Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs update from Chris Mason:
 "This is a large pull, with the bulk of the updates coming from:

   - Hole punching

   - send/receive fixes

   - fsync performance

   - Disk format extension allowing more hardlinks inside a single
     directory (btrfs-progs patch required to enable the compat bit for
     this one)

  I'm cooking more unrelated RAID code, but I wanted to make sure this
  original batch makes it in.  The largest updates here are relatively
  old and have been in testing for some time."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (121 commits)
  btrfs: init ref_index to zero in add_inode_ref
  Btrfs: remove repeated eb->pages check in, disk-io.c/csum_dirty_buffer
  Btrfs: fix page leakage
  Btrfs: do not warn_on when we cannot alloc a page for an extent buffer
  Btrfs: don't bug on enomem in readpage
  Btrfs: cleanup pages properly when ENOMEM in compression
  Btrfs: make filesystem read-only when submitting barrier fails
  Btrfs: detect corrupted filesystem after write I/O errors
  Btrfs: make compress and nodatacow mount options mutually exclusive
  btrfs: fix message printing
  Btrfs: don't bother committing delayed inode updates when fsyncing
  btrfs: move inline function code to header file
  Btrfs: remove unnecessary IS_ERR in bio_readpage_error()
  btrfs: remove unused function btrfs_insert_some_items()
  Btrfs: don't commit instead of overcommitting
  Btrfs: confirmation of value is added before trace_btrfs_get_extent() is called
  Btrfs: be smarter about dropping things from the tree log
  Btrfs: don't lookup csums for prealloc extents
  Btrfs: cache extent state when writing out dirty metadata pages
  Btrfs: do not hold the file extent leaf locked when adding extent item
  ...

fs/btrfs/backref.c

@@ -16,6 +16,7 @@
  * Boston, MA 021110-1307, USA.
  */
 
+#include <linux/vmalloc.h>
 #include "ctree.h"
 #include "disk-io.h"
 #include "backref.h"
@@ -231,7 +232,7 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 			}
 			if (!ret) {
 				ret = ulist_add(parents, eb->start,
-						(unsigned long)eie, GFP_NOFS);
+						(uintptr_t)eie, GFP_NOFS);
 				if (ret < 0)
 					break;
 				if (!extent_item_pos) {
@@ -363,8 +364,8 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 		ULIST_ITER_INIT(&uiter);
 		node = ulist_next(parents, &uiter);
 		ref->parent = node ? node->val : 0;
-		ref->inode_list =
-			node ? (struct extent_inode_elem *)node->aux : 0;
+		ref->inode_list = node ?
+			(struct extent_inode_elem *)(uintptr_t)node->aux : 0;
 
 		/* additional parents require new refs being added here */
 		while ((node = ulist_next(parents, &uiter))) {
@@ -375,8 +376,8 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 			}
 			memcpy(new_ref, ref, sizeof(*ref));
 			new_ref->parent = node->val;
-			new_ref->inode_list =
-					(struct extent_inode_elem *)node->aux;
+			new_ref->inode_list = (struct extent_inode_elem *)
+							(uintptr_t)node->aux;
 			list_add(&new_ref->list, &ref->list);
 		}
 		ulist_reinit(parents);
@@ -914,8 +915,8 @@ again:
 				free_extent_buffer(eb);
 			}
 			ret = ulist_add_merge(refs, ref->parent,
-					      (unsigned long)ref->inode_list,
-					      (unsigned long *)&eie, GFP_NOFS);
+					      (uintptr_t)ref->inode_list,
+					      (u64 *)&eie, GFP_NOFS);
 			if (!ret && extent_item_pos) {
 				/*
 				 * we've recorded that parent, so we must extend
@@ -959,7 +960,7 @@ static void free_leaf_list(struct ulist *blocks)
 	while ((node = ulist_next(blocks, &uiter))) {
 		if (!node->aux)
 			continue;
-		eie = (struct extent_inode_elem *)node->aux;
+		eie = (struct extent_inode_elem *)(uintptr_t)node->aux;
 		for (; eie; eie = eie_next) {
 			eie_next = eie->next;
 			kfree(eie);
@@ -1108,26 +1109,80 @@ static int inode_ref_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
 				found_key);
 }
 
