Skip to contentSkip to conversation

nvme_fc: add support for new ctrl_loss_tmo

PATCH New [email protected] 1 participant 3 messages Apr 25, 2017

Subsystems (2)
NVMe NVMe FC transport
lore
Share & tools
lore

[email protected]

author
#0 lore raw

cover nvme_fc: add support for new ctrl_loss_tmo

From: James Smart <[email protected]>

This patch adds support for the new ctrl_loss_tmo option.

It also addresses the MAX_RECONNECTS comment in Sagi's review.
http://lists.infradead.org/pipermail/linux-nvme/2017-April/009261.html

James Smart (2):
nvme_fc: get rid of local reconnect_delay
nvme_fc: Support ctrl_loss_tmo

 drivers/nvme/host/fc.c | 118 ++++++++++++++++++++-----------------------------
 1 file changed, 49 insertions(+), 69 deletions(-)
Signature · 1 line
2.9.3
author
#1 lore raw

PATCH 1/2 nvme_fc: get rid of local reconnect_delay

From: James Smart <[email protected]>

Remove the local copy of reconnect_delay.
Use the value in the controller options directly.

Signed-off-by James Smart <james.smart at broadcom.com>
---
 drivers/nvme/host/fc.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)
drivers/nvme/host/fc.c +4 -6
index e73862e..bab9959 100644
@@ -165,7 +165,6 @@ struct nvme_fc_ctrl {
 	struct work_struct	delete_work;
 	struct work_struct	reset_work;
 	struct delayed_work	connect_work;
-	int			reconnect_delay;
 	int			connect_attempts;
 
 	struct kref		ref;

@@ -2612,9 +2611,9 @@ nvme_fc_reset_ctrl_work(struct work_struct *work)
 
 		dev_warn(ctrl->ctrl.device,
 			"NVME-FC{%d}: Reconnect attempt in %d seconds.\n",
-			ctrl->cnum, ctrl->reconnect_delay);
+			ctrl->cnum, ctrl->ctrl.opts->reconnect_delay);
 		queue_delayed_work(nvme_fc_wq, &ctrl->connect_work,
-				ctrl->reconnect_delay * HZ);
+				ctrl->ctrl.opts->reconnect_delay * HZ);
 	} else
 		dev_info(ctrl->ctrl.device,
 			"NVME-FC{%d}: controller reset complete\n", ctrl->cnum);

@@ -2692,9 +2691,9 @@ nvme_fc_connect_ctrl_work(struct work_struct *work)
 
 		dev_warn(ctrl->ctrl.device,
 			"NVME-FC{%d}: Reconnect attempt in %d seconds.\n",
-			ctrl->cnum, ctrl->reconnect_delay);
+			ctrl->cnum, ctrl->ctrl.opts->reconnect_delay);
 		queue_delayed_work(nvme_fc_wq, &ctrl->connect_work,
-				ctrl->reconnect_delay * HZ);
+				ctrl->ctrl.opts->reconnect_delay * HZ);
 	} else
 		dev_info(ctrl->ctrl.device,
 			"NVME-FC{%d}: controller reconnect complete\n",

@@ -2746,7 +2745,6 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 	INIT_WORK(&ctrl->delete_work, nvme_fc_delete_ctrl_work);
 	INIT_WORK(&ctrl->reset_work, nvme_fc_reset_ctrl_work);
 	INIT_DELAYED_WORK(&ctrl->connect_work, nvme_fc_connect_ctrl_work);
-	ctrl->reconnect_delay = opts->reconnect_delay;
 	spin_lock_init(&ctrl->lock);
 
 	/* io queue count */
Signature · 1 line
2.9.3
author
#2 lore raw

PATCH 2/2 nvme_fc: Support ctrl_loss_tmo

In reply to [email protected]

From: James Smart <[email protected]>

Sync with Sagi's recent addition of ctrl_loss_tmo in the core fabrics layer.

Remove local connect limits and connect_attempts variable. Use fabrics new nr_connects variable and use of nvmf_should_reconnect() Refactor duplicate reconnect failure code.

