William Montgomery wrote:
>
> The scenario you describe _is_ fixed by Ingo's patch, see below:
>
> -----------------
> in linux/kernel/sched.c - reschedule_idle_slow
> int this_cpu = smp_processor_id();
> struct task_struct *tsk;
>
> - tsk = current;
> + tsk = cpu_curr(this_cpu);
> if (preemption_goodness(tsk, p, this_cpu) > 0)
> tsk->need_resched = 1;
> -----------------
>
> This eliminates the race and is a bug which should be fixed in 2.2.x.
>
> The latest lowlatency patch can be found at:
> http://www.redhat.com/~mingo/lowlatency-patches/lowlatency-2.2.15-C0
>
> Wm
The above code DOES fix the problem I was describing.
Your email seems to suggest the above code also fix some other problems?
If we choose the above way to fix the lost need_resched flag, we would
also
need to apply the similar fix to two other functions for the SMP version
:
1. smp_local_timer_interrupt()
2. update_process_times()
(I might miss one or two, but essentially all functions that a) can be
called
from interrupt context and b) need_resched flag is set to the CURRENT
process
need to be modified in a similar way.)
++++
If we were to fix *ONLY* the lost need_resched flag problem, I still
like
my fix a little better. However, if Ingo's fix has other benefits,
we probably should take his.
Jun
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Mar 23 2000 - 21:00:34 EST