linux/block/blk-mq-sched.c
Jens Axboe 8e8320c931 blk-mq: fix performance regression with shared tags
If we have shared tags enabled, then every IO completion will trigger
a full loop of every queue belonging to a tag set, and every hardware
queue for each of those queues, even if nothing needs to be done.
This causes a massive performance regression if you have a lot of
shared devices.

Instead of doing this huge full scan on every IO, add an atomic
counter to the main queue that tracks how many hardware queues have
been marked as needing a restart. With that, we can avoid looking for
restartable queues, if we don't have to.

Max reports that this restores performance. Before this patch, 4K
IOPS was limited to 22-23K IOPS. With the patch, we are running at
950-970K IOPS.

Fixes: 6d8c6c0f97 ("blk-mq: Restart a single queue if tag sets are shared")
Reported-by: Max Gurtovoy <maxg@mellanox.com>
Tested-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Tested-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-21 10:17:49 -06:00

613 lines
15 KiB
C

/*
* blk-mq scheduling framework
*
* Copyright (C) 2016 Jens Axboe
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/blk-mq.h>
#include <trace/events/block.h>
#include "blk.h"
#include "blk-mq.h"
#include "blk-mq-debugfs.h"
#include "blk-mq-sched.h"
#include "blk-mq-tag.h"
#include "blk-wbt.h"
void blk_mq_sched_free_hctx_data(struct request_queue *q,
void (*exit)(struct blk_mq_hw_ctx *))
{
struct blk_mq_hw_ctx *hctx;
int i;
queue_for_each_hw_ctx(q, hctx, i) {
if (exit && hctx->sched_data)
exit(hctx);
kfree(hctx->sched_data);
hctx->sched_data = NULL;
}
}
EXPORT_SYMBOL_GPL(blk_mq_sched_free_hctx_data);
static void __blk_mq_sched_assign_ioc(struct request_queue *q,
struct request *rq,
struct bio *bio,
struct io_context *ioc)
{
struct io_cq *icq;
spin_lock_irq(q->queue_lock);
icq = ioc_lookup_icq(ioc, q);
spin_unlock_irq(q->queue_lock);
if (!icq) {
icq = ioc_create_icq(ioc, q, GFP_ATOMIC);
if (!icq)
return;
}
rq->elv.icq = icq;
if (!blk_mq_sched_get_rq_priv(q, rq, bio)) {
rq->rq_flags |= RQF_ELVPRIV;
get_io_context(icq->ioc);
return;
}
rq->elv.icq = NULL;
}
static void blk_mq_sched_assign_ioc(struct request_queue *q,
struct request *rq, struct bio *bio)
{
struct io_context *ioc;
ioc = rq_ioc(bio);
if (ioc)
__blk_mq_sched_assign_ioc(q, rq, bio, ioc);
}
/*
* Mark a hardware queue as needing a restart. For shared queues, maintain
* a count of how many hardware queues are marked for restart.
*/
static void blk_mq_sched_mark_restart_hctx(struct blk_mq_hw_ctx *hctx)
{
if (test_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state))
return;
if (hctx->flags & BLK_MQ_F_TAG_SHARED) {
struct request_queue *q = hctx->queue;
if (!test_and_set_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state))
atomic_inc(&q->shared_hctx_restart);
} else
set_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
}
static bool blk_mq_sched_restart_hctx(struct blk_mq_hw_ctx *hctx)
{
if (!test_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state))
return false;
if (hctx->flags & BLK_MQ_F_TAG_SHARED) {
struct request_queue *q = hctx->queue;
if (test_and_clear_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state))
atomic_dec(&q->shared_hctx_restart);
} else
clear_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
if (blk_mq_hctx_has_pending(hctx)) {
blk_mq_run_hw_queue(hctx, true);
return true;
}
return false;
}
struct request *blk_mq_sched_get_request(struct request_queue *q,
struct bio *bio,
unsigned int op,
struct blk_mq_alloc_data *data)
{
struct elevator_queue *e = q->elevator;
struct request *rq;
blk_queue_enter_live(q);
data->q = q;
if (likely(!data->ctx))
data->ctx = blk_mq_get_ctx(q);
if (likely(!data->hctx))
data->hctx = blk_mq_map_queue(q, data->ctx->cpu);
if (e) {
data->flags |= BLK_MQ_REQ_INTERNAL;
/*
* Flush requests are special and go directly to the
* dispatch list.
