Re: OOPS: 2.6.16-rc6 cpufreq_conservative
From: Alexander Clouter
Date: Sun Mar 19 2006 - 06:58:09 EST
Parag Warudkar <kernel-stuff@xxxxxxxxxxx> [20060319 01:34:01 -0500]:
> On Saturday 18 March 2006 19:53, Andrew Morton wrote:
> > I might stick a once-off WARN_ON() in there so someone gets in
> > and works out why we keep on having to graft mysterious null-pointer
> > avoidances into cpufreq.
> cpufreq_conservative should be marked broken on SMP - I have used it on UP
> boxes without trouble but I can't even safely modprobe it on SMP - it nearly
> ate my filesystem.
> And there seem to be multiple different problems with it - I get different
> oopses depending upon whether or not I have loaded it before or after the
> ondemand module. Weird enough - cpufreq_conservative shares much of it's
> code with cpufreq_ondemand, which works without any problem.
Well its drifted a bit, however I submitted a number of patches here about
two weeks ago to bring it back into line and hopefully make it HOTPLUG safe.
The new set of patches pretty much make conservative's codebase identical to
ondemands....as no one has posted back having used these or anything what am
I to do?!
> Let me know if anyone has objection to marking cpufreq_conservative
> depends !SMP - I am planning to submit a patch soon.
Doesn't bother me, what does is no one is trying to updates to conservative
before deciding to declare it borked?
/ Misfortunes arrive on wings and leave \
\ on foot. /
Description: Digital signature