Skip to contentSkip to conversation

nvme: update firmware version after commit

PATCH Superseded Daniel Wagner 4 participants 8 messages Oct 13, 2023 – Oct 30, 2023 NVMe

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 nvme: update firmware version after commit

The firmware version sysfs entry needs to be updated after a successfully firmware activation.

nvme-cli stopped issuing an Identify Controller command to list the current firmware information and relies on sysfs showing the current firmware version.

---

Only compile tested. So you might want to wait until I am back from my vacation.

 drivers/nvme/host/core.c | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)
drivers/nvme/host/core.c +16 -1
index 21783aa2ee8e..686478555585 100644
@@ -4067,14 +4067,29 @@ static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl)
 static void nvme_get_fw_slot_info(struct nvme_ctrl *ctrl)
 {
 	struct nvme_fw_slot_info_log *log;
+	u64 afi;
 
 	log = kmalloc(sizeof(*log), GFP_KERNEL);
 	if (!log)
 		return;
 
 	if (nvme_get_log(ctrl, NVME_NSID_ALL, NVME_LOG_FW_SLOT, 0, NVME_CSI_NVM,
-			log, sizeof(*log), 0))
+			 log, sizeof(*log), 0)) {
 		dev_warn(ctrl->device, "Get FW SLOT INFO log error\n");
+		goto out_free_log;
+	}
+
+	afi = le64_to_cpu(log->afi);
+	if (afi & 0x30) {
+		dev_info(ctrl->device,
+			 "Firmware is activated after next Controller Level Reset\n");
+		goto out_free_log;
+	}
+
+	memcpy(ctrl->subsys->firmware_rev, &log->frs[afi & 0x3],
+		sizeof(ctrl->subsys->firmware_rev));
+
+out_free_log:
 	kfree(log);
 }
 
Signature · 1 line
2.42.0
maintainer
#1 lore raw

drivers/nvme/host/core.c static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl) · line 4083

Quoted 9 lines · Daniel Wagner
drivers/nvme/host/core.c +6 -1
 	if (nvme_get_log(ctrl, NVME_NSID_ALL, NVME_LOG_FW_SLOT, 0, NVME_CSI_NVM,
-			log, sizeof(*log), 0))
+			 log, sizeof(*log), 0)) {
 		dev_warn(ctrl->device, "Get FW SLOT INFO log error\n");
+		goto out_free_log;
+	}
+
+	afi = le64_to_cpu(log->afi);
+	if (afi & 0x30) {

That should be 'afi & 0x70'.

drivers/nvme/host/core.c static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl) · line 4089

Quoted 6 lines
drivers/nvme/host/core.c +6 -0
+		dev_info(ctrl->device,
+			 "Firmware is activated after next Controller Level Reset\n");
+		goto out_free_log;
+	}
+
+	memcpy(ctrl->subsys->firmware_rev, &log->frs[afi & 0x3],

and 'afi & 0x7'.

#2 lore raw
Quoted 12 lines · Keith Busch

drivers/nvme/host/core.c static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl) · line 4083

Quoted 9 lines · Daniel Wagner
drivers/nvme/host/core.c +6 -1
 	if (nvme_get_log(ctrl, NVME_NSID_ALL, NVME_LOG_FW_SLOT, 0, NVME_CSI_NVM,
-			log, sizeof(*log), 0))
+			 log, sizeof(*log), 0)) {
 		dev_warn(ctrl->device, "Get FW SLOT INFO log error\n");
+		goto out_free_log;
+	}
+
+	afi = le64_to_cpu(log->afi);
+	if (afi & 0x30) {

That should be 'afi & 0x70'.

Personally, I like GENMASK().
In this specific case it would be GENMASK_ULL(6, 4);

I find that more readable than 0x70.
Although for some reason GENMASK()/GENMASK_ULL() does not seem to be used in drivers/nvme/

Kind regards,
Niklas

#3 lore raw
In reply to Keith Busch
Quoted 9 lines · Keith Busch

drivers/nvme/host/core.c static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl) · line 4082

Quoted 8 lines · Daniel Wagner
drivers/nvme/host/core.c +5 -1
 	if (nvme_get_log(ctrl, NVME_NSID_ALL, NVME_LOG_FW_SLOT, 0, NVME_CSI_NVM,
-			log, sizeof(*log), 0))
+			 log, sizeof(*log), 0)) {
 		dev_warn(ctrl->device, "Get FW SLOT INFO log error\n");
+		goto out_free_log;
+	}
+
+	afi = le64_to_cpu(log->afi);

BTW, this field is a single byte, so there should be no need to use *_to_cpu() on this. (Byte order is only important when the field is more than one byte.)

Kind regards,
Niklas

#4 lore raw
Quoted 14 lines · Niklas Cassel
Quoted 9 lines · Keith Busch

drivers/nvme/host/core.c static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl) · line 4082

Quoted 8 lines · Daniel Wagner
drivers/nvme/host/core.c +5 -1
 	if (nvme_get_log(ctrl, NVME_NSID_ALL, NVME_LOG_FW_SLOT, 0, NVME_CSI_NVM,
-			log, sizeof(*log), 0))
+			 log, sizeof(*log), 0)) {
 		dev_warn(ctrl->device, "Get FW SLOT INFO log error\n");
+		goto out_free_log;
+	}
+
+	afi = le64_to_cpu(log->afi);

BTW, this field is a single byte, so there should be no need to use *_to_cpu() on this. (Byte order is only important when the field is more than one byte.)

Indeed, I somehow thought afi is of type __le64.

#5 lore raw
In reply to Keith Busch
Quoted 12 lines · Keith Busch

drivers/nvme/host/core.c static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl) · line 4083

Quoted 9 lines · Daniel Wagner
drivers/nvme/host/core.c +6 -1
 	if (nvme_get_log(ctrl, NVME_NSID_ALL, NVME_LOG_FW_SLOT, 0, NVME_CSI_NVM,
-			log, sizeof(*log), 0))
+			 log, sizeof(*log), 0)) {
 		dev_warn(ctrl->device, "Get FW SLOT INFO log error\n");
+		goto out_free_log;
+	}
+
+	afi = le64_to_cpu(log->afi);
+	if (afi & 0x30) {

That should be 'afi & 0x70'.

I've update the patch accordingly and send Kenij and one of our customer to test it.

#6 lore raw
In reply to Daniel Wagner
Quoted 8 lines · Daniel Wagner

The firmware version sysfs entry needs to be updated after a successfully firmware activation.

nvme-cli stopped issuing an Identify Controller command to list the current firmware information and relies on sysfs showing the current firmware version.

why did nvme-cli stopped using id-ctrl ?

-ck

maintainer
#7 lore raw
Quoted 11 lines · Chaitanya Kulkarni
Quoted 8 lines · Daniel Wagner

The firmware version sysfs entry needs to be updated after a successfully firmware activation.

nvme-cli stopped issuing an Identify Controller command to list the current firmware information and relies on sysfs showing the current firmware version.

why did nvme-cli stopped using id-ctrl ?

We have exported attributes. We should be able to use them so that we're not interrupting the device to provide info that the driver already caches. The driver just needs to make sure the contents are reliable.