-/*
- * this iterates to turn a btrfs_inode_ref into a full filesystem path. elements
- * of the path are separated by '/' and the path is guaranteed to be
- * 0-terminated. the path is only given within the current file system.
- * Therefore, it never starts with a '/'. the caller is responsible to provide
- * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
- * the start point of the resulting string is returned. this pointer is within
- * dest, normally.
- * in case the path buffer would overflow, the pointer is decremented further
- * as if output was written to the buffer, though no more output is actually
- * generated. that way, the caller can determine how much space would be
- * required for the path to fit into the buffer. in that case, the returned
- * value will be smaller than dest. callers must check this!
- */
-char *btrfs_iref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
-			 struct btrfs_inode_ref *iref,
+int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
+			  u64 start_off, struct btrfs_path *path,
+			  struct btrfs_inode_extref **ret_extref,
+			  u64 *found_off)
+{
+	int ret, slot;
+	struct btrfs_key key;
+	struct btrfs_key found_key;
+	struct btrfs_inode_extref *extref;
+	struct extent_buffer *leaf;
+	unsigned long ptr;
+
+	key.objectid = inode_objectid;
+	btrfs_set_key_type(&key, BTRFS_INODE_EXTREF_KEY);
+	key.offset = start_off;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+
+	while (1) {
+		leaf = path->nodes[0];
+		slot = path->slots[0];
+		if (slot >= btrfs_header_nritems(leaf)) {
+			/*
+			 * If the item at offset is not found,
+			 * btrfs_search_slot will point us to the slot
+			 * where it should be inserted. In our case
+			 * that will be the slot directly before the
+			 * next INODE_REF_KEY_V2 item. In the case
+			 * that we're pointing to the last slot in a
+			 * leaf, we must move one leaf over.
+			 */
+			ret = btrfs_next_leaf(root, path);
+			if (ret) {
+				if (ret >= 1)
+					ret = -ENOENT;
+				break;
+			}
+			continue;
+		}
+
+		btrfs_item_key_to_cpu(leaf, &found_key, slot);
+
+		/*
+		 * Check that we're still looking at an extended ref key for
+		 * this particular objectid. If we have different
+		 * objectid or type then there are no more to be found
+		 * in the tree and we can exit.
+		 */
+		ret = -ENOENT;
+		if (found_key.objectid != inode_objectid)
+			break;
+		if (btrfs_key_type(&found_key) != BTRFS_INODE_EXTREF_KEY)
+			break;
+
+		ret = 0;
+		ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
+		extref = (struct btrfs_inode_extref *)ptr;
+		*ret_extref = extref;
+		if (found_off)
+			*found_off = found_key.offset;
+		break;
+	}
+
+	return ret;
+}
+
+static char *ref_to_path(struct btrfs_root *fs_root,
+			 struct btrfs_path *path,
+			 u32 name_len, unsigned long name_off,
 			 struct extent_buffer *eb_in, u64 parent,
 			 char *dest, u32 size)
 {
-	u32 len;
 	int slot;
 	u64 next_inum;
 	int ret;
@@ -1135,17 +1190,17 @@ char *btrfs_iref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 	struct extent_buffer *eb = eb_in;
 	struct btrfs_key found_key;
 	int leave_spinning = path->leave_spinning;
+	struct btrfs_inode_ref *iref;
 
 	if (bytes_left >= 0)
 		dest[bytes_left] = '\0';
 
 	path->leave_spinning = 1;
 	while (1) {
-		len = btrfs_inode_ref_name_len(eb, iref);
-		bytes_left -= len;
+		bytes_left -= name_len;
 		if (bytes_left >= 0)
 			read_extent_buffer(eb, dest + bytes_left,
-						(unsigned long)(iref + 1), len);
+					   name_off, name_len);
 		if (eb != eb_in) {
 			btrfs_tree_read_unlock_blocking(eb);
 			free_extent_buffer(eb);
@@ -1155,6 +1210,7 @@ char *btrfs_iref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 			ret = -ENOENT;
 		if (ret)
 			break;
+
 		next_inum = found_key.offset;
 
 		/* regular exit ahead */
@@ -1170,8 +1226,11 @@ char *btrfs_iref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 			btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
 		}
 		btrfs_release_path(path);
-
 		iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
+
+		name_len = btrfs_inode_ref_name_len(eb, iref);
+		name_off = (unsigned long)(iref + 1);
+
 		parent = next_inum;
 		--bytes_left;
 		if (bytes_left >= 0)
@@ -1187,13 +1246,40 @@ char *btrfs_iref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 	return dest + bytes_left;
 }
 
+/*
+ * this iterates to turn a btrfs_inode_ref into a full filesystem path. elements
+ * of the path are separated by '/' and the path is guaranteed to be
+ * 0-terminated. the path is only given within the current file system.
+ * Therefore, it never starts with a '/'. the caller is responsible to provide
+ * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
+ * the start point of the resulting string is returned. this pointer is within
+ * dest, normally.
+ * in case the path buffer would overflow, the pointer is decremented further
+ * as if output was written to the buffer, though no more output is actually
+ * generated. that way, the caller can determine how much space would be
+ * required for the path to fit into the buffer. in that case, the returned
+ * value will be smaller than dest. callers must check this!
+ */
+char *btrfs_iref_to_path(struct btrfs_root *fs_root,
+			 struct btrfs_path *path,
+			 struct btrfs_inode_ref *iref,
+			 struct extent_buffer *eb_in, u64 parent,
+			 char *dest, u32 size)
+{
+	return ref_to_path(fs_root, path,
+			   btrfs_inode_ref_name_len(eb_in, iref),
+			   (unsigned long)(iref + 1),
+			   eb_in, parent, dest, size);
+}
+
 /*
  * this makes the path point to (logical EXTENT_ITEM *)
  * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for
  * tree blocks and <0 on error.
  */
 int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
-			struct btrfs_path *path, struct btrfs_key *found_key)
+			struct btrfs_path *path, struct btrfs_key *found_key,
+			u64 *flags_ret)
 {
 	int ret;
 	u64 flags;
@@ -1237,10 +1323,17 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
 		 (unsigned long long)found_key->objectid,
 		 (unsigned long long)found_key->offset,
 		 (unsigned long long)flags, item_size);
-	if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
-		return BTRFS_EXTENT_FLAG_TREE_BLOCK;
-	if (flags & BTRFS_EXTENT_FLAG_DATA)
-		return BTRFS_EXTENT_FLAG_DATA;
+
+	WARN_ON(!flags_ret);
+	if (flags_ret) {
+		if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+			*flags_ret = BTRFS_EXTENT_FLAG_TREE_BLOCK;
+		else if (flags & BTRFS_EXTENT_FLAG_DATA)
+			*flags_ret = BTRFS_EXTENT_FLAG_DATA;
+		else
+			BUG_ON(1);
+		return 0;
+	}
 
