From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1750706AbWFEHtj (ORCPT ); Mon, 5 Jun 2006 03:49:39 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1750713AbWFEHtj (ORCPT ); Mon, 5 Jun 2006 03:49:39 -0400 Received: from gate.crashing.org ([63.228.1.57]:42453 "EHLO gate.crashing.org") by vger.kernel.org with ESMTP id S1750706AbWFEHti (ORCPT ); Mon, 5 Jun 2006 03:49:38 -0400 Subject: Re: [RFC][PATCH] request_irq(...,SA_BOOTMEM); From: Benjamin Herrenschmidt To: Andrew Morton Cc: linux-kernel@vger.kernel.org, mingo@elte.hu, tglx@linutronix.de, torvalds@osdl.org In-Reply-To: <20060605003127.fc1ea37a.akpm@osdl.org> References: <1149486009.8543.42.camel@localhost.localdomain> <1149491309.8543.54.camel@localhost.localdomain> <20060605003127.fc1ea37a.akpm@osdl.org> Content-Type: text/plain Date: Mon, 05 Jun 2006 17:48:10 +1000 Message-Id: <1149493691.8543.57.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 > I don't immediately see anything in there which would prevent us from > running these: > > vfs_caches_init_early(); > cpuset_init_early(); > mem_init(); > kmem_cache_init(); > setup_per_cpu_pageset(); > > just after sort_main_extable(). > > But things will explode ;) > > I suggest you run up a patch, test it on whatever machines you have, send > it over and I'll do the same. But please make sure it has a config option > to restore the old sequence for now. a) So people can work out that it was > this patch which broke things and b) so it doesn't adversely affect testing > of other things too much. Good ideas. I'll give these things a spin. One thing that may explode is that all that code is running with local_irq_disable() (since local irqs aren't enabled before init_IRQ()) and that means possible use of some types of semaphores may trigger warn-on's (or worse as I think some implementations of down_read() might even force-enable irqs). But there is no fundamental reasons to do so ... that's the trick :) If that happens, those semaphores are still ok as they should never get into contention that early. Anyway, I'll give it a spin on ppc and maybe x86 if I can find a victim to test on here, and will send something. Cheers, Ben.