blk-mq: fix race between updating nr_hw_queues and switching io sched
authorMing Lei <ming.lei@redhat.com>
Sat, 6 Jan 2018 08:27:40 +0000 (16:27 +0800)
committerJens Axboe <axboe@kernel.dk>
Sat, 6 Jan 2018 16:25:36 +0000 (09:25 -0700)
commitfb350e0ad99359768e1e80b4784692031ec340e4
tree25d87e34b68ada63eb442c6269b031f037011520
parent7d4901a90d02500c8011472a060f9b2e60e6e605
blk-mq: fix race between updating nr_hw_queues and switching io sched

In both elevator_switch_mq() and blk_mq_update_nr_hw_queues(), sched tags
can be allocated, and q->nr_hw_queue is used, and race is inevitable, for
example: blk_mq_init_sched() may trigger use-after-free on hctx, which is
freed in blk_mq_realloc_hw_ctxs() when nr_hw_queues is decreased.

This patch fixes the race be holding q->sysfs_lock.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reported-by: Yi Zhang <yi.zhang@redhat.com>
Tested-by: Yi Zhang <yi.zhang@redhat.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
block/blk-mq.c