 	return -EIO;
 }
@@ -1404,12 +1497,13 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 		ULIST_ITER_INIT(&root_uiter);
 		while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
 			pr_debug("root %llu references leaf %llu, data list "
-				 "%#lx\n", root_node->val, ref_node->val,
-				 ref_node->aux);
-			ret = iterate_leaf_refs(
-				(struct extent_inode_elem *)ref_node->aux,
-				root_node->val, extent_item_objectid,
-				iterate, ctx);
+				 "%#llx\n", root_node->val, ref_node->val,
+				 (long long)ref_node->aux);
+			ret = iterate_leaf_refs((struct extent_inode_elem *)
+						(uintptr_t)ref_node->aux,
+						root_node->val,
+						extent_item_objectid,
+						iterate, ctx);
 		}
 		ulist_free(roots);
 		roots = NULL;
@@ -1432,15 +1526,15 @@ int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
 {
 	int ret;
 	u64 extent_item_pos;
+	u64 flags = 0;
 	struct btrfs_key found_key;
 	int search_commit_root = path->search_commit_root;
 
-	ret = extent_from_logical(fs_info, logical, path,
-					&found_key);
+	ret = extent_from_logical(fs_info, logical, path, &found_key, &flags);
 	btrfs_release_path(path);
 	if (ret < 0)
 		return ret;
-	if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+	if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
 		return -EINVAL;
 
 	extent_item_pos = logical - found_key.objectid;
@@ -1451,9 +1545,12 @@ int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
 	return ret;
 }
 
-static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
-				struct btrfs_path *path,
-				iterate_irefs_t *iterate, void *ctx)
+typedef int (iterate_irefs_t)(u64 parent, u32 name_len, unsigned long name_off,
+			      struct extent_buffer *eb, void *ctx);
+
+static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
+			      struct btrfs_path *path,
+			      iterate_irefs_t *iterate, void *ctx)
 {
 	int ret = 0;
 	int slot;
@@ -1470,7 +1567,7 @@ static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
 	while (!ret) {
 		path->leave_spinning = 1;
 		ret = inode_ref_info(inum, parent ? parent+1 : 0, fs_root, path,
-					&found_key);
+				     &found_key);
 		if (ret < 0)
 			break;
 		if (ret) {
@@ -1498,7 +1595,8 @@ static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
 				 "tree %llu\n", cur,
 				 (unsigned long long)found_key.objectid,
 				 (unsigned long long)fs_root->objectid);
-			ret = iterate(parent, iref, eb, ctx);
+			ret = iterate(parent, name_len,
+				      (unsigned long)(iref + 1), eb, ctx);
 			if (ret)
 				break;
 			len = sizeof(*iref) + name_len;
@@ -1513,12 +1611,98 @@ static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
 	return ret;
 }
 
+static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
+				 struct btrfs_path *path,
+				 iterate_irefs_t *iterate, void *ctx)
+{
+	int ret;
+	int slot;
+	u64 offset = 0;
+	u64 parent;
+	int found = 0;
+	struct extent_buffer *eb;
+	struct btrfs_inode_extref *extref;
+	struct extent_buffer *leaf;
+	u32 item_size;
+	u32 cur_offset;
+	unsigned long ptr;
+
+	while (1) {
+		ret = btrfs_find_one_extref(fs_root, inum, offset, path, &extref,
+					    &offset);
+		if (ret < 0)
+			break;
+		if (ret) {
+			ret = found ? 0 : -ENOENT;
+			break;
+		}
+		++found;
+
+		slot = path->slots[0];
+		eb = path->nodes[0];
+		/* make sure we can use eb after releasing the path */
+		atomic_inc(&eb->refs);
+
+		btrfs_tree_read_lock(eb);
+		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
+		btrfs_release_path(path);
+
+		leaf = path->nodes[0];
+		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+		ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
+		cur_offset = 0;
+
+		while (cur_offset < item_size) {
+			u32 name_len;
+
+			extref = (struct btrfs_inode_extref *)(ptr + cur_offset);
+			parent = btrfs_inode_extref_parent(eb, extref);
+			name_len = btrfs_inode_extref_name_len(eb, extref);
+			ret = iterate(parent, name_len,
+				      (unsigned long)&extref->name, eb, ctx);
+			if (ret)
+				break;
+
+			cur_offset += btrfs_inode_extref_name_len(leaf, extref);
+			cur_offset += sizeof(*extref);
+		}
+		btrfs_tree_read_unlock_blocking(eb);
+		free_extent_buffer(eb);
+
+		offset++;
+	}
+
+	btrfs_release_path(path);
+
+	return ret;
+}
+
+static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
+			 struct btrfs_path *path, iterate_irefs_t *iterate,
+			 void *ctx)
+{
+	int ret;
+	int found_refs = 0;
+
+	ret = iterate_inode_refs(inum, fs_root, path, iterate, ctx);
+	if (!ret)
+		++found_refs;
+	else if (ret != -ENOENT)
+		return ret;
+
+	ret = iterate_inode_extrefs(inum, fs_root, path, iterate, ctx);
+	if (ret == -ENOENT && found_refs)
+		return 0;
+
+	return ret;
+}
+
 /*
  * returns 0 if the path could be dumped (probably truncated)
  * returns <0 in case of an error
  */
-static int inode_to_path(u64 inum, struct btrfs_inode_ref *iref,
-				struct extent_buffer *eb, void *ctx)
+static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
+			 struct extent_buffer *eb, void *ctx)
 {
 	struct inode_fs_paths *ipath = ctx;
 	char *fspath;
@@ -1531,20 +1715,17 @@ static int inode_to_path(u64 inum, struct btrfs_inode_ref *iref,
 					ipath->fspath->bytes_left - s_ptr : 0;
 