*/
if (!op_is_flush(op) && e->type->ops.mq.get_request) {
rq = e->type->ops.mq.get_request(q, op, data);
if (rq)
rq->rq_flags |= RQF_QUEUED;
} else
rq = __blk_mq_alloc_request(data, op);
} else {
rq = __blk_mq_alloc_request(data, op);
}
if (rq) {
if (!op_is_flush(op)) {
rq->elv.icq = NULL;
if (e && e->type->icq_cache)
blk_mq_sched_assign_ioc(q, rq, bio);
}
data->hctx->queued++;
return rq;
}
blk_queue_exit(q);
return NULL;
}
void blk_mq_sched_put_request(struct request *rq)
{
struct request_queue *q = rq->q;
struct elevator_queue *e = q->elevator;
if (rq->rq_flags & RQF_ELVPRIV) {
blk_mq_sched_put_rq_priv(rq->q, rq);
if (rq->elv.icq) {
put_io_context(rq->elv.icq->ioc);
rq->elv.icq = NULL;
}
}
if ((rq->rq_flags & RQF_QUEUED) && e && e->type->ops.mq.put_request)
e->type->ops.mq.put_request(rq);
else
blk_mq_finish_request(rq);
}
void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
{
struct request_queue *q = hctx->queue;
struct elevator_queue *e = q->elevator;
const bool has_sched_dispatch = e && e->type->ops.mq.dispatch_request;
bool did_work = false;
LIST_HEAD(rq_list);
if (unlikely(blk_mq_hctx_stopped(hctx)))
return;
hctx->run++;
/*
* If we have previous entries on our dispatch list, grab them first for
* more fair dispatch.
*/
if (!list_empty_careful(&hctx->dispatch)) {
spin_lock(&hctx->lock);
if (!list_empty(&hctx->dispatch))
list_splice_init(&hctx->dispatch, &rq_list);
spin_unlock(&hctx->lock);
}
/*
* Only ask the scheduler for requests, if we didn't have residual
* requests from the dispatch list. This is to avoid the case where
* we only ever dispatch a fraction of the requests available because
* of low device queue depth. Once we pull requests out of the IO
* scheduler, we can no longer merge or sort them. So it's best to
* leave them there for as long as we can. Mark the hw queue as
* needing a restart in that case.
*/
if (!list_empty(&rq_list)) {
blk_mq_sched_mark_restart_hctx(hctx);
did_work = blk_mq_dispatch_rq_list(q, &rq_list);
} else if (!has_sched_dispatch) {
blk_mq_flush_busy_ctxs(hctx, &rq_list);
blk_mq_dispatch_rq_list(q, &rq_list);
}
/*
* We want to dispatch from the scheduler if we had no work left
* on the dispatch list, OR if we did have work but weren't able
* to make progress.
