You are not logged in.
Hello folks, I'm hoping that someone can help point me in the right direction on this one.
Summary:
After applying updates the other day (see pacman log later) I now have corruption on the center monitor of 3, which is higher resolution than the two sidecars. The corruption looks for all the world like a hardware (e.g. cable) problem, but is is definitely NOT hardware as a) nothing changed across the update and b) a Fedora live USB doesn't show the fault. Manjaro live USB *does* show the fault once I configure the monitors appropriately, so I don't believe it to be an issue with my arch install (yes, actual installed OS is plain old Arch, not Manjaro).
It takes quite a specific set-up to reproduce the issue, so I'm betting that nobody else has seen it yet. I'm looking for some guidance as to which package the bug might be in so I can submit a report to the appropriate maintainers.
Probably important details:
- Arch / XFCE / XDE / lightdm
- AMDGPU open source drivers (Radeon RX580)
- 3 monitors in PLP: sidecars are 1680x1050 (but turned so they're vertical), center is 2560x1600
- Corruption DOESN'T happen if the large monitor is placed on the left! Nor does it happen at the login screen (which doesn't have any screen rotation and probably also has the big panel on the left)
- Corruption DOESN'T happen if I reduce the resolution of the center screen to 2048x1280 (but obviously that's not a solution)
- Corruption DOES happen on KDE (tested with Manjaro KDE live USB) as long as you arrange the monitors in the same way
- I promise it's not a hardware fault!
Here's a pic of the corruption, which takes the form of every other line being offset, with the effect getting worse lower on the screen.
https://i.imgur.com/0pQMznF.jpg
Details, details, details
Here's a paste of the pacman.log from the day it all went wrong:
https://pastebin.com/H3i5LuzU
or
https://hastebin.com/dixemisepo.php
Here's a paste of what's installed on the Fedora XFCE LiveUSB that does NOT show the bug:
https://pastebin.com/KK7ZxZTx
or
https://hastebin.com/koregatufu.nginx
Here's some more pics of the corruption, and showing that it works on Fedora LiveUSB
https://imgur.com/a/DIqAfu1
Can anyone suggest which package might be the culprit? Or what my next step should be?
Many thanks in advance!
Last edited by Funkster (2018-11-29 20:14:05)
Offline
1. Read the Code of Conduct and only post thumbnails http://wiki.archlinux.org/index.php/Cod … s_and_code
2. Do not -Sy $package: partial upgrades are unsupported.
Moving to NC...
Offline
1) Dang, sorry. Noted.
2) That's the blender install_deps.sh script, I know not to do that :o)
Thanks.
Offline
2) That's the blender install_deps.sh script, I know not to do that
)
Not on Arch it isn't. Where did this come from?
Offline
Not on Arch it isn't. Where did this come from?
Fresh clone of the blender git repo. Wasn't me, guv, I swear!
Annoyingly I can't remember if I rebooted in between installing the blender deps and doing the full update... there was an hour in between while I was trying to make Blender build, but I only know for sure that the corruption turned up after the full update.
Anyway since the live USBs show the same issue, I'm assuming that the -Sy isn't the culprit.
Thanks.
Offline
Blender is in [community] and it is packaged competently; unlike that bullshit script.
If it happens from the live media, then you make need to talk to upstream.
Offline
So who to talk to? Which packages are likely candidates?
Offline
Try passing "amdgpu.dc=0" to the kernel.
Otherwise please post the output of "xrandr -q" w/ and w/o a problematic config, the output of "glxinfo -l" (lots of lines, try https://wiki.archlinux.org/index.php/Pastebin) and your xorg log.
Online
Try passing "amdgpu.dc=0" to the kernel.
Yes, that fixes it (presumably at great cost to graphics performance?).
I guess that means it's an issue with mesa?
EDIT: Thank you!
Last edited by Funkster (2018-11-29 19:11:06)
Offline
You're welcome.
No, it's an issue w/ the amdgpu kernel module (well: "on the surface")
Ideally the impact isn't that big (if at all), but DC is the coming stack, so if this does not "fix itself" you should raise a bug (https://bugs.freedesktop.org/buglist.cgi?component=DRM%2FAMDgpu&product=DRI&resolution=--- - yeah, that's the mesa bugtracker ;-)
Please always remember to mark resolved threads by editing your initial posts subject - so others will know that there's no task left, but maybe a solution to find.
Thanks.
You can annotate that it's only "worked around" by suppressing DC.
Online
Marked solved (didn't have room to write workaround in the title, hopefully people will read this).
Link to the bug report:
https://bugs.freedesktop.org/show_bug.cgi?id=108907
Offline