 	fspath_min = (char *)ipath->fspath->val + (i + 1) * s_ptr;
-	fspath = btrfs_iref_to_path(ipath->fs_root, ipath->btrfs_path, iref, eb,
-				inum, fspath_min, bytes_left);
+	fspath = ref_to_path(ipath->fs_root, ipath->btrfs_path, name_len,
+			     name_off, eb, inum, fspath_min,
+			     bytes_left);
 	if (IS_ERR(fspath))
 		return PTR_ERR(fspath);
 
 	if (fspath > fspath_min) {
-		pr_debug("path resolved: %s\n", fspath);
 		ipath->fspath->val[i] = (u64)(unsigned long)fspath;
 		++ipath->fspath->elem_cnt;
 		ipath->fspath->bytes_left = fspath - fspath_min;
 	} else {
-		pr_debug("missed path, not enough space. missing bytes: %lu, "
-			 "constructed so far: %s\n",
-			 (unsigned long)(fspath_min - fspath), fspath_min);
 		++ipath->fspath->elem_missed;
 		ipath->fspath->bytes_missing += fspath_min - fspath;
 		ipath->fspath->bytes_left = 0;
@@ -1566,7 +1747,7 @@ static int inode_to_path(u64 inum, struct btrfs_inode_ref *iref,
 int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
 {
 	return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path,
-				inode_to_path, ipath);
+			     inode_to_path, ipath);
 }
 
 struct btrfs_data_container *init_data_container(u32 total_bytes)
@@ -1575,7 +1756,7 @@ struct btrfs_data_container *init_data_container(u32 total_bytes)
 	size_t alloc_bytes;
 
 	alloc_bytes = max_t(size_t, total_bytes, sizeof(*data));
-	data = kmalloc(alloc_bytes, GFP_NOFS);
+	data = vmalloc(alloc_bytes);
 	if (!data)
 		return ERR_PTR(-ENOMEM);
 
@@ -1626,6 +1807,6 @@ void free_ipath(struct inode_fs_paths *ipath)
 {
 	if (!ipath)
 		return;
-	kfree(ipath->fspath);
+	vfree(ipath->fspath);
 	kfree(ipath);
 }

fs/btrfs/backref.h

@@ -33,14 +33,13 @@ struct inode_fs_paths {
 
 typedef int (iterate_extent_inodes_t)(u64 inum, u64 offset, u64 root,
 		void *ctx);
-typedef int (iterate_irefs_t)(u64 parent, struct btrfs_inode_ref *iref,
-				struct extent_buffer *eb, void *ctx);
 
 int inode_item_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
 			struct btrfs_path *path);
 
 int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
-			struct btrfs_path *path, struct btrfs_key *found_key);
+			struct btrfs_path *path, struct btrfs_key *found_key,
+			u64 *flags);
 
 int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
 				struct btrfs_extent_item *ei, u32 item_size,
@@ -69,4 +68,9 @@ struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
 					struct btrfs_path *path);
 void free_ipath(struct inode_fs_paths *ipath);
 
+int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
+			  u64 start_off, struct btrfs_path *path,
+			  struct btrfs_inode_extref **ret_extref,
+			  u64 *found_off);
+
 #endif

fs/btrfs/btrfs_inode.h

@@ -38,6 +38,7 @@
 #define BTRFS_INODE_DELALLOC_META_RESERVED	4
 #define BTRFS_INODE_HAS_ORPHAN_ITEM		5
 #define BTRFS_INODE_HAS_ASYNC_EXTENT		6
+#define BTRFS_INODE_NEEDS_FULL_SYNC		7
 
