[XFS] Prevent a deadlock when xfslogd unpins inodes.

The previous fixes for the use after free in xfs_iunpin left a nasty log
deadlock when xfslogd unpinned the inode and dropped the last reference to
the inode. the ->clear_inode() method can issue transactions, and if the
log was full, the transaction could push on the log and get stuck trying
to push the inode it was currently unpinning.

To fix this, we provide xfs_iunpin a guarantee that it will always have a
valid xfs_inode <-> linux inode link or a particular flag will be set on
the inode. We then use log forces during lookup to ensure transactions are
completed before we recycle the inode. This ensures that xfs_iunpin will
never use the linux inode after it is being freed, and any lookup on an
inode on the reclaim list will wait until it is safe to attach a new linux
inode to the xfs inode.

SGI-PV: 956832
SGI-Modid: xfs-linux-melb:xfs-kern:27359a

Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Shailendra Tripathi <stripathi@agami.com>
Signed-off-by: Takenori Nagano <t-nagano@ah.jp.nec.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>

1 files changed, 14 insertions, 7 deletions

diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 4c5d73c..bda774a 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -3827,11 +3827,16 @@ xfs_reclaim(
 	 */
 	xfs_synchronize_atime(ip);
 
-	/* If we have nothing to flush with this inode then complete the
-	 * teardown now, otherwise break the link between the xfs inode
-	 * and the linux inode and clean up the xfs inode later. This
-	 * avoids flushing the inode to disk during the delete operation
-	 * itself.
+	/*
+	 * If we have nothing to flush with this inode then complete the
+	 * teardown now, otherwise break the link between the xfs inode and the
+	 * linux inode and clean up the xfs inode later. This avoids flushing
+	 * the inode to disk during the delete operation itself.
+	 *
+	 * When breaking the link, we need to set the XFS_IRECLAIMABLE flag
+	 * first to ensure that xfs_iunpin() will never see an xfs inode
+	 * that has a linux inode being reclaimed. Synchronisation is provided
+	 * by the i_flags_lock.
 	 */
 	if (!ip->i_update_core && (ip->i_itemp == NULL)) {
 		xfs_ilock(ip, XFS_ILOCK_EXCL);
@@ -3840,11 +3845,13 @@ xfs_reclaim(
 	} else {
 		xfs_mount_t	*mp = ip->i_mount;
 
-		/* Protect sync from us */
+		/* Protect sync and unpin from us */
 		XFS_MOUNT_ILOCK(mp);
+		spin_lock(&ip->i_flags_lock);
+		__xfs_iflags_set(ip, XFS_IRECLAIMABLE);
 		vn_bhv_remove(VN_BHV_HEAD(vp), XFS_ITOBHV(ip));
+		spin_unlock(&ip->i_flags_lock);
 		list_add_tail(&ip->i_reclaim, &mp->m_del_inodes);
-		xfs_iflags_set(ip, XFS_IRECLAIMABLE);
 		XFS_MOUNT_IUNLOCK(mp);
 	}
 	return 0;