nvme I/O errors after 6.11 upgrade

[dopeytree]

Noticed this new log message after upgrade to 6.11.

 

I run 2x nvme drives in a pool (btrfs raid1 mirrored)

 

Sep 25 19:35:57 Moulin-Rouge kernel: block nvme0n1: No UUID available providing old NGUID
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 
Sep 25 19:35:59 Moulin-Rouge kernel: nvme0: Admin Cmd(0x7f), I/O Error (sct 0x0 / sc 0x1) DNR 

 

moulin-rouge-diagnostics-20220926-0253.zip

[JorgeB]

Quote

It's harmless. The driver is querying to see if the controller supports a particular mode of Identification. The NVMe spec doesn't provide a great way for a driver to know ahead of time if a controller supports an optional identification or not so the driver just has to try it to find out. There is a TP in the workgroup that may improve that situation, but that may take some time to publish and for devices to implement it.

 

https://bugzilla.kernel.org/show_bug.cgi?id=215763