 /* in memory btrfs inode */
 struct btrfs_inode {
@@ -143,6 +144,9 @@ struct btrfs_inode {
 	/* flags field from the on disk inode */
 	u32 flags;
 
+	/* a local copy of root's last_log_commit */
+	unsigned long last_log_commit;
+
 	/*
 	 * Counters to keep track of the number of extent item's we may use due
 	 * to delalloc and such.  outstanding_extents is the number of extent
@@ -202,15 +206,10 @@ static inline bool btrfs_is_free_space_inode(struct inode *inode)
 
 static inline int btrfs_inode_in_log(struct inode *inode, u64 generation)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	int ret = 0;
-
-	mutex_lock(&root->log_mutex);
 	if (BTRFS_I(inode)->logged_trans == generation &&
-	    BTRFS_I(inode)->last_sub_trans <= root->last_log_commit)
-		ret = 1;
-	mutex_unlock(&root->log_mutex);
-	return ret;
+	    BTRFS_I(inode)->last_sub_trans <= BTRFS_I(inode)->last_log_commit)
+		return 1;
+	return 0;
 }
 
 #endif

fs/btrfs/check-integrity.c

@@ -37,8 +37,9 @@
  *        the file system was mounted, (i.e., they have been
  *        referenced by the super block) or they have been
  *        written since then and the write completion callback
- *        was called and a FLUSH request to the device where
- *        these blocks are located was received and completed.
+ *        was called and no write error was indicated and a
+ *        FLUSH request to the device where these blocks are
+ *        located was received and completed.
  *    2b. All referenced blocks need to have a generation
  *        number which is equal to the parent's number.
  *
@@ -2601,6 +2602,17 @@ static int btrfsic_check_all_ref_blocks(struct btrfsic_state *state,
 			       (unsigned long long)l->block_ref_to->dev_bytenr,
 			       l->block_ref_to->mirror_num);
 			ret = -1;
+		} else if (l->block_ref_to->iodone_w_error) {
+			printk(KERN_INFO "btrfs: attempt to write superblock"
+			       " which references block %c @%llu (%s/%llu/%d)"
+			       " which has write error!\n",
+			       btrfsic_get_block_type(state, l->block_ref_to),
+			       (unsigned long long)
+			       l->block_ref_to->logical_bytenr,
+			       l->block_ref_to->dev_state->name,
+			       (unsigned long long)l->block_ref_to->dev_bytenr,
+			       l->block_ref_to->mirror_num);
+			ret = -1;
 		} else if (l->parent_generation !=
 			   l->block_ref_to->generation &&
 			   BTRFSIC_GENERATION_UNKNOWN !=

fs/btrfs/compression.c

@@ -577,6 +577,7 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 	u64 em_start;
 	struct extent_map *em;
 	int ret = -ENOMEM;
+	int faili = 0;
 	u32 *sums;
 
 	tree = &BTRFS_I(inode)->io_tree;
@@ -626,9 +627,13 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 	for (pg_index = 0; pg_index < nr_pages; pg_index++) {
 		cb->compressed_pages[pg_index] = alloc_page(GFP_NOFS |
 							      __GFP_HIGHMEM);
-		if (!cb->compressed_pages[pg_index])
+		if (!cb->compressed_pages[pg_index]) {
+			faili = pg_index - 1;
+			ret = -ENOMEM;
 			goto fail2;
+		}
 	}
+	faili = nr_pages - 1;
 	cb->nr_pages = nr_pages;
 
 	add_ra_bio_pages(inode, em_start + em_len, cb);
@@ -713,8 +718,10 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 	return 0;
 
 fail2:
-	for (pg_index = 0; pg_index < nr_pages; pg_index++)
-		free_page((unsigned long)cb->compressed_pages[pg_index]);
+	while (faili >= 0) {
+		__free_page(cb->compressed_pages[faili]);
+		faili--;
+	}
 
 	kfree(cb->compressed_pages);
 fail1:

fs/btrfs/ctree.c

@@ -4401,149 +4401,6 @@ void btrfs_extend_item(struct btrfs_trans_handle *trans,
 	}
 }
 
