Merge commit 'v2.6.37' into for-2.6.38-incoming

I made a slight mess of Documentation/filesystems/Locking; resolve
conflicts with upstream before fixing it up.

20 files changed, 188 insertions, 87 deletions

diff --git a/fs/btrfs/export.c b/fs/btrfs/export.c
index 6f04444..659f532 100644
--- a/fs/btrfs/export.c
+++ b/fs/btrfs/export.c
@@ -166,7 +166,7 @@ static struct dentry *btrfs_fh_to_dentry(struct super_block *sb, struct fid *fh,
 static struct dentry *btrfs_get_parent(struct dentry *child)
 {
 	struct inode *dir = child->d_inode;
-	static struct dentry *dentry;
+	struct dentry *dentry;
 	struct btrfs_root *root = BTRFS_I(dir)->root;
 	struct btrfs_path *path;
 	struct extent_buffer *leaf;
diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 158c700..d902948 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -40,7 +40,8 @@ int ceph_init_dentry(struct dentry *dentry)
 	if (dentry->d_fsdata)
 		return 0;
 
-	if (ceph_snap(dentry->d_parent->d_inode) == CEPH_NOSNAP)
+	if (dentry->d_parent == NULL ||   /* nfs fh_to_dentry */
+	    ceph_snap(dentry->d_parent->d_inode) == CEPH_NOSNAP)
 		dentry->d_op = &ceph_dentry_ops;
 	else if (ceph_snap(dentry->d_parent->d_inode) == CEPH_SNAPDIR)
 		dentry->d_op = &ceph_snapdir_dentry_ops;
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 8d79b89..7d0e4a8 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -282,7 +282,8 @@ int ceph_release(struct inode *inode, struct file *file)
 static int striped_read(struct inode *inode,
 			u64 off, u64 len,
 			struct page **pages, int num_pages,
-			int *checkeof, bool align_to_pages)
+			int *checkeof, bool align_to_pages,
+			unsigned long buf_align)
 {
 	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 	struct ceph_inode_info *ci = ceph_inode(inode);
@@ -307,7 +308,7 @@ static int striped_read(struct inode *inode,
 
 more:
 	if (align_to_pages)
-		page_align = (pos - io_align) & ~PAGE_MASK;
+		page_align = (pos - io_align + buf_align) & ~PAGE_MASK;
 	else
 		page_align = pos & ~PAGE_MASK;
 	this_len = left;
@@ -376,16 +377,18 @@ static ssize_t ceph_sync_read(struct file *file, char __user *data,
 	struct inode *inode = file->f_dentry->d_inode;
 	struct page **pages;
 	u64 off = *poff;
-	int num_pages = calc_pages_for(off, len);
-	int ret;
+	int num_pages, ret;
 
 	dout("sync_read on file %p %llu~%u %s\n", file, off, len,
 	     (file->f_flags & O_DIRECT) ? "O_DIRECT" : "");
 
-	if (file->f_flags & O_DIRECT)
-		pages = ceph_get_direct_page_vector(data, num_pages);
-	else
+	if (file->f_flags & O_DIRECT) {
+		num_pages = calc_pages_for((unsigned long)data, len);
+		pages = ceph_get_direct_page_vector(data, num_pages, true);
+	} else {
+		num_pages = calc_pages_for(off, len);
 		pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
+	}
 	if (IS_ERR(pages))
 		return PTR_ERR(pages);
 
@@ -400,7 +403,8 @@ static ssize_t ceph_sync_read(struct file *file, char __user *data,
 		goto done;
 
 	ret = striped_read(inode, off, len, pages, num_pages, checkeof,
-			   file->f_flags & O_DIRECT);
+			   file->f_flags & O_DIRECT,
+			   (unsigned long)data & ~PAGE_MASK);
 
 	if (ret >= 0 && (file->f_flags & O_DIRECT) == 0)
 		ret = ceph_copy_page_vector_to_user(pages, data, off, ret);
@@ -409,7 +413,7 @@ static ssize_t ceph_sync_read(struct file *file, char __user *data,
 
 done:
 	if (file->f_flags & O_DIRECT)
-		ceph_put_page_vector(pages, num_pages);
+		ceph_put_page_vector(pages, num_pages, true);
 	else
 		ceph_release_page_vector(pages, num_pages);
 	dout("sync_read result %d\n", ret);
@@ -456,6 +460,7 @@ static ssize_t ceph_sync_write(struct file *file, const char __user *data,
 	int do_sync = 0;
 	int check_caps = 0;
 	int page_align, io_align;
+	unsigned long buf_align;
 	int ret;
 	struct timespec mtime = CURRENT_TIME;
 
