diff options
Diffstat (limited to 'block/bfq-cgroup.c')
-rw-r--r-- | block/bfq-cgroup.c | 779 |
1 files changed, 525 insertions, 254 deletions
diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c index b3796a40a61a..2c90e5de0acd 100644 --- a/block/bfq-cgroup.c +++ b/block/bfq-cgroup.c @@ -6,16 +6,92 @@ #include <linux/slab.h> #include <linux/blkdev.h> #include <linux/cgroup.h> -#include <linux/elevator.h> #include <linux/ktime.h> #include <linux/rbtree.h> #include <linux/ioprio.h> #include <linux/sbitmap.h> #include <linux/delay.h> +#include "elevator.h" #include "bfq-iosched.h" -#if defined(CONFIG_BFQ_GROUP_IOSCHED) && defined(CONFIG_DEBUG_BLK_CGROUP) +#ifdef CONFIG_BFQ_CGROUP_DEBUG +static int bfq_stat_init(struct bfq_stat *stat, gfp_t gfp) +{ + int ret; + + ret = percpu_counter_init(&stat->cpu_cnt, 0, gfp); + if (ret) + return ret; + + atomic64_set(&stat->aux_cnt, 0); + return 0; +} + +static void bfq_stat_exit(struct bfq_stat *stat) +{ + percpu_counter_destroy(&stat->cpu_cnt); +} + +/** + * bfq_stat_add - add a value to a bfq_stat + * @stat: target bfq_stat + * @val: value to add + * + * Add @val to @stat. The caller must ensure that IRQ on the same CPU + * don't re-enter this function for the same counter. + */ +static inline void bfq_stat_add(struct bfq_stat *stat, uint64_t val) +{ + percpu_counter_add_batch(&stat->cpu_cnt, val, BLKG_STAT_CPU_BATCH); +} + +/** + * bfq_stat_read - read the current value of a bfq_stat + * @stat: bfq_stat to read + */ +static inline uint64_t bfq_stat_read(struct bfq_stat *stat) +{ + return percpu_counter_sum_positive(&stat->cpu_cnt); +} + +/** + * bfq_stat_reset - reset a bfq_stat + * @stat: bfq_stat to reset + */ +static inline void bfq_stat_reset(struct bfq_stat *stat) +{ + percpu_counter_set(&stat->cpu_cnt, 0); + atomic64_set(&stat->aux_cnt, 0); +} + +/** + * bfq_stat_add_aux - add a bfq_stat into another's aux count + * @to: the destination bfq_stat + * @from: the source + * + * Add @from's count including the aux one to @to's aux count. + */ +static inline void bfq_stat_add_aux(struct bfq_stat *to, + struct bfq_stat *from) +{ + atomic64_add(bfq_stat_read(from) + atomic64_read(&from->aux_cnt), + &to->aux_cnt); +} + +/** + * blkg_prfill_stat - prfill callback for bfq_stat + * @sf: seq_file to print to + * @pd: policy private data of interest + * @off: offset to the bfq_stat in @pd + * + * prfill callback for printing a bfq_stat. + */ +static u64 blkg_prfill_stat(struct seq_file *sf, struct blkg_policy_data *pd, + int off) +{ + return __blkg_prfill_u64(sf, pd, bfq_stat_read((void *)pd + off)); +} /* bfqg stats flags */ enum bfqg_stats_flags { @@ -53,7 +129,7 @@ static void bfqg_stats_update_group_wait_time(struct bfqg_stats *stats) now = ktime_get_ns(); if (now > stats->start_group_wait_time) - blkg_stat_add(&stats->group_wait_time, + bfq_stat_add(&stats->group_wait_time, now - stats->start_group_wait_time); bfqg_stats_clear_waiting(stats); } @@ -82,14 +158,14 @@ static void bfqg_stats_end_empty_time(struct bfqg_stats *stats) now = ktime_get_ns(); if (now > stats->start_empty_time) - blkg_stat_add(&stats->empty_time, + bfq_stat_add(&stats->empty_time, now - stats->start_empty_time); bfqg_stats_clear_empty(stats); } void bfqg_stats_update_dequeue(struct bfq_group *bfqg) { - blkg_stat_add(&bfqg->stats.dequeue, 1); + bfq_stat_add(&bfqg->stats.dequeue, 1); } void bfqg_stats_set_start_empty_time(struct bfq_group *bfqg) @@ -119,7 +195,7 @@ void bfqg_stats_update_idle_time(struct bfq_group *bfqg) u64 now = ktime_get_ns(); if (now > stats->start_idle_time) - blkg_stat_add(&stats->idle_time, + bfq_stat_add(&stats->idle_time, now - stats->start_idle_time); bfqg_stats_clear_idling(stats); } @@ -137,60 +213,55 @@ 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, + bfq_stat_add(&stats->avg_queue_size_sum, blkg_rwstat_total(&stats->queued)); - blkg_stat_add(&stats->avg_queue_size_samples, 1); + bfq_stat_add(&stats->avg_queue_size_samples, 1); bfqg_stats_update_group_wait_time(stats); } void bfqg_stats_update_io_add(struct bfq_group *bfqg, struct bfq_queue *bfqq, - unsigned int op) + blk_opf_t opf) { - blkg_rwstat_add(&bfqg->stats.queued, op, 1); + blkg_rwstat_add(&bfqg->stats.queued, opf, 1); bfqg_stats_end_empty_time(&bfqg->stats); - if (!