Subject: Re: sa/pthread smp concurrency (includes patch)
To: Frank van der Linden <firstname.lastname@example.org>
From: Stephan Uphoff <email@example.com>
Date: 07/14/2003 17:38:01
Frank van der Linden wrote:
> On Mon, Jul 14, 2003 at 12:25:17PM -0400, Thor Lancelot Simon wrote:
> > I don't entirely agree. The features added by this patch are necessary
> > to address the issues with SA in combination with SMP, are they not?
> Stephan's patch addresses the crash issues. Christian, in addition,
> made some scheduling and affinity improvements. They're not needed
> to avoid the SMP+SA crashes.
Just for clarification:
Christian's and my patches are fundamentally different.
My patch targets problems with low memory conditions for single
processor systems AND allowing SA to work on SMP.
My patch does NOT allow user space concurrency.
( Only one CPU per process is used for userspace processing)
My patch is NOT intended to be a final solution - just a temporary
fix until a new infrastructure is in place.
I think it would be good idea to check in my patch to make pthreads work
more reliable out of the box for current users.
I am working with Nathan to resolve some architectural problems with SA
that show up with multiple virtual CPUs (user space concurrency).
As soon as we agree on fixes (Or Nathan just trusts me to do the right thing
I will try to get a new SA kernel infrastructure in place.
However this will require quite a bit of work - so don't expect it to show
up next week.