@@ -471,6 +476,7 @@ static ssize_t ceph_sync_write(struct file *file, const char __user *data,
 		pos = *offset;
 
 	io_align = pos & ~PAGE_MASK;
+	buf_align = (unsigned long)data & ~PAGE_MASK;
 
 	ret = filemap_write_and_wait_range(inode->i_mapping, pos, pos + left);
 	if (ret < 0)
@@ -496,12 +502,15 @@ static ssize_t ceph_sync_write(struct file *file, const char __user *data,
 	 */
 more:
 	len = left;
-	if (file->f_flags & O_DIRECT)
+	if (file->f_flags & O_DIRECT) {
 		/* write from beginning of first page, regardless of
 		   io alignment */
-		page_align = (pos - io_align) & ~PAGE_MASK;
-	else
+		page_align = (pos - io_align + buf_align) & ~PAGE_MASK;
+		num_pages = calc_pages_for((unsigned long)data, len);
+	} else {
 		page_align = pos & ~PAGE_MASK;
+		num_pages = calc_pages_for(pos, len);
+	}
 	req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
 				    ceph_vino(inode), pos, &len,
 				    CEPH_OSD_OP_WRITE, flags,
@@ -512,10 +521,8 @@ more:
 	if (!req)
 		return -ENOMEM;
 
-	num_pages = calc_pages_for(pos, len);
-
 	if (file->f_flags & O_DIRECT) {
-		pages = ceph_get_direct_page_vector(data, num_pages);
+		pages = ceph_get_direct_page_vector(data, num_pages, false);
 		if (IS_ERR(pages)) {
 			ret = PTR_ERR(pages);
 			goto out;
@@ -565,7 +572,7 @@ more:
 	}
 
 	if (file->f_flags & O_DIRECT)
-		ceph_put_page_vector(pages, num_pages);
+		ceph_put_page_vector(pages, num_pages, false);
 	else if (file->f_flags & O_SYNC)
 		ceph_release_page_vector(pages, num_pages);
 
diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c
index dc96392..981c847 100644
--- a/fs/ext4/resize.c
+++ b/fs/ext4/resize.c
@@ -232,6 +232,8 @@ static int setup_new_group_blocks(struct super_block *sb,
 			       GFP_NOFS);
 	if (err)
 		goto exit_bh;
+	for (i = 0, bit = gdblocks + 1; i < reserved_gdb; i++, bit++)
+		ext4_set_bit(bit, bh->b_data);
 
 	ext4_debug("mark block bitmap %#04llx (+%llu)\n", input->block_bitmap,
 		   input->block_bitmap - start);
@@ -247,6 +249,9 @@ static int setup_new_group_blocks(struct super_block *sb,
 	err = sb_issue_zeroout(sb, block, sbi->s_itb_per_group, GFP_NOFS);
 	if (err)
 		goto exit_bh;
+	for (i = 0, bit = input->inode_table - start;
+	     i < sbi->s_itb_per_group; i++, bit++)
+		ext4_set_bit(bit, bh->b_data);
 
 	if ((err = extend_or_restart_transaction(handle, 2, bh)))
 		goto exit_bh;
diff --git a/fs/logfs/journal.c b/fs/logfs/journal.c
index f46ee8b..9da2970 100644
--- a/fs/logfs/journal.c
+++ b/fs/logfs/journal.c
@@ -828,7 +828,7 @@ void do_logfs_journal_wl_pass(struct super_block *sb)
 		super->s_journal_seg[i] = segno;
 		super->s_journal_ec[i] = ec;
 		logfs_set_segment_reserved(sb, segno);
-		err = btree_insert32(head, segno, (void *)1, GFP_KERNEL);
+		err = btree_insert32(head, segno, (void *)1, GFP_NOFS);
 		BUG_ON(err); /* mempool should prevent this */
 		err = logfs_erase_segment(sb, segno, 1);
 		BUG_ON(err); /* FIXME: remount-ro would be nicer */
diff --git a/fs/logfs/readwrite.c b/fs/logfs/readwrite.c
index 6127baf..ee99a9f 100644
--- a/fs/logfs/readwrite.c
+++ b/fs/logfs/readwrite.c
@@ -1994,6 +1994,9 @@ static int do_write_inode(struct inode *inode)
 
 	/* FIXME: transaction is part of logfs_block now.  Is that enough? */
 	err = logfs_write_buf(master_inode, page, 0);
+	if (err)
+		move_page_to_inode(inode, page);
+
 	logfs_put_write_page(page);
 	return err;
 }
diff --git a/fs/namei.c b/fs/namei.c
index 5362af9..4ff7ca5 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -1748,6 +1748,9 @@ struct file *do_filp_open(int dfd, const char *pathname,
 	if (!(open_flag & O_CREAT))
 		mode = 0;
 
