Skip to contentSkip to conversation

nvme: improve logging

PATCH New Chaitanya Kulkarni 5 participants 12 messages Apr 29, 2019 – Apr 30, 2019 NVMe

lore
Share & tools
lore
#0 lore raw

cover nvme: improve logging

These code changes are originally posted by
Hannes Reinecke <hare at suse.com> :-
http://lists.infradead.org/pipermail/linux-nvme/2019-April/023336.html

I've added all the comments posted on this patch-series along with my signoff to the patches. Please keep my sign off if Hannes is okay with it.

Regards,
-Chaitanya

Changes from V1:
1. Move the NVMe ctrl state array of string to the header file whee all controller states are defined.
2. Remvoe extra state_name variable and use the nvme_get_ctrl_state_name() when reporting the ctrl state.
3. Add "nvme-mp" prefix to the debug message for multipating code. Chaitanya Kulkarni (3):
nvme: introduce nvme-ctrl state name string array
nvme: improve logging
nvme-multipath: improve logging

 drivers/nvme/host/core.c      | 41 ++++++++++++++++++++---------------
 drivers/nvme/host/multipath.c |  5 ++++-
 drivers/nvme/host/nvme.h      | 10 +++++++++
 3 files changed, 37 insertions(+), 19 deletions(-)
Signature · 1 line
2.17.0
#1 lore raw

PATCH v2 1/3 nvme: introduce nvme-ctrl state name string array

From: Hannes Reinecke <[email protected]>

This patch intoduces nvme-ctrl state name array which is used inthe the later patch to improve the logging of the ctrl state.

Signed-off-by Hannes Reinecke <hare at suse.com>
Signed-off-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
---
 drivers/nvme/host/nvme.h | 10 ++++++++++
 1 file changed, 10 insertions(+)
drivers/nvme/host/nvme.h +10 -0
index 527d64545023..01a36bbafed6 100644
@@ -146,6 +146,16 @@ enum nvme_ctrl_state {
 	NVME_CTRL_DEAD,
 };
 
+static const char *const nvme_ctrl_state_name[] = {
+	[NVME_CTRL_NEW]         = "new",
+	[NVME_CTRL_LIVE]        = "live",
+	[NVME_CTRL_ADMIN_ONLY]  = "only-admin",
+	[NVME_CTRL_RESETTING]   = "resetting",
+	[NVME_CTRL_CONNECTING]  = "connecting",
+	[NVME_CTRL_DELETING]    = "deleting",
+	[NVME_CTRL_DEAD]        = "dead",
+};
+
 struct nvme_ctrl {
 	bool comp_seen;
 	enum nvme_ctrl_state state;
Signature · 1 line
2.17.0
#2 lore raw

PATCH v2 1/3 nvme: introduce nvme-ctrl state name string array

Quoted 3 lines · Chaitanya Kulkarni

From: Hannes Reinecke <hare at suse.com>

This patch intoduces nvme-ctrl state name array which is used inthe the

s/intoduces /introduces/
s/inthe/in the/

Quoted 15 lines

later patch to improve the logging of the ctrl state.

Signed-off-by Hannes Reinecke <hare at suse.com>
Signed-off-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
---
  drivers/nvme/host/nvme.h | 10 ++++++++++
  1 file changed, 10 insertions(+)
drivers/nvme/host/nvme.h +0 -0
index 527d64545023..01a36bbafed6 100644
@@ -146,6 +146,16 @@ enum nvme_ctrl_state {
  	NVME_CTRL_DEAD,
  };

drivers/nvme/host/nvme.h enum nvme_ctrl_state { · line 157

Quoted 10 lines
drivers/nvme/host/nvme.h +9 -0
 
+static const char *const nvme_ctrl_state_name[] = {
+	[NVME_CTRL_NEW]         = "new",
+	[NVME_CTRL_LIVE]        = "live",
+	[NVME_CTRL_ADMIN_ONLY]  = "only-admin",
+	[NVME_CTRL_RESETTING]   = "resetting",
+	[NVME_CTRL_CONNECTING]  = "connecting",
+	[NVME_CTRL_DELETING]    = "deleting",
+	[NVME_CTRL_DEAD]        = "dead",
+};

I haven't been paying attention if this was bought up before but won't this create multiple copies of this data for every source file the includes the header file?

Quoted 5 lines
drivers/nvme/host/nvme.h +1 -0
+
  struct nvme_ctrl {
  	bool comp_seen;
  	enum nvme_ctrl_state state;
#3 lore raw

PATCH v2 1/3 nvme: introduce nvme-ctrl state name string array

Quoted 25 lines · Kenneth
Quoted 3 lines · Chaitanya Kulkarni

From: Hannes Reinecke <hare at suse.com>

This patch intoduces nvme-ctrl state name array which is used inthe the

s/intoduces /introduces/
s/inthe/in the/

Quoted 16 lines

later patch to improve the logging of the ctrl state.

