You are not logged in.
mesa-test-git 24.3.0_devel.195300.b01d76027d3-1 seems fine.
Online
116 revs left
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
mesa-test-git 24.3.0_devel.195417.1c156f7fa9a-1 is broken.
Online
58 revisions left
[1e0c502a77cf9beb6bec53e1e19d2741691eb8ff] panfrost: Don't turn 3D/cube images into 2D arrays
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
mesa-test-git 24.3.0_devel.195358.1e0c502a77c-1 seems fine.
Online
29 revisions left
[5b8f27d3d7503c501e7b7b12f37ff7feafc37b56] freedreno/ci: Uprev kernel to 6.11
I'm going to sleep soon, so this is last one for now.
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
mesa-test-git 24.3.0_devel.195387.5b8f27d3d75-1 is broken.
Sounds good. At this rate, we should finish bisecting tomorrow.
Online
14 revs left
[40d9616bd3bebb19e4cdb9115ce6028dc79d88ad] radeonsi: don't pad esgs_vertex_stride if it's 0
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
Upps, I missed the show. Still using
mesa-test-git-25.0.0_devel.199748.72351afe24f
without problems, but as explained before, the number won't tell you all ...
So I lean back until anyone gives me hints how to help.
Offline
kclisp was able to break that version, can you try putting a heavy load on the graphics card to verify ?
This is how kclisp did it
I believe I've found a quick way to reproduce the issue. I use i3; I start two terminals in one workspace, and hold Super + Shift + Space on one of them, which rapidly alternates it between a floating window and a tiling window, producing a lot of graphical window resizing. This can produce a freeze on broken builds in a few seconds.
Quick switching betwee graphics-heavy applications (especially if they're on different virtual workspaces) is also worth a try.
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
I don't want to spoil the party ...
kclisp was able to break that version, can you try putting a heavy load on the graphics card to verify ?
.
Would firing up up to 8 instances of glmark2 count for "heavy load"? They are running in the background while I am writing this. Up to now, as during the last sessions yesterday and the day before, nothing unexpected here with
pacman -Qs mesa
local/glu 9.0.3-2.1
Mesa OpenGL utility library
local/lib32-glu 9.0.3-2.1
Mesa OpenGL utility library (32 bits)
local/lib32-mesa 1:24.2.7-1.1
Open-source OpenGL drivers - 32-bit
local/mesa-demos 9.0.0-5.1
Mesa demos
local/mesa-test-git 25.0.0_devel.199748.72351afe24f-1
an open-source implementation of the OpenGL specification, stripped down git
version
local/mesa-utils 9.0.0-5.1
Essential Mesa utilities
Writing the rest of this message with this online gpu stress test running in the background really (!!!) slow things down nearly to be unusable, but no freeze.
Offline
Oh dear.
In normal operation, it never took me longer than 10 hours for the system to freeze.
If the 25.0.0. test version is still online, I can also test it if necessary. But this cannot start before late afternoon (utc+1).
Maybe other users can participate. Just to be on the safe side.
Offline
In normal operation, it never took me longer than 10 hours for the system to freeze.
This was my observation too. So up to now, I felt quit happy with this version. 24.3.0 from the repo failed within a few hours while listening to an audio stream.
Maybe other users can participate. Just to be on the safe side
I would be very happen to see this happen. My system is no fresh install - in fact I don't remember when I set it up and it migrates from pc to pc over the years. So some more test results would rule out the possibility of chasing ghosts.
The link is still online:
very recent trunk commit 72351afe24f
Last edited by bernd_b (2025-01-08 15:26:40)
Offline
> Would firing up up to 8 instances of glmark2 count for "heavy load"?
This hasn't worked for me, and neither has the aforementioned in-browser GPU stress test, or running multiple instances of vkmark. I've not yet tried the i3 method, but attempting something that sounds similar within GNOME has not reproduced the issue. I can spend more time experimenting later today.
Offline
In case it's helpful in identifying the issue, I've shared the complete specs of my system and associated logs at https://gitlab.freedesktop.org/drm/amd/ … te_2727135
Offline
The recent posts strongly suggest there is more then 1 bug present.
kclisp : you are sofar the only one who has found a reproducable way to produce a crash.
Continue the bisecting tests, that way we can atleast find the cause of the bug you can reproduce .
All others : keep trying to break very recent trunk commit 72351afe24f in a reproducible way.
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
I'm now using mesa-test-git 25.0.0_devel.199748.72351afe24f-1 with AMD Radeon Vega 8 Graphics (Picasso/Raven 2)...
$ inxi -G
Graphics:
Device-1: Advanced Micro Devices [AMD/ATI] Picasso/Raven 2 [Radeon Vega
Series / Radeon Mobile Series] driver: amdgpu v: kernel
Display: x11 server: X.Org v: 21.1.15 driver: X: loaded: amdgpu
unloaded: modesetting dri: radeonsi gpu: amdgpu resolution: 1680x1050~60Hz
API: EGL v: 1.5 drivers: kms_swrast,radeonsi,swrast
platforms: gbm,x11,surfaceless,device
API: OpenGL v: 4.6 compat-v: 4.5 vendor: amd mesa v: 25.0.0-devel
renderer: AMD Radeon Vega 8 Graphics (radeonsi raven ACO DRM 3.54
6.6.69-1-lts)
API: Vulkan v: 1.4.303 drivers: N/A surfaces: xcb,xlib
Test
1)
The side 'Stress My GPU' (gpu only) on https://mprep.info/gpu/ doesn't stress the vega 8 gpu enouph (monitored by nvtop - https://github.com/Syllo/nvtop)...
2)
Quickly switching workspaces with 'ctrl+alt+cursor right/left' (Xfce) shows no effect...
Last edited by orbit-oc (2025-01-08 16:13:46)
Offline
mesa-test-git 24.3.0_devel.195372.40d9616bd3b-1 seems fine.
I wouldn't be surprised at all if there were multiple bugs. For reference, I'm using AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx (Raven Ridge).
Online
@bernd_b
just to check, did you restart your system?
I'm not sure how much GPU load by itself affects the frequency of freezes - maybe certain operations are more prone? The reproducer I use only uses around 30% of the GPU.
Online
7 revs left to check
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
mesa-test-git 24.3.0_devel.195379.653bcd85e02-1 is broken.
Online
Only 3 revisions left, getting close.
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
mesa-test-git 24.3.0_devel.195375.0112fd7d40e-1 seems fine.
Online
1 rev left to test after this one, almost there.
[c99b55092ff70f283e1cb8d112bf7bdb587236a4] radeonsi: move barriers out of si_execute_clears
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
mesa-test-git 24.3.0_devel.195377.c99b55092ff-1 seems fine.
Online