+	/* Must never be set by userspace */
+	open_flag &= ~FMODE_NONOTIFY;
+
 	/*
 	 * O_SYNC is implemented as __O_SYNC|O_DSYNC.  As many places only
 	 * check for O_DSYNC if the need any syncing at all we enforce it's
diff --git a/fs/nilfs2/gcinode.c b/fs/nilfs2/gcinode.c
index 33ad25d..caf9a6a 100644
--- a/fs/nilfs2/gcinode.c
+++ b/fs/nilfs2/gcinode.c
@@ -176,7 +176,6 @@ int nilfs_gccache_wait_and_mark_dirty(struct buffer_head *bh)
 int nilfs_init_gcinode(struct inode *inode)
 {
 	struct nilfs_inode_info *ii = NILFS_I(inode);
-	struct the_nilfs *nilfs = NILFS_SB(inode->i_sb)->s_nilfs;
 
 	inode->i_mode = S_IFREG;
 	mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
@@ -186,14 +185,6 @@ int nilfs_init_gcinode(struct inode *inode)
 	ii->i_flags = 0;
 	nilfs_bmap_init_gc(ii->i_bmap);
 
-	/*
-	 * Add the inode to GC inode list. Garbage Collection
-	 * is serialized and no two processes manipulate the
-	 * list simultaneously.
-	 */
-	igrab(inode);
-	list_add(&NILFS_I(inode)->i_dirty, &nilfs->ns_gc_inodes);
-
 	return 0;
 }
 
diff --git a/fs/nilfs2/ioctl.c b/fs/nilfs2/ioctl.c
index e00d945..b185e93 100644
--- a/fs/nilfs2/ioctl.c
+++ b/fs/nilfs2/ioctl.c
@@ -337,6 +337,7 @@ static int nilfs_ioctl_move_blocks(struct super_block *sb,
 				   struct nilfs_argv *argv, void *buf)
 {
 	size_t nmembs = argv->v_nmembs;
+	struct the_nilfs *nilfs = NILFS_SB(sb)->s_nilfs;
 	struct inode *inode;
 	struct nilfs_vdesc *vdesc;
 	struct buffer_head *bh, *n;
@@ -353,6 +354,17 @@ static int nilfs_ioctl_move_blocks(struct super_block *sb,
 			ret = PTR_ERR(inode);
 			goto failed;
 		}
+		if (list_empty(&NILFS_I(inode)->i_dirty)) {
+			/*
+			 * Add the inode to GC inode list. Garbage Collection
+			 * is serialized and no two processes manipulate the
+			 * list simultaneously.
+			 */
+			igrab(inode);
+			list_add(&NILFS_I(inode)->i_dirty,
+				 &nilfs->ns_gc_inodes);
+		}
+
 		do {
 			ret = nilfs_ioctl_move_inode_block(inode, vdesc,
 							   &buffers);
diff --git a/fs/notify/fanotify/fanotify.c b/fs/notify/fanotify/fanotify.c
index b04f88e..f35794b 100644
--- a/fs/notify/fanotify/fanotify.c
+++ b/fs/notify/fanotify/fanotify.c
@@ -92,7 +92,11 @@ static int fanotify_get_response_from_access(struct fsnotify_group *group,
 
 	pr_debug("%s: group=%p event=%p\n", __func__, group, event);
 
-	wait_event(group->fanotify_data.access_waitq, event->response);
+	wait_event(group->fanotify_data.access_waitq, event->response ||
+				atomic_read(&group->fanotify_data.bypass_perm));
+
+	if (!event->response) /* bypass_perm set */
+		return 0;
 
 	/* userspace responded, convert to something usable */
 	spin_lock(&event->lock);
diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c
index 0632248..8b61220 100644
--- a/fs/notify/fanotify/fanotify_user.c
+++ b/fs/notify/fanotify/fanotify_user.c
@@ -106,20 +106,29 @@ static int create_fd(struct fsnotify_group *group, struct fsnotify_event *event)
 	return client_fd;
 }
 
-static ssize_t fill_event_metadata(struct fsnotify_group *group,
+static int fill_event_metadata(struct fsnotify_group *group,
 				   struct fanotify_event_metadata *metadata,
 				   struct fsnotify_event *event)
 {
+	int ret = 0;
+
 	pr_debug("%s: group=%p metadata=%p event=%p\n", __func__,
 		 group, metadata, event);
 
 	metadata->event_len = FAN_EVENT_METADATA_LEN;
+	metadata->metadata_len = FAN_EVENT_METADATA_LEN;
 	metadata->vers = FANOTIFY_METADATA_VERSION;
 	metadata->mask = event->mask & FAN_ALL_OUTGOING_EVENTS;
 	metadata->pid = pid_vnr(event->tgid);
-	metadata->fd = create_fd(group, event);
+	if (unlikely(event->mask & FAN_Q_OVERFLOW))
+		metadata->fd = FAN_NOFD;
+	else {
+		metadata->fd = create_fd(group, event);
+		if (metadata->fd < 0)
+			ret = metadata->fd;
+	}
 
