5 files changed, 253 insertions, 10 deletions

diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index de5e4f2adfea..0ef5cc13fae2 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -66,7 +66,11 @@ struct btrfs_inode {
 	 */
 	struct btrfs_key location;
 
-	/* Lock for counters */
+	/*
+	 * Lock for counters and all fields used to determine if the inode is in
+	 * the log or not (last_trans, last_sub_trans, last_log_commit,
+	 * logged_trans).
+	 */
 	spinlock_t lock;
 
 	/* the extent_tree has caches of all the extent mappings to disk */
@@ -250,6 +254,9 @@ static inline bool btrfs_is_free_space_inode(struct inode *inode)
 
 static inline int btrfs_inode_in_log(struct inode *inode, u64 generation)
 {
+	int ret = 0;
+
+	spin_lock(&BTRFS_I(inode)->lock);
 	if (BTRFS_I(inode)->logged_trans == generation &&
 	    BTRFS_I(inode)->last_sub_trans <=
 	    BTRFS_I(inode)->last_log_commit &&
@@ -263,9 +270,10 @@ static inline int btrfs_inode_in_log(struct inode *inode, u64 generation)
 		 */
 		smp_mb();
 		if (list_empty(&BTRFS_I(inode)->extent_tree.modified_extents))
-			return 1;
+			ret = 1;
 	}
-	return 0;
+	spin_unlock(&BTRFS_I(inode)->lock);
+	return ret;
 }
 
 #define BTRFS_DIO_ORIG_BIO_SUBMITTED	0x1
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 150db5e50c2d..fd105c172c8b 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1811,7 +1811,9 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 	 * otherwise subsequent syncs to a file that's been synced in this
 	 * transaction will appear to have already occured.
 	 */
+	spin_lock(&BTRFS_I(inode)->lock);
 	BTRFS_I(inode)->last_sub_trans = root->log_transid;
+	spin_unlock(&BTRFS_I(inode)->lock);
 	if (num_written > 0) {
 		err = generic_write_sync(file, pos, num_written);
 		if (err < 0)
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 937050a2b68e..96b189b8898a 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -136,9 +136,11 @@ struct btrfs_pending_snapshot {
 static inline void btrfs_set_inode_last_trans(struct btrfs_trans_handle *trans,
 					      struct inode *inode)
 {
+	spin_lock(&BTRFS_I(inode)->lock);
 	BTRFS_I(inode)->last_trans = trans->transaction->transid;
 	BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
 	BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->root->last_log_commit;
+	spin_unlock(&BTRFS_I(inode)->lock);
 }
 
 int btrfs_end_transaction(struct btrfs_trans_handle *trans,
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 6c95159302dd..016c90fc85db 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -492,11 +492,19 @@ insert:
 
 		if (btrfs_inode_generation(eb, src_item) == 0) {
 			struct extent_buffer *dst_eb = path->nodes[0];
+			const u64 ino_size = btrfs_inode_size(eb, src_item);
 
+			/*
+			 * For regular files an ino_size == 0 is used only when
+			 * logging that an inode exists, as part of a directory
+			 * fsync, and the inode wasn't fsynced before. In this
+			 * case don't set the size of the inode in the fs/subvol
+			 * tree, otherwise we would be throwing valid data away.
+			 */
 			if (S_ISREG(btrfs_inode_mode(eb, src_item)) &&
-			    S_ISREG(btrfs_inode_mode(dst_eb, dst_item))) {
+			    S_ISREG(btrfs_inode_mode(dst_eb, dst_item)) &&
+			    ino_size != 0) {
 				struct btrfs_map_token token;
-				u64 ino_size = btrfs_inode_size(eb, src_item);
 
 				btrfs_init_map_token(&token);
 				btrfs_set_token_inode_size(dst_eb, dst_item,
@@ -3124,6 +3132,7 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root, struct inode *inode,
 			  struct btrfs_path *path,
 			  struct btrfs_path *dst_path, int key_type,
+			  struct btrfs_log_ctx *ctx,
 			  u64 min_offset, u64 *last_offset_ret)
 {
 	struct btrfs_key min_key;
@@ -3208,6 +3217,8 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 		src = path->nodes[0];
 		nritems = btrfs_header_nritems(src);
 		for (i = path->slots[0]; i < nritems; i++) {
+			struct btrfs_dir_item *di;
+
 			btrfs_item_key_to_cpu(src, &min_key, i);
 
