lkml.org 
[lkml]   [2000]   [Aug]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
Subject[patch] sched-2.4.0-test8-B0

On Tue, 29 Aug 2000, David S. Miller wrote:

> i agree fully. The attached patch against test8-pre1 changes
> schedule_idle() from inline to FASTCALL - things look much nicer with this
> one applied.
>
> Ingo, the local flags stuff on Sparc, remember?

ouch, i thought this has been taken care of. The attached
sched-2.4.0-test8-B0 patch gets rid of flags-passing in reschedule_idle().
(the patch is relative to test8-pre1 + the previous scheduler patch) While
i liked smp_send_reschedule() being executed in parallel (it stresses the
APIC code harder), putting it into the critical section is just a small
increase of critical section length and the code gets cleaner (and
slightly faster) as a result. Patch applies, compiles, works here just
fine. This should solve your problem, right?

Ingo
--- linux/kernel/sched.c.orig Tue Aug 29 12:30:18 2000
+++ linux/kernel/sched.c Tue Aug 29 12:33:32 2000
@@ -213,9 +213,9 @@
* This function must be inline as anything that saves and restores
* flags has to do so within the same register window on sparc (Anton)
*/
-static FASTCALL(void reschedule_idle(struct task_struct *, unsigned long));
+static FASTCALL(void reschedule_idle(struct task_struct * p));

-static void reschedule_idle(struct task_struct * p, unsigned long flags)
+static void reschedule_idle(struct task_struct * p)
{
#ifdef CONFIG_SMP
int this_cpu = smp_processor_id();
@@ -286,7 +286,6 @@
goto preempt_now;
}

- spin_unlock_irqrestore(&runqueue_lock, flags);
return;

send_now_idle:
@@ -298,12 +297,10 @@
if ((tsk->processor != current->processor) && !tsk->need_resched)
smp_send_reschedule(tsk->processor);
tsk->need_resched = 1;
- spin_unlock_irqrestore(&runqueue_lock, flags);
return;

preempt_now:
tsk->need_resched = 1;
- spin_unlock_irqrestore(&runqueue_lock, flags);
/*
* the APIC stuff can go outside of the lock because
* it uses no task information, only CPU#.
@@ -318,7 +315,6 @@
tsk = cpu_curr(this_cpu);
if (preemption_goodness(tsk, p, this_cpu) > 1)
tsk->need_resched = 1;
- spin_unlock_irqrestore(&runqueue_lock, flags);
#endif
}

@@ -367,9 +363,7 @@
if (task_on_runqueue(p))
goto out;
add_to_runqueue(p);
- reschedule_idle(p, flags); // spin_unlocks runqueue
-
- return;
+ reschedule_idle(p);
out:
spin_unlock_irqrestore(&runqueue_lock, flags);
}
@@ -482,10 +476,9 @@
* current process as well.)
*/
running_again:
- if (prev == idle_task(smp_processor_id()))
- goto out_unlock;
- reschedule_idle(prev, flags); // spin_unlocks runqueue
- return;
+ if (prev != idle_task(smp_processor_id()))
+ reschedule_idle(prev);
+ goto out_unlock;
#endif /* CONFIG_SMP */
}
\
 
 \ /
  Last update: 2005-03-22 12:38    [W:0.072 / U:2.432 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site