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

[ Upstream commit ff9fc7e ]

While nvme_tcp_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.

Tested-by: Chris Leech <cleech@redhat.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drivers/nvme/host/tcp.c

@@ -2105,6 +2105,7 @@ static void nvme_tcp_error_recovery_work(struct work_struct *work)
 	struct nvme_ctrl *ctrl = &tcp_ctrl->ctrl;
 
 	nvme_stop_keep_alive(ctrl);
+	flush_work(&ctrl->async_event_work);
 	nvme_tcp_teardown_io_queues(ctrl, false);
 	/* unquiesce to fail fast pending requests */
 	nvme_start_queues(ctrl);