[Bug 201957] amdgpu: ring gfx timeout

[bugzilla-daemon at kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=201957

--- Comment #75 from Danil (s48gs.w at gmail.com) ---
> Did you try with the latest Linux Kernel? I had a lot of gpu lockups like
> this. Also try these kernel parameters : "amdgpu.ppfeaturemask=0xffffbffb 
> amdgpu.noretry=0 amdgpu.lockup_timeout=0 amdgpu.gpu_recovery=1
> amdgpu.audio=0 amdgpu.deep_color=1 amd_iommu=on iommu=pt"" ( you might also
> try with amdgpu.ppfeaturemask=0xfffd7fff or amdgpu.ppfeaturemask=0xffffffff )

I tried.

my kernel:
"Linux 5.17.4-1-default #1 SMP PREEMPT Wed Apr 20 07:43:03 UTC 2022 (75e9961)
x86_64 x86_64 x86_64 GNU/Linux"

(this video linked above - were not able to freeze integrated AMD GPU for me, I
mean before I tested with no kernel parameters)

Result is surprising - no crash/freeze for 4+ hours already, I did launch lots
of apps that were reason of freeze for me before. 

As I described above - https://bugzilla.kernel.org/show_bug.cgi?id=201957#c68
for me this freeze happening only when I used OpenGL/Vulkan and video on
background(everything on integrated GPU), and how it was looking from user
experience - when bug triggered(randomly) everything just slowly become lower
and lower FPS, apps that was working on 60fps on fullscreen drop to 5 FPS, and
video also drop to 5-10fps (UI still was responsible)... and freeze in next few
mins/seconds.

Full kernel boot option now: "splash=silent quiet
amdgpu.ppfeaturemask=0xffffbffb amdgpu.noretry=0 amdgpu.lockup_timeout=0
amdgpu.gpu_recovery=1 amdgpu.audio=0 amdgpu.deep_color=1 amd_iommu=on iommu=pt
"

Now, after boot with these options, I see:

Just after boot everything working (OpenGL/Vulkan acceleration by integrated
GPU) with expected performance.

After trying to "trigger bug" (opening multiple OpenGL apps with Vulkan and
WebGL and playing many videos) - OpenGL and Vulkan drops FPS to 20(constant for
single triangle in fullscreen), WebGL2 does not work anymore in webbrowser(even
after browser restart), but Video - still playing with 60 fps with no lag, and
system UI also does not lag.

So GPU graphics acceleration just drop to very low performance mode look like,
but everything else works fine. (also launching graphic apps(native only) using
Nvidia GPU works with 60fps as expected).

Interesting - since FPS droped 20 I can no longer launch "anything" in Wine
(any version include Proton) (after boot it was working), I launched few apps
after boot and check them when GPU FPS drops wine always crash with:
"wine: Unhandled page fault on execute access to 00007F894E200460 at address
00007F894E200460 (thread 0070), starting debugger..."
(not being able to use Wine is a big disadvantage)

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.