*/
if (!did_work && has_sched_dispatch) {
do {
struct request *rq;
rq = e->type->ops.mq.dispatch_request(hctx);
if (!rq)
break;
list_add(&rq->queuelist, &rq_list);
} while (blk_mq_dispatch_rq_list(q, &rq_list));
}
}
bool blk_mq_sched_try_merge(struct request_queue *q, struct bio *bio,
struct request **merged_request)
{
struct request *rq;
switch (elv_merge(q, &rq, bio)) {
case ELEVATOR_BACK_MERGE:
if (!blk_mq_sched_allow_merge(q, rq, bio))
return false;
if (!bio_attempt_back_merge(q, rq, bio))
return false;
*merged_request = attempt_back_merge(q, rq);
if (!*merged_request)
elv_merged_request(q, rq, ELEVATOR_BACK_MERGE);
return true;
case ELEVATOR_FRONT_MERGE:
if (!blk_mq_sched_allow_merge(q, rq, bio))
return false;
if (!bio_attempt_front_merge(q, rq, bio))
return false;
*merged_request = attempt_front_merge(q, rq);
if (!*merged_request)
elv_merged_request(q, rq, ELEVATOR_FRONT_MERGE);
return true;
default:
return false;
}
}
EXPORT_SYMBOL_GPL(blk_mq_sched_try_merge);
bool __blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio)
{
struct elevator_queue *e = q->elevator;
if (e->type->ops.mq.bio_merge) {
struct blk_mq_ctx *ctx = blk_mq_get_ctx(q);
struct blk_mq_hw_ctx *hctx = blk_mq_map_queue(q, ctx->cpu);
blk_mq_put_ctx(ctx);
return e->type->ops.mq.bio_merge(hctx, bio);
}
return false;
}
bool blk_mq_sched_try_insert_merge(struct request_queue *q, struct request *rq)
{
return rq_mergeable(rq) && elv_attempt_insert_merge(q, rq);
}
EXPORT_SYMBOL_GPL(blk_mq_sched_try_insert_merge);
void blk_mq_sched_request_inserted(struct request *rq)
{
trace_block_rq_insert(rq->q, rq);
}
EXPORT_SYMBOL_GPL(blk_mq_sched_request_inserted);
static bool blk_mq_sched_bypass_insert(struct blk_mq_hw_ctx *hctx,
struct request *rq)
{
if (rq->tag == -1) {
rq->rq_flags |= RQF_SORTED;
return false;
}
/*
* If we already have a real request tag, send directly to
* the dispatch list.
*/
spin_lock(&hctx->lock);
list_add(&rq->queuelist, &hctx->dispatch);
spin_unlock(&hctx->lock);
return true;
}
/**
* list_for_each_entry_rcu_rr - iterate in a round-robin fashion over rcu list
* @pos: loop cursor.
* @skip: the list element that will not be examined. Iteration starts at
* @skip->next.
* @head: head of the list to examine. This list must have at least one
* element, namely @skip.
* @member: name of the list_head structure within typeof(*pos).
*/
#define list_for_each_entry_rcu_rr(pos, skip, head, member) \
for ((pos) = (skip); \
(pos = (pos)->member.next != (head) ? list_entry_rcu( \
(pos)->member.next, typeof(*pos), member) : \
list_entry_rcu((pos)->member.next->next, typeof(*pos), member)), \
(pos) != (skip); )
/*
* Called after a driver tag has been freed to check whether a hctx needs to
* be restarted. Restarts @hctx if its tag set is not shared. Restarts hardware
* queues in a round-robin fashion if the tag set of @hctx is shared with other
* hardware queues.
*/
void blk_mq_sched_restart(struct blk_mq_hw_ctx *const hctx)
{
struct blk_mq_tags *const tags = hctx->tags;
struct blk_mq_tag_set *const set = hctx->queue->tag_set;
struct request_queue *const queue = hctx->queue, *q;
struct blk_mq_hw_ctx *hctx2;
unsigned int i, j;
if (set->flags & BLK_MQ_F_TAG_SHARED) {
/*
* If this is 0, then we know that no hardware queues
* have RESTART marked. We're done.