 			if (min_key.objectid != ino || min_key.type != key_type)
@@ -3218,6 +3229,37 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 				err = ret;
 				goto done;
 			}
+
+			/*
+			 * We must make sure that when we log a directory entry,
+			 * the corresponding inode, after log replay, has a
+			 * matching link count. For example:
+			 *
+			 * touch foo
+			 * mkdir mydir
+			 * sync
+			 * ln foo mydir/bar
+			 * xfs_io -c "fsync" mydir
+			 * <crash>
+			 * <mount fs and log replay>
+			 *
+			 * Would result in a fsync log that when replayed, our
+			 * file inode would have a link count of 1, but we get
+			 * two directory entries pointing to the same inode.
+			 * After removing one of the names, it would not be
+			 * possible to remove the other name, which resulted
+			 * always in stale file handle errors, and would not
+			 * be possible to rmdir the parent directory, since
+			 * its i_size could never decrement to the value
+			 * BTRFS_EMPTY_DIR_SIZE, resulting in -ENOTEMPTY errors.
+			 */
+			di = btrfs_item_ptr(src, i, struct btrfs_dir_item);
+			btrfs_dir_item_key_to_cpu(src, di, &tmp);
+			if (ctx &&
+			    (btrfs_dir_transid(src, di) == trans->transid ||
+			     btrfs_dir_type(src, di) == BTRFS_FT_DIR) &&
+			    tmp.type != BTRFS_ROOT_ITEM_KEY)
+				ctx->log_new_dentries = true;
 		}
 		path->slots[0] = nritems;
 
@@ -3279,7 +3321,8 @@ done:
 static noinline int log_directory_changes(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root, struct inode *inode,
 			  struct btrfs_path *path,
-			  struct btrfs_path *dst_path)
+			  struct btrfs_path *dst_path,
+			  struct btrfs_log_ctx *ctx)
 {
 	u64 min_key;
 	u64 max_key;
@@ -3291,7 +3334,7 @@ again:
 	max_key = 0;
 	while (1) {
 		ret = log_dir_items(trans, root, inode, path,
-				    dst_path, key_type, min_key,
+				    dst_path, key_type, ctx, min_key,
 				    &max_key);
 		if (ret)
 			return ret;
@@ -4067,7 +4110,7 @@ static int logged_inode_size(struct btrfs_root *log, struct inode *inode,
 	if (ret < 0) {
 		return ret;
 	} else if (ret > 0) {
-		*size_ret = i_size_read(inode);
+		*size_ret = 0;
 	} else {
 		struct btrfs_inode_item *item;
 
@@ -4374,15 +4417,18 @@ log_extents:
 	}
 
 	if (inode_only == LOG_INODE_ALL && S_ISDIR(inode->i_mode)) {
-		ret = log_directory_changes(trans, root, inode, path, dst_path);
+		ret = log_directory_changes(trans, root, inode, path, dst_path,
+					    ctx);
 		if (ret) {
 			err = ret;
 			goto out_unlock;
 		}
 	}
 
+	spin_lock(&BTRFS_I(inode)->lock);
 	BTRFS_I(inode)->logged_trans = trans->transid;
 	BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->last_sub_trans;
+	spin_unlock(&BTRFS_I(inode)->lock);
 out_unlock:
 	if (unlikely(err))
 		btrfs_put_logged_extents(&logged_list);
@@ -4469,6 +4515,181 @@ out:
 	return ret;
 }
 