-	return metadata->fd;
+	return ret;
 }
 
 #ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
@@ -200,7 +209,7 @@ static int prepare_for_access_response(struct fsnotify_group *group,
 
 	mutex_lock(&group->fanotify_data.access_mutex);
 
-	if (group->fanotify_data.bypass_perm) {
+	if (atomic_read(&group->fanotify_data.bypass_perm)) {
 		mutex_unlock(&group->fanotify_data.access_mutex);
 		kmem_cache_free(fanotify_response_event_cache, re);
 		event->response = FAN_ALLOW;
@@ -257,24 +266,34 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 
 	pr_debug("%s: group=%p event=%p\n", __func__, group, event);
 
-	fd = fill_event_metadata(group, &fanotify_event_metadata, event);
-	if (fd < 0)
-		return fd;
+	ret = fill_event_metadata(group, &fanotify_event_metadata, event);
+	if (ret < 0)
+		goto out;
 
+	fd = fanotify_event_metadata.fd;
 	ret = prepare_for_access_response(group, event, fd);
 	if (ret)
 		goto out_close_fd;
 
 	ret = -EFAULT;
-	if (copy_to_user(buf, &fanotify_event_metadata, FAN_EVENT_METADATA_LEN))
+	if (copy_to_user(buf, &fanotify_event_metadata,
+			 fanotify_event_metadata.event_len))
 		goto out_kill_access_response;
 
-	return FAN_EVENT_METADATA_LEN;
+	return fanotify_event_metadata.event_len;
 
 out_kill_access_response:
 	remove_access_response(group, event, fd);
 out_close_fd:
-	sys_close(fd);
+	if (fd != FAN_NOFD)
+		sys_close(fd);
+out:
+#ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
+	if (event->mask & FAN_ALL_PERM_EVENTS) {
+		event->response = FAN_DENY;
+		wake_up(&group->fanotify_data.access_waitq);
+	}
+#endif
 	return ret;
 }
 
@@ -382,7 +401,7 @@ static int fanotify_release(struct inode *ignored, struct file *file)
 
 	mutex_lock(&group->fanotify_data.access_mutex);
 
-	group->fanotify_data.bypass_perm = true;
+	atomic_inc(&group->fanotify_data.bypass_perm);
 
 	list_for_each_entry_safe(re, lre, &group->fanotify_data.access_list, list) {
 		pr_debug("%s: found group=%p re=%p event=%p\n", __func__, group,
@@ -586,11 +605,10 @@ static int fanotify_add_vfsmount_mark(struct fsnotify_group *group,
 {
 	struct fsnotify_mark *fsn_mark;
 	__u32 added;
+	int ret = 0;
 
 	fsn_mark = fsnotify_find_vfsmount_mark(group, mnt);
 	if (!fsn_mark) {
-		int ret;
-
 		if (atomic_read(&group->num_marks) > group->fanotify_data.max_marks)
 			return -ENOSPC;
 
@@ -600,17 +618,16 @@ static int fanotify_add_vfsmount_mark(struct fsnotify_group *group,
 
 		fsnotify_init_mark(fsn_mark, fanotify_free_mark);
 		ret = fsnotify_add_mark(fsn_mark, group, NULL, mnt, 0);
-		if (ret) {
-			fanotify_free_mark(fsn_mark);
-			return ret;
-		}
+		if (ret)
+			goto err;
 	}
 	added = fanotify_mark_add_to_mask(fsn_mark, mask, flags);
-	fsnotify_put_mark(fsn_mark);
+
 	if (added & ~mnt->mnt_fsnotify_mask)
 		fsnotify_recalc_vfsmount_mask(mnt);
-
-	return 0;
+err:
+	fsnotify_put_mark(fsn_mark);
+	return ret;
 }
 
 static int fanotify_add_inode_mark(struct fsnotify_group *group,
@@ -619,6 +636,7 @@ static int fanotify_add_inode_mark(struct fsnotify_group *group,
 {
 	struct fsnotify_mark *fsn_mark;
 	__u32 added;
+	int ret = 0;
 
 	pr_debug("%s: group=%p inode=%p\n", __func__, group, inode);
 
@@ -634,8 +652,6 @@ static int fanotify_add_inode_mark(struct fsnotify_group *group,
 
 	fsn_mark = fsnotify_find_inode_mark(group, inode);
 	if (!fsn_mark) {
-		int ret;
-
 		if (atomic_read(&group->num_marks) > group->fanotify_data.max_marks)
 			return -ENOSPC;
 
