Skip to contentSkip to conversation

nvme: add bug report info for global duplicate id

PATCH Applied Keith Busch 3 participants 5 messages Jun 7, 2022 – Jun 13, 2022 NVMe

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 nvme: add bug report info for global duplicate id

From: Keith Busch <[email protected]>

The recent global id check is finding poorly implemented devices in the wild. Include relavant device information in the output to help quicken an appropriate quirk patch.

---
v1->v2:

Allow transport drivers to define their own device info op. For PCI, this will let us print the device id in addition to the vendor.

 drivers/nvme/host/core.c |  1 +
 drivers/nvme/host/nvme.h | 28 ++++++++++++++++++++++++++++
 drivers/nvme/host/pci.c  | 16 ++++++++++++++++
 3 files changed, 45 insertions(+)
drivers/nvme/host/core.c +1 -0
index 24165daee3c8..5c9f3648c7d6 100644
@@ -3863,6 +3863,7 @@ static int nvme_init_ns_head(struct nvme_ns *ns, unsigned nsid,
 	if (ret) {
 		dev_err(ctrl->device,
 			"globally duplicate IDs for nsid %d\n", nsid);
+		nvme_print_device_info(ctrl);
 		return ret;
 	}
 
drivers/nvme/host/nvme.h +28 -0
index 9b72b6ecf33c..0da94b233fed 100644
@@ -503,6 +503,7 @@ struct nvme_ctrl_ops {
 	void (*submit_async_event)(struct nvme_ctrl *ctrl);
 	void (*delete_ctrl)(struct nvme_ctrl *ctrl);
 	int (*get_address)(struct nvme_ctrl *ctrl, char *buf, int size);
+	void (*print_device_info)(struct nvme_ctrl *ctrl);
 };
 
 /*
@@ -548,6 +549,33 @@ static inline struct request *nvme_cid_to_rq(struct blk_mq_tags *tags,
 	return blk_mq_tag_to_rq(tags, nvme_tag_from_cid(command_id));
 }
 
+/*
+ * Return the length of the string without the space padding
+ */
+static inline int nvme_strlen(char *s, int len)
+{
+	while (s[len - 1] == ' ')
+		len--;
+	return len;
+}
+
+static inline void nvme_print_device_info(struct nvme_ctrl *ctrl)
+{
+	struct nvme_subsystem *subsys = ctrl->subsys;
+
+	if (ctrl->ops->print_device_info) {
+		ctrl->ops->print_device_info(ctrl);
+		return;
+	}
+
+	dev_err(ctrl->device,
+		"VID:%04x model:%.*s firmware:%.*s\n", subsys->vendor_id,
+		nvme_strlen(subsys->model, sizeof(subsys->model)),
+		subsys->model, nvme_strlen(subsys->firmware_rev,
+					   sizeof(subsys->firmware_rev)),
+		subsys->firmware_rev);
+}
+
 #ifdef CONFIG_FAULT_INJECTION_DEBUG_FS
 void nvme_fault_inject_init(struct nvme_fault_inject *fault_inj,
 			    const char *dev_name);
drivers/nvme/host/pci.c +16 -0
index 9d963f6cdbae..be053d943731 100644
@@ -2984,6 +2984,21 @@ static int nvme_pci_get_address(struct nvme_ctrl *ctrl, char *buf, int size)
 	return snprintf(buf, size, "%s\n", dev_name(&pdev->dev));
 }
 
+
+static void nvme_pci_print_device_info(struct nvme_ctrl *ctrl)
+{
+	struct pci_dev *pdev = to_pci_dev(to_nvme_dev(ctrl)->dev);
+	struct nvme_subsystem *subsys = ctrl->subsys;
+
+	dev_err(ctrl->device,
+		"VID:DID %04x:%04x model:%.*s firmware:%.*s\n",
+		pdev->vendor, pdev->device,
+		nvme_strlen(subsys->model, sizeof(subsys->model)),
+		subsys->model, nvme_strlen(subsys->firmware_rev,
+					   sizeof(subsys->firmware_rev)),
+		subsys->firmware_rev);
+}
+
 static const struct nvme_ctrl_ops nvme_pci_ctrl_ops = {
 	.name			= "pcie",
 	.module			= THIS_MODULE,
@@ -2995,6 +3010,7 @@ static const struct nvme_ctrl_ops nvme_pci_ctrl_ops = {
 	.free_ctrl		= nvme_pci_free_ctrl,
 	.submit_async_event	= nvme_pci_submit_async_event,
 	.get_address		= nvme_pci_get_address,
+	.print_device_info	= nvme_pci_print_device_info,
 };
 
 static int nvme_dev_map(struct nvme_dev *dev)
Signature · 1 line
2.30.2
maintainer
#1 lore raw
Quoted 7 lines · Keith Busch

From: Keith Busch <[email protected]>

The recent global id check is finding poorly implemented devices in the wild. Include relavant device information in the output to help quicken an appropriate quirk patch.

This problem seems more widespread than expected, so I think it would help to apply this patch now.

And since this appears to be a gap in compliance testing, I feel I should report this. I think UNH is still handling compliance certification.

#2 lore raw
Quoted 2 lines · Keith Busch

This problem seems more widespread than expected, so I think it would help to apply this patch now.

Yes, that's the plan. But as Jens was out for last week I didn't bother with a pull request.

Quoted 2 lines

And since this appears to be a gap in compliance testing, I feel I should report this. I think UNH is still handling compliance certification.

UNH provides the test suite, but the specs are handled in the nvme interoperability workgroup. I already asked them to test this, but nothing happened, but I think you asking again is definitively a good idea.

maintainer
#3 lore raw
Quoted 9 lines · Christoph Hellwig
Quoted 2 lines · Keith Busch

And since this appears to be a gap in compliance testing, I feel I should report this. I think UNH is still handling compliance certification.

UNH provides the test suite, but the specs are handled in the nvme interoperability workgroup. I already asked them to test this, but nothing happened, but I think you asking again is definitively a good idea.

Message sent, and I cc'ed the email you use for nvme workgroup discussion.