lkml.org 
[lkml]   [2014]   [Apr]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: 3.15-rc2: longhaul cpufreq stalls tasks for 120s+
From
On 25 April 2014 01:14, Srivatsa S. Bhat
<srivatsa.bhat@linux.vnet.ibm.com> wrote:
> On 04/25/2014 12:33 AM, Meelis Roos wrote:
>>> I see traces of mutex_lock_slowpath() etc in your logs.. Can you please
>>> enable lockdep and sleep-inside-atomic-section check and let us know if
>>> it complains? Specifically, enable these config options:
>>>
>>> CONFIG_LOCKDEP_SUPPORT=y
>>> CONFIG_DEBUG_RT_MUTEXES=y
>>> CONFIG_DEBUG_PI_LIST=y
>>> CONFIG_DEBUG_SPINLOCK=y
>>> CONFIG_DEBUG_MUTEXES=y
>>> CONFIG_DEBUG_LOCK_ALLOC=y
>>> CONFIG_PROVE_LOCKING=y
>>> CONFIG_LOCKDEP=y
>>> CONFIG_DEBUG_ATOMIC_SLEEP=y
>>
>> At boot, nothing special was in dmesg. After running cpufreq-info (which
>> hangs indefinitely), I get the following dmesg with hangs and lock info:
>
>
> Thank you for the logs, but unfortunately they don't give us much
> further insight into this issue.
>
> Just to be sure that our suspicion is right, can you try reverting
> commit 12478cf0c55 (cpufreq: Make sure frequency transitions are
> serialized) and check whether reverting the patch solves the problem?

You need to revert these three patches in the mentioned order.

236a980 cpufreq: Make cpufreq_notify_transition &
cpufreq_notify_post_transition static
8fec051 cpufreq: Convert existing drivers to use
cpufreq_freq_transition_{begin|end}
12478cf cpufreq: Make sure frequency transitions are serialized


\
 
 \ /
  Last update: 2014-04-25 09:21    [W:0.816 / U:0.028 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site