*/
if (!atomic_read(&queue->shared_hctx_restart))
return;
rcu_read_lock();
list_for_each_entry_rcu_rr(q, queue, &set->tag_list,
tag_set_list) {
queue_for_each_hw_ctx(q, hctx2, i)
if (hctx2->tags == tags &&
blk_mq_sched_restart_hctx(hctx2))
goto done;
}
j = hctx->queue_num + 1;
for (i = 0; i < queue->nr_hw_queues; i++, j++) {
if (j == queue->nr_hw_queues)
j = 0;
hctx2 = queue->queue_hw_ctx[j];
if (hctx2->tags == tags &&
blk_mq_sched_restart_hctx(hctx2))
break;
}
done:
rcu_read_unlock();
} else {
blk_mq_sched_restart_hctx(hctx);
}
}
/*
* Add flush/fua to the queue. If we fail getting a driver tag, then
* punt to the requeue list. Requeue will re-invoke us from a context
* that's safe to block from.
*/
static void blk_mq_sched_insert_flush(struct blk_mq_hw_ctx *hctx,
struct request *rq, bool can_block)
{
if (blk_mq_get_driver_tag(rq, &hctx, can_block)) {
blk_insert_flush(rq);
blk_mq_run_hw_queue(hctx, true);
} else
blk_mq_add_to_requeue_list(rq, false, true);
}
void blk_mq_sched_insert_request(struct request *rq, bool at_head,
bool run_queue, bool async, bool can_block)
{
struct request_queue *q = rq->q;
struct elevator_queue *e = q->elevator;
struct blk_mq_ctx *ctx = rq->mq_ctx;
struct blk_mq_hw_ctx *hctx = blk_mq_map_queue(q, ctx->cpu);
if (rq->tag == -1 && op_is_flush(rq->cmd_flags)) {
blk_mq_sched_insert_flush(hctx, rq, can_block);
return;
}
if (e && blk_mq_sched_bypass_insert(hctx, rq))
goto run;
if (e && e->type->ops.mq.insert_requests) {
LIST_HEAD(list);
list_add(&rq->queuelist, &list);
e->type->ops.mq.insert_requests(hctx, &list, at_head);
} else {
spin_lock(&ctx->lock);
__blk_mq_insert_request(hctx, rq, at_head);
spin_unlock(&ctx->lock);
}
run:
if (run_queue)
blk_mq_run_hw_queue(hctx, async);
}
void blk_mq_sched_insert_requests(struct request_queue *q,
struct blk_mq_ctx *ctx,
struct list_head *list, bool run_queue_async)
{
struct blk_mq_hw_ctx *hctx = blk_mq_map_queue(q, ctx->cpu);
struct elevator_queue *e = hctx->queue->elevator;
if (e) {
struct request *rq, *next;
/*
* We bypass requests that already have a driver tag assigned,
* which should only be flushes. Flushes are only ever inserted
* as single requests, so we shouldn't ever hit the
* WARN_ON_ONCE() below (but let's handle it just in case).
*/
list_for_each_entry_safe(rq, next, list, queuelist) {
if (WARN_ON_ONCE(rq->tag != -1)) {
list_del_init(&rq->queuelist);
blk_mq_sched_bypass_insert(hctx, rq);
}
}
}
if (e && e->type->ops.mq.insert_requests)
e->type->ops.mq.insert_requests(hctx, list, false);
else
blk_mq_insert_requests(hctx, ctx, list);
blk_mq_run_hw_queue(hctx, run_queue_async);
}
static void blk_mq_sched_free_tags(struct blk_mq_tag_set *set,
struct blk_mq_hw_ctx *hctx,
unsigned int hctx_idx)
{
if (hctx->sched_tags) {
blk_mq_free_rqs(set, hctx->sched_tags, hctx_idx);
blk_mq_free_rq_map(hctx->sched_tags);
hctx->sched_tags = NULL;
}
}
static int blk_mq_sched_alloc_tags(struct request_queue *q,
