Re: [PATCH][RFC] Proposal For A More Scalable Scheduler ...

Mike Fedyk (mfedyk@matchmail.com)
Tue, 30 Oct 2001 15:04:29 -0800


On Tue, Oct 30, 2001 at 09:02:54AM -0800, Davide Libenzi wrote:
> On Tue, 30 Oct 2001, Mike Fedyk wrote:
>
> > On Mon, Oct 29, 2001 at 09:38:07PM -0800, Davide Libenzi wrote:
> > > 2) My Linux Scheduler Stuff Page:
> > > http://www.xmailserver.org/linux-patches/lnxsched.html
> > >
> >
> > Anyone know if this is preempt safe? It's using processor specific lists,
> > and might not be.
>
> Processor specific lists ?
> The mss scheduler patch in for x86 but it's trivial ( about 10 lines of
> code ) to port it to other arcs.
>

The proposed implementation uses a runqueue-per-cpu scheduler where,
inside each CPU, the scheduler code is exactly the same of the current one.
The big runqueue_lock has been substituted by locks that protects CPU run
queues.
By having separate run queues the length/cost of the goodness() loop
has been divided by N ( N == number of CPUs ) and the presence of
per-runqueue locks gives the scheduler a full parallelism between the CPUs.

-------------

Looking at this again, it probably is preempt safe... I probably merged it
wrong.

I'll try to fit it into my next kernel...
-
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/