Random freezing (AMD) [SOLVED]

[east]

Hello!
Recently I reinstalled Gentoo (+ rootfs encryption) and started receiving a error i did not with any of my previous Gentoo installations.
My XFCE4 freezes, seemingly at random, except for my mouse cursor and audio.
If i switch tty (with ctrl+ alt + F1-12) and switch back without doing anything else, it stops being frozen until the next random soft crash (which can be hours or minutes).
I've reinstalled Gentoo again to see if it'd fix the issue. It did not.

I have a suspicion this error in Xorg.0.log is the cause.

Code: Select all

[    53.443] (WW) AMDGPU(0): flip queue failed: Invalid argument
[    53.443] (WW) AMDGPU(0): Page flip failed: Invalid argument

I tried disabling page flipping which introduced performance issues and mouse lag but from my short test seemed to remove the error code in Xorg.0.log.
I'm unsure if it resolved the crashing issue as i did not test for very long due to aforementioned issues.

/etc/X11/xorg.conf.d/20-amdgpu.conf

Code: Select all

Section "Device"
    Identifier "AMD"
    Driver "amdgpu"
    Option "TearFree" "true"
    Option "PageFlip" "false"
EndSection

I do not have swap (previously did), kernel is 6.12.47-gentoo-dist (gentoo-kernel-bin), I have 64gb ram, a Ryzen 9 7950X, and a AMD Radeon RX 7900 XTX.

I have two monitors, a 240hz (via Display Port into GPU), and a 60hz (via HDMI into Motherboard).

The user i startx with is in the groups : tty disk wheel floppy video kvm input users pipewire user

Xorg.0.log :
https://paste.gentoo.zip/YtP8w9aT

Maybe it's related to the kernel or i did something improperly?
I'm unsure the reason, I did not have this issue on my previous Gentoo installation or on a liveUSB (of gentoo or mint).
I'm at a loss and haven't really found much helpful info about it online.
While writing this post I found this forum post which seems to have a similar issue without being resolved.

Apologies if this is the improper avenue to receive support or i did not include proper information.
Thank you for your time.

[NeddySeagoon]

east,

Welcome to Gentoo.

A few things ...
You need two Section "Device" entries as you have two differerent video cards that both use amdgpu.
The Identifiers must be unique and they bind to the right card because you mus use the BusID option.
I'm struggling to provide an example as I'm on my phone right now.
This allows you to use differerent options on each card.

I'm not sure that Xorg is using your config at all, as it is still doing the automatic driver detection and that should be missing from your Xorg.0.log when a driver is given.

It's also worth trying a testing kernel and testing linux-firmware.
The latter may not exist but if it does, try it.
mesa is vaguely implicated but thats a long shot.

I'll post more when I'm at a real PC.

-- edit --

Code: Select all

       BusID  "bus-id"
              This specifies the bus location of the graphics card.  For PCI/AGP cards, the  bus-id  string  has  the  form  PCI:bus@do‐
              main:device:function  (e.g.,  “PCI:1@0:0:0”  might be appropriate for an AGP card). The "@domain" part can be left out for
              PCI domain 0. This field is usually optional in single-head configurations when  using  the  primary  graphics  card.   In
              multi-head  configurations,  or  when using a secondary graphics card in a single-head configuration, this entry is manda‐
              tory.  Its main purpose is to make an unambiguous connection between the device section and the hardware it is  represent‐
              ing.  This information can usually be found by running the pciaccess tool scanpci.

Its also in your Xorg.0.log anh in the lspci output.

Code: Select all

[    52.121] (--) PCI:*(3@0:0:0) 1002:744c:1da2:e471 rev 200, Mem @ 0xf000000000/34359738368, 0xf800000000/268435456, 0xdf000000/1048576, I/O @ 0x0000f000/256, BIOS @ 0x????????/131072
[    52.121] (--) PCI: (89@0:0:0) 1002:164e:1849:364e rev 193, Mem @ 0xf870000000/268435456, 0xf880000000/2097152, 0xdef00000/524288, I/O @ 0x0000e000/256

Code: Select all

man amdgpu

will show you all the options to turn on/off acceleration selectively

[east]

Thanks for your reply it's been quite helpful in troubleshooting.