-/*
- * Given a key and some data, insert items into the tree.
- * This does all the path init required, making room in the tree if needed.
- * Returns the number of keys that were inserted.
- */
-int btrfs_insert_some_items(struct btrfs_trans_handle *trans,
-			    struct btrfs_root *root,
-			    struct btrfs_path *path,
-			    struct btrfs_key *cpu_key, u32 *data_size,
-			    int nr)
-{
-	struct extent_buffer *leaf;
-	struct btrfs_item *item;
-	int ret = 0;
-	int slot;
-	int i;
-	u32 nritems;
-	u32 total_data = 0;
-	u32 total_size = 0;
-	unsigned int data_end;
-	struct btrfs_disk_key disk_key;
-	struct btrfs_key found_key;
-	struct btrfs_map_token token;
-
-	btrfs_init_map_token(&token);
-
-	for (i = 0; i < nr; i++) {
-		if (total_size + data_size[i] + sizeof(struct btrfs_item) >
-		    BTRFS_LEAF_DATA_SIZE(root)) {
-			break;
-			nr = i;
-		}
-		total_data += data_size[i];
-		total_size += data_size[i] + sizeof(struct btrfs_item);
-	}
-	BUG_ON(nr == 0);
-
-	ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1);
-	if (ret == 0)
-		return -EEXIST;
-	if (ret < 0)
-		goto out;
-
-	leaf = path->nodes[0];
-
-	nritems = btrfs_header_nritems(leaf);
-	data_end = leaf_data_end(root, leaf);
-
-	if (btrfs_leaf_free_space(root, leaf) < total_size) {
-		for (i = nr; i >= 0; i--) {
-			total_data -= data_size[i];
-			total_size -= data_size[i] + sizeof(struct btrfs_item);
-			if (total_size < btrfs_leaf_free_space(root, leaf))
-				break;
-		}
-		nr = i;
-	}
-
-	slot = path->slots[0];
-	BUG_ON(slot < 0);
-
-	if (slot != nritems) {
-		unsigned int old_data = btrfs_item_end_nr(leaf, slot);
-
-		item = btrfs_item_nr(leaf, slot);
-		btrfs_item_key_to_cpu(leaf, &found_key, slot);
-
-		/* figure out how many keys we can insert in here */
-		total_data = data_size[0];
-		for (i = 1; i < nr; i++) {
-			if (btrfs_comp_cpu_keys(&found_key, cpu_key + i) <= 0)
-				break;
-			total_data += data_size[i];
-		}
-		nr = i;
-
-		if (old_data < data_end) {
-			btrfs_print_leaf(root, leaf);
-			printk(KERN_CRIT "slot %d old_data %d data_end %d\n",
-			       slot, old_data, data_end);
-			BUG_ON(1);
-		}
-		/*
-		 * item0..itemN ... dataN.offset..dataN.size .. data0.size
-		 */
-		/* first correct the data pointers */
-		for (i = slot; i < nritems; i++) {
-			u32 ioff;
-
-			item = btrfs_item_nr(leaf, i);
-			ioff = btrfs_token_item_offset(leaf, item, &token);
-			btrfs_set_token_item_offset(leaf, item,
-						    ioff - total_data, &token);
-		}
-		/* shift the items */
-		memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
-			      btrfs_item_nr_offset(slot),
-			      (nritems - slot) * sizeof(struct btrfs_item));
-
-		/* shift the data */
-		memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
-			      data_end - total_data, btrfs_leaf_data(leaf) +
-			      data_end, old_data - data_end);
-		data_end = old_data;
-	} else {
-		/*
-		 * this sucks but it has to be done, if we are inserting at
-		 * the end of the leaf only insert 1 of the items, since we
-		 * have no way of knowing whats on the next leaf and we'd have
-		 * to drop our current locks to figure it out
-		 */
-		nr = 1;
-	}
-
-	/* setup the item for the new data */
-	for (i = 0; i < nr; i++) {
-		btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
-		btrfs_set_item_key(leaf, &disk_key, slot + i);
-		item = btrfs_item_nr(leaf, slot + i);
-		btrfs_set_token_item_offset(leaf, item,
-					    data_end - data_size[i], &token);
-		data_end -= data_size[i];
-		btrfs_set_token_item_size(leaf, item, data_size[i], &token);
-	}
-	btrfs_set_header_nritems(leaf, nritems + nr);
-	btrfs_mark_buffer_dirty(leaf);
-
-	ret = 0;
-	if (slot == 0) {
-		btrfs_cpu_key_to_disk(&disk_key, cpu_key);
-		fixup_low_keys(trans, root, path, &disk_key, 1);
-	}
-
-	if (btrfs_leaf_free_space(root, leaf) < 0) {
-		btrfs_print_leaf(root, leaf);
-		BUG();
-	}
-out:
-	if (!ret)
-		ret = nr;
-	return ret;
-}
-
 /*
  * this is a helper for btrfs_insert_empty_items, the main goal here is
  * to save stack depth by doing the bulk of the work in a function
@@ -5073,6 +4930,7 @@ static void tree_move_down(struct btrfs_root *root,
 			   struct btrfs_path *path,
 			   int *level, int root_level)
 {
+	BUG_ON(*level == 0);
 	path->nodes[*level - 1] = read_node_slot(root, path->nodes[*level],
 					path->slots[*level]);
 	path->slots[*level - 1] = 0;
@@ -5089,7 +4947,7 @@ static int tree_move_next_or_upnext(struct btrfs_root *root,
 
 	path->slots[*level]++;
 
-	while (path->slots[*level] == nritems) {
+	while (path->slots[*level] >= nritems) {
 		if (*level == root_level)
 			return -1;
 
@@ -5433,9 +5291,11 @@ int btrfs_compare_trees(struct btrfs_root *left_root,
 					goto out;
 				advance_right = ADVANCE;
 			} else {
+				WARN_ON(!extent_buffer_uptodate(left_path->nodes[0]));
 				ret = tree_compare_item(left_root, left_path,
 						right_path, tmp_buf);
 				if (ret) {
+					WARN_ON(!extent_buffer_uptodate(left_path->nodes[0]));
 					ret = changed_cb(left_root, right_root,
 						left_path, right_path,
 						&left_key,

fs/btrfs/ctree.h

@@ -154,6 +154,13 @@ struct btrfs_ordered_sum;
  */
 #define BTRFS_NAME_LEN 255
 
+/*
+ * Theoretical limit is larger, but we keep this down to a sane
+ * value. That should limit greatly the possibility of collisions on
+ * inode ref items.
+ */
+#define BTRFS_LINK_MAX 65535U
+
 /* 32 bytes in various csum fields */
 #define BTRFS_CSUM_SIZE 32
 
