From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1160998AbXAaXXF (ORCPT ); Wed, 31 Jan 2007 18:23:05 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1161028AbXAaXXF (ORCPT ); Wed, 31 Jan 2007 18:23:05 -0500 Received: from www.osadl.org ([213.239.205.134]:42451 "EHLO mail.tglx.de" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1160998AbXAaXXD (ORCPT ); Wed, 31 Jan 2007 18:23:03 -0500 Subject: Re: [PATCH 14/23] clocksource: increase initcall priority From: Thomas Gleixner Reply-To: tglx@linutronix.de To: Daniel Walker Cc: David Brownell , linux-kernel@vger.kernel.org, john stultz , Ingo Molnar In-Reply-To: <1170283629.9781.163.camel@imap.mvista.com> References: <20070131194311.4B8381FD833@adsl-69-226-248-13.dsl.pltn13.pacbell.net> <1170283629.9781.163.camel@imap.mvista.com> Content-Type: text/plain Date: Thu, 01 Feb 2007 00:23:46 +0100 Message-Id: <1170285826.29240.227.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.6.1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Daniel, On Wed, 2007-01-31 at 14:47 -0800, Daniel Walker wrote: > > So don't assume any platform doesn't use clocksource initcalls. > > What does your OMAP clocksource do now ? I thought one of the changes > that you made was to have both 32k and mpu both registered .. It is up to the clocksource driver, when the clocksource_register() call is done. This may happen in early boot as well as after initializing some other things first. Johns clocksource code works with ARM which does the register call in timer_init() as well as with some other hardware which gets initialized late in the boot process. clocksource_initcall is simply superfluid. Adjusting the initcalls of a particular device to the point where it becomes available is a seperate topic. I'm not in the way of that and I think John's patch draft to solve the sound issue is heading into exactly this direction. We do not need another initcall level, especially not one which just aliases an existing one. The only thing we really want to control explicitely is when we actually switch away from jiffies_clocksource. But this does not need an artificial initcall at all. tglx