Manjaro stable: Ryzen system unbootable with kernel 4.15+

[R33D3M33R]

Dear developers,

I'm using Manjaro Stable with Plasma Desktop on a Ryzen 5 1500X build with Gigabyte GA-AB350M-GAMING 3 motherboard.

Since the update to kernel 4.15 my Ryzen machine doesn't boot properly anymore. The boot stops when the login window should appear. I can only see the mouse cursor, which can be normally moved around for like 2 seconds and then it completely freezes. Keyboard doesn't respond at all. The HDD activity led is constantly lit. It almost looks like some kind of a memleak.

I forced the boot with xforcevesa to rule out any amdgpu glitches, but it still freezes. Everything works completely fine with the LTS kernel.

Please help me debug this.

Best Regards,
Andrej

[philmmanjaro]

Issue should have been fixed already with 4.15.5-1 or later. Please check once more. Most possible it was due to the framebuffer changes. Else tell me which kernel version created your issues first. Does the current release work for you?

[R33D3M33R]

I first noticed this with the 4.15.5-1 kernel. I used 4.14 LTS before and 4.15.5-1 was the first version of 4.15 that I installed. This issue also occurs on 4.15.7-1 and 4.16.0-1 kernels.

I cannot reproduce this problem on a Kabini laptop with A4-5000 APU.

[philmmanjaro]

First change to framebuffers was made with v4.15.4. Then we reverted it with v4.15.5 in a different way. Maybe a rebuild of the current kernel with config v4.15.2-2 might fix your issue. Which modules are loaded on your end anyway? lsmod.

[R33D3M33R]

I'm attaching lsmod output:

lsmod.txt

I also tried to clear the cache; ~/.cache/ and /var/lib/sddm/.cache/ but it didn't help.

[R33D3M33R]

I have finally been able to debug this.

As I suspected, this problem is related to the hard drive problems on B350 Promontory chipset with 4.15+ kernels.

When I disabled SDDM and booted to shell, I was finally able to see errors. All of them were related to my SATA HDD(s):

failed command: WRITE FPDMA QUEUED
failed command: WRITE DMA
status: { DRDY }
SError CommWake 10B8B

etc.

I attach sample output from log.

b350errors.log

What I did to temporarily solve the problem is move the hard drives from ASMedia SATA ports (labeled ASATA3) to the ports provided by the B350 chipset (labeled SATA). So whatever changed from 4.14 to 4.15 broke the SATA port compatibility with my HDD(s).

Best Regards,
Andrej