You are not logged in.
Happens un current stable kernel once an application goes fullscreen the screen turns black for seconds and then returns. The screen connected to a laptop via hdmi. Nothing in journalctl nor dmesg. Peculiar thing is only happen once before the screen goes sleep or the laptop after suspend. Meaning that when using system it happens only once. In lts 6.12 the issue not presemt
Last edited by matryoshka (2026-01-16 08:37:41)
Offline
Please post your complete system journal for the boot:
sudo journalctl -b | curl -F 'file=@-' 0x0.stPossibly https://bbs.archlinux.org/viewtopic.php … 9#p2282749
Offline
Not sure it was like that before but to the bug does not appear after reboot and appears after screen timeout once before another screen timeout
Offline
This sounds even more like https://bbs.archlinux.org/viewtopic.php … 9#p2282749
Can you attach the output via displayport (and/or try to build the kernel w/ c97da4785b3bbc60c24cfd1ffea1d7c8b90ed743 reverted)?
Offline
No dp port nor dispay to test it out.
It sounds like the issue you mentioned but still my behavior is different but is related to hdmi.
I do not know how to build the kernel.
Should I get the source for latest stable then git revert "c97da4785b3bbc60c24cfd1ffea1d7c8b90ed743" and after that do steps from this: https://wiki.archlinux.org/title/Kernel … figuration ?
Offline
Also no other monitor (though that would produce inconclusive results only anyway)
You can use https://wiki.archlinux.org/title/Kernel … ild_system
Offline
Then I wont be able to use git revert as this use different history
Offline
https://github.com/archlinux/linux/comm … c8b90ed743 is very much a commit in the archlinux kernel git history?
Offline
Yeah, this turns into something different.
I thought that pkgctl repo clone --protocol=https linux will get linux code but it only contains a git repo with arch package file and Im do not know how to work with it.
Should have I called a git revert somewhere in that __PKGBUILD file? Wiki also says it applies all patches found in "src". I have got patch file from the code
Well and some another issue I have got after calling makepkg:
...
==> Verifying source file signatures with gpg...
linux-6.18.5.tar ... FAILED (unknown public key 38DBBDC86092693E)
linux-v6.18.5-arch1.patch.zst ... FAILED (unknown public key B8AC08600F108CDF)
==> ERROR: One or more PGP signatures could not be verified!
~/D/b/linux ((6.18.5.arch1-1)|✚1) $
It really turns into something else so am not sure whether should I continue to fight with. I will try tomoroww again as it really late. Do you know how can I revert the commit when using arch pkg build or why it wasnt enough to put patch file alongside with PKGBUILD file?
Offline
Ah,
Reason: The linux PKGBUILD no longer uses git [2][3] to fetch its sources. (Discuss in Talk:Kernel/Arch build system)
You'd fetch https://github.com/archlinux/linux/comm … ed743.diff and revert that w/ patch -R resp. you can just use interdiff from https://archlinux.org/packages/extra/x86_64/patchutils/ to invert the patch
interdiff -q c97da4785b3bbc60c24cfd1ffea1d7c8b90ed743.diff.diff /dev/null > invert_c97da4785b3bbc60c24cfd1ffea1d7c8b90ed743.diff.patchand put that w/ the other patches in the package (the PKGBUILD should apply them all but iirc the extension has to be "patch", not "diff")
Offline
Thanks for the note. Patch file did not get discovered by itself so I had to add it to sources array:
==> Starting prepare()...
Setting version...
Applying patch linux-v6.18.5-arch1.patch...
patching file Makefile
patching file drivers/android/binder/page_range.rs
patching file drivers/android/binder/thread.rs
patching file drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
patching file drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
patching file drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
patching file fs/btrfs/delayed-inode.c
patching file kernel/fork.c
patching file net/ipv4/esp4_offload.c
patching file net/ipv6/esp6_offload.c
Applying patch revert.patch...
patching file drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
patching file drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
Setting config...
#
# No change to .config
#
Prepared linux-test version 6.18.5-arch1-1-testAlso, I got patch from linux remote source
As I did not have a reproducible way to see the bug before I went with these steps:
1. log in into gnome
2. set a very short timieout for display
3. wait for the timeout
4. wake up by interacting and test by going to fullscreen view of youtube video
On stable kernel black screen appears as I have described earlier.
With reverted commit no issue is seen.
I will test by using this kernel for sometime to see if it truly gone.
Thanks for your time, seth. Haven't touched PKGBUILD before
Aslo, patch from here https://gitlab.freedesktop.org/drm/amd/-/issues/4859 also passes my test so I instead try to run with it for some time
Last edited by matryoshka (2026-01-15 05:05:27)
Offline
\o/
Seems the "feature" will disabled by default in 6.18.5(?)
In case and 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.
Offline
Using mentioned patch for more than a day and no issues.
According to gitlab comments the patch should arrive in 6.19: https://gitlab.freedesktop.org/drm/amd/ … te_3286536
Offline