Signed-off-by Hannes Reinecke <hare at suse.com>
Signed-off-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
---
   drivers/nvme/host/nvme.h | 10 ++++++++++
   1 file changed, 10 insertions(+)
drivers/nvme/host/nvme.h +0 -0
index 527d64545023..01a36bbafed6 100644
@@ -146,6 +146,16 @@ enum nvme_ctrl_state {
   	NVME_CTRL_DEAD,
   };
 
Quoted 13 lines

drivers/nvme/host/nvme.h enum nvme_ctrl_state { · line 157

Quoted 9 lines
drivers/nvme/host/nvme.h +9 -0
+static const char *const nvme_ctrl_state_name[] = {
+	[NVME_CTRL_NEW]         = "new",
+	[NVME_CTRL_LIVE]        = "live",
+	[NVME_CTRL_ADMIN_ONLY]  = "only-admin",
+	[NVME_CTRL_RESETTING]   = "resetting",
+	[NVME_CTRL_CONNECTING]  = "connecting",
+	[NVME_CTRL_DELETING]    = "deleting",
+	[NVME_CTRL_DEAD]        = "dead",
+};

I haven't been paying attention if this was bought up before but won't this create multiple copies of this data for every source file the includes the header file?

Yes, we did discuss this.

Quoted 12 lines

drivers/nvme/host/nvme.h enum nvme_ctrl_state { · line 161

Quoted 6 lines
drivers/nvme/host/nvme.h +1 -0
+
   struct nvme_ctrl {
   	bool comp_seen;
   	enum nvme_ctrl_state state;

_______________________________________________
Linux-nvme mailing list
Linux-nvme at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

#4 lore raw

PATCH v2 1/3 nvme: introduce nvme-ctrl state name string array

Quoted 41 lines · Chaitanya Kulkarni
Quoted 26 lines · Kenneth
Quoted 3 lines · Chaitanya Kulkarni

From: Hannes Reinecke <hare at suse.com>

This patch intoduces nvme-ctrl state name array which is used inthe the

s/intoduces /introduces/
s/inthe/in the/

Quoted 16 lines

later patch to improve the logging of the ctrl state.

Signed-off-by Hannes Reinecke <hare at suse.com>
Signed-off-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
---
   drivers/nvme/host/nvme.h | 10 ++++++++++
   1 file changed, 10 insertions(+)
drivers/nvme/host/nvme.h +0 -0
index 527d64545023..01a36bbafed6 100644
@@ -146,6 +146,16 @@ enum nvme_ctrl_state {
   	NVME_CTRL_DEAD,
   };
 
Quoted 13 lines

drivers/nvme/host/nvme.h enum nvme_ctrl_state { · line 157

Quoted 9 lines
drivers/nvme/host/nvme.h +9 -0
+static const char *const nvme_ctrl_state_name[] = {
+	[NVME_CTRL_NEW]         = "new",
+	[NVME_CTRL_LIVE]        = "live",
+	[NVME_CTRL_ADMIN_ONLY]  = "only-admin",
+	[NVME_CTRL_RESETTING]   = "resetting",
+	[NVME_CTRL_CONNECTING]  = "connecting",
+	[NVME_CTRL_DELETING]    = "deleting",
+	[NVME_CTRL_DEAD]        = "dead",
+};

I haven't been paying attention if this was bought up before but won't this create multiple copies of this data for every source file the includes the header file?

Quoted 1 line

Yes, we did discuss this.

Hi Chaitanya,

In an e-mail from Christoph I found the following recommendation with regard to the nvme_ctrl_state_name[] array: "Just keep it near the top of multipath.c." and you replied "Okay, that works too."

Did I perhaps miss something?

Thanks,

Bart.

maintainer
#5 lore raw

PATCH v2 1/3 nvme: introduce nvme-ctrl state name string array

Quoted 7 lines
Quoted 6 lines
Quoted 4 lines

I haven't been paying attention if this was bought up before but won't this create multiple copies of this data for every source file the includes the header file?

Yes, we did discuss this.

Quoted 7 lines

Hi Chaitanya,

In an e-mail from Christoph I found the following recommendation with regard to the nvme_ctrl_state_name[] array: "Just keep it near the top of multipath.c." and you replied "Okay, that works too."

Did I perhaps miss something?

That was my recollection too...

#6 lore raw

PATCH v2 1/3 nvme: introduce nvme-ctrl state name string array

Seems like I missed it. Let me send the V3 along with other comments from Sagi.

