disk enters read-only state after resuming from suspend on Huawei RLEF-X laptop

FlyingPotato_131 · 03-30-2023, 04:47 AM

I have installed pop_os on a laptop I sometimes use, but I found many things break upon returning from sleep. I've done some digging and found this in the systemd journal while in the broken state

Code:

Mar 30 12:42:13 pop-os systemd-sleep[5012]: System returned from sleep state.
Mar 30 12:42:13 pop-os kernel: PM: suspend exit
Mar 30 12:42:13 pop-os kernel: EXT4-fs error (device nvme0n1p3): __ext4_find_entry:1663: inode #3690611: comm hdparm: reading directory lblock 0
Mar 30 12:42:13 pop-os kernel: Buffer I/O error on dev nvme0n1p3, logical block 0, lost sync page write
Mar 30 12:42:13 pop-os kernel: EXT4-fs (nvme0n1p3): I/O error while writing superblock
Mar 30 12:42:13 pop-os systemd-sleep[5123]: /usr/lib/systemd/system-sleep/hdparm: 5: /usr/lib/pm-utils/power.d/95hdparm-apm: Input/output error
Mar 30 12:42:13 pop-os [5120]: /usr/lib/systemd/system-sleep/pop-default-settings_bluetooth-suspend failed with exit status 126.
Mar 30 12:42:13 pop-os [5120]: /usr/lib/systemd/system-sleep/hdparm failed with exit status 126.
Mar 30 12:42:13 pop-os systemd[1]: systemd-suspend.service: Deactivated successfully.
Mar 30 12:42:13 pop-os systemd[1]: Finished System Suspend.
Mar 30 12:42:13 pop-os systemd[1]: Stopped target Sleep.
Mar 30 12:42:13 pop-os systemd[1]: Reached target Suspend.
Mar 30 12:42:13 pop-os systemd-logind[763]: Operation 'sleep' finished.
Mar 30 12:42:13 pop-os systemd[1]: Stopped target Suspend.
Mar 30 12:42:13 pop-os ModemManager[808]: <info>  [sleep-monitor-systemd] system is resuming

for some reason the disk seems to be in a readonly state after waking up. I have tested it on a liveusb endeavouros and have the same problem. What may be causing this issue?

here's the output of inxi

Code:

CPU: 12-core (4-mt/8-st) 12th Gen Intel Core i5-12500H (-MST AMCP-)
speed/min/max: 3043/400/4500:3300 MHz Kernel: 6.2.0-76060200-generic x86_64
Up: 5m Mem: 3194.8/15721.9 MiB (20.3%) Storage: 476.94 GiB (14.1% used)

smallpond · 03-31-2023, 03:48 PM

What is the output of:

Code:

sudo lspci -vv -s $(lspci |awk '/NVMe/ {print $1}')

uteck · 03-31-2023, 04:04 PM

I had a similar issue on my Chromebook running Linux from SD card. It was sporadic when resuming from sleep, but from what I could determine it was a power issue with the SD card not getting powered up fast enough as the system resumed from memory.
Some links for reading:
https://www.kernel.org/doc/html/v5.4...m/devices.html
https://01.org/blogs/2018/improving-...me-performance
https://www.linuxtechmore.com/2022/0...-on-linux.html

The last link has a possible fix, others are just informative.

FlyingPotato_131 · 04-01-2023, 11:12 AM

Quote:

Originally Posted by smallpond

What is the output of:

Code:

sudo lspci -vv -s $(lspci |awk '/NVMe/ {print $1}')

for some reason $(lspci |awk '/NVMe/ {print $1}') returns nothing (i.e. the output is empty)

computersavvy · 04-01-2023, 02:03 PM

Why not try simply

Code:

 lspci | grep -i nvme

and see what it returns.

Once you know the actual name then put it into the previous command as the system sees it.

To test the output of

Code:

lspci |awk '/NVMe/ {print $1}'

you would need to use only that or

Code:

echo $(lspci |awk '/NVMe/ {print $1}')

. That part only gives the device to be selected from the output of "lspci -vv -s"

FlyingPotato_131 · 04-02-2023, 03:16 PM

Quote:

Originally Posted by computersavvy

Why not try simply

Code:

 lspci | grep -i nvme

and see what it returns.

Once you know the actual name then put it into the previous command as the system sees it.

To test the output of

Code:

lspci |awk '/NVMe/ {print $1}'

you would need to use only that or

Code:

echo $(lspci |awk '/NVMe/ {print $1}')

. That part only gives the device to be selected from the output of "lspci -vv -s"

as I expected, lspci | grep -i nvme shows nothing, i.e. the drive doesn't appear as a pci device (it does appear as a block device though)

FlyingPotato_131 · 04-06-2023, 07:31 AM

Quote:

Originally Posted by uteck

I had a similar issue on my Chromebook running Linux from SD card. It was sporadic when resuming from sleep, but from what I could determine it was a power issue with the SD card not getting powered up fast enough as the system resumed from memory.
Some links for reading:
https://www.kernel.org/doc/html/v5.4...m/devices.html
https://01.org/blogs/2018/improving-...me-performance
https://www.linuxtechmore.com/2022/0...-on-linux.html

The last link has a possible fix, others are just informative.

I have tried setting iommu=soft but it has not worked, in fact the system stopped loading at all

toliak · 07-12-2023, 05:53 PM

Hello.

I have ran into the same problem on RLEF-X laptop.

Briefly:
The kernel puts the disk into D3Cold state when it goes to suspend.
On wake up, kernel returns back the disk from the D3Cold state into D0 and fails.

Highly likely, that the NVMe disk does not support D3Cold.
For example, there is a reddit post (about IdeaPad laptop with the same issue) where the one replaced NVMe by another one and suspend began to work correctly.

I have made a "research" over that issue and found, that the only way is to modify kernel sources and recompile the kernel.
Full research located at https://gist.github.com/Toliak/86340...4337ba6d8e971b.

FlyingPotato_131 · 12-04-2023, 11:49 PM

Quote:

Originally Posted by toliak

Hello.

I have ran into the same problem on RLEF-X laptop.

Briefly:
The kernel puts the disk into D3Cold state when it goes to suspend.
On wake up, kernel returns back the disk from the D3Cold state into D0 and fails.

Highly likely, that the NVMe disk does not support D3Cold.
For example, there is a reddit post (about IdeaPad laptop with the same issue) where the one replaced NVMe by another one and suspend began to work correctly.

I have made a "research" over that issue and found, that the only way is to modify kernel sources and recompile the kernel.
Full research located at https://gist.github.com/Toliak/86340...4337ba6d8e971b.

yoooooo I shall try this to see if it works