+struct btrfs_dir_list {
+	u64 ino;
+	struct list_head list;
+};
+
+/*
+ * Log the inodes of the new dentries of a directory. See log_dir_items() for
+ * details about the why it is needed.
+ * This is a recursive operation - if an existing dentry corresponds to a
+ * directory, that directory's new entries are logged too (same behaviour as
+ * ext3/4, xfs, f2fs, reiserfs, nilfs2). Note that when logging the inodes
+ * the dentries point to we do not lock their i_mutex, otherwise lockdep
+ * complains about the following circular lock dependency / possible deadlock:
+ *
+ *        CPU0                                        CPU1
+ *        ----                                        ----
+ * lock(&type->i_mutex_dir_key#3/2);
+ *                                            lock(sb_internal#2);
+ *                                            lock(&type->i_mutex_dir_key#3/2);
+ * lock(&sb->s_type->i_mutex_key#14);
+ *
+ * Where sb_internal is the lock (a counter that works as a lock) acquired by
+ * sb_start_intwrite() in btrfs_start_transaction().
+ * Not locking i_mutex of the inodes is still safe because:
+ *
+ * 1) For regular files we log with a mode of LOG_INODE_EXISTS. It's possible
+ *    that while logging the inode new references (names) are added or removed
+ *    from the inode, leaving the logged inode item with a link count that does
+ *    not match the number of logged inode reference items. This is fine because
+ *    at log replay time we compute the real number of links and correct the
+ *    link count in the inode item (see replay_one_buffer() and
+ *    link_to_fixup_dir());
+ *
+ * 2) For directories we log with a mode of LOG_INODE_ALL. It's possible that
+ *    while logging the inode's items new items with keys BTRFS_DIR_ITEM_KEY and
+ *    BTRFS_DIR_INDEX_KEY are added to fs/subvol tree and the logged inode item
+ *    has a size that doesn't match the sum of the lengths of all the logged
+ *    names. This does not result in a problem because if a dir_item key is
+ *    logged but its matching dir_index key is not logged, at log replay time we
+ *    don't use it to replay the respective name (see replay_one_name()). On the
+ *    other hand if only the dir_index key ends up being logged, the respective
+ *    name is added to the fs/subvol tree with both the dir_item and dir_index
+ *    keys created (see replay_one_name()).
+ *    The directory's inode item with a wrong i_size is not a problem as well,
+ *    since we don't use it at log replay time to set the i_size in the inode
+ *    item of the fs/subvol tree (see overwrite_item()).
+ */
+static int log_new_dir_dentries(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root,
+				struct inode *start_inode,
+				struct btrfs_log_ctx *ctx)
+{
+	struct btrfs_root *log = root->log_root;
+	struct btrfs_path *path;
+	LIST_HEAD(dir_list);
+	struct btrfs_dir_list *dir_elem;
+	int ret = 0;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	dir_elem = kmalloc(sizeof(*dir_elem), GFP_NOFS);
+	if (!dir_elem) {
+		btrfs_free_path(path);
+		return -ENOMEM;
+	}
+	dir_elem->ino = btrfs_ino(start_inode);
+	list_add_tail(&dir_elem->list, &dir_list);
+
+	while (!list_empty(&dir_list)) {
+		struct extent_buffer *leaf;
+		struct btrfs_key min_key;
+		int nritems;
+		int i;
+
+		dir_elem = list_first_entry(&dir_list, struct btrfs_dir_list,
+					    list);
+		if (ret)
+			goto next_dir_inode;
+
+		min_key.objectid = dir_elem->ino;
+		min_key.type = BTRFS_DIR_ITEM_KEY;
+		min_key.offset = 0;
+again:
+		btrfs_release_path(path);
+		ret = btrfs_search_forward(log, &min_key, path, trans->transid);
+		if (ret < 0) {
+			goto next_dir_inode;
+		} else if (ret > 0) {
+			ret = 0;
+			goto next_dir_inode;
+		}
+
+process_leaf:
+		leaf = path->nodes[0];
+		nritems = btrfs_header_nritems(leaf);
+		for (i = path->slots[0]; i < nritems; i++) {
+			struct btrfs_dir_item *di;
+			struct btrfs_key di_key;
+			struct inode *di_inode;
+			struct btrfs_dir_list *new_dir_elem;
+			int log_mode = LOG_INODE_EXISTS;
+			int type;
+
+			btrfs_item_key_to_cpu(leaf, &min_key, i);
+			if (min_key.objectid != dir_elem->ino ||
+			    min_key.type != BTRFS_DIR_ITEM_KEY)
+				goto next_dir_inode;
+
+			di = btrfs_item_ptr(leaf, i, struct btrfs_dir_item);
+			type = btrfs_dir_type(leaf, di);
+			if (btrfs_dir_transid(leaf, di) < trans->transid &&
+			    type != BTRFS_FT_DIR)
+				continue;
+			btrfs_dir_item_key_to_cpu(leaf, di, &di_key);
+			if (di_key.type == BTRFS_ROOT_ITEM_KEY)
+				continue;
+
+			di_inode = btrfs_iget(root->fs_info->sb, &di_key,
+					      root, NULL);
+			if (IS_ERR(di_inode)) {
+				ret = PTR_ERR(di_inode);
+				goto next_dir_inode;
+			}
+
+			if (btrfs_inode_in_log(di_inode, trans->transid)) {
+				iput(di_inode);
+				continue;
+			}
+
+			ctx->log_new_dentries = false;
+			if (type == BTRFS_FT_DIR)
+				log_mode = LOG_INODE_ALL;
+			btrfs_release_path(path);
+			ret = btrfs_log_inode(trans, root, di_inode,
+					      log_mode, 0, LLONG_MAX, ctx);
+			iput(di_inode);
+			if (ret)
+				goto next_dir_inode;
+			if (ctx->log_new_dentries) {
+				new_dir_elem = kmalloc(sizeof(*new_dir_elem),
+						       GFP_NOFS);
+				if (!new_dir_elem) {
+					ret = -ENOMEM;
+					goto next_dir_inode;
+				}
+				new_dir_elem->ino = di_key.objectid;
+				list_add_tail(&new_dir_elem->list, &dir_list);
+			}
+			break;
+		}
+		if (i == nritems) {
+			ret = btrfs_next_leaf(log, path);
+			if (ret < 0) {
+				goto next_dir_inode;
+			} else if (ret > 0) {
+				ret = 0;
+				goto next_dir_inode;
+			}
+			goto process_leaf;
+		}
+		if (min_key.offset < (u64)-1) {
+			min_key.offset++;
+			goto again;
+		}
+next_dir_inode:
+		list_del(&dir_elem->list);
+		kfree(dir_elem);
+	}
+
+	btrfs_free_path(path);
+	return ret;
+}
+
 /*
  * helper function around btrfs_log_inode to make sure newly created
  * parent directories also end up in the log.  A minimal inode and backref
@@ -4491,6 +4712,8 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 	const struct dentry * const first_parent = parent;
 	const bool did_unlink = (BTRFS_I(inode)->last_unlink_trans >
 				 last_committed);
+	bool log_dentries = false;
+	struct inode *orig_inode = inode;
 
