To: Arjan van de Ven <arjan@intel.linux.com>
Cc: linux-kernel@vger.kernel.org, akpm@osdl.org, ak@suse.de
Subject: Re: [Patch 1/3] prefetch the mmap_sem in the fault path
References: <1140686152.2972.25.camel@laptopd505.fenrus.org>
	<1140687007.4672.6.camel@laptopd505.fenrus.org>
From: Jes Sorensen <jes@sgi.com>
Date: 23 Feb 2006 07:29:00 -0500
In-Reply-To: <1140687007.4672.6.camel@laptopd505.fenrus.org>
Message-ID: <yq0zmkigsxf.fsf@jaguar.mkp.net>
User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.4
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1056
Lines: 19

>>>>> "Arjan" == Arjan van de Ven <arjan@intel.linux.com> writes:

Arjan> In a micro-benchmark that stresses the pagefault path, the
Arjan> down_read_trylock on the mmap_sem showed up quite high on the
Arjan> profile. Turns out this lock is bouncing between cpus quite a
Arjan> bit and thus is cache-cold a lot. This patch prefetches the
Arjan> lock (for write) as early as possible (and before some other
Arjan> somewhat expensive operations). With this patch, the
Arjan> down_read_trylock basically fell out of the top of profile.

Out of curiousity, how big was the box used for testing? It might be
worth investigating if anything can be done to reduce the number of
times that lock is taken in the first place.

After all, what's a pain on a 4-way tends to be an utter nightmare on
a 16-way ;(

Cheers,
Jes
-
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/