I've since switched from kernel version 6.12.47 to version 6.16.9 which on it's own didn't seem to impact the issue.
I've also made two files in /etc/X11/xorg.conf.d/ one for dgpu (1-amdgpu.conf) and another for igpu (2-amdgpu.conf).

(1-amdgpu.conf)

Code: Select all

Section "Device"
       	### Available Driver options are:-
       	### Values: <i>: integer, <f>: float, <bool>: "True"/"False",
       	### <string>: "String", <freq>: "<f> Hz/kHz/MHz",
       	### <percent>: "<f>%"
       	### [arg]: arg optional
          #Option     "Accel"                     "false"
       	Option     "SWcursor"                   "true"
          Option     "EnablePageFlip"            "false"
       	#Option     "SubPixelOrder"             # [<str>]
       	#Option     "ZaphodHeads"               # <str>
       	#Option     "AccelMethod"               # <str>
       	#Option     "DRI3"                      # [<bool>]
       	#Option     "DRI"                       # <i>
       	#Option     "ShadowPrimary"             # [<bool>]
       	Option     "TearFree"                   "true"
       	#Option     "DeleteUnusedDP12Displays"  # [<bool>]
       	#Option     "VariableRefresh"           # [<bool>]
       	#Option     "AsyncFlipSecondaries"       "true"
       	Identifier  "Card0"
       	Driver      "amdgpu"
       	BusID       "PCI:03:00:0"
EndSection

(2-amdgpu.conf)

Code: Select all

Section "Device"
       	### Available Driver options are:-
       	### Values: <i>: integer, <f>: float, <bool>: "True"/"False",
       	### <string>: "String", <freq>: "<f> Hz/kHz/MHz",
       	### <percent>: "<f>%"
       	### [arg]: arg optional
       	#Option     "Accel"                     "false"
       	#Option     "SWcursor"                  # [<bool>]
       	#Option     "EnablePageFlip"            "false"
       	#Option     "SubPixelOrder"             # [<str>]
       	#Option     "ZaphodHeads"               # <str>
       	#Option     "AccelMethod"               # <str>
       	#Option     "DRI3"                      # [<bool>]
       	#Option     "DRI"                       # <i>
       	#Option     "ShadowPrimary"             # [<bool>]
       	#Option     "TearFree"                   "false"
       	#Option     "DeleteUnusedDP12Displays"  # [<bool>]
       	#Option     "VariableRefresh"           # [<bool>]
       	#Option     "AsyncFlipSecondaries"       "true"
       	Identifier  "Card1"
       	Driver      "amdgpu"
       	BusID       "PCI:59:00:0"
EndSection

I tried a myriad of different combinations, but the only way thus far I've seen to remove the error is to outright disable "EnablePageFlip" on my dGPU (1-amdgpu.conf) but I'm unsure the full implications of these settings on system performance and stability which is something that is quite important to me.

I also enabled "SWcursor" to remedy the errors below which seemed to have worked(?) :

Code: Select all

[   114.688] (EE) event3  - Logitech USB Receiver: client bug: event processing lagging behind by 35ms, your system is too slow
[   114.857] (II) event3  - Logitech USB Receiver: SYN_DROPPED event - some input events have been lost.

The mouse errors were present with "EnablePageFlip" on and off.

Updated Xorg.0.log:
https://paste.gentoo.zip/pVyrsIan

I am still in the process of testing and will subsequently edit this post with any findings.
Forgive my inexperience and thank you for the help.


edit :
I've experienced another crash while having PageFlip disabled, unsure if it's related to previous crash issue or a entirely new one.
Here is the offending log : https://paste.gentoo.zip/MFAgOeLI

[east]

I fixed the issue by adding the following kernel parameters

/etc/default/grub

GRUB_CMDLINE_LINUX="amdgpu.runpm=0 amdgpu.dcdebugmask=0x10"

Code: Select all

grub-mkconfig -o /boot/grub/grub.cfg

If anyone else is having this issue read through the amd gitlab issue(s) listed below; Multiple different proposed fixes, this is the one that helped in my case.
I've also since switched from X11 - Wayland, this along with kernel parameters fixed a lot of related problems like my mouse dropping inputs and graphical errors.

Resources:
https://gitlab.freedesktop.org/drm/amd/-/issues/4141
https://gitlab.freedesktop.org/drm/amd/-/issues/2950