 	sb = inode->i_sb;
 
@@ -4546,6 +4769,9 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 		goto end_trans;
 	}
 
+	if (S_ISDIR(inode->i_mode) && ctx && ctx->log_new_dentries)
+		log_dentries = true;
+
 	while (1) {
 		if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
 			break;
@@ -4582,7 +4808,10 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 		dput(old_parent);
 		old_parent = parent;
 	}
-	ret = 0;
+	if (log_dentries)
+		ret = log_new_dir_dentries(trans, root, orig_inode, ctx);
+	else
+		ret = 0;
 end_trans:
 	dput(old_parent);
 	if (ret < 0) {
diff --git a/fs/btrfs/tree-log.h b/fs/btrfs/tree-log.h
index 154990c26dcb..6916a781ea02 100644
--- a/fs/btrfs/tree-log.h
+++ b/fs/btrfs/tree-log.h
@@ -29,6 +29,7 @@ struct btrfs_log_ctx {
 	int log_ret;
 	int log_transid;
 	int io_err;
+	bool log_new_dentries;
 	struct list_head list;
 };
 
@@ -37,6 +38,7 @@ static inline void btrfs_init_log_ctx(struct btrfs_log_ctx *ctx)
 	ctx->log_ret = 0;
 	ctx->log_transid = 0;
 	ctx->io_err = 0;
+	ctx->log_new_dentries = false;
 	INIT_LIST_HEAD(&ctx->list);
 }