Date: Sun, 19 Feb 2006 00:41:03 +0100
From: Sam Ravnborg <sam@ravnborg.org>
To: Andi Kleen <ak@suse.de>
Cc: linux-kernel@vger.kernel.org
Subject: Re: kbuild:
Message-ID: <20060218234103.GA9091@mars.ravnborg.org>
References: <20060217214855.GA5563@mars.ravnborg.org> <p73y80848qb.fsf@verdi.suse.de> <20060218223835.GA21068@mars.ravnborg.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20060218223835.GA21068@mars.ravnborg.org>
User-Agent: Mutt/1.5.11
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1442
Lines: 32

On Sat, Feb 18, 2006 at 11:38:36PM +0100, Sam Ravnborg wrote:
> On Sat, Feb 18, 2006 at 11:12:28PM +0100, Andi Kleen wrote:
> > Sam Ravnborg <sam@ravnborg.org> writes:
> > 
> > > I have moved the functionality of reference_init + reference_discarded
> > > to modpost to secure a much wider use of this check.
> > 
> > How much does that slow the build down?
> It obviously depends on number of modules / size of vmlinux.
> With x86_64 and defconfig + all oss drivers configured as modules the
> modpost stage takes around 0.2 sec in total. So this is down in the
> noise level. Building allmodconfig kernel atm and if I see > 2 sec
> difference with and without the check I will do a follow-up post.
> 
> Obviously the more modules with problems the longer time. So I will
> skip the warning messages in the measurements since I assume they will
> be taken care of.

Have figures for an allmodconfig now - this is a bit more than 1500
modules.
Running modpost alone takes 8,1 seconds without the checks and 8,9
seconds with the checks enabled (output disabled).

Considering that a a make allmodconfig with no updates takes
roughly 33 seconds the added overhead of 0,8 seconds is acceptable.

	Sam
-
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/