From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758823AbYAUOuQ (ORCPT ); Mon, 21 Jan 2008 09:50:16 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753720AbYAUOuC (ORCPT ); Mon, 21 Jan 2008 09:50:02 -0500 Received: from mx2.mail.elte.hu ([157.181.151.9]:41730 "EHLO mx2.mail.elte.hu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753475AbYAUOuA (ORCPT ); Mon, 21 Jan 2008 09:50:00 -0500 Date: Mon, 21 Jan 2008 15:49:23 +0100 From: Ingo Molnar To: Mel Gorman Cc: KOSAKI Motohiro , linux-mm@kvack.org, linux-kernel@vger.kernel.org, apw@shadowen.org Subject: Re: [PATCH 0/2] Relax restrictions on setting CONFIG_NUMA on x86 Message-ID: <20080121144923.GA8959@elte.hu> References: <20080118153529.12646.5260.sendpatchset@skynet.skynet.ie> <20080121093702.8FC2.KOSAKI.MOTOHIRO@jp.fujitsu.com> <20080121143508.GA8485@csn.ul.ie> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20080121143508.GA8485@csn.ul.ie> User-Agent: Mutt/1.5.17 (2007-11-01) X-ELTE-VirusStatus: clean X-ELTE-SpamScore: -1.5 X-ELTE-SpamLevel: X-ELTE-SpamCheck: no X-ELTE-SpamVersion: ELTE 2.0 X-ELTE-SpamCheck-Details: score=-1.5 required=5.9 tests=BAYES_00 autolearn=no SpamAssassin version=3.2.3 -1.5 BAYES_00 BODY: Bayesian spam probability is 0 to 1% [score: 0.0000] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org * Mel Gorman wrote: > > I think this patch become easy to the porting of fakenuma. > > It would be great if that was available, particularly if it could fake > memoryless nodes as that is a place where we've found a few > difficult-to-reproduce bugs. yeah. Your previous patch (see below) had build problems - are those resolved meanwhile? Ingo -----------> Subject: x86: allow any x86 sub-architecture type to set CONFIG_NUMA From: Mel Gorman While there are a limited number of x86 sub-architecture types that can really support NUMA, there is nothing stopping other machines booting that type of kernel. The fact that X86_GENERICARCH can set NUMA currently is an indicator of that. This restriction only limits potential testing coverage. This patch allows any sub-architecture to set CONFIG_NUMA if they wish. Signed-off-by: Mel Gorman Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 2 +- include/asm-x86/acpi.h | 1 + include/linux/acpi.h | 1 - 3 files changed, 2 insertions(+), 2 deletions(-) Index: linux-x86.q/arch/x86/Kconfig =================================================================== --- linux-x86.q.orig/arch/x86/Kconfig +++ linux-x86.q/arch/x86/Kconfig @@ -816,7 +816,7 @@ config X86_PAE config NUMA bool "Numa Memory Allocation and Scheduler Support (EXPERIMENTAL)" depends on SMP - depends on X86_64 || (X86_32 && (X86_NUMAQ || (X86_SUMMIT || X86_GENERICARCH) && ACPI) && EXPERIMENTAL) + depends on X86_64 || (X86_32 && ACPI && EXPERIMENTAL) default n if X86_PC default y if (X86_NUMAQ || X86_SUMMIT) help Index: linux-x86.q/include/asm-x86/acpi.h =================================================================== --- linux-x86.q.orig/include/asm-x86/acpi.h +++ linux-x86.q/include/asm-x86/acpi.h @@ -79,6 +79,7 @@ int __acpi_release_global_lock(unsigned :"0"(n_hi), "1"(n_lo)) #ifdef CONFIG_ACPI +#define NR_NODE_MEMBLKS MAX_NUMNODES extern int acpi_lapic; extern int acpi_ioapic; extern int acpi_noirq; Index: linux-x86.q/include/linux/acpi.h =================================================================== --- linux-x86.q.orig/include/linux/acpi.h +++ linux-x86.q/include/linux/acpi.h @@ -95,7 +95,6 @@ void acpi_table_print_madt_entry (struct /* the following four functions are architecture-dependent */ #ifdef CONFIG_HAVE_ARCH_PARSE_SRAT -#define NR_NODE_MEMBLKS MAX_NUMNODES #define acpi_numa_slit_init(slit) do {} while (0) #define acpi_numa_processor_affinity_init(pa) do {} while (0) #define acpi_numa_memory_affinity_init(ma) do {} while (0)