Re: [PATCH] sched: shorten setting the allowed cpu mask of task

[Yong Zhang]

On Mon, May 09, 2011 at 08:52:53PM +0800, Hillf Danton wrote:
> On Mon, May 9, 2011 at 12:39 PM, Yong Zhang <yong.zhang0@xxxxxxxxx> wrote:
> > On Fri, May 6, 2011 at 8:52 PM, Hillf Danton <dhillf@xxxxxxxxx> wrote:
> >> When setting the allowed cpu mask for a given task, if the task is
> >> already bound to certain cpu, after checking the validity of the new
> >
> > Maybe we don't need to restrict it only on task bound to certain cpu.
> >
> Hi Yong
> 
> The original code guards, I guess, casual change in the mask of
> allowed CPUs, if bounded,
> for tasks such as the workers of work queue. So the restriction looks necessary.

Yeah, that is true; but I don't think we need to go ahead for
unbounded task if cpu_allowed will not be changed.

My thought is like below:

---
Subject: [PATCH] sched: avoid going ahead if cpu_allowed will not be changed

If cpumask_equal(&p->cpus_allowed, new_mask) is true, seems
there is no reason to prevent set_cpus_allowed_ptr() return
directly.

Signed-off-by: Yong Zhang <yong.zhang0@xxxxxxxxx>
---
 kernel/sched.c |    6 ++++--
 1 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/sched.c b/kernel/sched.c
index da93381..56bc1fa 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -5946,13 +5946,15 @@ int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
 
 	rq = task_rq_lock(p, &flags);
 
+	if (cpumask_equal(&p->cpus_allowed, new_mask))
+		goto out;
+
 	if (!cpumask_intersects(new_mask, cpu_active_mask)) {
 		ret = -EINVAL;
 		goto out;
 	}
 
-	if (unlikely((p->flags & PF_THREAD_BOUND) && p != current &&
-		     !cpumask_equal(&p->cpus_allowed, new_mask))) {
+	if (unlikely((p->flags & PF_THREAD_BOUND) && p != current)) {
 		ret = -EINVAL;
 		goto out;
 	}
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/