@@ -489,6 +496,8 @@ struct btrfs_super_block {
  */
 #define BTRFS_FEATURE_INCOMPAT_BIG_METADATA	(1ULL << 5)
 
+#define BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF	(1ULL << 6)
+
 #define BTRFS_FEATURE_COMPAT_SUPP		0ULL
 #define BTRFS_FEATURE_COMPAT_RO_SUPP		0ULL
 #define BTRFS_FEATURE_INCOMPAT_SUPP			\
@@ -496,7 +505,8 @@ struct btrfs_super_block {
 	 BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL |	\
 	 BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS |		\
 	 BTRFS_FEATURE_INCOMPAT_BIG_METADATA |		\
-	 BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO)
+	 BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO |		\
+	 BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF)
 
 /*
  * A leaf is full of items. offset and size tell us where to find
@@ -643,6 +653,14 @@ struct btrfs_inode_ref {
 	/* name goes here */
 } __attribute__ ((__packed__));
 
+struct btrfs_inode_extref {
+	__le64 parent_objectid;
+	__le64 index;
+	__le16 name_len;
+	__u8   name[0];
+	/* name goes here */
+} __attribute__ ((__packed__));
+
 struct btrfs_timespec {
 	__le64 sec;
 	__le32 nsec;
@@ -1028,12 +1046,22 @@ struct btrfs_space_info {
 	wait_queue_head_t wait;
 };
 
+#define	BTRFS_BLOCK_RSV_GLOBAL		1
+#define	BTRFS_BLOCK_RSV_DELALLOC	2
+#define	BTRFS_BLOCK_RSV_TRANS		3
+#define	BTRFS_BLOCK_RSV_CHUNK		4
+#define	BTRFS_BLOCK_RSV_DELOPS		5
+#define	BTRFS_BLOCK_RSV_EMPTY		6
+#define	BTRFS_BLOCK_RSV_TEMP		7
+
 struct btrfs_block_rsv {
 	u64 size;
 	u64 reserved;
 	struct btrfs_space_info *space_info;
 	spinlock_t lock;
-	unsigned int full;
+	unsigned short full;
+	unsigned short type;
+	unsigned short failfast;
 };
 
 /*
@@ -1127,6 +1155,9 @@ struct btrfs_block_group_cache {
 	 * Today it will only have one thing on it, but that may change
 	 */
 	struct list_head cluster_list;
+
+	/* For delayed block group creation */
+	struct list_head new_bg_list;
 };
 
 /* delayed seq elem */
@@ -1240,7 +1271,6 @@ struct btrfs_fs_info {
 	struct mutex reloc_mutex;
 
 	struct list_head trans_list;
-	struct list_head hashers;
 	struct list_head dead_roots;
 	struct list_head caching_block_groups;
 
@@ -1366,9 +1396,6 @@ struct btrfs_fs_info {
 	struct rb_root defrag_inodes;
 	atomic_t defrag_running;
 
-	spinlock_t ref_cache_lock;
-	u64 total_ref_cache_size;
-
 	/*
 	 * these three are in extended format (availability of single
 	 * chunks is denoted by BTRFS_AVAIL_ALLOC_BIT_SINGLE bit, other
@@ -1441,6 +1468,8 @@ struct btrfs_fs_info {
 
 	/* next backup root to be overwritten */
 	int backup_root_index;
+
+	int num_tolerated_disk_barrier_failures;
 };
 
 /*
@@ -1481,9 +1510,9 @@ struct btrfs_root {
 	wait_queue_head_t log_commit_wait[2];
 	atomic_t log_writers;
 	atomic_t log_commit[2];
+	atomic_t log_batch;
 	unsigned long log_transid;
 	unsigned long last_log_commit;
-	unsigned long log_batch;
 	pid_t log_start_pid;
 	bool log_multiple_pids;
 
@@ -1592,6 +1621,7 @@ struct btrfs_ioctl_defrag_range_args {
  */
 #define BTRFS_INODE_ITEM_KEY		1
 #define BTRFS_INODE_REF_KEY		12
+#define BTRFS_INODE_EXTREF_KEY		13
 #define BTRFS_XATTR_ITEM_KEY		24
 #define BTRFS_ORPHAN_ITEM_KEY		48
 /* reserve 2-15 close to the inode for later flexibility */
@@ -1978,6 +2008,13 @@ BTRFS_SETGET_STACK_FUNCS(block_group_flags,
 BTRFS_SETGET_FUNCS(inode_ref_name_len, struct btrfs_inode_ref, name_len, 16);
 BTRFS_SETGET_FUNCS(inode_ref_index, struct btrfs_inode_ref, index, 64);
 