Addresses review comment by Sagi on controller reset support:
http://lists.infradead.org/pipermail/linux-nvme/2017-April/009261.html

Signed-off-by James Smart <james.smart at broadcom.com>
---
 drivers/nvme/host/fc.c | 116 +++++++++++++++++++++----------------------------
 1 file changed, 49 insertions(+), 67 deletions(-)
drivers/nvme/host/fc.c +49 -67
index bab9959..4d4b902 100644
@@ -45,8 +45,6 @@ enum nvme_fc_queue_flags {
 
 #define NVMEFC_QUEUE_DELAY	3		/* ms units */
 
-#define NVME_FC_MAX_CONNECT_ATTEMPTS	1
-
 struct nvme_fc_queue {
 	struct nvme_fc_ctrl	*ctrl;
 	struct device		*dev;

@@ -165,7 +163,6 @@ struct nvme_fc_ctrl {
 	struct work_struct	delete_work;
 	struct work_struct	reset_work;
 	struct delayed_work	connect_work;
-	int			connect_attempts;
 
 	struct kref		ref;
 	u32			flags;

@@ -2302,7 +2299,7 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	int ret;
 	bool changed;
 
-	ctrl->connect_attempts++;
+	++ctrl->ctrl.opts->nr_reconnects;
 
 	/*
 	 * Create the admin queue

@@ -2399,7 +2396,7 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
 	WARN_ON_ONCE(!changed);
 
-	ctrl->connect_attempts = 0;
+	ctrl->ctrl.opts->nr_reconnects = 0;
 
 	kref_get(&ctrl->ctrl.kref);
 

@@ -2542,16 +2539,22 @@ nvme_fc_delete_ctrl_work(struct work_struct *work)
 	nvme_put_ctrl(&ctrl->ctrl);
 }
 
-static int
-__nvme_fc_del_ctrl(struct nvme_fc_ctrl *ctrl)
+static bool
+__nvme_fc_schedule_delete_work(struct nvme_fc_ctrl *ctrl)
 {
 	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_DELETING))
-		return -EBUSY;
+		return true;
 
 	if (!queue_work(nvme_fc_wq, &ctrl->delete_work))
-		return -EBUSY;
+		return true;
 
-	return 0;
+	return false;
+}
+
+static int
+__nvme_fc_del_ctrl(struct nvme_fc_ctrl *ctrl)
+{
+	return __nvme_fc_schedule_delete_work(ctrl) ? -EBUSY : 0;
 }
 
 /*

@@ -2577,6 +2580,35 @@ nvme_fc_del_nvme_ctrl(struct nvme_ctrl *nctrl)
 }
 
 static void
+nvme_fc_reconnect_or_delete(struct nvme_fc_ctrl *ctrl, int status)
+{
+	/* If we are resetting/deleting then do nothing */
+	if (ctrl->ctrl.state != NVME_CTRL_RECONNECTING) {
+		WARN_ON_ONCE(ctrl->ctrl.state == NVME_CTRL_NEW ||
+			ctrl->ctrl.state == NVME_CTRL_LIVE);
+		return;
+	}
+
+	dev_warn(ctrl->ctrl.device,
+		"NVME-FC{%d}: reset: Reconnect attempt failed (%d)\n",
+		ctrl->cnum, status);
+
+	if (nvmf_should_reconnect(&ctrl->ctrl)) {
+		dev_info(ctrl->ctrl.device,
+			"NVME-FC{%d}: Reconnect attempt in %d seconds.\n",
+			ctrl->cnum, ctrl->ctrl.opts->reconnect_delay);
+		queue_delayed_work(nvme_fc_wq, &ctrl->connect_work,
+				ctrl->ctrl.opts->reconnect_delay * HZ);
+	} else {
+		dev_info(ctrl->ctrl.device,
+				"NVME-FC{%d}: Max reconnect attempts (%d) "
+				"reached. Removing controller\n",
+				ctrl->cnum, ctrl->ctrl.opts->nr_reconnects);
+		WARN_ON(__nvme_fc_schedule_delete_work(ctrl));
+	}
+}
+
+static void
 nvme_fc_reset_ctrl_work(struct work_struct *work)
 {
 	struct nvme_fc_ctrl *ctrl =

