Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757896AbYFPRWU (ORCPT ); Mon, 16 Jun 2008 13:22:20 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754299AbYFPRVd (ORCPT ); Mon, 16 Jun 2008 13:21:33 -0400 Received: from bohort.kerlabs.com ([62.160.40.57]:50842 "EHLO bohort.kerlabs.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754352AbYFPRVc (ORCPT ); Mon, 16 Jun 2008 13:21:32 -0400 Message-Id: <20080616171058.722251454@kerlabs.com> References: <20080616170057.399713501@kerlabs.com> User-Agent: quilt/0.46-1 Date: Mon, 16 Jun 2008 19:00:59 +0200 From: Louis Rilling To: Joel.Becker@oracle.com Cc: Louis.Rilling@kerlabs.com, linux-kernel@vger.kernel.org, ocfs2-devel@oss.oracle.com Subject: [PATCH 2/5][BUGFIX] configfs: Protect configfs_dirent s_links list mutations Content-Disposition: inline; filename=configfs-fix-symlink-dirent-linkage-locking.patch Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 3116 Lines: 80 Symlinks to a config_item are listed under its configfs_dirent s_links, but the list mutations are not protected by any common lock. This patch uses the configfs_dirent_lock spinlock to add the necessary protection. Note: we should also protect the list_empty() test in configfs_detach_prep() but 1/ the lock should not be released immediately because nothing would prevent the list from being filled after a successful list_empty() test, making the problem tricky, 2/ this will be solved by the rmdir() vs rename() deadlock bugfix. Signed-off-by: Louis Rilling --- fs/configfs/dir.c | 5 +++-- fs/configfs/symlink.c | 8 ++++++-- 2 files changed, 9 insertions(+), 4 deletions(-) Index: b/fs/configfs/symlink.c =================================================================== --- a/fs/configfs/symlink.c 2008-06-16 17:42:31.000000000 +0200 +++ b/fs/configfs/symlink.c 2008-06-16 17:42:43.000000000 +0200 @@ -77,12 +77,15 @@ static int create_link(struct config_ite sl = kmalloc(sizeof(struct configfs_symlink), GFP_KERNEL); if (sl) { sl->sl_target = config_item_get(item); - /* FIXME: needs a lock, I'd bet */ + spin_lock(&configfs_dirent_lock); list_add(&sl->sl_list, &target_sd->s_links); + spin_unlock(&configfs_dirent_lock); ret = configfs_create_link(sl, parent_item->ci_dentry, dentry); if (ret) { + spin_lock(&configfs_dirent_lock); list_del_init(&sl->sl_list); + spin_unlock(&configfs_dirent_lock); config_item_put(item); kfree(sl); } @@ -186,8 +189,9 @@ int configfs_unlink(struct inode *dir, s type->ct_item_ops->drop_link(parent_item, sl->sl_target); - /* FIXME: Needs lock */ + spin_lock(&configfs_dirent_lock); list_del_init(&sl->sl_list); + spin_unlock(&configfs_dirent_lock); /* Put reference from create_link() */ config_item_put(sl->sl_target); Index: b/fs/configfs/dir.c =================================================================== --- a/fs/configfs/dir.c 2008-06-16 17:42:31.000000000 +0200 +++ b/fs/configfs/dir.c 2008-06-16 18:54:20.000000000 +0200 @@ -37,10 +37,11 @@ DECLARE_RWSEM(configfs_rename_sem); /* * Protects mutations of configfs_dirent linkage together with proper i_mutex + * Also protects mutations of symlinks linkage to target configfs_dirent * Mutators of configfs_dirent linkage must *both* have the proper inode locked * and configfs_dirent_lock locked, in that order. - * This allows one to safely traverse configfs_dirent trees without having to - * lock inodes. + * This allows one to safely traverse configfs_dirent trees and symlinks without + * having to lock inodes. */ DEFINE_SPINLOCK(configfs_dirent_lock); -- Dr Louis Rilling Kerlabs Skype: louis.rilling Batiment Germanium Phone: (+33|0) 6 80 89 08 23 80 avenue des Buttes de Coesmes http://www.kerlabs.com/ 35700 Rennes -- 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/