[PATCH 4/6] nvme: rdma: unquiesce queues until ctrl state is changed

[Ming Lei]

Unquiesce queues until controller state is changed successfully.

Prepare for moving start freeze into reconnect work for preventing new
request from entering queue in LIVE state.

Before removing namespace, unquiescing queues is done unconditionally, so
it is safe to do this way in case of state change failure.

Signed-off-by: Ming Lei <ming.lei at redhat.com>
---
 drivers/nvme/host/rdma.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c
index d433b2ec07a6..cdfa93d67b80 100644
--- a/drivers/nvme/host/rdma.c
+++ b/drivers/nvme/host/rdma.c
@@ -1117,9 +1117,7 @@ static void nvme_rdma_error_recovery_work(struct work_struct *work)
 	nvme_stop_keep_alive(&ctrl->ctrl);
 	flush_work(&ctrl->ctrl.async_event_work);
 	nvme_rdma_teardown_io_queues(ctrl, false);
-	nvme_unquiesce_io_queues(&ctrl->ctrl);
 	nvme_rdma_teardown_admin_queue(ctrl, false);
-	nvme_unquiesce_admin_queue(&ctrl->ctrl);
 	nvme_auth_stop(&ctrl->ctrl);
 
 	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_CONNECTING)) {
@@ -1129,6 +1127,8 @@ static void nvme_rdma_error_recovery_work(struct work_struct *work)
 		return;
 	}
 
+	nvme_unquiesce_io_queues(&ctrl->ctrl);
+	nvme_unquiesce_admin_queue(&ctrl->ctrl);
 	nvme_rdma_reconnect_or_remove(ctrl);
 }
 
-- 
2.40.1