@@ -645,16 +661,16 @@ static int fanotify_add_inode_mark(struct fsnotify_group *group,
 
 		fsnotify_init_mark(fsn_mark, fanotify_free_mark);
 		ret = fsnotify_add_mark(fsn_mark, group, inode, NULL, 0);
-		if (ret) {
-			fanotify_free_mark(fsn_mark);
-			return ret;
-		}
+		if (ret)
+			goto err;
 	}
 	added = fanotify_mark_add_to_mask(fsn_mark, mask, flags);
-	fsnotify_put_mark(fsn_mark);
+
 	if (added & ~inode->i_fsnotify_mask)
 		fsnotify_recalc_inode_mask(inode);
-	return 0;
+err:
+	fsnotify_put_mark(fsn_mark);
+	return ret;
 }
 
 /* fanotify syscalls */
@@ -687,8 +703,10 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
 
 	/* fsnotify_alloc_group takes a ref.  Dropped in fanotify_release */
 	group = fsnotify_alloc_group(&fanotify_fsnotify_ops);
-	if (IS_ERR(group))
+	if (IS_ERR(group)) {
+		free_uid(user);
 		return PTR_ERR(group);
+	}
 
 	group->fanotify_data.user = user;
 	atomic_inc(&user->fanotify_listeners);
@@ -698,6 +716,7 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
 	mutex_init(&group->fanotify_data.access_mutex);
 	init_waitqueue_head(&group->fanotify_data.access_waitq);
 	INIT_LIST_HEAD(&group->fanotify_data.access_list);
+	atomic_set(&group->fanotify_data.bypass_perm, 0);
 #endif
 	switch (flags & FAN_ALL_CLASS_BITS) {
 	case FAN_CLASS_NOTIF:
@@ -764,8 +783,10 @@ SYSCALL_DEFINE(fanotify_mark)(int fanotify_fd, unsigned int flags,
 	if (flags & ~FAN_ALL_MARK_FLAGS)
 		return -EINVAL;
 	switch (flags & (FAN_MARK_ADD | FAN_MARK_REMOVE | FAN_MARK_FLUSH)) {
-	case FAN_MARK_ADD:
+	case FAN_MARK_ADD:		/* fallthrough */
 	case FAN_MARK_REMOVE:
+		if (!mask)
+			return -EINVAL;
 	case FAN_MARK_FLUSH:
 		break;
 	default:
diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c
index 444c305..4cd5d5d 100644
--- a/fs/notify/inotify/inotify_user.c
+++ b/fs/notify/inotify/inotify_user.c
@@ -752,6 +752,7 @@ SYSCALL_DEFINE1(inotify_init1, int, flags)
 	if (ret >= 0)
 		return ret;
 
+	fsnotify_put_group(group);
 	atomic_dec(&user->inotify_devs);
 out_free_uid:
 	free_uid(user);
diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index f1e962c..0d7c554 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -573,11 +573,14 @@ static void ocfs2_dio_end_io(struct kiocb *iocb,
 	/* this io's submitter should not have unlocked this before we could */
 	BUG_ON(!ocfs2_iocb_is_rw_locked(iocb));
 
+	if (ocfs2_iocb_is_sem_locked(iocb)) {
+		up_read(&inode->i_alloc_sem);
+		ocfs2_iocb_clear_sem_locked(iocb);
+	}
+
 	ocfs2_iocb_clear_rw_locked(iocb);
 
 	level = ocfs2_iocb_rw_locked_level(iocb);
-	if (!level)
-		up_read(&inode->i_alloc_sem);
 	ocfs2_rw_unlock(inode, level);
 
