ceph: do not access the kiocb after aio requests
authorXiubo Li <xiubli@redhat.com>
Mon, 6 Jul 2020 12:51:35 +0000 (08:51 -0400)
committerIlya Dryomov <idryomov@gmail.com>
Mon, 3 Aug 2020 09:05:25 +0000 (11:05 +0200)
commitd1d9655052606fd9078e896668ec90191372d513
treedcdb21734df1009fd848b91d7d3d54fbd57387b0
parent585d72f33e7083972030fac7792ea3050a4a8dff
ceph: do not access the kiocb after aio requests

In aio case, if the completion comes very fast just before the
ceph_read_iter() returns to fs/aio.c, the kiocb will be freed in
the completion callback, then if ceph_read_iter() access again
we will potentially hit the use-after-free bug.

[ jlayton: initialize direct_lock early, and use it everywhere ]

URL: https://tracker.ceph.com/issues/45649
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
fs/ceph/file.c