From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752661AbXDDXG0 (ORCPT ); Wed, 4 Apr 2007 19:06:26 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752662AbXDDXG0 (ORCPT ); Wed, 4 Apr 2007 19:06:26 -0400 Received: from netops-testserver-3-out.sgi.com ([192.48.171.28]:56718 "EHLO relay.sgi.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1752661AbXDDXGZ (ORCPT ); Wed, 4 Apr 2007 19:06:25 -0400 From: Christoph Lameter To: akpm@linux-foundation.org Cc: linux-ia64@vger.kernel.org, linux-kernel@vger.kernel.org, Martin Bligh , linux-mm@kvack.org, Andi Kleen , Christoph Lameter , Dave Hansen , KAMEZAWA Hiroyuki Message-Id: <20070404230624.20292.32311.sendpatchset@schroedinger.engr.sgi.com> In-Reply-To: <20070404230619.20292.4475.sendpatchset@schroedinger.engr.sgi.com> References: <20070404230619.20292.4475.sendpatchset@schroedinger.engr.sgi.com> Subject: [PATCH 2/4] x86_64: SPARSE_VIRTUAL 2M page size support Date: Wed, 4 Apr 2007 16:06:24 -0700 (PDT) Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org x86_64 implement SPARSE_VIRTUAL x86_64 is using 2M page table entries to map its 1-1 kernel space. We implement the virtual memmap also using 2M page table entries. So there is no difference at all to FLATMEM. Both schemes require a page table and a TLB for each 2MB. FLATMEM still references memory since the mem_map pointer itself a variable. SPARSE_VIRTUAL uses a constant for vmemmap. Thus no memory reference. SPARSE_VIRTUAL should be superior to even FLATMEM. With this SPARSEMEM becomes the most efficient way of handling virt_to_page, pfn_to_page and friends for UP, SMP and NUMA on x86_64. Signed-off-by: Christoph Lameter Index: linux-2.6.21-rc5-mm4/include/asm-x86_64/page.h =================================================================== --- linux-2.6.21-rc5-mm4.orig/include/asm-x86_64/page.h 2007-04-03 18:41:06.000000000 -0700 +++ linux-2.6.21-rc5-mm4/include/asm-x86_64/page.h 2007-04-03 18:41:59.000000000 -0700 @@ -128,6 +128,7 @@ extern unsigned long phys_base; VM_READ | VM_WRITE | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC) #define __HAVE_ARCH_GATE_AREA 1 +#define vmemmap ((struct page *)0xffffe20000000000UL) #include #include Index: linux-2.6.21-rc5-mm4/Documentation/x86_64/mm.txt =================================================================== --- linux-2.6.21-rc5-mm4.orig/Documentation/x86_64/mm.txt 2007-04-03 18:41:06.000000000 -0700 +++ linux-2.6.21-rc5-mm4/Documentation/x86_64/mm.txt 2007-04-03 18:41:59.000000000 -0700 @@ -9,6 +9,7 @@ ffff800000000000 - ffff80ffffffffff (=40 ffff810000000000 - ffffc0ffffffffff (=46 bits) direct mapping of all phys. memory ffffc10000000000 - ffffc1ffffffffff (=40 bits) hole ffffc20000000000 - ffffe1ffffffffff (=45 bits) vmalloc/ioremap space +ffffe20000000000 - ffffe2ffffffffff (=40 bits) virtual memory map ... unused hole ... ffffffff80000000 - ffffffff82800000 (=40 MB) kernel text mapping, from phys 0 ... unused hole ... Index: linux-2.6.21-rc5-mm4/arch/x86_64/Kconfig =================================================================== --- linux-2.6.21-rc5-mm4.orig/arch/x86_64/Kconfig 2007-04-03 18:41:06.000000000 -0700 +++ linux-2.6.21-rc5-mm4/arch/x86_64/Kconfig 2007-04-03 18:41:59.000000000 -0700 @@ -392,6 +392,12 @@ config ARCH_SPARSEMEM_ENABLE def_bool y depends on (NUMA || EXPERIMENTAL) +config SPARSE_VIRTUAL + def_bool y + +config ARCH_SUPPORTS_PMD_MAPPING + def_bool y + config ARCH_MEMORY_PROBE def_bool y depends on MEMORY_HOTPLUG