 	if (is_async)
diff --git a/fs/ocfs2/aops.h b/fs/ocfs2/aops.h
index 76bfdfd..eceb456 100644
--- a/fs/ocfs2/aops.h
+++ b/fs/ocfs2/aops.h
@@ -68,8 +68,27 @@ static inline void ocfs2_iocb_set_rw_locked(struct kiocb *iocb, int level)
 	else
 		clear_bit(1, (unsigned long *)&iocb->private);
 }
+
+/*
+ * Using a named enum representing lock types in terms of #N bit stored in
+ * iocb->private, which is going to be used for communication bewteen
+ * ocfs2_dio_end_io() and ocfs2_file_aio_write/read().
+ */
+enum ocfs2_iocb_lock_bits {
+	OCFS2_IOCB_RW_LOCK = 0,
+	OCFS2_IOCB_RW_LOCK_LEVEL,
+	OCFS2_IOCB_SEM,
+	OCFS2_IOCB_NUM_LOCKS
+};
+
 #define ocfs2_iocb_clear_rw_locked(iocb) \
-	clear_bit(0, (unsigned long *)&iocb->private)
+	clear_bit(OCFS2_IOCB_RW_LOCK, (unsigned long *)&iocb->private)
 #define ocfs2_iocb_rw_locked_level(iocb) \
-	test_bit(1, (unsigned long *)&iocb->private)
+	test_bit(OCFS2_IOCB_RW_LOCK_LEVEL, (unsigned long *)&iocb->private)
+#define ocfs2_iocb_set_sem_locked(iocb) \
+	set_bit(OCFS2_IOCB_SEM, (unsigned long *)&iocb->private)
+#define ocfs2_iocb_clear_sem_locked(iocb) \
+	clear_bit(OCFS2_IOCB_SEM, (unsigned long *)&iocb->private)
+#define ocfs2_iocb_is_sem_locked(iocb) \
+	test_bit(OCFS2_IOCB_SEM, (unsigned long *)&iocb->private)
 #endif /* OCFS2_FILE_H */
diff --git a/fs/ocfs2/cluster/masklog.c b/fs/ocfs2/cluster/masklog.c
index c7fba39..6c61771 100644
--- a/fs/ocfs2/cluster/masklog.c
+++ b/fs/ocfs2/cluster/masklog.c
@@ -113,10 +113,11 @@ static struct mlog_attribute mlog_attrs[MLOG_MAX_BITS] = {
 	define_mask(QUOTA),
 	define_mask(REFCOUNT),
 	define_mask(BASTS),
+	define_mask(RESERVATIONS),
+	define_mask(CLUSTER),
 	define_mask(ERROR),
 	define_mask(NOTICE),
 	define_mask(KTHREAD),
-	define_mask(RESERVATIONS),
 };
 
 static struct attribute *mlog_attr_ptrs[MLOG_MAX_BITS] = {NULL, };
diff --git a/fs/ocfs2/cluster/masklog.h b/fs/ocfs2/cluster/masklog.h
index ea2ed9f..34d6544 100644
--- a/fs/ocfs2/cluster/masklog.h
+++ b/fs/ocfs2/cluster/masklog.h
@@ -81,7 +81,7 @@
 #include <linux/sched.h>
 
 /* bits that are frequently given and infrequently matched in the low word */
-/* NOTE: If you add a flag, you need to also update mlog.c! */
+/* NOTE: If you add a flag, you need to also update masklog.c! */
 #define ML_ENTRY	0x0000000000000001ULL /* func call entry */
 #define ML_EXIT		0x0000000000000002ULL /* func call exit */
 #define ML_TCP		0x0000000000000004ULL /* net cluster/tcp.c */
@@ -114,13 +114,14 @@
 #define ML_XATTR	0x0000000020000000ULL /* ocfs2 extended attributes */
 #define ML_QUOTA	0x0000000040000000ULL /* ocfs2 quota operations */
 #define ML_REFCOUNT	0x0000000080000000ULL /* refcount tree operations */
-#define ML_BASTS	0x0000001000000000ULL /* dlmglue asts and basts */
+#define ML_BASTS	0x0000000100000000ULL /* dlmglue asts and basts */
+#define ML_RESERVATIONS	0x0000000200000000ULL /* ocfs2 alloc reservations */
+#define ML_CLUSTER	0x0000000400000000ULL /* cluster stack */
+
 /* bits that are infrequently given and frequently matched in the high word */
-#define ML_ERROR	0x0000000100000000ULL /* sent to KERN_ERR */
-#define ML_NOTICE	0x0000000200000000ULL /* setn to KERN_NOTICE */
-#define ML_KTHREAD	0x0000000400000000ULL /* kernel thread activity */
-#define ML_RESERVATIONS	0x0000000800000000ULL /* ocfs2 alloc reservations */
-#define ML_CLUSTER	0x0000001000000000ULL /* cluster stack */
+#define ML_ERROR	0x1000000000000000ULL /* sent to KERN_ERR */
+#define ML_NOTICE	0x2000000000000000ULL /* setn to KERN_NOTICE */
+#define ML_KTHREAD	0x4000000000000000ULL /* kernel thread activity */
 
 #define MLOG_INITIAL_AND_MASK (ML_ERROR|ML_NOTICE)
 #define MLOG_INITIAL_NOT_MASK (ML_ENTRY|ML_EXIT)
diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c
index c49f6de..d417b3f 100644
--- a/fs/ocfs2/dir.c
+++ b/fs/ocfs2/dir.c
@@ -2461,8 +2461,10 @@ static int ocfs2_dx_dir_attach_index(struct ocfs2_super *osb,
 
 	di->i_dx_root = cpu_to_le64(dr_blkno);
 
