Regression (bisected): Lots of wakeups when using battery

Regression (bisected): Lots of wakeups when using battery

[Jaime Velasco Juan]

Hi, testing 2.6.35-rc5 I noticed that when my laptop is on battery,
there are bursts of wakeups (they seem random, but usually happens once
each 5 sec. or so; they cause some audible noise). If the laptop is plugged
it behaves well. In single user mode, there are usually 2 wakeups/sec with
the system idle, but on battery an with the bug, it goes to more than 100 w/s.

Bisection points to:

commit 3310d4d38fbc514e7b18bd3b1eea8effdd63b5aa
Author: Peter Zijlstra <peterz@infradead.org>
Date:   Thu Jun 17 18:02:37 2010 +0200

    nohz: Fix nohz ratelimit
    
    Chris Wedgwood reports that 39c0cbe (sched: Rate-limit nohz) causes a
    serial console regression, unresponsiveness, and indeed it does. The
    reason is that the nohz code is skipped even when the tick was already
    stopped before the nohz_ratelimit(cpu) condition changed.
    
    Move the nohz_ratelimit() check to the other conditions which prevent
    long idle sleeps.

Reverting said commit seems to fix the issue.

Regards

Re: Regression (bisected): Lots of wakeups when using battery

[Mike Galbraith]

(corrects cc)

On Fri, 2010-07-30 at 22:45 +0100, Jaime Velasco Juan wrote:
> Hi, testing 2.6.35-rc5 I noticed that when my laptop is on battery,
> there are bursts of wakeups (they seem random, but usually happens once
> each 5 sec. or so; they cause some audible noise). If the laptop is plugged
> it behaves well. In single user mode, there are usually 2 wakeups/sec with
> the system idle, but on battery an with the bug, it goes to more than 100 w/s.
> 
> Bisection points to:
> 
> commit 3310d4d38fbc514e7b18bd3b1eea8effdd63b5aa
> Author: Peter Zijlstra <peterz@infradead.org>
> Date:   Thu Jun 17 18:02:37 2010 +0200
> 
>     nohz: Fix nohz ratelimit
>     
>     Chris Wedgwood reports that 39c0cbe (sched: Rate-limit nohz) causes a
>     serial console regression, unresponsiveness, and indeed it does. The
>     reason is that the nohz code is skipped even when the tick was already
>     stopped before the nohz_ratelimit(cpu) condition changed.
>     
>     Move the nohz_ratelimit() check to the other conditions which prevent
>     long idle sleeps.
> 
> Reverting said commit seems to fix the issue.

And if you revert that, using a serial console becomes.. unpleasant.

The whole ratelimit nohz thing has been dropped for now.  It's a nice
little optimization for high frequency switchers on an underutilized
box, but has caused more irritation than it's worth.

	-Mike