Re: [git] CFS-devel, group scheduler, fixes

[Willy Tarreau <w@1wt.eu>]

On Thu, Sep 20, 2007 at 09:15:07AM +0200, Mike Galbraith wrote:
> But, I did just manage to trigger some horrid behavior, and log it.  I
> modified the kernel to print task's actual tree key instead of their
> current vruntime, and was watching that while make -j2 was running (and
> not seeing anything very interesting), when on a lark, I restarted
> SuSE's system updater thingy.  That beast chews 100% CPU for so long at
> startup that I long ago got annoyed, and changed it to run at nice 19.
> Anyway, when it started, interactivity went to hell in the proverbial
> hand-basket, and the sched_debug log shows some interesting results..
> like spread0 hitting -13659412644, and cc1 being keyed at -3867063305.
> 
> cpu#0, 2992.608 MHz
>   .nr_running                    : 4
>   .load                          : 4096
>   .nr_switches                   : 1105882
>   .nr_load_updates               : 735146
>   .nr_uninterruptible            : 4294966399

(...)

> cpu#1, 2992.608 MHz
>   .nr_running                    : 5
>   .load                          : 6208
>   .nr_switches                   : 1012995
>   .nr_load_updates               : 747540
>   .nr_uninterruptible            : 897

I don't know if this is relevant, but 4294966399 in nr_uninterruptible
for cpu#0 equals -897, exactly the negation of cpu1.nr_uninterruptible.
I don't know if this rings a bell for someone or if it's a completely
useless comment, but just in case...

HTH,
Willy