+	spin_lock(&OCFS2_I(dir)->ip_lock);
 	OCFS2_I(dir)->ip_dyn_features |= OCFS2_INDEXED_DIR_FL;
 	di->i_dyn_features = cpu_to_le16(OCFS2_I(dir)->ip_dyn_features);
+	spin_unlock(&OCFS2_I(dir)->ip_lock);
 
 	ocfs2_journal_dirty(handle, di_bh);
 
@@ -4466,8 +4468,10 @@ static int ocfs2_dx_dir_remove_index(struct inode *dir,
 		goto out_commit;
 	}
 
+	spin_lock(&OCFS2_I(dir)->ip_lock);
 	OCFS2_I(dir)->ip_dyn_features &= ~OCFS2_INDEXED_DIR_FL;
 	di->i_dyn_features = cpu_to_le16(OCFS2_I(dir)->ip_dyn_features);
+	spin_unlock(&OCFS2_I(dir)->ip_lock);
 	di->i_dx_root = cpu_to_le64(0ULL);
 
 	ocfs2_journal_dirty(handle, di_bh);
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index f564b0e..59f0f6b 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -2346,7 +2346,8 @@ static void dlm_deref_lockres_worker(struct dlm_work_item *item, void *data)
  */
 static int dlm_is_lockres_migrateable(struct dlm_ctxt *dlm,
 				      struct dlm_lock_resource *res,
-				      int *numlocks)
+				      int *numlocks,
+				      int *hasrefs)
 {
 	int ret;
 	int i;
@@ -2356,6 +2357,9 @@ static int dlm_is_lockres_migrateable(struct dlm_ctxt *dlm,
 
 	assert_spin_locked(&res->spinlock);
 
+	*numlocks = 0;
+	*hasrefs = 0;
+
 	ret = -EINVAL;
 	if (res->owner == DLM_LOCK_RES_OWNER_UNKNOWN) {
 		mlog(0, "cannot migrate lockres with unknown owner!\n");
@@ -2386,7 +2390,13 @@ static int dlm_is_lockres_migrateable(struct dlm_ctxt *dlm,
 	}
 
 	*numlocks = count;
-	mlog(0, "migrateable lockres having %d locks\n", *numlocks);
+
+	count = find_next_bit(res->refmap, O2NM_MAX_NODES, 0);
+	if (count < O2NM_MAX_NODES)
+		*hasrefs = 1;
+
+	mlog(0, "%s: res %.*s, Migrateable, locks %d, refs %d\n", dlm->name,
+	     res->lockname.len, res->lockname.name, *numlocks, *hasrefs);
 
 leave:
 	return ret;
@@ -2408,7 +2418,7 @@ static int dlm_migrate_lockres(struct dlm_ctxt *dlm,
 	const char *name;
 	unsigned int namelen;
 	int mle_added = 0;
-	int numlocks;
+	int numlocks, hasrefs;
 	int wake = 0;
 
 	if (!dlm_grab(dlm))
@@ -2417,13 +2427,13 @@ static int dlm_migrate_lockres(struct dlm_ctxt *dlm,
 	name = res->lockname.name;
 	namelen = res->lockname.len;
 
-	mlog(0, "migrating %.*s to %u\n", namelen, name, target);
+	mlog(0, "%s: Migrating %.*s to %u\n", dlm->name, namelen, name, target);
 
 	/*
 	 * ensure this lockres is a proper candidate for migration
 	 */
 	spin_lock(&res->spinlock);
-	ret = dlm_is_lockres_migrateable(dlm, res, &numlocks);
+	ret = dlm_is_lockres_migrateable(dlm, res, &numlocks, &hasrefs);
 	if (ret < 0) {
 		spin_unlock(&res->spinlock);
 		goto leave;
@@ -2431,10 +2441,8 @@ static int dlm_migrate_lockres(struct dlm_ctxt *dlm,
 	spin_unlock(&res->spinlock);
 
 	/* no work to do */
-	if (numlocks == 0) {
-		mlog(0, "no locks were found on this lockres! done!\n");
+	if (numlocks == 0 && !hasrefs)
 		goto leave;
-	}
 
 	/*
 	 * preallocate up front
@@ -2459,14 +2467,14 @@ static int dlm_migrate_lockres(struct dlm_ctxt *dlm,
 	 * find a node to migrate the lockres to
 	 */
 