struct blk_mq_hw_ctx *hctx,
unsigned int hctx_idx)
{
struct blk_mq_tag_set *set = q->tag_set;
int ret;
hctx->sched_tags = blk_mq_alloc_rq_map(set, hctx_idx, q->nr_requests,
set->reserved_tags);
if (!hctx->sched_tags)
return -ENOMEM;
ret = blk_mq_alloc_rqs(set, hctx->sched_tags, hctx_idx, q->nr_requests);
if (ret)
blk_mq_sched_free_tags(set, hctx, hctx_idx);
return ret;
}
static void blk_mq_sched_tags_teardown(struct request_queue *q)
{
struct blk_mq_tag_set *set = q->tag_set;
struct blk_mq_hw_ctx *hctx;
int i;
queue_for_each_hw_ctx(q, hctx, i)
blk_mq_sched_free_tags(set, hctx, i);
}
int blk_mq_sched_init_hctx(struct request_queue *q, struct blk_mq_hw_ctx *hctx,
unsigned int hctx_idx)
{
struct elevator_queue *e = q->elevator;
int ret;
if (!e)
return 0;
ret = blk_mq_sched_alloc_tags(q, hctx, hctx_idx);
if (ret)
return ret;
if (e->type->ops.mq.init_hctx) {
ret = e->type->ops.mq.init_hctx(hctx, hctx_idx);
if (ret) {
blk_mq_sched_free_tags(q->tag_set, hctx, hctx_idx);
return ret;
}
}
blk_mq_debugfs_register_sched_hctx(q, hctx);
return 0;
}
void blk_mq_sched_exit_hctx(struct request_queue *q, struct blk_mq_hw_ctx *hctx,
unsigned int hctx_idx)
{
struct elevator_queue *e = q->elevator;
if (!e)
return;
blk_mq_debugfs_unregister_sched_hctx(hctx);
if (e->type->ops.mq.exit_hctx && hctx->sched_data) {
e->type->ops.mq.exit_hctx(hctx, hctx_idx);
hctx->sched_data = NULL;
}
blk_mq_sched_free_tags(q->tag_set, hctx, hctx_idx);
}
int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
{
struct blk_mq_hw_ctx *hctx;
struct elevator_queue *eq;
unsigned int i;
int ret;
if (!e) {
q->elevator = NULL;
return 0;
}
/*
* Default to 256, since we don't split into sync/async like the
* old code did. Additionally, this is a per-hw queue depth.
*/
q->nr_requests = 2 * BLKDEV_MAX_RQ;
queue_for_each_hw_ctx(q, hctx, i) {
ret = blk_mq_sched_alloc_tags(q, hctx, i);
if (ret)
goto err;
}
ret = e->ops.mq.init_sched(q, e);
if (ret)
goto err;
blk_mq_debugfs_register_sched(q);
queue_for_each_hw_ctx(q, hctx, i) {
if (e->ops.mq.init_hctx) {
ret = e->ops.mq.init_hctx(hctx, i);
if (ret) {
eq = q->elevator;
blk_mq_exit_sched(q, eq);
kobject_put(&eq->kobj);
return ret;
}
}
blk_mq_debugfs_register_sched_hctx(q, hctx);
}
return 0;
err:
blk_mq_sched_tags_teardown(q);
q->elevator = NULL;
return ret;
}
void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e)
{
struct blk_mq_hw_ctx *hctx;
unsigned int i;
queue_for_each_hw_ctx(q, hctx, i) {
blk_mq_debugfs_unregister_sched_hctx(hctx);
if (e->type->ops.mq.exit_hctx && hctx->sched_data) {
e->type->ops.mq.exit_hctx(hctx, i);
hctx->sched_data = NULL;
}
}
blk_mq_debugfs_unregister_sched(q);
if (e->type->ops.mq.exit_sched)
e->type->ops.mq.exit_sched(e);
blk_mq_sched_tags_teardown(q);
q->elevator = NULL;
}
int blk_mq_sched_init(struct request_queue *q)
{
int ret;
mutex_lock(&q->sysfs_lock);
ret = elevator_init(q, NULL);
mutex_unlock(&q->sysfs_lock);
return ret;
}