Quoted 6 lines · Linux-nvme on behalf of Sagi Grimberg
Quoted 5 lines
Quoted 3 lines

I haven't been paying attention if this was bought up before but won't this create multiple copies of this data for every source file the includes the header file?

Yes, we did discuss this.

Quoted 7 lines

Hi Chaitanya,

In an e-mail from Christoph I found the following recommendation with regard to the nvme_ctrl_state_name[] array: "Just keep it near the top of multipath.c." and you replied "Okay, that works too."

Did I perhaps miss something?

That was my recollection too...

_______________________________________________
Linux-nvme mailing list
Linux-nvme at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

#7 lore raw

PATCH v2 2/3 nvme: improve logging

In reply to Chaitanya Kulkarni

From: Hannes Reinecke <[email protected]>

Currently nvme is very reluctant if it comes to logging anything, _unless_ it's an ANA AEN. So this patch tries to remedy this by decreasing the priority for the ANA AEN logging message, and improve the logging when calling nvme_reset_ctrl().

Also, print ctrl state transition info.

Signed-off-by Hannes Reinecke <hare at suse.com>
Signed-off-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
---
 drivers/nvme/host/core.c | 41 ++++++++++++++++++++++------------------
 1 file changed, 23 insertions(+), 18 deletions(-)
drivers/nvme/host/core.c +23 -18
index 3dd043aa6d1f..c17e731f53e3 100644
@@ -94,6 +94,13 @@ static void nvme_put_subsystem(struct nvme_subsystem *subsys);
 static void nvme_remove_invalid_namespaces(struct nvme_ctrl *ctrl,
 					   unsigned nsid);
 
+static const char *nvme_get_ctrl_state_name(unsigned int state)
+{
+	if ((unsigned int) state < ARRAY_SIZE(nvme_ctrl_state_name))
+		return nvme_ctrl_state_name[state];
+	return "unknown";
+}
+
 static void nvme_set_queue_dying(struct nvme_ns *ns)
 {
 	/*
@@ -119,10 +126,16 @@ static void nvme_queue_scan(struct nvme_ctrl *ctrl)
 
 int nvme_reset_ctrl(struct nvme_ctrl *ctrl)
 {
-	if (!nvme_change_ctrl_state(ctrl, NVME_CTRL_RESETTING))
+	if (!nvme_change_ctrl_state(ctrl, NVME_CTRL_RESETTING)) {
+		dev_warn(ctrl->device, "cannot reset ctrl in state %s\n",
+				nvme_get_ctrl_state_name(ctrl->state));
 		return -EBUSY;
-	if (!queue_work(nvme_reset_wq, &ctrl->reset_work))
+	}
+	if (!queue_work(nvme_reset_wq, &ctrl->reset_work)) {
+		dev_dbg(ctrl->device, "ctrl reset already queued\n");
 		return -EBUSY;
+	}
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(nvme_reset_ctrl);
@@ -370,8 +383,13 @@ bool nvme_change_ctrl_state(struct nvme_ctrl *ctrl,
 		break;
 	}
 
-	if (changed)
+	if (changed) {
 		ctrl->state = new_state;
+		dev_info(ctrl->device, "state change from (%s -> %s)\n",
+				nvme_get_ctrl_state_name(new_state),
+					nvme_get_ctrl_state_name(ctrl->state));
+
+	}
 
 	spin_unlock_irqrestore(&ctrl->lock, flags);
 	if (changed && ctrl->state == NVME_CTRL_LIVE)
@@ -2960,21 +2978,8 @@ static ssize_t nvme_sysfs_show_state(struct device *dev,
 				     char *buf)
 {
 	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
-	static const char *const state_name[] = {
-		[NVME_CTRL_NEW]		= "new",
-		[NVME_CTRL_LIVE]	= "live",
-		[NVME_CTRL_ADMIN_ONLY]	= "only-admin",
-		[NVME_CTRL_RESETTING]	= "resetting",
-		[NVME_CTRL_CONNECTING]	= "connecting",
-		[NVME_CTRL_DELETING]	= "deleting",
-		[NVME_CTRL_DEAD]	= "dead",
-	};
-
-	if ((unsigned)ctrl->state < ARRAY_SIZE(state_name) &&
-	    state_name[ctrl->state])
-		return sprintf(buf, "%s\n", state_name[ctrl->state]);
-
-	return sprintf(buf, "unknown state\n");
+
+	return sprintf(buf, "%s\n", nvme_get_ctrl_state_name(ctrl->state));
 }
 
 static DEVICE_ATTR(state, S_IRUGO, nvme_sysfs_show_state, NULL);
Signature · 1 line
2.17.0
maintainer
#8 lore raw