(bfqq == ((struct bfq_data *)bfqg->bfqd)->in_service_queue)) + if (!(bfqq == bfqg->bfqd->in_service_queue)) bfqg_stats_set_start_group_wait_time(bfqg, bfqq_group(bfqq)); } -void bfqg_stats_update_io_remove(struct bfq_group *bfqg, unsigned int op) +void bfqg_stats_update_io_remove(struct bfq_group *bfqg, blk_opf_t opf) { - blkg_rwstat_add(&bfqg->stats.queued, op, -1); + blkg_rwstat_add(&bfqg->stats.queued, opf, -1); } -void bfqg_stats_update_io_merged(struct bfq_group *bfqg, unsigned int op) +void bfqg_stats_update_io_merged(struct bfq_group *bfqg, blk_opf_t opf) { - blkg_rwstat_add(&bfqg->stats.merged, op, 1); + blkg_rwstat_add(&bfqg->stats.merged, opf, 1); } void bfqg_stats_update_completion(struct bfq_group *bfqg, u64 start_time_ns, - u64 io_start_time_ns, unsigned int op) + u64 io_start_time_ns, blk_opf_t opf) { struct bfqg_stats *stats = &bfqg->stats; u64 now = ktime_get_ns(); if (now > io_start_time_ns) - blkg_rwstat_add(&stats->service_time, op, + blkg_rwstat_add(&stats->service_time, opf, now - io_start_time_ns); if (io_start_time_ns > start_time_ns) - blkg_rwstat_add(&stats->wait_time, op, + blkg_rwstat_add(&stats->wait_time, opf, io_start_time_ns - start_time_ns); } -#else /* CONFIG_BFQ_GROUP_IOSCHED && CONFIG_DEBUG_BLK_CGROUP */ +#else /* CONFIG_BFQ_CGROUP_DEBUG */ -void bfqg_stats_update_io_add(struct bfq_group *bfqg, struct bfq_queue *bfqq, - unsigned int op) { } -void bfqg_stats_update_io_remove(struct bfq_group *bfqg, unsigned int op) { } -void bfqg_stats_update_io_merged(struct bfq_group *bfqg, unsigned int op) { } +void bfqg_stats_update_io_remove(struct bfq_group *bfqg, blk_opf_t opf) { } +void bfqg_stats_update_io_merged(struct bfq_group *bfqg, blk_opf_t opf) { } void bfqg_stats_update_completion(struct bfq_group *bfqg, u64 start_time_ns, - u64 io_start_time_ns, unsigned int op) { } + u64 io_start_time_ns, blk_opf_t opf) { } void bfqg_stats_update_dequeue(struct bfq_group *bfqg) { } -void bfqg_stats_set_start_empty_time(struct bfq_group *bfqg) { } -void bfqg_stats_update_idle_time(struct bfq_group *bfqg) { } void bfqg_stats_set_start_idle_time(struct bfq_group *bfqg) { } -void bfqg_stats_update_avg_queue_size(struct bfq_group *bfqg) { } -#endif /* CONFIG_BFQ_GROUP_IOSCHED && CONFIG_DEBUG_BLK_CGROUP */ +#endif /* CONFIG_BFQ_CGROUP_DEBUG */ #ifdef CONFIG_BFQ_GROUP_IOSCHED @@ -245,14 +316,12 @@ struct bfq_group *bfqq_group(struct bfq_queue *bfqq) static void bfqg_get(struct bfq_group *bfqg) { - bfqg->ref++; + refcount_inc(&bfqg->ref); } static void bfqg_put(struct bfq_group *bfqg) { - bfqg->ref--; - - if (bfqg->ref == 0) + if (refcount_dec_and_test(&bfqg->ref)) kfree(bfqg); } @@ -271,21 +340,32 @@ void bfqg_and_blkg_put(struct bfq_group *bfqg) bfqg_put(bfqg); } +void bfqg_stats_update_legacy_io(struct request_queue *q, struct request *rq) +{ + struct bfq_group *bfqg = blkg_to_bfqg(rq->bio->bi_blkg); + + if (!bfqg) + return; + + blkg_rwstat_add(&bfqg->stats.bytes, rq->cmd_flags, blk_rq_bytes(rq)); + blkg_rwstat_add(&bfqg->stats.ios, rq->cmd_flags, 1); +} + /* @stats = 0 */ static void bfqg_stats_reset(struct bfqg_stats *stats) { -#ifdef CONFIG_DEBUG_BLK_CGROUP +#ifdef CONFIG_BFQ_CGROUP_DEBUG /* 