Re: [RFC PATCH] sched_domains: Make SD_NODE_INIT per-arch

[Matthew Dobson]

On Thu, 2004-09-30 at 12:23, Andrew Morton wrote:
> Matthew Dobson <colpatch@xxxxxxxxxx> wrote:
> >
> > I would like to try to get this in before then, unless this will really
> >  make things difficult for you.
> 
> It's about three weeks late for 2.6.9.  I already have a string of CPU
> scheduler patches awaiting the 2.6.10 stream and once we're at -rc2 we
> really should only be looking at bugfixes.

Yeah, that's entirely my fault for slacking on sending this out...  I
should have sent this a while ago.  It is a small portion of some larger
sched_domains changes that I am working on, but at some point I realized
my larger changeset will be far more controversial and have a much
larger impact than some of the smaller bits, as well as not being ready
for prime time yet.  Plus, like I said earlier, this allows
arch-specific tweaking with minimal intrusiveness from the application
of this patch forward.


> Grumble, mutter..  it looks like one of those "if it compiled, it works"
> things.  Problem is, any time anyone touches that particular piece of the
> kernel, half the architectures stop compiing.

It *should* be.  I'd be quite happy if you just picked it up in -mm to
assure it far wider testing.  I've compiled and booted it on x86, x86_64
& ppc64.  I've got no access to ia64 right now, or I'd test it there. 
But the patch *will* spit out #errors for any arch that doesn't have
SD_NODE_INIT defined if they also have NUMA defined.  I'm don't know of
anyone else (ie: *not* x86, x86_64, ppc64 & ia64) that is building NUMA
kernels, but if they are, it's a trivial patch to their
include/asm/topology.h to make the arch build.

Of course, the ultimate decision is yours, Andrew...

-Matt

-
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/