Control: tags -1 + moreinfo
Hi Klaus,
Sorry for the late reply.
On Sun, May 10, 2026 at 08:40:53PM +0200, Klaus Singvogel wrote:
> Hello Salvatore,
>
> Salvatore Bonaccorso wrote:
> > On Sat, May 09, 2026 at 11:51:17PM +0200, Klaus Singvogel wrote:
> > > Package: linux-image-amd64
> > > Version: 6.12.86-1
> > > Severity: important
> > >
> >
> > Sorry to hear about your regression. So since you seem to have an easy
> > "reproducer" might you be able to bisect the problem? In this case
> > since some major versions are involved it might be worth first trying
> > to narrow the range a bit, by fetching kernel images built for Debian
> > from the snaphots service where it is bumped to major upstream
> > version. Then once you have arange from a 6.n to a 6.m version where
> > things regress, to bisect it.
>
> I tested the following kernel versions on Debian Trixie, and the issue is
> present with all of them:
>
> - 6.12.85 (trixie)
> - 6.12.86 (trixie)
> - 6.19.14 (trixie-backports)
> - 7.0.4 (trixie-backports)
>
> With 6.19.14, the issue occurred regularly, even when the HDMI-connected
> display was idle and no video playback was active, but not 7.0.4
>
> With the 6.12.8x and 7.0.4-1 kernels, the issue appears to be triggered mainly
> during video playback, for example with YouTube, video advertisements in
> websites, or VLC.
>
> The problem did not occur when running Debian Bookworm with:
>
> - 6.1.0 (bookworm)
>
> I could not perform a bisection, but based on the above results the regression
> appears to have been introduced somewhere between 6.1.x and 6.12.x.
Is the reason you could not do a bisect because needing help? if that
is the reason here are some instructions, so we have 6.1.y is good and
6.12.y is bad as I understand. Then you can do the follwoing which
would need a couple of times compiling the kernel and test it:
git clone
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
cd linux-stable
git checkout v6.1
cp /boot/config-$(uname -r) .config
yes '' | make localmodconfig
make savedefconfig
mv defconfig arch/x86/configs/my_defconfig
# test 6.1 to ensure this is "good"
make my_defconfig
make -j $(nproc) bindeb-pkg
... install the resulting .deb package and confirm problem does not exist
# test 6.12 to ensure this is "bad"
git checkout v6.12
make my_defconfig
make -j $(nproc) bindeb-pkg
... install the resulting .deb package and confirm problem exists
With that confirmed, the bisection can start:
git bisect start
git bisect good v6.1
git bisect bad v6.12
In each bisection step git checks out a state between the oldest
known-bad and the newest known-good commit. In each step test using:
make my_defconfig
make -j $(nproc) bindeb-pkg
... install, verify if problem exists
and if the problem is hit run:
git bisect bad
and if the problem doesn't trigger run:
git bisect good
. Please pay attention to always select the just built kernel for
booting, it won't always be the default kernel picked up by grub.
Iterate until git announces to have identified the first bad commit.
Then provide the output of
git bisect log
In the course of the bisection you might have to uninstall previous
kernels again to not exhaust the disk space in /boot. Also in the end
uninstall all self-built kernels again.
One note: The range is still substantial. So it still would be
beneficial to narrow down first the range by manual "bisecting" the
available linux images released in the process of developing trixie,
they can be fetched via the snapshot service:
https://snapshot.debian.org/ .
I hope this is bit of help to narrow down better the issue.
Regards,
Salvatore
