bdi: Fix oops in wb_workfn()
authorJan Kara <jack@suse.cz>
Thu, 3 May 2018 16:26:26 +0000 (18:26 +0200)
committerJens Axboe <axboe@kernel.dk>
Thu, 3 May 2018 22:11:37 +0000 (16:11 -0600)
commitb8b784958eccbf8f51ebeee65282ca3fd59ea391
treee2173e7ba2206c19a06c0bc3fa59430b9b33f92f
parent8bfc3b4c6f9de815de4ab73784b9419348266a65
bdi: Fix oops in wb_workfn()

Syzbot has reported that it can hit a NULL pointer dereference in
wb_workfn() due to wb->bdi->dev being NULL. This indicates that
wb_workfn() was called for an already unregistered bdi which should not
happen as wb_shutdown() called from bdi_unregister() should make sure
all pending writeback works are completed before bdi is unregistered.
Except that wb_workfn() itself can requeue the work with:

mod_delayed_work(bdi_wq, &wb->dwork, 0);

and if this happens while wb_shutdown() is waiting in:

flush_delayed_work(&wb->dwork);

the dwork can get executed after wb_shutdown() has finished and
bdi_unregister() has cleared wb->bdi->dev.

Make wb_workfn() use wakeup_wb() for requeueing the work which takes all
the necessary precautions against racing with bdi unregistration.

CC: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
CC: Tejun Heo <tj@kernel.org>
Fixes: 839a8e8660b6777e7fe4e80af1a048aebe2b5977
Reported-by: syzbot <syzbot+9873874c735f2892e7e9@syzkaller.appspotmail.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
fs/fs-writeback.c