On Wed, 19 Nov 2025 17:57:29 +0100 Uwe =?utf-8?Q?Kleine-K=C3=B6nig?=
<[email protected]> wrote:
Control: tag -1 + moreinfo
Hallo Bernhard,
On Thu, Nov 06, 2025 at 10:21:14PM +0100, Bernhard Übelacker wrote:
> Package: src:linux
> Version: 6.12.57-1
> Severity: normal
> X-Debbugs-Cc: [email protected]
>
> Dear Maintainer,
> I recently find my desktop system crashing with this below message.
>
> I have a crashkernel setup, and attached a dmesg from collected by it.
>
> This happens now always when invoking a make of a wine tree.
I see you have some xen(?) and vhba traces in the logs you provided.
Does this problem occur in a virtual environment or on the native host?
Can you reproduce the problem without vhba?
Best regards
Uwe
Hello,
sorry for my late reply.
This system is a bare metal installation,
not running inside any virtualisation I am aware of.
However, at this time a few KVM guest VMs could have been running.
The vhba is a "virtual host bus adapter" from package vhba-dkms, used by cdemu.
I hesitated to deliberately try to crash the system,
therefore did not yet test again without the vhba module.
But I changed my opinion now, it may be simply a "thermal" issue.
I got another crash during the summer months,
so I changed the bios to some more aggressive cooling.
And added to my plasma desktop a widget to show CPU temperature.
With this I noticed there are from time to time some spikes,
instantly growing by ca. 10 degrees, then slowly going down again (see attached
picture).
I got the impression these CPU boosts appear independent of the CPU temperature,
so could the issue be when the CPU is already quite hot and a boost kicks in
the temperature crosses some threshold?
Therefore I started to disable this CPU boost by this command after the boot:
echo "0" | tee /sys/devices/system/cpu/cpufreq/boost
With this I am doing builds again with "make --load-average" and got no more
crashes,
so this connection to make was probably a wrong conclusion.
But I am now also following the kernel line from trixie backports since June.
So I am not sure if there is still a point in having this bug report open,
except if acpi_cpufreq should avoid boosts when over some temperature?
Thanks for all your work.
Kind regards,
Bernhard
