Re: readprofile from 2.5.25 web server benchmark

[Andrew Morton <akpm@zip.com.au>]

kuznet@ms2.inr.ac.ru wrote:
> 
> Hello!
> 
> > --- 2.5.25/kernel/timer.c~timer-speedup       Tue Jul  9 00:04:33 2002
> > +++ 2.5.25-akpm/kernel/timer.c        Tue Jul  9 00:06:09 2002
> > @@ -211,6 +211,9 @@ int mod_timer(struct timer_list *timer,
> >       int ret;
> >       unsigned long flags;
> >
> > +     if (timer_pending(timer) && timer->expires == expires)
> > +             return 1;
> > +
> >       spin_lock_irqsave(&timerlist_lock, flags);
> >       timer->expires = expires;
> >       ret = detach_timer(timer);
> 
> Does this not lose timer sometimes?
> 

Well I've thought about it long and hard, but I can't see a
hole.  But there probably is one.  Somewhere.

It reduces mod_timer() load by 30% in a netperf run, so it's
worth thinking about.

-