RE: [>=2.6.23][BUG] Oops on power disconnection

["Pallipadi, Venkatesh" <venkatesh.pallipadi@intel.com>]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 4219 bytes --]

 

>-----Original Message-----
>From: Thomas Renninger [mailto:trenn@suse.de] 
>Sent: Thursday, April 03, 2008 2:05 PM
>To: Sanjeev Aditya Naga
>Cc: Andrew Morton; linux-kernel@vger.kernel.org; 
>linux-acpi@vger.kernel.org; Pallipadi, Venkatesh; Li, Shaohua
>Subject: Re: [>=2.6.23][BUG] Oops on power disconnection
>
>
>On Thu, 2008-04-03 at 21:45 +0530, Sanjeev Aditya Naga wrote:
>> Hello Thomas,
>> 
>> On Wed, Apr 2, 2008 at 11:42 PM, Thomas Renninger 
><trenn@suse.de> wrote:
>> >
>> > On Wed, 2008-04-02 at 21:43 +0530, Sanjeev Aditya Naga wrote:
>> >  > Hello Thomas,
>> >  >
>> >  > > Hi,
>> >  > >
>> >  > >
>> >  > >  On Wed, 2008-04-02 at 18:38 +0530, Sanjeev Aditya Naga wrote:
>> >  > >  > Hello Andrew Morton,
>> >  > >  >
>> >  > >  > Greetings!
>> >  > >  >
>> >  > >  > Thank you for the update. Is there anything I can do
>> >  > >  > from my side?
>> >  > >  >
>> >  > >  > I thought it was a acpi (dsdt) problem. And based on
>> >  > >  > a tutorial, I have tried to extract, fix, 
>recompile the dsdt
>> >  > >  > and use it with the kernel. But still I have the 
>same problem.
>> >  > >  > Let me know if I shall attach the dsdt (original) 
>decompiled
>> >  > >  > code, if that helps.
>> >  > >
>> >  > >  this could be due a general memory corruption 
>problem through ACPICA.
>> >  > >  If you get different backtraces on reboots even you 
>only modified things
>> >  > >  that do not have to do with the problem, it's 
>probably that and related
>> >  > >  to:
>> >  > >  http://bugzilla.kernel.org/show_bug.cgi?id=10339
>> >  > >
>> >  > >  You might want to try the latest kernel or the patch 
>posted there.
>> >  >
>> >  > Thank you for the update.
>> >  > I have checked the bug and unfortunately its not the
>> >  > same issue. Things work absolutely fine, when I'm
>> >  > running on AC power. It even displays the exact
>> >  > battery (and charging) status to me. It messes up
>> >  > when suddenly AC power gets disconnected and
>> >  > switches to battery mode (The time when I get Oops).
>> >  > The system is still usable after switching to battery
>> >  > mode and I still get correct battery stats until its
>> >  > completely discharged. However most of the commands
>> >  > like kill, poweroff, java doesn't work after the Oops.
>> >  The bug is not related to battery, but to AML parsing and 
>can therefore
>> >  affect anything which is ACPI related.
>> >
>> >
>> >  > BTW there is one similarity with the referenced bug.
>> >  > If I boot the computer without AC Power, it gives the
>> >  > same Oops and stops during booting itself.
>> >  >
>> >  > I shall try the latest kernel once and shall update you.
>> >  That would be great.
>> >  If it works, please give the patch there a try, IMO this 
>one should see
>> >  2.6.2[34].X stable kernels soon.
>> 
>> I have got the lastest kernel 2.6.25-rc8 today. I observed that
>> the referenced patch is already in the kernel. However this didn't
>> solve the problem in question. I get the same Oops on this kernel
>> as well. Find the latest dmesg along with the Oops:
>Ok.
>Maybe best is you document the backtraces/oopses with kernel 
>versions at
>bugzilla.kernel.org and add dmesg and acpidump.
>It seems your machine notifies OS that the C-state table changed.
>AFAIK this is rare and there might be a general bug in the 
>cpuidle layer
>which I do not know well.
>Best you add Venkatesh and Shaohua Li <shaohua.li@intel.com> to CC of
>the bug.
>While the backtrace shows a lot, cpuidle IMO is missing a general debug
>option like in the cpufreq layer.
>I couldn't find a single printk in the whole cpuidle/{cpuidle,sysfs}.c
>files, even on error paths. Also in the cpuidle specific parts of
>drivers/acpi/processor_idle.c some debug printks may help for 
>future bug
>reports. It is very hard to guess what happened...
>

This looks like cpuidle and kobject interaction.
The latest oops looks different from the original one. Latest one is a
warn_on in lib/kref.c:43

We (Me or Shaohua) will take a deeper look at get back on this.

Thanks,
Venki

ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dʇڙë,j\a­¢f£¢·hšïêÿ‘êçz_è®\x03(­éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨è­Ú&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