@@ -2587,34 +2619,9 @@ nvme_fc_reset_ctrl_work(struct work_struct *work)
 	nvme_fc_delete_association(ctrl);
 
 	ret = nvme_fc_create_association(ctrl);
-	if (ret) {
-		dev_warn(ctrl->ctrl.device,
-			"NVME-FC{%d}: reset: Reconnect attempt failed (%d)\n",
-			ctrl->cnum, ret);
-		if (ctrl->connect_attempts >= NVME_FC_MAX_CONNECT_ATTEMPTS) {
-			dev_warn(ctrl->ctrl.device,
-				"NVME-FC{%d}: Max reconnect attempts (%d) "
-				"reached. Removing controller\n",
-				ctrl->cnum, ctrl->connect_attempts);
-
-			if (!nvme_change_ctrl_state(&ctrl->ctrl,
-				NVME_CTRL_DELETING)) {
-				dev_err(ctrl->ctrl.device,
-					"NVME-FC{%d}: failed to change state "
-					"to DELETING\n", ctrl->cnum);
-				return;
-			}
-
-			WARN_ON(!queue_work(nvme_fc_wq, &ctrl->delete_work));
-			return;
-		}
-
-		dev_warn(ctrl->ctrl.device,
-			"NVME-FC{%d}: Reconnect attempt in %d seconds.\n",
-			ctrl->cnum, ctrl->ctrl.opts->reconnect_delay);
-		queue_delayed_work(nvme_fc_wq, &ctrl->connect_work,
-				ctrl->ctrl.opts->reconnect_delay * HZ);
-	} else
+	if (ret)
+		nvme_fc_reconnect_or_delete(ctrl, ret);
+	else
 		dev_info(ctrl->ctrl.device,
 			"NVME-FC{%d}: controller reset complete\n", ctrl->cnum);
 }

@@ -2667,34 +2674,9 @@ nvme_fc_connect_ctrl_work(struct work_struct *work)
 				struct nvme_fc_ctrl, connect_work);
 
 	ret = nvme_fc_create_association(ctrl);
-	if (ret) {
-		dev_warn(ctrl->ctrl.device,
-			"NVME-FC{%d}: Reconnect attempt failed (%d)\n",
-			ctrl->cnum, ret);
-		if (ctrl->connect_attempts >= NVME_FC_MAX_CONNECT_ATTEMPTS) {
-			dev_warn(ctrl->ctrl.device,
-				"NVME-FC{%d}: Max reconnect attempts (%d) "
-				"reached. Removing controller\n",
-				ctrl->cnum, ctrl->connect_attempts);
-
-			if (!nvme_change_ctrl_state(&ctrl->ctrl,
-				NVME_CTRL_DELETING)) {
-				dev_err(ctrl->ctrl.device,
-					"NVME-FC{%d}: failed to change state "
-					"to DELETING\n", ctrl->cnum);
-				return;
-			}
-
-			WARN_ON(!queue_work(nvme_fc_wq, &ctrl->delete_work));
-			return;
-		}
-
-		dev_warn(ctrl->ctrl.device,
-			"NVME-FC{%d}: Reconnect attempt in %d seconds.\n",
-			ctrl->cnum, ctrl->ctrl.opts->reconnect_delay);
-		queue_delayed_work(nvme_fc_wq, &ctrl->connect_work,
-				ctrl->ctrl.opts->reconnect_delay * HZ);
-	} else
+	if (ret)
+		nvme_fc_reconnect_or_delete(ctrl, ret);
+	else
 		dev_info(ctrl->ctrl.device,
 			"NVME-FC{%d}: controller reconnect complete\n",
 			ctrl->cnum);

@@ -2959,7 +2941,7 @@ nvme_fc_create_ctrl(struct device *dev, struct nvmf_ctrl_options *opts)
 static struct nvmf_transport_ops nvme_fc_transport = {
 	.name		= "fc",
 	.required_opts	= NVMF_OPT_TRADDR | NVMF_OPT_HOST_TRADDR,
-	.allowed_opts	= NVMF_OPT_RECONNECT_DELAY,
+	.allowed_opts	= NVMF_OPT_RECONNECT_DELAY | NVMF_OPT_CTRL_LOSS_TMO,
 	.create_ctrl	= nvme_fc_create_ctrl,
 };
Signature · 1 line
2.9.3