From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262049AbVGVHuY (ORCPT ); Fri, 22 Jul 2005 03:50:24 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262032AbVGVHuY (ORCPT ); Fri, 22 Jul 2005 03:50:24 -0400 Received: from dsl027-180-168.sfo1.dsl.speakeasy.net ([216.27.180.168]:18911 "EHLO sunset.davemloft.net") by vger.kernel.org with ESMTP id S262049AbVGVHuW (ORCPT ); Fri, 22 Jul 2005 03:50:22 -0400 Date: Fri, 22 Jul 2005 00:50:25 -0700 (PDT) Message-Id: <20050722.005025.26277081.davem@davemloft.net> To: ncunningham@cyclades.com Cc: linux-kernel@vger.kernel.org Subject: Re: [QN/PATCH] Why do some archs allocate stack via kmalloc, others via get_free_pages? From: "David S. Miller" In-Reply-To: <1122005477.3033.56.camel@localhost> References: <1122005477.3033.56.camel@localhost> X-Mailer: Mew version 4.2 on Emacs 21.4 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org From: Nigel Cunningham Date: Fri, 22 Jul 2005 14:11:17 +1000 > In making some modifications to Suspend, we've discovered that some > arches use kmalloc and others use get_free_pages to allocate the stack. > Is there a reason for the variation? If not, could the following patch > be considered for inclusion (tested on x86 only). Some platforms really need it to be page aligned (sparc32 sun4c needs to virtually map the resulting pages into a specific place, for example). But, for the ones that don't have this requirement, they want the cache coloring.