-	mlog(0, "picking a migration node\n");
 	spin_lock(&dlm->spinlock);
 	/* pick a new node */
 	if (!test_bit(target, dlm->domain_map) ||
 	    target >= O2NM_MAX_NODES) {
 		target = dlm_pick_migration_target(dlm, res);
 	}
-	mlog(0, "node %u chosen for migration\n", target);
+	mlog(0, "%s: res %.*s, Node %u chosen for migration\n", dlm->name,
+	     namelen, name, target);
 
 	if (target >= O2NM_MAX_NODES ||
 	    !test_bit(target, dlm->domain_map)) {
@@ -2667,7 +2675,7 @@ int dlm_empty_lockres(struct dlm_ctxt *dlm, struct dlm_lock_resource *res)
 {
 	int ret;
 	int lock_dropped = 0;
-	int numlocks;
+	int numlocks, hasrefs;
 
 	spin_lock(&res->spinlock);
 	if (res->owner != dlm->node_num) {
@@ -2681,8 +2689,8 @@ int dlm_empty_lockres(struct dlm_ctxt *dlm, struct dlm_lock_resource *res)
 	}
 
 	/* No need to migrate a lockres having no locks */
-	ret = dlm_is_lockres_migrateable(dlm, res, &numlocks);
-	if (ret >= 0 && numlocks == 0) {
+	ret = dlm_is_lockres_migrateable(dlm, res, &numlocks, &hasrefs);
+	if (ret >= 0 && numlocks == 0 && !hasrefs) {
 		spin_unlock(&res->spinlock);
 		goto leave;
 	}
@@ -2915,6 +2923,12 @@ static u8 dlm_pick_migration_target(struct dlm_ctxt *dlm,
 		}
 		queue++;
 	}
+
+	nodenum = find_next_bit(res->refmap, O2NM_MAX_NODES, 0);
+	if (nodenum < O2NM_MAX_NODES) {
+		spin_unlock(&res->spinlock);
+		return nodenum;
+	}
 	spin_unlock(&res->spinlock);
 	mlog(0, "have not found a suitable target yet! checking domain map\n");
 
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 77b4c04..f6cba56 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -2241,11 +2241,15 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 
 	mutex_lock(&inode->i_mutex);
 
+	ocfs2_iocb_clear_sem_locked(iocb);
+
 relock:
 	/* to match setattr's i_mutex -> i_alloc_sem -> rw_lock ordering */
 	if (direct_io) {
 		down_read(&inode->i_alloc_sem);
 		have_alloc_sem = 1;
+		/* communicate with ocfs2_dio_end_io */
+		ocfs2_iocb_set_sem_locked(iocb);
 	}
 
 	/*
@@ -2382,8 +2386,10 @@ out:
 		ocfs2_rw_unlock(inode, rw_level);
 
 out_sems:
-	if (have_alloc_sem)
+	if (have_alloc_sem) {
 		up_read(&inode->i_alloc_sem);
+		ocfs2_iocb_clear_sem_locked(iocb);
+	}
 
 	mutex_unlock(&inode->i_mutex);
 
@@ -2527,6 +2533,8 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
 		goto bail;
 	}
 
+	ocfs2_iocb_clear_sem_locked(iocb);
+
 	/*
 	 * buffered reads protect themselves in ->readpage().  O_DIRECT reads
 	 * need locks to protect pending reads from racing with truncate.
@@ -2534,6 +2542,7 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
 	if (filp->f_flags & O_DIRECT) {
 		down_read(&inode->i_alloc_sem);
 		have_alloc_sem = 1;
+		ocfs2_iocb_set_sem_locked(iocb);
 
 		ret = ocfs2_rw_lock(inode, 0);
 		if (ret < 0) {
@@ -2575,8 +2584,10 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
 	}
 
 bail:
-	if (have_alloc_sem)
+	if (have_alloc_sem) {
 		up_read(&inode->i_alloc_sem);
+		ocfs2_iocb_clear_sem_locked(iocb);
+	}
 	if (rw_level != -1)
 		ocfs2_rw_unlock(inode, rw_level);
 	mlog_exit(ret);
diff --git a/fs/ocfs2/ocfs2_fs.h b/fs/ocfs2/ocfs2_fs.h
index c2e4f82..bf2e776 100644
--- a/fs/ocfs2/ocfs2_fs.h
+++ b/fs/ocfs2/ocfs2_fs.h
@@ -350,7 +350,7 @@ enum {
 #define OCFS2_LAST_LOCAL_SYSTEM_INODE LOCAL_GROUP_QUOTA_SYSTEM_INODE
 	NUM_SYSTEM_INODES
 };
-#define NUM_GLOBAL_SYSTEM_INODES OCFS2_LAST_GLOBAL_SYSTEM_INODE
+#define NUM_GLOBAL_SYSTEM_INODES OCFS2_FIRST_LOCAL_SYSTEM_INODE
 #define NUM_LOCAL_SYSTEM_INODES	\
 		(NUM_SYSTEM_INODES - OCFS2_FIRST_LOCAL_SYSTEM_INODE)