s390/qdio: (re-)initialize tiqdio list entries
authorJulian Wiedmann <jwi@linux.ibm.com>
Tue, 18 Jun 2019 09:25:59 +0000 (11:25 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sun, 21 Jul 2019 07:06:08 +0000 (09:06 +0200)
commit355dbc42bfb47a9033d740b69c1ea4dcfbe009f8
treeba57affd26bc07125106d1f404abe6c4f925dd41
parent6c016de667d828cf3d9b72be152c8118545bfd67
s390/qdio: (re-)initialize tiqdio list entries

commit e54e4785cb5cb4896cf4285964aeef2125612fb2 upstream.

When tiqdio_remove_input_queues() removes a queue from the tiq_list as
part of qdio_shutdown(), it doesn't re-initialize the queue's list entry
and the prev/next pointers go stale.

If a subsequent qdio_establish() fails while sending the ESTABLISH cmd,
it calls qdio_shutdown() again in QDIO_IRQ_STATE_ERR state and
tiqdio_remove_input_queues() will attempt to remove the queue entry a
second time. This dereferences the stale pointers, and bad things ensue.
Fix this by re-initializing the list entry after removing it from the
list.

For good practice also initialize the list entry when the queue is first
allocated, and remove the quirky checks that papered over this omission.
Note that prior to
commit e521813468f7 ("s390/qdio: fix access to uninitialized qdio_q fields"),
these checks were bogus anyway.

setup_queues_misc() clears the whole queue struct, and thus needs to
re-init the prev/next pointers as well.

Fixes: 779e6e1c724d ("[S390] qdio: new qdio driver.")
Cc: <stable@vger.kernel.org>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/s390/cio/qdio_setup.c
drivers/s390/cio/qdio_thinint.c