Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-16 Thread Srikar Dronamraju
* Peter Zijlstra [2012-08-13 09:51:13]: > On Fri, 2012-08-10 at 21:54 +0530, Srikar Dronamraju wrote: > > > This change worked well on the 2 node machine > > but on the 8 node machine it hangs with repeated messages > > > > Pid: 60935, comm: numa01 Tainted: GW3.5.0-numasched_v2_020

Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-16 Thread Srikar Dronamraju
* Peter Zijlstra [2012-08-13 10:11:28]: > On Mon, 2012-08-13 at 09:51 +0200, Peter Zijlstra wrote: > > On Fri, 2012-08-10 at 21:54 +0530, Srikar Dronamraju wrote: > > > > > This change worked well on the 2 node machine > > > but on the 8 node machine it hangs with repeated messages > > > > > >

Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-13 Thread Peter Zijlstra
On Mon, 2012-08-13 at 09:51 +0200, Peter Zijlstra wrote: > On Fri, 2012-08-10 at 21:54 +0530, Srikar Dronamraju wrote: > > > This change worked well on the 2 node machine > > but on the 8 node machine it hangs with repeated messages > > > > Pid: 60935, comm: numa01 Tainted: GW3.5.0-n

Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-13 Thread Peter Zijlstra
On Fri, 2012-08-10 at 21:54 +0530, Srikar Dronamraju wrote: > This change worked well on the 2 node machine > but on the 8 node machine it hangs with repeated messages > > Pid: 60935, comm: numa01 Tainted: GW3.5.0-numasched_v2_020812+ #4 > Call Trace: > [] ? rcu_check_callback s+0x

Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-10 Thread Srikar Dronamraju
> --- > --- a/include/linux/sched.h > +++ b/include/linux/sched.h > @@ -1539,6 +1539,7 @@ struct task_struct { > #ifdef CONFIG_SMP > u64 node_stamp; /* migration stamp */ > unsigned long numa_contrib; > + struct callback_head numa_work; > #endif /* CONFIG_SMP */

Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-08 Thread Peter Zijlstra
On Tue, 2012-08-07 at 22:49 +0530, Srikar Dronamraju wrote: > Are you referring to this the commit 158e1645e (trim task_work: get rid of > hlist) No, to something like the below.. > I am also able to reproduce this on another 8 node machine too. Ship me one ;-) > Just to update, I had to rever

Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-07 Thread Srikar Dronamraju
* Peter Zijlstra [2012-08-07 15:52:48]: > On Tue, 2012-08-07 at 18:03 +0530, Srikar Dronamraju wrote: > > Hi, > > > > INFO: rcu_sched self-detected stall on CPU { 7} (t=105182911 jiffies) > > Pid: 5173, comm: qpidd Tainted: GW3.5.0numasched_v2_020812+ #1 > > Call Trace: > >[

Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-07 Thread Srikar Dronamraju
* John Stultz [2012-08-07 10:08:51]: > On 08/07/2012 05:33 AM, Srikar Dronamraju wrote: > >Hi, > > > >I saw this while I was running the 2nd August -tip kernel + Peter's > >numasched patches. > > > >Top showed load average to be 240, there was one cpu (cpu 7) which > >showed 100% while all other

Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-07 Thread John Stultz
On 08/07/2012 05:33 AM, Srikar Dronamraju wrote: Hi, I saw this while I was running the 2nd August -tip kernel + Peter's numasched patches. Top showed load average to be 240, there was one cpu (cpu 7) which showed 100% while all other cpus were idle. The system showed some sluggishness. Befor

Re: rcu stalls seen with numasched_v2 patches applied.

2012-08-07 Thread Peter Zijlstra
On Tue, 2012-08-07 at 18:03 +0530, Srikar Dronamraju wrote: > Hi, > > I saw this while I was running the 2nd August -tip kernel + Peter's > numasched patches. > > Top showed load average to be 240, there was one cpu (cpu 7) which > showed 100% while all other cpus were idle. The system showed

rcu stalls seen with numasched_v2 patches applied.

2012-08-07 Thread Srikar Dronamraju
Hi, I saw this while I was running the 2nd August -tip kernel + Peter's numasched patches. Top showed load average to be 240, there was one cpu (cpu 7) which showed 100% while all other cpus were idle. The system showed some sluggishness. Before I saw this I ran Andrea's autonuma benchmark cou