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); + bfq_stat_reset(&stats->time); + bfq_stat_reset(&stats->avg_queue_size_sum); + bfq_stat_reset(&stats->avg_queue_size_samples); + bfq_stat_reset(&stats->dequeue); + bfq_stat_reset(&stats->group_wait_time); + bfq_stat_reset(&stats->idle_time); + bfq_stat_reset(&stats->empty_time); #endif } @@ -295,19 +375,19 @@ static void bfqg_stats_add_aux(struct bfqg_stats *to, struct bfqg_stats *from) if (!to || !from) return; -#ifdef CONFIG_DEBUG_BLK_CGROUP +#ifdef CONFIG_BFQ_CGROUP_DEBUG /* 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, + bfq_stat_add_aux(&from->time, &from->time); + bfq_stat_add_aux(&to->avg_queue_size_sum, &from->avg_queue_size_sum); + bfq_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); + bfq_stat_add_aux(&to->dequeue, &from->dequeue); + bfq_stat_add_aux(&to->group_wait_time, &from->group_wait_time); + bfq_stat_add_aux(&to->idle_time, &from->idle_time); + bfq_stat_add_aux(&to->empty_time, &from->empty_time); #endif } @@ -355,41 +435,49 @@ void bfq_init_entity(struct bfq_entity *entity, struct bfq_group *bfqg) static void bfqg_stats_exit(struct bfqg_stats *stats) { -#ifdef CONFIG_DEBUG_BLK_CGROUP + blkg_rwstat_exit(&stats->bytes); + blkg_rwstat_exit(&stats->ios); +#ifdef CONFIG_BFQ_CGROUP_DEBUG 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); + bfq_stat_exit(&stats->time); + bfq_stat_exit(&stats->avg_queue_size_sum); + bfq_stat_exit(&stats->avg_queue_size_samples); + bfq_stat_exit(&stats->dequeue); + bfq_stat_exit(&stats->group_wait_time); + bfq_stat_exit(&stats->idle_time); + bfq_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->bytes, gfp) || + blkg_rwstat_init(&stats->ios, gfp)) + goto error; + +#ifdef CONFIG_BFQ_CGROUP_DEBUG 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; - } + bfq_stat_init(&stats->time, gfp) || + bfq_stat_init(&stats->avg_queue_size_sum, gfp) || + bfq_stat_init(&stats->avg_queue_size_samples, gfp) || + bfq_stat_init(&stats->dequeue, gfp) || + bfq_stat_init(&stats->group_wait_time, gfp) || + bfq_stat_init(&stats->idle_time, gfp) || + bfq_stat_init(&stats->empty_time, gfp)) + goto error; #endif return 0; + +error: + bfqg_stats_exit(stats); + return -ENOMEM; } static struct bfq_group_data *cpd_to_bfqgd(struct blkcg_policy_data *cpd) @@ -409,15 +497,9 @@ static struct blkcg_policy_data *bfq_cpd_alloc(gfp_t gfp) 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; + bgd->weight = CGROUP_WEIGHT_DFL; + return &bgd->pd; } static void bfq_cpd_free(struct blkcg_policy_data *cpd) @@ -425,11 +507,12 @@ 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) +static struct blkg_policy_data *bfq_pd_alloc(struct gendisk *disk, + struct blkcg *blkcg, gfp_t gfp) { struct bfq_group *bfqg; - bfqg = kzalloc_node(sizeof(*bfqg), gfp, node); + bfqg = kzalloc_node(sizeof(*bfqg), gfp, disk->node_id); if (!bfqg) return NULL; @@ -439,7 +522,7 @@ static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, int node) } /* see comments in bfq_bic_update_cgroup for why refcounting */ - bfqg_get(bfqg); + refcount_set(&bfqg->ref, 1); return &bfqg->pd; } @@ -453,12 +536,15 @@ static void bfq_pd_init(struct blkg_policy_data *pd) entity->orig_weight = entity->weight = entity->new_weight = d->weight; entity->my_sched_data = &bfqg->sched_data; + entity->last_bfqq_created = NULL; + 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->num_queues_with_pending_reqs = 0; bfqg->rq_pos_tree = RB_ROOT; } @@ -487,28 +573,11 @@ static void bfq_group_set_parent(struct bfq_group *bfqg, entity->sched_data = &parent->sched_data; } -static struct bfq_group *bfq_lookup_bfqg(struct bfq_data *bfqd, - struct blkcg *blkcg) +static void bfq_link_bfqg(struct bfq_data *bfqd, struct bfq_group *bfqg) { - struct blkcg_gq *blkg; - - blkg = blkg_lookup(blkcg, bfqd->queue); - if (likely(blkg)) - return blkg_to_bfqg(blkg); - return NULL; -} - -struct bfq_group *bfq_find_set_group(struct