@@ -127,6 +127,13 @@ unsigned long blk_mq_get_tags(struct blk_mq_alloc_data *data, int nr_tags,
return ret;
}
+static inline bool preempt_tag(struct blk_mq_alloc_data *data,
+ struct sbitmap_queue *bt)
+{
+ return data->preempt ||
+ atomic_read(&bt->ws_active) <= SBQ_WAIT_QUEUES;
+}
+
unsigned int blk_mq_get_tag(struct blk_mq_alloc_data *data)
{
struct blk_mq_tags *tags = blk_mq_tags_from_data(data);
@@ -148,12 +155,14 @@ unsigned int blk_mq_get_tag(struct blk_mq_alloc_data *data)
tag_offset = tags->nr_reserved_tags;
}
- tag = __blk_mq_get_tag(data, bt);
- if (tag != BLK_MQ_NO_TAG)
- goto found_tag;
+ if (data->flags & BLK_MQ_REQ_NOWAIT || preempt_tag(data, bt)) {
+ tag = __blk_mq_get_tag(data, bt);
+ if (tag != BLK_MQ_NO_TAG)
+ goto found_tag;
- if (data->flags & BLK_MQ_REQ_NOWAIT)
- return BLK_MQ_NO_TAG;
+ if (data->flags & BLK_MQ_REQ_NOWAIT)
+ return BLK_MQ_NO_TAG;
+ }
do {
struct sbitmap_queue *bt_prev;
@@ -169,21 +178,26 @@ unsigned int blk_mq_get_tag(struct blk_mq_alloc_data *data)
* Retry tag allocation after running the hardware queue,
* as running the queue may also have found completions.
*/
- tag = __blk_mq_get_tag(data, bt);
- if (tag != BLK_MQ_NO_TAG)
- break;
+ if (preempt_tag(data, bt)) {
+ tag = __blk_mq_get_tag(data, bt);
+ if (tag != BLK_MQ_NO_TAG)
+ break;
+ }
if (!ws)
ws = bt_wait_ptr(bt, data->hctx);
sbitmap_prepare_to_wait(bt, ws, &wait, TASK_UNINTERRUPTIBLE);
- tag = __blk_mq_get_tag(data, bt);
- if (tag != BLK_MQ_NO_TAG)
- break;
+ if (preempt_tag(data, bt)) {
+ tag = __blk_mq_get_tag(data, bt);
+ if (tag != BLK_MQ_NO_TAG)
+ break;
+ }
bt_prev = bt;
io_schedule();
+ data->preempt = true;
sbitmap_finish_wait(bt, ws, &wait);
data->ctx = blk_mq_get_ctx(data->q);
@@ -1446,6 +1446,34 @@ static bool blk_mq_check_expired(struct request *rq, void *priv, bool reserved)
return true;
}
+static void blk_mq_check_tag_waiters(struct blk_mq_hw_ctx *hctx)
+{
+ bool warn = false;
+ struct blk_mq_tags *tags = hctx->tags;
+
+again:
+ if (atomic_read(&tags->bitmap_tags.ws_active)) {
+ warn = true;
+ sbitmap_queue_wake_all(&tags->bitmap_tags);
+ }
+
+ if (atomic_read(&tags->breserved_tags.ws_active)) {
+ warn = true;
+ sbitmap_queue_wake_all(&tags->breserved_tags);
+ }
+
+ if (hctx->sched_tags && tags != hctx->sched_tags) {
+ tags = hctx->sched_tags;
+ goto again;
+ }
+
+ /*
+ * This is problematic because someone is still waiting for tag while
+ * no tag is used.
+ */
+ WARN_ON_ONCE(warn);
+}
+
static void blk_mq_timeout_work(struct work_struct *work)
{
struct request_queue *q =
@@ -1482,6 +1510,7 @@ static void blk_mq_timeout_work(struct work_struct *work)
* each hctx as idle.
*/
queue_for_each_hw_ctx(q, hctx, i) {
+ blk_mq_check_tag_waiters(hctx);
/* the hctx may be unmapped, so check it here */
if (blk_mq_hw_queue_mapped(hctx))
blk_mq_tag_idle(hctx);
@@ -156,6 +156,8 @@ struct blk_mq_alloc_data {
/* allocate multiple requests/tags in one go */
unsigned int nr_tags;
+ /* true if blk_mq_get_tag() will try to preempt tag */
+ bool preempt;
struct request **cached_rq;
/* input & output parameter */
Tag preemption is the default behaviour, specifically blk_mq_get_tag() will try to get tag unconditionally, which means a new io can preempt tag even if there are lots of ios that are waiting for tags. Such behaviour doesn't make sense when the disk is under heavy load, because it will intensify competition without improving performance, especially for huge io as split ios are unlikely to be issued continuously. The ideal way to disable tag preemption is to track how many tags are available, and wait directly in blk_mq_get_tag() if free tags are very little. However, this is out of reality because fast path is affected. As 'ws_active' is only updated in slow path, this patch disable tag preemption if 'ws_active' is greater than 8, which means there are many threads waiting for tags already. Once tag preemption is disabled, there is a situation that can cause performance degradation(or io hung in extreme scenarios): the waitqueue doesn't have 'wake_batch' threads, thus wake up on this waitqueue might cause the concurrency of ios to be decreased. The next patch will fix this problem. This patch also add a detection in blk_mq_timeout_work(), just in case io hung is triggered due to waiters can't awakened in some corner cases. Signed-off-by: Yu Kuai <yukuai3@huawei.com> --- block/blk-mq-tag.c | 36 +++++++++++++++++++++++++----------- block/blk-mq.c | 29 +++++++++++++++++++++++++++++ block/blk-mq.h | 2 ++ 3 files changed, 56 insertions(+), 11 deletions(-)