summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorYu Kuai <yukuai3@huawei.com>2023-06-10 10:30:43 +0800
committerJens Axboe <axboe@kernel.dk>2023-06-12 09:55:53 -0600
commit4f1731df60f9033669f024d06ae26a6301260b55 (patch)
tree8b3d6dbff9e497e9b2ecf9603284499be48a2bde
parent0733ad8002916b9dbbbcfe6e92ad44d2657de1c1 (diff)
downloadlinux-rpi-4f1731df60f9033669f024d06ae26a6301260b55.tar.gz
linux-rpi-4f1731df60f9033669f024d06ae26a6301260b55.tar.bz2
linux-rpi-4f1731df60f9033669f024d06ae26a6301260b55.zip
blk-mq: fix potential io hang by wrong 'wake_batch'
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>
-rw-r--r--block/blk-mq-debugfs.c2
-rw-r--r--block/blk-mq-tag.c15
-rw-r--r--block/blk-mq.h3
-rw-r--r--include/linux/blk-mq.h3
4 files changed, 13 insertions, 10 deletions
diff --git a/block/blk-mq-debugfs.c b/block/blk-mq-debugfs.c
index 68165a50951b..c3b5930106b2 100644
--- a/block/blk-mq-debugfs.c
+++ b/block/blk-mq-debugfs.c
@@ -401,7 +401,7 @@ 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 d6af9d431dc6..426197312069 100644
--- a/block/blk-mq-tag.c
+++ b/block/blk-mq-tag.c
@@ -38,6 +38,7 @@ 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;
if (blk_mq_is_shared_tags(hctx->flags)) {
struct request_queue *q = hctx->queue;
@@ -51,9 +52,11 @@ void __blk_mq_tag_busy(struct blk_mq_hw_ctx *hctx)
set_bit(BLK_MQ_S_TAG_ACTIVE, &hctx->state);
}
- 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);
}
/*
@@ -86,9 +89,11 @@ 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 8c642e9f32f1..1743857e0b01 100644
--- a/block/blk-mq.h
+++ b/block/blk-mq.h
@@ -412,8 +412,7 @@ 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;
diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h
index 59b52ec155b1..f401067ac03a 100644
--- a/include/linux/blk-mq.h
+++ b/include/linux/blk-mq.h
@@ -739,8 +739,7 @@ struct request *blk_mq_alloc_request_hctx(struct request_queue *q,
struct blk_mq_tags {
unsigned int nr_tags;
unsigned int nr_reserved_tags;
-
- atomic_t active_queues;
+ unsigned int active_queues;
struct sbitmap_queue bitmap_tags;
struct sbitmap_queue breserved_tags;