Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933071Ab2FVOft (ORCPT ); Fri, 22 Jun 2012 10:35:49 -0400 Received: from relay3.sgi.com ([192.48.152.1]:49818 "EHLO relay.sgi.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S932575Ab2FVOfP (ORCPT ); Fri, 22 Jun 2012 10:35:15 -0400 Date: Fri, 22 Jun 2012 09:35:13 -0500 From: Nathan Zimmer To: Christoph Lameter , Nick Piggin , Hugh Dickins , Lee Schermerhorn , KOSAKI Motohiro Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org, riel@redhat.com Subject: [PATCH v3] tmpfs not interleaving properly Message-ID: <20120622143512.GA18468@gulag1.americas.sgi.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.17 (2007-11-01) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2966 Lines: 76 When tmpfs has the memory policy interleaved it always starts allocating at each file at node 0. When there are many small files the lower nodes fill up disproportionately. This patch attempts to spread out node usage by starting files at nodes other then 0. I disturbed the addr parameter since alloc_pages_vma will only use it when the policy is MPOL_INTERLEAVE. A files preferred node is selected by the cpu_mem_spread_node rotor. v2: passed preferred node via addr v3: using current->cpuset_mem_spread_rotor instead of random_node Cc: Christoph Lameter Cc: Nick Piggin Cc: Hugh Dickins Cc: Lee Schermerhorn Cc: KOSAKI Motohiro Acked-by: Rik van Riel Signed-off-by: Nathan T Zimmer --- include/linux/shmem_fs.h | 1 + mm/shmem.c | 9 +++++++-- 2 files changed, 8 insertions(+), 2 deletions(-) diff --git a/include/linux/shmem_fs.h b/include/linux/shmem_fs.h index bef2cf0..cfe8a34 100644 --- a/include/linux/shmem_fs.h +++ b/include/linux/shmem_fs.h @@ -17,6 +17,7 @@ struct shmem_inode_info { char *symlink; /* unswappable short symlink */ }; struct shared_policy policy; /* NUMA memory alloc policy */ + unsigned long node_offset; /* bias for interleaved nodes */ struct list_head swaplist; /* chain of maybes on swap */ struct list_head xattr_list; /* list of shmem_xattr */ struct inode vfs_inode; diff --git a/mm/shmem.c b/mm/shmem.c index a15a466..93801b3 100644 --- a/mm/shmem.c +++ b/mm/shmem.c @@ -64,6 +64,7 @@ static struct vfsmount *shm_mnt; #include #include #include +#include #include #include @@ -938,9 +939,12 @@ static struct page *shmem_alloc_page(gfp_t gfp, pvma.vm_policy = mpol_shared_policy_lookup(&info->policy, index); /* - * alloc_page_vma() will drop the shared policy reference + * alloc_page_vma() will drop the shared policy reference. + * + * To avoid allocating all tmpfs pages on node 0, we fake up a virtual + * address based on this file's predetermined preferred node. */ - return alloc_page_vma(gfp, &pvma, 0); + return alloc_page_vma(gfp, &pvma, info->node_offset << PAGE_SHIFT); } #else /* !CONFIG_NUMA */ #ifdef CONFIG_TMPFS @@ -1374,6 +1378,7 @@ static struct inode *shmem_get_inode(struct super_block *sb, const struct inode inode->i_fop = &shmem_file_operations; mpol_shared_policy_init(&info->policy, shmem_get_sbmpol(sbinfo)); + info->node_offset = cpuset_mem_spread_node(); break; case S_IFDIR: inc_nlink(inode); -- 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/