nvme-rdma: fix possible use-after-free in transport error_recovery work

[ Upstream commit b6bb172 ]

While nvme_rdma_submit_async_event_work is checking the ctrl and queue
state before preparing the AER command and scheduling io_work, in order
to fully prevent a race where this check is not reliable the error
recovery work must flush async_event_work before continuing to destroy
the admin queue after setting the ctrl state to RESETTING such that
there is no race .submit_async_event and the error recovery handler
itself changing the ctrl state.

Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drivers/nvme/host/rdma.c

@@ -1200,6 +1200,7 @@ static void nvme_rdma_error_recovery_work(struct work_struct *work)
 			struct nvme_rdma_ctrl, err_work);
 
 	nvme_stop_keep_alive(&ctrl->ctrl);
+	flush_work(&ctrl->ctrl.async_event_work);
 	nvme_rdma_teardown_io_queues(ctrl, false);
 	nvme_start_queues(&ctrl->ctrl);
 	nvme_rdma_teardown_admin_queue(ctrl, false);