PATCH v2 2/3 nvme: improve logging

drivers/nvme/host/core.c bool nvme_change_ctrl_state(struct nvme_ctrl *ctrl, · line 390

Quoted 7 lines
drivers/nvme/host/core.c +4 -1
-	if (changed)
+	if (changed) {
  		ctrl->state = new_state;
+		dev_info(ctrl->device, "state change from (%s -> %s)\n",
+				nvme_get_ctrl_state_name(new_state),
+					nvme_get_ctrl_state_name(ctrl->state));

dev_info is too chatty... dev_dbg please...

drivers/nvme/host/core.c static ssize_t nvme_sysfs_show_state(struct device *dev, · line 2971

Quoted 13 lines
drivers/nvme/host/core.c +0 -9
@@ -2960,21 +2978,8 @@ static ssize_t nvme_sysfs_show_state(struct device *dev,
  				     char *buf)
  {
  	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
-	static const char *const state_name[] = {
-		[NVME_CTRL_NEW]		= "new",
-		[NVME_CTRL_LIVE]	= "live",
-		[NVME_CTRL_ADMIN_ONLY]	= "only-admin",
-		[NVME_CTRL_RESETTING]	= "resetting",
-		[NVME_CTRL_CONNECTING]	= "connecting",
-		[NVME_CTRL_DELETING]	= "deleting",
-		[NVME_CTRL_DEAD]	= "dead",
-	};

Patch #1 should erase this table here.

Quoted 12 lines
drivers/nvme/host/core.c +2 -6
-
-	if ((unsigned)ctrl->state < ARRAY_SIZE(state_name) &&
-	    state_name[ctrl->state])
-		return sprintf(buf, "%s\n", state_name[ctrl->state]);
-
-	return sprintf(buf, "unknown state\n");
+
+	return sprintf(buf, "%s\n", nvme_get_ctrl_state_name(ctrl->state));
  }
 
  static DEVICE_ATTR(state, S_IRUGO, nvme_sysfs_show_state, NULL);
#9 lore raw

PATCH v2 3/3 nvme-multipath: improve logging

In reply to Chaitanya Kulkarni

From: Hannes Reinecke <[email protected]>

This patch improves the logging for nvme-multipath code.

Signed-off-by Hannes Reinecke <hare at suse.com>
Signed-off-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
---
 drivers/nvme/host/multipath.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/nvme/host/multipath.c +4 -1
index f0716f6ce41f..e4929b7b42f9 100644
@@ -81,6 +81,9 @@ void nvme_failover_req(struct request *req)
 		 * Reset the controller for any non-ANA error as we don't know
 		 * what caused the error.
 		 */
+		dev_info(ns->ctrl->device,
+				"nvme-mp: status 0x%04x, resetting controller\n",
+				status);
 		nvme_reset_ctrl(ns->ctrl);
 		break;
 	}
@@ -421,7 +424,7 @@ static int nvme_update_ana_state(struct nvme_ctrl *ctrl,
 	unsigned *nr_change_groups = data;
 	struct nvme_ns *ns;
 
-	dev_info(ctrl->device, "ANA group %d: %s.\n",
+	dev_dbg(ctrl->device, "ANA group %d: %s.\n",
 			le32_to_cpu(desc->grpid),
 			nvme_ana_state_names[desc->state]);
 
Signature · 1 line
2.17.0
maintainer
#10 lore raw

PATCH v2 3/3 nvme-multipath: improve logging

Looks fine,

Reviewed-by Sagi Grimberg <sagi at grimberg.me>
#11 lore raw

PATCH v2 3/3 nvme-multipath: improve logging

In reply to Chaitanya Kulkarni

drivers/nvme/host/multipath.c void nvme_failover_req(struct request *req) · line 86

Quoted 3 lines
drivers/nvme/host/multipath.c +3 -0
+		dev_info(ns->ctrl->device,
+				"nvme-mp: status 0x%04x, resetting controller\n",
+				status);

This message still looks a little gibberish to me.

Why not:

	"resetting path due to unhandled error (0x%04x")"

drivers/nvme/host/multipath.c static int nvme_update_ana_state(struct nvme_ctrl *ctrl, · line 429

Quoted 4 lines
drivers/nvme/host/multipath.c +1 -1
-	dev_info(ctrl->device, "ANA group %d: %s.\n",
+	dev_dbg(ctrl->device, "ANA group %d: %s.\n",
 			le32_to_cpu(desc->grpid),
 			nvme_ana_state_names[desc->state]);

I've applied this hunk to nvme-5.2, thanks.