loop: Fix deadlock when calling blkdev_reread_part()
authorJan Kara <jack@suse.cz>
Thu, 8 Nov 2018 13:01:14 +0000 (14:01 +0100)
committerJens Axboe <axboe@kernel.dk>
Thu, 8 Nov 2018 13:30:32 +0000 (06:30 -0700)
commit0da03cab87e6323ff2e05b14bc7d5c6fcc531efd
treedd5886fab804057fe9fe61a69db3176c91220a65
parent85b0a54a82e4fbceeb1aebb7cb6909edd1a24668
loop: Fix deadlock when calling blkdev_reread_part()

Calling blkdev_reread_part() under loop_ctl_mutex causes lockdep to
complain about circular lock dependency between bdev->bd_mutex and
lo->lo_ctl_mutex. The problem is that on loop device open or close
lo_open() and lo_release() get called with bdev->bd_mutex held and they
need to acquire loop_ctl_mutex. OTOH when loop_reread_partitions() is
called with loop_ctl_mutex held, it will call blkdev_reread_part() which
acquires bdev->bd_mutex. See syzbot report for details [1].

Move call to blkdev_reread_part() in __loop_clr_fd() from under
loop_ctl_mutex to finish fixing of the lockdep warning and the possible
deadlock.

[1] https://syzkaller.appspot.com/bug?id=bf154052f0eea4bc7712499e4569505907d1588

Reported-by: syzbot <syzbot+4684a000d5abdade83fac55b1e7d1f935ef1936e@syzkaller.appspotmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
drivers/block/loop.c