ++ &pages_shared_attr.attr, ++ &pages_sharing_attr.attr, ++ &pages_unshared_attr.attr, ++ &full_scans_attr.attr, ++ &pages_scanned_attr.attr, ++ &hash_strength_attr.attr, ++ &sleep_times_attr.attr, ++ &thrash_threshold_attr.attr, ++ &abundant_threshold_attr.attr, ++ &cpu_ratios_attr.attr, ++ &eval_intervals_attr.attr, ++ NULL, ++}; ++ ++static struct attribute_group uksm_attr_group = { ++ .attrs = uksm_attrs, ++ .name = "uksm", ++}; ++#endif /* CONFIG_SYSFS */ ++ ++static inline void init_scan_ladder(void) ++{ ++ int i; ++ struct scan_rung *rung; ++ ++ for (i = 0; i < SCAN_LADDER_SIZE; i++) { ++ rung = uksm_scan_ladder + i; ++ slot_tree_init_root(&rung->vma_root); ++ } ++ ++ init_performance_values(); ++ uksm_calc_scan_pages(); ++} ++ ++static inline int cal_positive_negative_costs(void) ++{ ++ struct page *p1, *p2; ++ unsigned char *addr1, *addr2; ++ unsigned long i, time_start, hash_cost; ++ unsigned long loopnum = 0; ++ ++ /*IMPORTANT: volatile is needed to prevent over-optimization by gcc. */ ++ volatile u32 hash; ++ volatile int ret; ++ ++ p1 = alloc_page(GFP_KERNEL); ++ if (!p1) ++ return -ENOMEM; ++ ++ p2 = alloc_page(GFP_KERNEL); ++ if (!p2) ++ return -ENOMEM; ++ ++ addr1 = kmap_atomic(p1); ++ addr2 = kmap_atomic(p2); ++ memset(addr1, prandom_u32(), PAGE_SIZE); ++ memcpy(addr2, addr1, PAGE_SIZE); ++ ++ /* make sure that the two pages differ in last byte */ ++ addr2[PAGE_SIZE-1] = ~addr2[PAGE_SIZE-1]; ++ kunmap_atomic(addr2); ++ kunmap_atomic(addr1); ++ ++ time_start = jiffies; ++ while (jiffies - time_start < 100) { ++ for (i = 0; i < 100; i++) ++ hash = page_hash(p1, HASH_STRENGTH_FULL, 0); ++ loopnum += 100; ++ } ++ hash_cost = (jiffies - time_start); ++ ++ time_start = jiffies; ++ for (i = 0; i < loopnum; i++) ++ ret = pages_identical(p1, p2); ++ memcmp_cost = HASH_STRENGTH_FULL * (jiffies - time_start); ++ memcmp_cost /= hash_cost; ++ pr_info("UKSM: relative memcmp_cost = %lu " ++ "hash=%u cmp_ret=%d.\n", ++ memcmp_cost, hash, ret); ++ ++ __free_page(p1); ++ __free_page(p2); ++ return 0; ++} ++ ++static int init_zeropage_hash_table(void) ++{ ++ struct page *page; ++ char *addr; ++ int i; ++ ++ page = alloc_page(GFP_KERNEL); ++ if (!page) ++ return -ENOMEM; ++ ++ addr = kmap_atomic(page); ++ memset(addr, 0, PAGE_SIZE); ++ kunmap_atomic(addr); ++ ++ zero_hash_table = kmalloc_array(HASH_STRENGTH_MAX, sizeof(u32), ++ GFP_KERNEL); ++ if (!zero_hash_table) ++ return -ENOMEM; ++ ++ for (i = 0; i < HASH_STRENGTH_MAX; i++) ++ zero_hash_table[i] = page_hash(page, i, 0); ++ ++ __free_page(page); ++ ++ return 0; ++} ++ ++static inline int init_random_sampling(void) ++{ ++ unsigned long i; ++ ++ random_nums = kmalloc(PAGE_SIZE, GFP_KERNEL); ++ if (!random_nums) ++ return -ENOMEM; ++ ++ for (i = 0; i < HASH_STRENGTH_FULL; i++) ++ random_nums[i] = i; ++ ++ for (i = 0; i < HASH_STRENGTH_FULL; i++) { ++ unsigned long rand_range, swap_index, tmp; ++ ++ rand_range = HASH_STRENGTH_FULL - i; ++ swap_index = i + prandom_u32() % rand_range; ++ tmp = random_nums[i]; ++ random_nums[i] = random_nums[swap_index]; ++ random_nums[swap_index] = tmp; ++ } ++ ++ rshash_state.state = RSHASH_NEW; ++ rshash_state.below_count = 0; ++ rshash_state.lookup_window_index = 0; ++ ++ return cal_positive_negative_costs(); ++} ++ ++static int __init uksm_slab_init(void) ++{ ++ rmap_item_cache = UKSM_KMEM_CACHE(rmap_item, 0); ++ if (!rmap_item_cache) ++ goto out; ++ ++ stable_node_cache = UKSM_KMEM_CACHE(stable_node, 0); ++ if (!stable_node_cache) ++ goto out_free1; ++ ++ node_vma_cache = UKSM_KMEM_CACHE(node_vma, 0); ++ if (!node_vma_cache) ++ goto out_free2; ++ ++ vma_slot_cache = UKSM_KMEM_CACHE(vma_slot, 0); ++ if (!vma_slot_cache) ++ goto out_free3; ++ ++ tree_node_cache = UKSM_KMEM_CACHE(tree_node, 0); ++ if (!tree_node_cache) ++ goto out_free4; ++ ++ return 0; ++ ++out_free4: ++ kmem_cache_destroy(vma_slot_cache); ++out_free3: ++ kmem_cache_destroy(node_vma_cache); ++out_free2: ++ kmem_cache_destroy(stable_node_cache); ++out_free1: ++ kmem_cache_destroy(rmap_item_cache); ++out: ++ return -ENOMEM; ++} ++ ++static void __init uksm_slab_free(void) ++{ ++ kmem_cache_destroy(stable_node_cache); ++ kmem_cache_destroy(rmap_item_cache); ++ kmem_cache_destroy(node_vma_cache); ++ kmem_cache_destroy(vma_slot_cache); ++ kmem_cache_destroy(tree_node_cache); ++} ++ ++/* Common interface to ksm, different to it. */ ++int ksm_madvise(struct vm_area_struct *vma, unsigned long start, ++ unsigned long end, int advice, unsigned long *vm_flags) ++{ ++ int err; ++ ++ switch (advice) { ++ case MADV_MERGEABLE: ++ return 0; /* just ignore the advice */ ++ ++ case MADV_UNMERGEABLE: ++ if (!(*vm_flags & VM_MERGEABLE) || !uksm_flags_can_scan(*vm_flags)) ++ return 0; /* just ignore the advice */ ++ ++ if (vma->anon_vma) { ++ err = unmerge_uksm_pages(vma, start, end); ++ if (err) ++ return err; ++ } ++ ++ uksm_remove_vma(vma); ++ *vm_flags &= ~VM_MERGEABLE; ++ break; ++ } ++ ++ return 0; ++} ++ ++/* Common interface to ksm, actually the same. */ ++struct page *ksm_might_need_to_copy(struct page *page, ++ struct vm_area_struct *vma, unsigned long address) ++{ ++ struct anon_vma *anon_vma = page_anon_vma(page); ++ struct page *new_page; ++ ++ if (PageKsm(page)) { ++ if (page_stable_node(page)) ++ return page; /* no need to copy it */ ++ } else if (!anon_vma) { ++ return page; /* no need to copy it */ ++ } else if (anon_vma->root == vma->anon_vma->root && ++ page->index == linear_page_index(vma, address)) { ++ return page; /* still no need to copy it */ ++ } ++ if (!PageUptodate(page)) ++ return page; /* let do_swap_page report the error */ ++ ++ new_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address); ++ if (new_page) { ++ copy_user_highpage(new_page, page, address, vma); ++ ++ SetPageDirty(new_page); ++ __SetPageUptodate(new_page); ++ __SetPageLocked(new_page); ++ } ++ ++ return new_page; ++} ++ ++static int __init uksm_init(void) ++{ ++ struct task_struct *uksm_thread; ++ int err; ++ ++ uksm_sleep_jiffies = msecs_to_jiffies(100); ++ uksm_sleep_saved = uksm_sleep_jiffies; ++ ++ slot_tree_init(); ++ init_scan_ladder(); ++ ++ ++ err = init_random_sampling(); ++ if (err) ++ goto out_free2; ++ ++ err = uksm_slab_init(); ++ if (err) ++ goto out_free1; ++ ++ err = init_zeropage_hash_table(); ++ if (err) ++ goto out_free0; ++ ++ uksm_thread = kthread_run(uksm_scan_thread, NULL, "uksmd"); ++ if (IS_ERR(uksm_thread)) { ++ pr_err("uksm: creating kthread failed\n"); ++ err = PTR_ERR(uksm_thread); ++ goto out_free; ++ } ++ ++#ifdef CONFIG_SYSFS ++ err = sysfs_create_group(mm_kobj, &uksm_attr_group); ++ if (err) { ++ pr_err("uksm: register sysfs failed\n"); ++ kthread_stop(uksm_thread); ++ goto out_free; ++ } ++#else ++ uksm_run = UKSM_RUN_MERGE; /* no way for user to start it */ ++ ++#endif /* CONFIG_SYSFS */ ++ ++#ifdef CONFIG_MEMORY_HOTREMOVE ++ /* ++ * Choose a high priority since the callback takes uksm_thread_mutex: ++ * later callbacks could only be taking locks which nest within that. ++ */ ++ hotplug_memory_notifier(uksm_memory_callback, 100); ++#endif ++ return 0; ++ ++out_free: ++ kfree(zero_hash_table); ++out_free0: ++ uksm_slab_free(); ++out_free1: ++ kfree(random_nums); ++out_free2: ++ kfree(uksm_scan_ladder); ++ return err; ++} ++ ++#ifdef MODULE ++subsys_initcall(ksm_init); ++#else ++late_initcall(uksm_init); ++#endif ++ +diff -Nur a/mm/vmstat.c b/mm/vmstat.c +--- a/mm/vmstat.c 2018-05-25 15:18:02.000000000 +0100 ++++ b/mm/vmstat.c 2018-05-26 19:30:55.791140570 +0100 +@@ -1091,6 +1091,9 @@ + "nr_dirtied", + "nr_written", + ++#ifdef CONFIG_UKSM ++ "nr_uksm_zero_pages", ++#endif + /* enum writeback_stat_item counters */ + "nr_dirty_threshold", + "nr_dirty_background_threshold", -- cgit v1.2.3