Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759902Ab0GVS1j (ORCPT ); Thu, 22 Jul 2010 14:27:39 -0400 Received: from rcsinet10.oracle.com ([148.87.113.121]:42541 "EHLO rcsinet10.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1759599Ab0GVSWh (ORCPT ); Thu, 22 Jul 2010 14:22:37 -0400 From: Yinghai Lu To: Ingo Molnar , Thomas Gleixner , "H. Peter Anvin" , Andrew Morton , David Miller , Benjamin Herrenschmidt , Linus Torvalds Cc: Johannes Weiner , linux-kernel@vger.kernel.org, linux-arch@vger.kernel.org, Benjamin Herrenschmidt Subject: [PATCH 22/31] memblock: NUMA allocate can now use early_pfn_map Date: Thu, 22 Jul 2010 11:20:55 -0700 Message-Id: <1279822864-17154-23-git-send-email-yinghai@kernel.org> X-Mailer: git-send-email 1.6.4.2 In-Reply-To: <1279822864-17154-1-git-send-email-yinghai@kernel.org> References: <1279822864-17154-1-git-send-email-yinghai@kernel.org> X-Source-IP: acsmt353.oracle.com [141.146.40.153] X-Auth-Type: Internal IP X-CT-RefId: str=0001.0A090209.4C488C35.0216,ss=1,fgs=0 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 3572 Lines: 92 From: Benjamin Herrenschmidt We now provide a default (weak) implementation of memblock_nid_range() which uses the early_pfn_map[] if CONFIG_ARCH_POPULATES_NODE_MAP is set. Sparc still needs to use its own method due to the way the pages can be scattered between nodes. This implementation is inefficient due to our main algorithm and callback construct wanting to work on an ascending addresses bases while early_pfn_map[] would rather work with nid's (it's unsorted at that stage). But it should work and we can look into improving it subsequently, possibly using arch compile options to chose a different algorithm alltogether. Signed-off-by: Benjamin Herrenschmidt --- include/linux/memblock.h | 3 +++ mm/memblock.c | 28 +++++++++++++++++++++++++++- 2 files changed, 30 insertions(+), 1 deletions(-) diff --git a/include/linux/memblock.h b/include/linux/memblock.h index 15da7d9..b69c243 100644 --- a/include/linux/memblock.h +++ b/include/linux/memblock.h @@ -47,6 +47,9 @@ extern long memblock_remove(phys_addr_t base, phys_addr_t size); extern long __init memblock_free(phys_addr_t base, phys_addr_t size); extern long __init memblock_reserve(phys_addr_t base, phys_addr_t size); +/* The numa aware allocator is only available if + * CONFIG_ARCH_POPULATES_NODE_MAP is set + */ extern phys_addr_t __init memblock_alloc_nid(phys_addr_t size, phys_addr_t align, int nid); extern phys_addr_t __init memblock_alloc(phys_addr_t size, phys_addr_t align); diff --git a/mm/memblock.c b/mm/memblock.c index b4870cf..a9c15a5 100644 --- a/mm/memblock.c +++ b/mm/memblock.c @@ -15,6 +15,7 @@ #include #include #include +#include #include struct memblock memblock; @@ -451,11 +452,36 @@ phys_addr_t __init memblock_alloc(phys_addr_t size, phys_addr_t align) /* * Additional node-local allocators. Search for node memory is bottom up * and walks memblock regions within that node bottom-up as well, but allocation - * within an memblock region is top-down. + * within an memblock region is top-down. XXX I plan to fix that at some stage + * + * WARNING: Only available after early_node_map[] has been populated, + * on some architectures, that is after all the calls to add_active_range() + * have been done to populate it. */ phys_addr_t __weak __init memblock_nid_range(phys_addr_t start, phys_addr_t end, int *nid) { +#ifdef CONFIG_ARCH_POPULATES_NODE_MAP + /* + * This code originates from sparc which really wants use to walk by addresses + * and returns the nid. This is not very convenient for early_pfn_map[] users + * as the map isn't sorted yet, and it really wants to be walked by nid. + * + * For now, I implement the inefficient method below which walks the early + * map multiple times. Eventually we may want to use an ARCH config option + * to implement a completely different method for both case. + */ + unsigned long start_pfn, end_pfn; + int i; + + for (i = 0; i < MAX_NUMNODES; i++) { + get_pfn_range_for_nid(i, &start_pfn, &end_pfn); + if (start < PFN_PHYS(start_pfn) || start >= PFN_PHYS(end_pfn)) + continue; + *nid = i; + return min(end, PFN_PHYS(end_pfn)); + } +#endif *nid = 0; return end; -- 1.6.4.2 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/