From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from relay.sgi.com (relay1.corp.sgi.com [137.38.102.111]) by oss.sgi.com (Postfix) with ESMTP id 56ADF8002 for ; Tue, 18 Feb 2014 22:17:21 -0600 (CST) Received: from cuda.sgi.com (cuda2.sgi.com [192.48.176.25]) by relay1.corp.sgi.com (Postfix) with ESMTP id 4B5F98F8078 for ; Tue, 18 Feb 2014 20:17:18 -0800 (PST) Received: from ipmail06.adl2.internode.on.net (ipmail06.adl2.internode.on.net [150.101.137.129]) by cuda.sgi.com with ESMTP id YgkDvdKNRlZguR2k for ; Tue, 18 Feb 2014 20:17:16 -0800 (PST) Received: from disappointment.disaster.area ([192.168.1.110] helo=disappointment) by dastard with esmtp (Exim 4.76) (envelope-from ) id 1WFyaT-00022k-7S for xfs@oss.sgi.com; Wed, 19 Feb 2014 15:16:45 +1100 Received: from dave by disappointment with local (Exim 4.80) (envelope-from ) id 1WFyaT-0001Fa-5S for xfs@oss.sgi.com; Wed, 19 Feb 2014 15:16:45 +1100 From: Dave Chinner Subject: [PATCH 2/3] xfs: fix directory inode iolock lockdep false positive Date: Wed, 19 Feb 2014 15:16:41 +1100 Message-Id: <1392783402-4726-3-git-send-email-david@fromorbit.com> In-Reply-To: <1392783402-4726-1-git-send-email-david@fromorbit.com> References: <1392783402-4726-1-git-send-email-david@fromorbit.com> List-Id: XFS Filesystem from SGI List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Errors-To: xfs-bounces@oss.sgi.com Sender: xfs-bounces@oss.sgi.com To: xfs@oss.sgi.com From: Dave Chinner The change to add the IO lock to protect the directory extent map during readdir operations has cause lockdep to have a heart attack as it now sees a different locking order on inodes w.r.t. the mmap_sem because readdir has a different ordering to write(). Add a new lockdep class for directory inodes to avoid this false positive. Signed-off-by: Dave Chinner --- fs/xfs/xfs_iops.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c index 9ddfb81..bb3bb65 100644 --- a/fs/xfs/xfs_iops.c +++ b/fs/xfs/xfs_iops.c @@ -48,6 +48,18 @@ #include #include +/* + * Directories have different lock order w.r.t. mmap_sem compared to regular + * files. This is due to readdir potentially triggering page faults on a user + * buffer inside filldir(), and this happens with the ilock on the directory + * held. For regular files, the lock order is the other way around - the + * mmap_sem is taken during the page fault, and then we lock the ilock to do + * block mapping. Hence we need a different class for the directory ilock so + * that lockdep can tell them apart. + */ +static struct lock_class_key xfs_nondir_ilock_class; +static struct lock_class_key xfs_dir_ilock_class; + static int xfs_initxattrs( struct inode *inode, @@ -1191,6 +1203,7 @@ xfs_setup_inode( xfs_diflags_to_iflags(inode, ip); ip->d_ops = ip->i_mount->m_nondir_inode_ops; + lockdep_set_class(&ip->i_lock.mr_lock, &xfs_nondir_ilock_class); switch (inode->i_mode & S_IFMT) { case S_IFREG: inode->i_op = &xfs_inode_operations; @@ -1198,6 +1211,7 @@ xfs_setup_inode( inode->i_mapping->a_ops = &xfs_address_space_operations; break; case S_IFDIR: + lockdep_set_class(&ip->i_lock.mr_lock, &xfs_dir_ilock_class); if (xfs_sb_version_hasasciici(&XFS_M(inode->i_sb)->m_sb)) inode->i_op = &xfs_dir_ci_inode_operations; else -- 1.8.4.rc3 _______________________________________________ xfs mailing list xfs@oss.sgi.com http://oss.sgi.com/mailman/listinfo/xfs