bfq_data *bfqd, - struct blkcg *blkcg) -{ - struct bfq_group *bfqg, *parent; + struct bfq_group *parent; struct bfq_entity *entity; - 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 @@ -516,16 +585,37 @@ struct bfq_group *bfq_find_set_group(struct bfq_data *bfqd, */ entity = &bfqg->entity; for_each_entity(entity) { - bfqg = container_of(entity, struct bfq_group, entity); - if (bfqg != bfqd->root_group) { - parent = bfqg_parent(bfqg); + struct bfq_group *curr_bfqg = container_of(entity, + struct bfq_group, entity); + if (curr_bfqg != bfqd->root_group) { + parent = bfqg_parent(curr_bfqg); if (!parent) parent = bfqd->root_group; - bfq_group_set_parent(bfqg, parent); + bfq_group_set_parent(curr_bfqg, parent); } } +} - return bfqg; +struct bfq_group *bfq_bio_bfqg(struct bfq_data *bfqd, struct bio *bio) +{ + struct blkcg_gq *blkg = bio->bi_blkg; + struct bfq_group *bfqg; + + while (blkg) { + if (!blkg->online) { + blkg = blkg->parent; + continue; + } + bfqg = blkg_to_bfqg(blkg); + if (bfqg->pd.online) { + bio_associate_blkg_from_css(bio, &blkg->blkcg->css); + return bfqg; + } + blkg = blkg->parent; + } + bio_associate_blkg_from_css(bio, + &bfqg_to_blkg(bfqd->root_group)->blkcg->css); + return bfqd->root_group; } /** @@ -546,6 +636,32 @@ void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, struct bfq_group *bfqg) { struct bfq_entity *entity = &bfqq->entity; + struct bfq_group *old_parent = bfqq_group(bfqq); + bool has_pending_reqs = false; + + /* + * No point to move bfqq to the same group, which can happen when + * root group is offlined + */ + if (old_parent == bfqg) + return; + + /* + * oom_bfqq is not allowed to move, oom_bfqq will hold ref to root_group + * until elevator exit. + */ + if (bfqq == &bfqd->oom_bfqq) + return; + /* + * Get extra reference to prevent bfqq from being freed in + * next possible expire or deactivate. + */ + bfqq->ref++; + + if (entity->in_groups_with_pending_reqs) { + has_pending_reqs = true; + bfq_del_bfqq_in_groups_with_pending_reqs(bfqq); + } /* If bfqq is empty, then bfq_bfqq_expire also invokes * bfq_del_bfqq_busy, thereby removing bfqq and its entity @@ -559,91 +675,128 @@ void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq, if (bfq_bfqq_busy(bfqq)) bfq_deactivate_bfqq(bfqd, bfqq, false, false); - else if (entity->on_st) + else if (entity->on_st_or_in_serv) bfq_put_idle_entity(bfq_entity_service_tree(entity), entity); - bfqg_and_blkg_put(bfqq_group(bfqq)); + bfqg_and_blkg_put(old_parent); + + if (entity->parent && + entity->parent->last_bfqq_created == bfqq) + entity->parent->last_bfqq_created = NULL; + else if (bfqd->last_bfqq_created == bfqq) + bfqd->last_bfqq_created = NULL; entity->parent = bfqg->my_entity; entity->sched_data = &bfqg->sched_data; /* pin down bfqg and its associated blkg */ bfqg_and_blkg_get(bfqg); + if (has_pending_reqs) + bfq_add_bfqq_in_groups_with_pending_reqs(bfqq); + if (bfq_bfqq_busy(bfqq)) { if (unlikely(!bfqd->nonrot_with_queueing)) bfq_pos_tree_add_move(bfqd, bfqq); bfq_activate_bfqq(bfqd, bfqq); } - if (!bfqd->in_service_queue && !bfqd->rq_in_driver) + if (!bfqd->in_service_queue && !bfqd->tot_rq_in_driver) bfq_schedule_dispatch(bfqd); + /* release extra ref taken above, bfqq may happen to be freed now */ + bfq_put_queue(bfqq); +} + +static void bfq_sync_bfqq_move(struct bfq_data *bfqd, + struct bfq_queue *sync_bfqq, + struct bfq_io_cq *bic, + struct bfq_group *bfqg, + unsigned int act_idx) +{ + struct bfq_queue *bfqq; + + if (!sync_bfqq->new_bfqq && !bfq_bfqq_coop(sync_bfqq)) { + /* We are the only user of this bfqq, just move it */ + if (sync_bfqq->entity.sched_data != &bfqg->sched_data) + bfq_bfqq_move(bfqd, sync_bfqq, bfqg); + return; + } + + /* + * The queue was merged to a different queue. Check + * that the merge