[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <200712290255.40233.ak@suse.de>
Date: Sat, 29 Dec 2007 02:55:39 +0100
From: Andi Kleen <ak@...e.de>
To: Mike Travis <travis@....com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Christoph Lameter <clameter@....com>, linux-mm@...ck.org,
linux-kernel@...r.kernel.org,
Rusty Russell <rusty@...tcorp.com.au>, tglx@...utronix.de,
mingo@...hat.com
Subject: Re: [PATCH 05/10] x86_64: Use generic percpu
On Friday 28 December 2007 23:05:05 Mike Travis wrote:
> Andi Kleen wrote:
> > On Friday 28 December 2007 01:10:51 travis@....com wrote:
> >> x86_64 provides an optimized way to determine the local per cpu area
> >> offset through the pda and determines the base by accessing a remote
> >> pda.
> >
> > And? The rationale for this patch seems to be incomplete.
> >
> > As far as I can figure out you're replacing an optimized percpu
> > implementation which a dumber generic one. Which needs
> > at least some description why.
>
> The specific intent for the next wave of changes coming are to reduce
[...] That should be in the changelog of the patch.
Anyways the difference between the x86 percpu.h and the generic one is
that x86-64 uses a short cut through the PDA to get the current cpu
offset for the current CPU case. The generic one goes through
smp_processor_id()->array reference instead.
I would request that this optimization is not being removed
without suitable replacement in the same patchkit.
-Andi
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists