Continuation of: Are there any NVME/SSD related bug fixes in Linux 5.6 vis-a-vis 5.4?

Support
1

Hi all,

I was earlier having a lot of problems (system freeze) with Manjaro Linux when I attempted to have it boot off an NVME SSD. There were many people following my original post which unfortunately went down in the Manjaro website crash.

So I am posting again to the new forum for anyone interested.

My problems have been solved. My Manjaro 5.4.x does not freeze anymore. Unfortunately and strangely, I do not know which fix actually worked. I applied several fixes:

  1. Switched NVME from Kingston to Samsung 970 Evo Plus - did not help

  2. Switched motherboard from Gigabyte B450 AORUS Elite ATX to ASRock B450 Steel Legend ATX - did not help

  3. { Disabled hibernation + uninstalled Timeshift + switched Manjaro root filesystem from ext4 to XFS } - after this set of fixes, the nature of the problem changed. Instead of random system freeze once a day, I started experiencing random system restart once a day.

  4. Then I effected my final set of fixes:

    a) Disabled some ACPI components in BIOS.

    b) Renamed /boot/amd-ucode.img as /boot/amd-ucode.bin,
    and then ran mkinitrd (to create a new initram) followed by update-grub.
    The point is I kept /boot/amd-ucode.img out of the initial ramdisk contents.

    c) Swapped my power cord with another power cord.

    d) Swapped my USB mouse with another (new) mouse.

After I applied the set of four fixes above, my Manjaro has been rock-solid stable.

I wish I could offer more help to people suffering the syndrome, but I can’t.

If you are facing NVME freeze, perhaps try the final set of fixes above one step at a time, and see which one works for you.

Regards,
Manish Jain

2

Is your processor microcode actually updating now?

3

have you try theses options

add in boot kernel ( or /etc/default/grub )
GRUB_DECLINE_LINUX_DEFAULT=

“pcie=aspm=off”
or
“nvme_core.default_ps_max_latency_us=0”

4

It looks like my problems are gone for good, and I won’t alter any variables till I see signs of any problem. Just yesterday, I upgraded the system with ‘pacman -Syyu’, which pulled in a fresh amd-ucode.img. I am running the system now with amd-ucode.img bundled into grub’s initrd, and everything is okay.

The most unfortunate part now is I do not know which particular fix worked for my system. If somebody else is facing similar NVME freeze, he can at least look up this thread to try figure out what might work for him/her.

Thanks all,
Manish Jain

5

After I did a system upgrade day before yesterday (which pulled in a fresh /boot/amd-ucode.img), my system suddenly restarted today again. I have now renamed /boot/amd-ucode.img as /boot/amd-ucode.bin, and removed /boot/amd-ucode.img from grub’s configuration for initramfs. I will update in a week or so if my system restart problem is solved by this simple step.

6

what is exactly the trouble with amd-ucode.img and .bin ?

7

Whenever the package amd-ucode gets installed (normally as part of a system upgrade), the file /boot/amd-ucode.img gets created as well as gets pulled into the initramfs of grub.

About 10 days back, suspecting /boot/amd-ucode.img as my system’s problem on account of which random freeze/restart occurs, I:

  1. Renamed /boot/amd-ucode.img as /boot/amd-ucode.bin;
  2. Rebooted the system. At the grub prompt, I removed /boot/amd-ucode.img from the initrd parameter list, and then booted the system;
  3. Ran update-grub to make the changes permanent.

My system then worked stably for a week - no random freeze/restart.

Day before yesterday, pamac indicated a system upgrade was available. I upgraded my system with pacman -Syyu, as part of which /boot/amd-ucode.img again got created again and pulled into initramfs.

Within 48 hours, my system today morning abruptly restarted.

I strongly suspect /boot/amd-ucode.img is the problem - it does not go down well with my AMD Ryzen 3 1200 processor. I have today again renamed /boot/amd-ucode.img as /boot/amd-ucode.bin and removed it from initramfs using the sequence detailed above.

My system is now working stably. If my system continues to work stably for a week, I shall file a bug report at kernel.org specifying the amd-ucode package as bug-ridden.

Perhaps Manjaro/Arch developers too should keep a watch on this thread for the next few days.

Thanks,
Manish Jain

8