LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
* Why one of the 2 patches to fix "autogroup" sched problem on UP machines is missing in -rc3?
@ 2011-02-01  9:05 Kazuhisa Ichikawa
  2011-02-01 10:46 ` Yong Zhang
  0 siblings, 1 reply; 3+ messages in thread
From: Kazuhisa Ichikawa @ 2011-02-01  9:05 UTC (permalink / raw)
  To: Yong Zhang; +Cc: Ingo Molnar, linux-kernel

Hi Yong,

Now 2.6.38-rc3 is out, and I think one of the 2 patches you wrote to
fix "autogroup" sched problem on UP machines is missing there.

More precisely, your first patch for kernel/sched.c (titled: sched:
tg->se->load should be initialised to tg->shares) is missing, while
the second one for kernel/sched_fair.c (titled: sched: Fix poor
interactivity on UP systems due to group scheduler nice tune bug) is
in -rc3.

So what happened to the first patch? Was it superseded and made
obsolete by the second patch?

# Sorry for not replying to your original post. I'm not subscribing to
linux kernel mailing list.

Regards,

- Kazuhisa

^ permalink raw reply	[flat|nested] 3+ messages in thread
* Re: Why one of the 2 patches to fix "autogroup" sched problem on UP machines is missing in -rc3?
  2011-02-01  9:05 Why one of the 2 patches to fix "autogroup" sched problem on UP machines is missing in -rc3? Kazuhisa Ichikawa
@ 2011-02-01 10:46 ` Yong Zhang
  2011-02-01 11:19   ` Kazuhisa Ichikawa
  0 siblings, 1 reply; 3+ messages in thread
From: Yong Zhang @ 2011-02-01 10:46 UTC (permalink / raw)
  To: Kazuhisa Ichikawa; +Cc: Ingo Molnar, linux-kernel

On Tue, Feb 01, 2011 at 06:05:23PM +0900, Kazuhisa Ichikawa wrote:
> Hi Yong,
> 
> Now 2.6.38-rc3 is out, and I think one of the 2 patches you wrote to
> fix "autogroup" sched problem on UP machines is missing there.

Actually the merged one hide the initialising problem.
Because when we enqueue a process to an empty group, update_cfs_shares()
will be called, thus tg->se->load is updated.
enqueue_task_fair()
  update_cfs_shares()
    reweight_entity()
      update_load_set()

> 
> More precisely, your first patch for kernel/sched.c (titled:?sched:
> tg->se->load should be initialised to tg->shares) is missing, while
> the second one for kernel/sched_fair.c (titled:?sched: Fix poor
> interactivity on UP systems due to group scheduler nice tune bug) is
> in -rc3.

See my above answer.

> 
> So what happened to the first patch? Was it superseded and made
> obsolete by the second patch?

Yes.

> 
> # Sorry for not replying to your original post. I'm not subscribing to
> linux kernel mailing list.

Thanks for taking care of this issue :)

Thanks,
Yong

^ permalink raw reply	[flat|nested] 3+ messages in thread
* Re: Why one of the 2 patches to fix "autogroup" sched problem on UP machines is missing in -rc3?
  2011-02-01 10:46 ` Yong Zhang
@ 2011-02-01 11:19   ` Kazuhisa Ichikawa
  0 siblings, 0 replies; 3+ messages in thread
From: Kazuhisa Ichikawa @ 2011-02-01 11:19 UTC (permalink / raw)
  To: Yong Zhang; +Cc: Ingo Molnar, linux-kernel

> Actually the merged one hide the initialising problem.
> Because when we enqueue a process to an empty group, update_cfs_shares()
> will be called, thus tg->se->load is updated.
> enqueue_task_fair()
>  update_cfs_shares()
>    reweight_entity()
>      update_load_set()

Ah, I see. Thanks for the clarification!

- Kazuhisa


2011/2/1 Yong Zhang <yong.zhang0@gmail.com>:
> On Tue, Feb 01, 2011 at 06:05:23PM +0900, Kazuhisa Ichikawa wrote:
>> Hi Yong,
>>
>> Now 2.6.38-rc3 is out, and I think one of the 2 patches you wrote to
>> fix "autogroup" sched problem on UP machines is missing there.
>
> Actually the merged one hide the initialising problem.
> Because when we enqueue a process to an empty group, update_cfs_shares()
> will be called, thus tg->se->load is updated.
> enqueue_task_fair()
>  update_cfs_shares()
>    reweight_entity()
>      update_load_set()
>
>>
>> More precisely, your first patch for kernel/sched.c (titled:?sched:
>> tg->se->load should be initialised to tg->shares) is missing, while
>> the second one for kernel/sched_fair.c (titled:?sched: Fix poor
>> interactivity on UP systems due to group scheduler nice tune bug) is
>> in -rc3.
>
> See my above answer.
>
>>
>> So what happened to the first patch? Was it superseded and made
>> obsolete by the second patch?
>
> Yes.
>
>>
>> # Sorry for not replying to your original post. I'm not subscribing to
>> linux kernel mailing list.
>
> Thanks for taking care of this issue :)
>
> Thanks,
> Yong
>

^ permalink raw reply	[flat|nested] 3+ messages in thread
end of thread, other threads:[~2011-02-01 11:19 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-02-01  9:05 Why one of the 2 patches to fix "autogroup" sched problem on UP machines is missing in -rc3? Kazuhisa Ichikawa
2011-02-01 10:46 ` Yong Zhang
2011-02-01 11:19   ` Kazuhisa Ichikawa

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).