Re: [Lse-tech] Re: [ckrm-tech] Re: [PATCH 00/01] Move Exit Connectors

[Paul E. McKenney]

On Tue, Jan 17, 2006 at 09:55:06PM -0500, Lee Revell wrote:
> On Tue, 2006-01-17 at 18:49 -0800, Paul E. McKenney wrote:
> > - * softirq handlers will have completed, since in some kernels
> > + * softirq handlers will have completed, since in some kernels, these
> > + * handlers can run in process context, and can block.
> >   * 
> 
> I was under the impression that softirq handlers can run in process
> context in all kernels.  Specifically, in realtime variants softirqs
> always run in process context, and in mainline this only happens under
> high load.

We might be talking past each other on this one.  If I am not getting
too confused, it is possible to configure a mainline kernel so that
the load cannot rise high enough to force softirqs into process
context.  Although looking at 2.6.15, it appears that this would
require rebuilding after hand-editing the value of MAX_SOFTIRQ_RESTART,
which some might feel too-brutal of tweaking to be considered mere
"configuration".

In any case, the key point of the comment is that synchronize_sched()
is not guaranteed to wait for all pending softirq handlers to complete.
Does the comment make that sufficiently clear?

						Thanx, Paul
-
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/