Public bug reported:
Currently the cpufreq scaling ACPI P-states driver ("acpi-cpufreq" module) is
built-in into kernel.
Associated kernel key: CONFIG_X86_ACPI_CPUFREQ (Defaults to Y)
It is reasonable to modularize it in the upcoming version of default Karmic
kernel because of its bug and consequent Via CPUs problems. The best solution
out there is blacklisting it in favor of an open-source "cpufreq-via" module, a
good replacement for the buggy and deprecated "e_powersaver". Problem is that
in Karmic it is unable to insert "cpufreq-via" instead or even on par with
"acpi-cpufreq" because "Device or resource is busy". On a generic kernel, but
recompiled configured to use "acpi-cpufreq" as a module, it could be
blacklisted at once and won't cause problems of any kind. On current generic
kernel with the module built-in this can't be done. Not sure, but modularizing
won't probably affect its current users because the module is still here but
won't mess up anymore for those users who don't need it. This will definitely
increase the experience quality for Via-based systems users because currently
CPU frequency scaling is completely broken there. Moreover, for some stupid
reason, with the default P-states driver the Via Nano CPU frequency will stay
at the lowest grade available forever while driver reports wrong frequency.
This results in nothing but annoyance and abysmal overall system performance,
requiring whole kernel recompilation for a single small flag to be disabled.
** Affects: ubuntu
Importance: Undecided
Status: New
--
Built-in acpi-cpufreq causes problems on Via Nano CPUs
https://bugs.launchpad.net/bugs/488792
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