chain still belongs to the same + * cgroup. + */ + for (bfqq = sync_bfqq; bfqq; bfqq = bfqq->new_bfqq) + if (bfqq->entity.sched_data != &bfqg->sched_data) + break; + if (bfqq) { + /* + * Some queue changed cgroup so the merge is not valid + * anymore. We cannot easily just cancel the merge (by + * clearing new_bfqq) as there may be other processes + * using this queue and holding refs to all queues + * below sync_bfqq->new_bfqq. Similarly if the merge + * already happened, we need to detach from bfqq now + * so that we cannot merge bio to a request from the + * old cgroup. + */ + bfq_put_cooperator(sync_bfqq); + bic_set_bfqq(bic, NULL, true, act_idx); + bfq_release_process_ref(bfqd, sync_bfqq); + } } /** - * __bfq_bic_change_cgroup - move @bic to @cgroup. + * __bfq_bic_change_cgroup - move @bic to @bfqg. * @bfqd: the queue descriptor. * @bic: the bic to move. - * @blkcg: the blk-cgroup to move to. + * @bfqg: the group to move to. * * 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) - * - * 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) +static void __bfq_bic_change_cgroup(struct bfq_data *bfqd, + struct bfq_io_cq *bic, + struct bfq_group *bfqg) { - 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; - - bfqg = bfq_find_set_group(bfqd, blkcg); + unsigned int act_idx; - if (unlikely(!bfqg)) - bfqg = bfqd->root_group; + for (act_idx = 0; act_idx < bfqd->num_actuators; act_idx++) { + struct bfq_queue *async_bfqq = bic_to_bfqq(bic, false, act_idx); + struct bfq_queue *sync_bfqq = bic_to_bfqq(bic, true, act_idx); - 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 (async_bfqq && + async_bfqq->entity.sched_data != &bfqg->sched_data) { + bic_set_bfqq(bic, NULL, false, act_idx); + bfq_release_process_ref(bfqd, async_bfqq); } - } - if (sync_bfqq) { - entity = &sync_bfqq->entity; - if (entity->sched_data != &bfqg->sched_data) - bfq_bfqq_move(bfqd, sync_bfqq, bfqg); + if (sync_bfqq) + bfq_sync_bfqq_move(bfqd, sync_bfqq, bic, bfqg, act_idx); } - - return bfqg; } 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; + struct bfq_group *bfqg = bfq_bio_bfqg(bfqd, bio); uint64_t serial_nr; - rcu_read_lock(); - serial_nr = __bio_blkcg(bio)->css.serial_nr; + serial_nr = bfqg_to_blkg(bfqg)->blkcg->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; + return; - bfqg = __bfq_bic_change_cgroup(bfqd, bic, __bio_blkcg(bio)); + /* + * New cgroup for this process. Make sure it is linked to bfq internal + * cgroup hierarchy. + */ + bfq_link_bfqg(bfqd, bfqg); + __bfq_bic_change_cgroup(bfqd, bic, bfqg); /* * Update blkg_path for bfq_log_* functions. We cache this * path, and update it here, for the following @@ -696,8 +849,6 @@ void bfq_bic_update_cgroup(struct bfq_io_cq *bic, struct bio *bio) */ blkg_path(bfqg_to_blkg(bfqg), bfqg->blkg_path, sizeof(bfqg->blkg_path)); bic->blkcg_serial_nr = serial_nr; -out: - rcu_read_unlock(); } /** @@ -715,39 +866,55 @@ static void bfq_flush_idle_tree(struct bfq_service_tree *st) /** * 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. + * @entity: the entity to move, if entity is a leaf; or the parent entity + * of an active leaf entity to move, if entity is not a leaf. + * @ioprio_class: I/O priority class to reparent. */ static void bfq_reparent_leaf_entity(struct bfq_data *bfqd, - struct bfq_entity *entity) + struct bfq_entity *entity, + int ioprio_class) { - struct bfq_queue *bfqq = bfq_entity_to_bfqq(entity); + struct bfq_queue *bfqq; + struct bfq_entity *child_entity = entity; + while (child_entity->my_sched_data) { /* leaf not reached yet */ + struct bfq_sched_data *child_sd = child_entity->my_sched_data; + struct bfq_service_tree *child_st = child_sd->service_tree + + ioprio_class; + struct rb_root *child_active = &child_st->active; + + child_entity = bfq_entity_of(rb_first(child_active)); + + if (!child_entity) + child_entity = child_sd->in_service_entity; + } + + bfqq = bfq_entity_to_bfqq(child_entity); bfq_bfqq_move(bfqd, bfqq, bfqd->root_group); } /** - * bfq_reparent_active_entities - move to the root group all active - * entities. + * bfq_reparent_active_queues - move to the root group all active queues. * @bfqd: the device data structure with the root group. * @bfqg: the group to move from. - * @st: the service tree with the entities. + * @st: the service tree to start the search from. + * @ioprio_class: I/O priority class to reparent. */ -static void bfq_reparent_active_entities(struct bfq_data *bfqd, - struct bfq_group *bfqg, - struct bfq_service_tree *st) +static void bfq_reparent_active_queues(struct bfq_data *bfqd, + struct bfq_group *bfqg, + struct bfq_service_tree *st, + int ioprio_class) { struct rb_root *active = &st->active; - struct bfq_entity *entity = NULL; - - if (!RB_EMPTY_ROOT(&st->active)) - entity = bfq_entity_of(rb_first(active)); + struct bfq_entity *entity; - for (; entity ; entity = bfq_entity_of(rb_first(active))) - bfq_reparent_leaf_entity(bfqd, entity); + while ((entity = bfq_entity_of(rb_first(active)))) + bfq_reparent_leaf_entity(bfqd, entity, ioprio_class); if (bfqg->sched_data.in_service_entity) bfq_reparent_leaf_entity(bfqd, - bfqg->sched_data.in_service_entity); + bfqg->sched_data.in_service_entity, + ioprio_class); } /** @@ -780,13 +947,6 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) 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. - */ - 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 @@ -798,7 +958,20 @@ static void bfq_pd_offline(struct blkg_policy_data *pd) * There is no need to put the sync queues, as the * scheduler has taken no reference. */ - bfq_reparent_active_entities(bfqd, bfqg, st); + bfq_reparent_active_queues(bfqd, bfqg, st, 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. In addition, even + * bfq_reparent_active_queues() may happen to add some + * entities to the idle tree. It happens if, in some + * of the calls to bfq_bfqq_move() performed by + * bfq_reparent_active_queues(), the queue to move is + * empty and gets expired. + */ + bfq_flush_idle_tree(st); } __bfq_deactivate_entity(entity, false); @@ -828,7 +1001,7 @@ void bfq_end_wr_async(struct bfq_data *bfqd) bfq_end_wr_async_queues(bfqd, bfqd->root_group); } -static int bfq_io_show_weight(struct seq_file *sf, void *v) +static int bfq_io_show_weight_legacy(struct seq_file *sf, void *v) { struct blkcg *blkcg = css_to_blkcg(seq_css(sf)); struct bfq_group_data *bfqgd = blkcg_to_bfqgd(blkcg); @@ -842,6 +1015,60 @@ static int bfq_io_show_weight(struct seq_file *sf, void *v) return 0; } +static u64 bfqg_prfill_weight_device(struct seq_file *sf, + struct blkg_policy_data *pd, int off) +{ + struct bfq_group *bfqg = pd_to_bfqg(pd); + + if (!bfqg->entity.dev_weight) + return 0; + return __blkg_prfill_u64(sf, pd, bfqg->entity.dev_weight); +} + +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); + + seq_printf(sf, "default %u\n", bfqgd->weight); + blkcg_print_blkgs(sf, blkcg, bfqg_prfill_weight_device, + &blkcg_policy_bfq, 0, false); + return 0; +} + +static void bfq_group_set_weight(struct bfq_group *bfqg, u64 weight, u64 dev_weight) +{ + weight = dev_weight ?: weight; + + bfqg->entity.dev_weight = dev_weight; + /* + * 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)weight != bfqg->entity.new_weight) { + bfqg->entity.new_weight = (unsigned short)weight; + /* + * 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; + } +} + static int bfq_io_set_weight_legacy(struct cgroup_subsys_state *css, struct cftype *cftype, u64 val) @@ -860,61 +1087,72 @@ static int bfq_io_set_weight_legacy(struct cgroup_subsys_state *css, 