RDMA/core: Fix kernel panic by always initializing qp->usecnt
authorBernd Schubert <bernd.schubert@itwm.fraunhofer.de>
Fri, 20 Jan 2012 18:43:54 +0000 (18:43 +0000)
committerRoland Dreier <roland@purestorage.com>
Fri, 27 Jan 2012 17:20:10 +0000 (09:20 -0800)
commite47e321a35c741ee41b67976f8c6a3a7a42bc5c0
treedecac0280849f82f81bcaaa9ba465bb512f5cd8b
parentdcd6c92267155e70a94b3927bce681ce74b80d1f
RDMA/core: Fix kernel panic by always initializing qp->usecnt

We have just been investigating kernel panics related to
cq->ibcq.event_handler() completion calls.  The problem is that
ib_destroy_qp() fails with -EBUSY.

Further investigation revealed qp->usecnt is not initialized.  This
counter was introduced in linux-3.2 by commit 0e0ec7e0638e
("RDMA/core: Export ib_open_qp() to share XRC TGT QPs") but it only
gets initialized for IB_QPT_XRC_TGT, but it is checked in
ib_destroy_qp() for any QP type.

Fix this by initializing qp->usecnt for every QP we create.

Signed-off-by: Bernd Schubert <bernd.schubert@itwm.fraunhofer.de>
Signed-off-by: Sven Breuner <sven.breuner@itwm.fraunhofer.de>

[ Initialize qp->usecnt in uverbs too.  - Sean ]

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>
drivers/infiniband/core/uverbs_cmd.c
drivers/infiniband/core/verbs.c