From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx2.suse.de ([195.135.220.15]:42640 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751479AbXBRRIX (ORCPT ); Sun, 18 Feb 2007 12:08:23 -0500 From: Andi Kleen Subject: Re: Which architectures need to sync vmalloc mappings between processes? Date: Sun, 18 Feb 2007 18:08:13 +0100 References: <45D7E438.4020902@goop.org> In-Reply-To: <45D7E438.4020902@goop.org> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200702181808.13576.ak@suse.de> Sender: linux-arch-owner@vger.kernel.org To: Jeremy Fitzhardinge Cc: linux-arch@vger.kernel.org, Linux Kernel Mailing List List-ID: > On i386 with PAE, this is not necessary because the kernel's mappings > are shared between all processes anyway, so it would be a no-op. > However, non-PAE i386 has a separate kernel mapping for each process, > and so needs to sync them - typically lazily on faults, but > vmalloc_sync_all exists to allow mass syncing when required. The only case that needs it right now on x86 is NMI handlers in modules to avoid early IRETs that could cause NMI recursion. That's a very obscure architecture specific quirk and I wouldn't expect any other architecture to suffer from this. -Andi