diff options
| author | Yu Kuai <yukuai3@huawei.com> | 2023-06-10 05:30:43 +0300 |
|---|---|---|
| committer | Greg Kroah-Hartman <gregkh@linuxfoundation.org> | 2023-07-19 17:20:55 +0300 |
| commit | 931bd6758bcc0f3d27d0492456798afae028e52f (patch) | |
| tree | e8404c4dcabce4ada87dfba4b6905b0cf2ba92db /block | |
| parent | c2a0eb3b2017bd1ab8bc68f9296619036a3d06df (diff) | |
| download | linux-931bd6758bcc0f3d27d0492456798afae028e52f.tar.xz | |
blk-mq: fix potential io hang by wrong 'wake_batch'
[ Upstream commit 4f1731df60f9033669f024d06ae26a6301260b55 ]
In __blk_mq_tag_busy/idle(), updating 'active_queues' and calculating
'wake_batch' is not atomic:
t1: t2:
_blk_mq_tag_busy blk_mq_tag_busy
inc active_queues
// assume 1->2
inc active_queues
// 2 -> 3
blk_mq_update_wake_batch
// calculate based on 3
blk_mq_update_wake_batch
/* calculate based on 2, while active_queues is actually 3. */
Fix this problem by protecting them wih 'tags->lock', this is not a hot
path, so performance should not be concerned. And now that all writers
are inside the lock, switch 'actives_queues' from atomic to unsigned
int.
Fixes: 180dccb0dba4 ("blk-mq: fix tag_get wait task can't be awakened")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20230610023043.2559121-1-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Diffstat (limited to 'block')
| -rw-r--r-- | block/blk-mq-debugfs.c | 2 | ||||
| -rw-r--r-- | block/blk-mq-tag.c | 15 | ||||
| -rw-r--r-- | block/blk-mq.h | 3 |
3 files changed, 12 insertions, 8 deletions
diff --git a/block/blk-mq-debugfs.c b/block/blk-mq-debugfs.c index bd942341b638..7675e663df36 100644 --- a/block/blk-mq-debugfs.c +++ b/block/blk-mq-debugfs.c @@ -427,7 +427,7 @@ static void blk_mq_debugfs_tags_show(struct seq_file *m, seq_printf(m, "nr_tags=%u\n", tags->nr_tags); seq_printf(m, "nr_reserved_tags=%u\n", tags->nr_reserved_tags); seq_printf(m, "active_queues=%d\n", - atomic_read(&tags->active_queues)); + READ_ONCE(tags->active_queues)); seq_puts(m, "\nbitmap_tags:\n"); sbitmap_queue_show(&tags->bitmap_tags, m); diff --git a/block/blk-mq-tag.c b/block/blk-mq-tag.c index a80d7c62bdfe..100889c276c3 100644 --- a/block/blk-mq-tag.c +++ b/block/blk-mq-tag.c @@ -40,6 +40,7 @@ static void blk_mq_update_wake_batch(struct blk_mq_tags *tags, void __blk_mq_tag_busy(struct blk_mq_hw_ctx *hctx) { unsigned int users; + struct blk_mq_tags *tags = hctx->tags; /* * calling test_bit() prior to test_and_set_bit() is intentional, @@ -57,9 +58,11 @@ void __blk_mq_tag_busy(struct blk_mq_hw_ctx *hctx) return; } - users = atomic_inc_return(&hctx->tags->active_queues); - - blk_mq_update_wake_batch(hctx->tags, users); + spin_lock_irq(&tags->lock); + users = tags->active_queues + 1; + WRITE_ONCE(tags->active_queues, users); + blk_mq_update_wake_batch(tags, users); + spin_unlock_irq(&tags->lock); } /* @@ -92,9 +95,11 @@ void __blk_mq_tag_idle(struct blk_mq_hw_ctx *hctx) return; } - users = atomic_dec_return(&tags->active_queues); - + spin_lock_irq(&tags->lock); + users = tags->active_queues - 1; + WRITE_ONCE(tags->active_queues, users); blk_mq_update_wake_batch(tags, users); + spin_unlock_irq(&tags->lock); blk_mq_tag_wakeup_all(tags, false); } diff --git a/block/blk-mq.h b/block/blk-mq.h index 0b2870839cdd..c6eca452ea2a 100644 --- a/block/blk-mq.h +++ b/block/blk-mq.h @@ -362,8 +362,7 @@ static inline bool hctx_may_queue(struct blk_mq_hw_ctx *hctx, return true; } - users = atomic_read(&hctx->tags->active_queues); - + users = READ_ONCE(hctx->tags->active_queues); if (!users) return true; |