> That's changing schedule(). Seems that I'd ruled out that
> option prematurely. As current->preempt_count() and PREEMPT_ACTIVE
> can both evaluate to constant zero if CONFIG_PREEMPT=n, it can
> be done ifdeflessly.
>
> Everything happens inside rq->lock. Looks solid to me.
It is solid - I did not just invent that approach, it was how we have
always done it until around 2.5.6-pre (Ingo sent a patch to change it).
The 2.4 patches do it this way (take a look) and 2.5 before the change
obviously worked like this.
It hasn't shown any problems in ~6 months of use.
I'll cook up a patch to do it, but I'd like to hear Linus opinion ...
Robert Love
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/