]> git.ipfire.org Git - thirdparty/kernel/linux.git/commit
blk-mq: pop cached request if it is usable
authorKeith Busch <kbusch@kernel.org>
Thu, 21 May 2026 19:02:53 +0000 (12:02 -0700)
committerJens Axboe <axboe@kernel.dk>
Thu, 21 May 2026 19:04:11 +0000 (13:04 -0600)
commitdc278e9bf2b9513a763353e6b9cc21e0f532954e
treee43d8e1f3ddbca290546c7759432cb9a6eee3468
parent94449463d2ab125067adccc2b0c2d7cd753edfb4
blk-mq: pop cached request if it is usable

When submitting a bio to blk-mq, if the task should sleep after peeking
a cached request, but before it pops it, the plug flushes and calls
blk_mq_free_plug_rqs, freeing the cached_rqs. This creates a
use-after-free bug. Fix this by popping the cached request before any
possible blocking calls if it is suitable for use.

Popping this request first holds a queue reference, so avoid any
serialization races with queue freezes and can safely proceed with
dispatching that request to the driver. This potentially increases a
timing window from when a driver wants to freeze its queue to when
requests stop being dispatched. That scenario is off the fast path
though, and drivers need to appropriately handle requests during a
freeze request anyway.

The downside is the popped element needs to be individually freed when
we performed a bio plug merge. The cached request would have had to be
freed later anyway, but this patch does it inline with building the plug
list instead of after flushing it.

Fixes: b0077e269f6c1 ("blk-mq: make sure active queue usage is held for bio_integrity_prep()")
Fixes: 7b4f36cd22a65 ("block: ensure we hold a queue reference when using queue limits")
Signed-off-by: Keith Busch <kbusch@kernel.org>
Link: https://patch.msgid.link/20260521190253.242065-1-kbusch@meta.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
block/blk-mq.c