ceph: fix potential mdsc use-after-free crash
authorXiubo Li <xiubli@redhat.com>
Wed, 1 Jul 2020 05:52:48 +0000 (01:52 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 3 Sep 2020 09:26:47 +0000 (11:26 +0200)
commit01540d5e7c1bcd0f5384abe5776d1731c747cc77
tree75e54ef772a1dbda966b49556ddbc60e373fb90c
parent9da791b5410e77dba2bd32350346b6c9ec9bce9c
ceph: fix potential mdsc use-after-free crash

[ Upstream commit fa9967734227b44acb1b6918033f9122dc7825b9 ]

Make sure the delayed work stopped before releasing the resources.

cancel_delayed_work_sync() will only guarantee that the work finishes
executing if the work is already in the ->worklist.  That means after
the cancel_delayed_work_sync() returns, it will leave the work requeued
if it was rearmed at the end. That can lead to a use after free once the
work struct is freed.

Fix it by flushing the delayed work instead of trying to cancel it, and
ensure that the work doesn't rearm if the mdsc is stopping.

URL: https://tracker.ceph.com/issues/46293
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
fs/ceph/mds_client.c