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; - } + if (bfqg) + bfq_group_set_weight(bfqg, val, 0); } 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) +static ssize_t bfq_io_set_device_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); + int ret; + struct blkg_conf_ctx ctx; + struct blkcg *blkcg = css_to_blkcg(of_css(of)); + struct bfq_group *bfqg; + u64 v; + blkg_conf_init(&ctx, buf); + + ret = blkg_conf_prep(blkcg, &blkcg_policy_bfq, &ctx); if (ret) - return ret; + goto out; + + if (sscanf(ctx.body, "%llu", &v) == 1) { + /* require "default" on dfl */ + ret = -ERANGE; + if (!v) + goto out; + } else if (!strcmp(strim(ctx.body), "default")) { + v = 0; + } else { + ret = -EINVAL; + goto out; + } - ret = bfq_io_set_weight_legacy(of_css(of), NULL, weight); + bfqg = blkg_to_bfqg(ctx.blkg); + + ret = -ERANGE; + if (!v || (v >= BFQ_MIN_WEIGHT && v <= BFQ_MAX_WEIGHT)) { + bfq_group_set_weight(bfqg, bfqg->entity.weight, v); + ret = 0; + } +out: + blkg_conf_exit(&ctx); return ret ?: nbytes; } -#ifdef CONFIG_DEBUG_BLK_CGROUP -static int bfqg_print_stat(struct seq_file *sf, void *v) +static ssize_t bfq_io_set_weight(struct kernfs_open_file *of, + char *buf, size_t nbytes, + loff_t off) { - blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), blkg_prfill_stat, - &blkcg_policy_bfq, seq_cft(sf)->private, false); - return 0; + char *endp; + int ret; + u64 v; + + buf = strim(buf); + + /* "WEIGHT" or "default WEIGHT" sets the default weight */ + v = simple_strtoull(buf, &endp, 0); + if (*endp == '\0' || sscanf(buf, "default %llu", &v) == 1) { + ret = bfq_io_set_weight_legacy(of_css(of), NULL, v); + return ret ?: nbytes; + } + + return bfq_io_set_device_weight(of, buf, nbytes, off); } static int bfqg_print_rwstat(struct seq_file *sf, void *v) @@ -924,43 +1162,69 @@ static int bfqg_print_rwstat(struct seq_file *sf, void *v) 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); + struct blkg_rwstat_sample sum; + + blkg_rwstat_recursive_sum(pd_to_blkg(pd), &blkcg_policy_bfq, off, &sum); return __blkg_prfill_rwstat(sf, pd, &sum); } -static int bfqg_print_stat_recursive(struct seq_file *sf, void *v) +static int bfqg_print_rwstat_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); + bfqg_prfill_rwstat_recursive, &blkcg_policy_bfq, + seq_cft(sf)->private, true); return 0; } -static int bfqg_print_rwstat_recursive(struct seq_file *sf, void *v) +#ifdef CONFIG_BFQ_CGROUP_DEBUG +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 u64 bfqg_prfill_stat_recursive(struct seq_file *sf, + struct blkg_policy_data *pd, int off) +{ + struct blkcg_gq *blkg = pd_to_blkg(pd); + struct blkcg_gq *pos_blkg; + struct cgroup_subsys_state *pos_css; + u64 sum = 0; + + lockdep_assert_held(&blkg->q->queue_lock); + + rcu_read_lock(); + blkg_for_each_descendant_pre(pos_blkg, pos_css, blkg) { + struct bfq_stat *stat; + + if (!pos_blkg->online) + continue; + + stat = (void *)blkg_to_pd(pos_blkg, &blkcg_policy_bfq) + off; + sum += bfq_stat_read(stat) + atomic64_read(&stat->aux_cnt); + } + rcu_read_unlock(); + + return __blkg_prfill_u64(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_rwstat_recursive, &blkcg_policy_bfq, - seq_cft(sf)->private, true); + bfqg_prfill_stat_recursive, &blkcg_policy_bfq, + seq_cft(sf)->private, false); 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); + struct bfq_group *bfqg = blkg_to_bfqg(pd->blkg); + u64 sum = blkg_rwstat_total(&bfqg->stats.bytes); return __blkg_prfill_u64(sf, pd, sum >> 9); } @@ -975,12 +1239,13 @@ static int bfqg_print_stat_sectors(struct seq_file *sf, void *v) 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]); + struct blkg_rwstat_sample tmp; - return __blkg_prfill_u64(sf, pd, sum >> 9); + blkg_rwstat_recursive_sum(pd->blkg, &blkcg_policy_bfq, + offsetof(struct