From: James Smart <james.smart@broadcom.com>
Date: Fri, 23 Oct 2020 15:26:04 -0700
Subject: nvme-fc: track error_recovery while connecting
Patch-mainline: v5.10-rc2
Git-commit: caf1cbe3677576ef860f7bbba9074d0e6a0bcabb
References: bsc#1187076

Whenever there are errors during CONNECTING, the driver recovers by
aborting all outstanding ios and counts on the io completion to fail them
and thus the connection/association they are on.  However, the connection
failure depends on a failure state from the core routines.  Not all
commands that are issued by the core routine are guaranteed to cause a
failure of the core routine. They may be treated as a failure status and
the status is then ignored.

As such, whenever the transport enters error_recovery while CONNECTING,
it will set a new flag indicating an association failed. The
create_association routine which creates and initializes the controller,
will monitor the state of the flag as well as the core routine error
status and ensure the association fails if there was an error.

Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Daniel Wagner <dwagner@suse.de>
---
 drivers/nvme/host/fc.c |   14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)

--- a/drivers/nvme/host/fc.c
+++ b/drivers/nvme/host/fc.c
@@ -140,7 +140,8 @@ struct nvme_fc_rport {
 
 /* fc_ctrl flags values - specified as bit positions */
 #define ASSOC_ACTIVE		0
-#define FCCTRL_TERMIO		1
+#define ASSOC_FAILED		1
+#define FCCTRL_TERMIO		2
 
 struct nvme_fc_ctrl {
 	spinlock_t		lock;
@@ -2660,6 +2661,8 @@ nvme_fc_create_association(struct nvme_f
 	if (nvme_fc_ctlr_active_on_rport(ctrl))
 		return -ENOTUNIQ;
 
+	clear_bit(ASSOC_FAILED, &ctrl->flags);
+
 	/*
 	 * Create the admin queue
 	 */
@@ -2698,7 +2701,7 @@ nvme_fc_create_association(struct nvme_f
 		min_t(int, NVME_CAP_MQES(ctrl->ctrl.cap), ctrl->ctrl.sqsize);
 
 	ret = nvme_enable_ctrl(&ctrl->ctrl, ctrl->ctrl.cap);
-	if (ret)
+	if (ret || test_bit(ASSOC_FAILED, &ctrl->flags))
 		goto out_disconnect_admin_queue;
 
 	ctrl->ctrl.max_segments = ctrl->lport->ops->max_sgl_segments;
@@ -2708,7 +2711,7 @@ nvme_fc_create_association(struct nvme_f
 	blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
 
 	ret = nvme_init_identify(&ctrl->ctrl);
-	if (ret)
+	if (ret || test_bit(ASSOC_FAILED, &ctrl->flags))
 		goto out_disconnect_admin_queue;
 
 	/* sanity checks */
@@ -2752,9 +2755,9 @@ nvme_fc_create_association(struct nvme_f
 			ret = nvme_fc_create_io_queues(ctrl);
 		else
 			ret = nvme_fc_recreate_io_queues(ctrl);
-		if (ret)
-			goto out_term_aen_ops;
 	}
+	if (ret || test_bit(ASSOC_FAILED, &ctrl->flags))
+		goto out_term_aen_ops;
 
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
 
@@ -2961,6 +2964,7 @@ static void
 	 */
 	if (ctrl->ctrl.state == NVME_CTRL_CONNECTING) {
 		__nvme_fc_abort_outstanding_ios(ctrl, true);
+		set_bit(ASSOC_FAILED, &ctrl->flags);
 		return;
 	}