nvme nvme0: controller is down

saddamus · 02-01-2024, 03:13 AM

I have bought a new laptop - Chuwi freebook 2023. It' intel n100 based machine.
It's shipped with Windows and they declare Windows only support, but it's not an option for me to work on Windows. The only, but serious problem i have with this machine is that it's freezing at random times - it can work fine for an hour or two, then freezes and needs to be force rebooted. As disk becomes read only, i had to enable remote logging to find the reason, which is:

Quote:

Feb 1 09:33:04 fedora kernel 04 kern warning kernel: nvme nvme0: controller is down; will reset: CSTS=0xffffffff, PCI_STATUS=0x10
Feb 1 09:33:04 fedora kernel 04 kern warning kernel: nvme nvme0: Does your device have a faulty power saving mode enabled?
Feb 1 09:33:04 fedora kernel 04 kern warning kernel: nvme nvme0: Try "nvme_core.default_ps_max_latency_us=0 pcie_aspm=off" and report a bug
Feb 1 09:33:04 fedora kernel 06 kern info kernel: nvme 0000:01:00.0: enabling device (0000 -> 0002)
Feb 1 09:33:04 fedora kernel 06 kern info kernel: nvme nvme0: 4/0/0 default/read/poll queues
Feb 1 09:33:04 fedora kernel 03 kern err kernel: nvme0n1: I/O Cmd(0x1) @ LBA 262687384, 8 blocks, I/O Error (sct 0x0 / sc 0x6)
Feb 1 09:33:04 fedora kernel 03 kern err kernel: I/O error, dev nvme0n1, sector 262687384 op 0x1

WRITE) flags 0x103000 phys_seg 1 prio class 2
Feb 1 09:33:04 fedora kernel 03 kern err kernel: Buffer I/O error on dev nvme0n1p5, logical block 9299, lost async page write
Feb 1 09:33:04 fedora kernel 03 kern err kernel: nvme0n1: I/O Cmd(0x1) @ LBA 677849992, 8 blocks, I/O Error (sct 0x0 / sc 0x6)
Feb 1 09:33:04 fedora kernel 03 kern err kernel: I/O error, dev nvme0n1, sector 677849992 op 0x1

WRITE) flags 0x103000 phys_seg 1 prio class 2

There is no relation whether disk is under load or idle - i can run heavy red-write operation without issue. Smartctl doesnt inform about any problem. What can i do to get closer the reason ?
This didnt help:

Quote:

root@adam-lap:/home/adam# cat /sys/module/nvme_core/parameters/default_ps_max_latency_us
0

business_kid · 02-01-2024, 03:34 AM

Hello, saddamus & welcome to LQ.

The power saving could be a thing. There's a 'powersave' governor in the kernel, acpi can be set up to dictate how your precious battery power gets distributed.

One faultfinding trick might be:

Code:

dmesg |tail -f >>some_file

Then when the crash comes, you'll hopefully have something to go on. I'd also look at getting a standard kernel from the Linus source, as Fedora patch theirs.

Arnulf · 02-01-2024, 12:47 PM

The "Chuwi freebook 2023" is quite new. Use a recent kernel, 6.6.15 LTS at minimum or current stable 6.7.3.