bfq_group, stats.bytes), &tmp); + + return __blkg_prfill_u64(sf, pd, + (tmp.cnt[BLKG_RWSTAT_READ] + tmp.cnt[BLKG_RWSTAT_WRITE]) >> 9); } static int bfqg_print_stat_sectors_recursive(struct seq_file *sf, void *v) @@ -995,11 +1260,11 @@ 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 samples = bfq_stat_read(&bfqg->stats.avg_queue_size_samples); u64 v = 0; if (samples) { - v = blkg_stat_read(&bfqg->stats.avg_queue_size_sum); + v = bfq_stat_read(&bfqg->stats.avg_queue_size_sum); v = div64_u64(v, samples); } __blkg_prfill_u64(sf, pd, v); @@ -1014,13 +1279,13 @@ static int bfqg_print_avg_queue_size(struct seq_file *sf, void *v) 0, false); return 0; } -#endif /* CONFIG_DEBUG_BLK_CGROUP */ +#endif /* CONFIG_BFQ_CGROUP_DEBUG */ struct bfq_group *bfq_create_group_hierarchy(struct bfq_data *bfqd, int node) { int ret; - ret = blkcg_activate_policy(bfqd->queue, &blkcg_policy_bfq); + ret = blkcg_activate_policy(bfqd->queue->disk, &blkcg_policy_bfq); if (ret) return NULL; @@ -1032,8 +1297,6 @@ struct blkcg_policy blkcg_policy_bfq = { .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, @@ -1047,22 +1310,28 @@ struct cftype bfq_blkcg_legacy_files[] = { { .name = "bfq.weight", .flags = CFTYPE_NOT_ON_ROOT, - .seq_show = bfq_io_show_weight, + .seq_show = bfq_io_show_weight_legacy, .write_u64 = bfq_io_set_weight_legacy, }, + { + .name = "bfq.weight_device", + .flags = CFTYPE_NOT_ON_ROOT, + .seq_show = bfq_io_show_weight, + .write = bfq_io_set_weight, + }, /* statistics, covers only the tasks in the bfqg */ { .name = "bfq.io_service_bytes", - .private = (unsigned long)&blkcg_policy_bfq, - .seq_show = blkg_print_stat_bytes, + .private = offsetof(struct bfq_group, stats.bytes), + .seq_show = bfqg_print_rwstat, }, { .name = "bfq.io_serviced", - .private = (unsigned long)&blkcg_policy_bfq, - .seq_show = blkg_print_stat_ios, + .private = offsetof(struct bfq_group, stats.ios), + .seq_show = bfqg_print_rwstat, }, -#ifdef CONFIG_DEBUG_BLK_CGROUP +#ifdef CONFIG_BFQ_CGROUP_DEBUG { .name = "bfq.time", .private = offsetof(struct bfq_group, stats.time), @@ -1092,20 +1361,20 @@ struct cftype bfq_blkcg_legacy_files[] = { .private = offsetof(struct bfq_group, stats.queued), .seq_show = bfqg_print_rwstat, }, -#endif /* CONFIG_DEBUG_BLK_CGROUP */ +#endif /* CONFIG_BFQ_CGROUP_DEBUG */ /* the same statistics which cover the bfqg and its descendants */ { .name = "bfq.io_service_bytes_recursive", - .private = (unsigned long)&blkcg_policy_bfq, - .seq_show = blkg_print_stat_bytes_recursive, + .private = offsetof(struct bfq_group, stats.bytes), + .seq_show = bfqg_print_rwstat_recursive, }, { .name = "bfq.io_serviced_recursive", - .private = (unsigned long)&blkcg_policy_bfq, - .seq_show = blkg_print_stat_ios_recursive, + .private = offsetof(struct bfq_group, stats.ios), + .seq_show = bfqg_print_rwstat_recursive, }, -#ifdef CONFIG_DEBUG_BLK_CGROUP +#ifdef CONFIG_BFQ_CGROUP_DEBUG { .name = "bfq.time_recursive", .private = offsetof(struct bfq_group, stats.time), @@ -1159,7 +1428,7 @@ struct cftype bfq_blkcg_legacy_files[] = { .private = offsetof(struct bfq_group, stats.dequeue), .seq_show = bfqg_print_stat, }, -#endif /* CONFIG_DEBUG_BLK_CGROUP */ +#endif /* CONFIG_BFQ_CGROUP_DEBUG */ { } /* terminate */ }; @@ -1198,7 +1467,7 @@ void bfq_end_wr_async(struct bfq_data *bfqd) bfq_end_wr_async_queues(bfqd, bfqd->root_group); } -struct bfq_group *bfq_find_set_group(struct bfq_data *bfqd, struct blkcg *blkcg) +struct bfq_group *bfq_bio_bfqg(struct bfq_data *bfqd, struct bio *bio) { return bfqd->root_group; } @@ -1208,6 +1477,8 @@ struct bfq_group *bfqq_group(struct bfq_queue *bfqq) return bfqq->bfqd->root_group; } +void bfqg_and_blkg_put(struct bfq_group *bfqg) {} + struct bfq_group *bfq_create_group_hierarchy(struct bfq_data *bfqd, int node) { struct bfq_group *bfqg; |