+/* struct btrfs_inode_extref */
+BTRFS_SETGET_FUNCS(inode_extref_parent, struct btrfs_inode_extref,
+		   parent_objectid, 64);
+BTRFS_SETGET_FUNCS(inode_extref_name_len, struct btrfs_inode_extref,
+		   name_len, 16);
+BTRFS_SETGET_FUNCS(inode_extref_index, struct btrfs_inode_extref, index, 64);
+
 /* struct btrfs_inode_item */
 BTRFS_SETGET_FUNCS(inode_generation, struct btrfs_inode_item, generation, 64);
 BTRFS_SETGET_FUNCS(inode_sequence, struct btrfs_inode_item, sequence, 64);
@@ -2858,6 +2895,8 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 			   u64 size);
 int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root, u64 group_start);
+void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans,
+				       struct btrfs_root *root);
 u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags);
 u64 btrfs_get_alloc_profile(struct btrfs_root *root, int data);
 void btrfs_clear_space_info_full(struct btrfs_fs_info *info);
@@ -2874,8 +2913,9 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes);
 void btrfs_delalloc_release_metadata(struct inode *inode, u64 num_bytes);
 int btrfs_delalloc_reserve_space(struct inode *inode, u64 num_bytes);
 void btrfs_delalloc_release_space(struct inode *inode, u64 num_bytes);
-void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv);
-struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_root *root);
+void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv, unsigned short type);
+struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_root *root,
+					      unsigned short type);
 void btrfs_free_block_rsv(struct btrfs_root *root,
 			  struct btrfs_block_rsv *rsv);
 int btrfs_block_rsv_add(struct btrfs_root *root,
@@ -3172,12 +3212,12 @@ int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root,
 			   const char *name, int name_len,
 			   u64 inode_objectid, u64 ref_objectid, u64 *index);
-struct btrfs_inode_ref *
-btrfs_lookup_inode_ref(struct btrfs_trans_handle *trans,
-			struct btrfs_root *root,
-			struct btrfs_path *path,
-			const char *name, int name_len,
-			u64 inode_objectid, u64 ref_objectid, int mod);
+int btrfs_get_inode_ref_index(struct btrfs_trans_handle *trans,
+			      struct btrfs_root *root,
+			      struct btrfs_path *path,
+			      const char *name, int name_len,
+			      u64 inode_objectid, u64 ref_objectid, int mod,
+			      u64 *ret_index);
 int btrfs_insert_empty_inode(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root,
 			     struct btrfs_path *path, u64 objectid);
@@ -3185,6 +3225,19 @@ int btrfs_lookup_inode(struct btrfs_trans_handle *trans, struct btrfs_root
 		       *root, struct btrfs_path *path,
 		       struct btrfs_key *location, int mod);
 
+struct btrfs_inode_extref *
+btrfs_lookup_inode_extref(struct btrfs_trans_handle *trans,
+			  struct btrfs_root *root,
+			  struct btrfs_path *path,
+			  const char *name, int name_len,
+			  u64 inode_objectid, u64 ref_objectid, int ins_len,
+			  int cow);
+
+int btrfs_find_name_in_ext_backref(struct btrfs_path *path,
+				   u64 ref_objectid, const char *name,
+				   int name_len,
+				   struct btrfs_inode_extref **extref_ret);
+
 /* file-item.c */
 int btrfs_del_csums(struct btrfs_trans_handle *trans,
 		    struct btrfs_root *root, u64 bytenr, u64 len);
@@ -3249,6 +3302,8 @@ int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
 			struct btrfs_root *root,
 			struct inode *dir, u64 objectid,
 			const char *name, int name_len);
+int btrfs_truncate_page(struct inode *inode, loff_t from, loff_t len,
+			int front);
 int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root,
 			       struct inode *inode, u64 new_size,
@@ -3308,16 +3363,27 @@ void btrfs_inherit_iflags(struct inode *inode, struct inode *dir);
 int btrfs_defrag_file(struct inode *inode, struct file *file,
 		      struct btrfs_ioctl_defrag_range_args *range,
 		      u64 newer_than, unsigned long max_pages);
+void btrfs_get_block_group_info(struct list_head *groups_list,
+				struct btrfs_ioctl_space_info *space);
+
 /* file.c */
 int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans,
 			   struct inode *inode);
 int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info);
 int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
-int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
-			    int skip_pinned);
+void btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
+			     int skip_pinned);
+int btrfs_replace_extent_cache(struct inode *inode, struct extent_map *replace,
+			       u64 start, u64 end, int skip_pinned,
+			       int modified);
 extern const struct file_operations btrfs_file_operations;
-int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode,
-		       u64 start, u64 end, u64 *hint_byte, int drop_cache);
+int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
+			 struct btrfs_root *root, struct inode *inode,
+			 struct btrfs_path *path, u64 start, u64 end,
+			 u64 *drop_end, int drop_cache);
+int btrfs_drop_extents(struct btrfs_trans_handle *trans,
+		       struct btrfs_root *root, struct inode *inode, u64 start,
+		       u64 end, int drop_cache);
 int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
 			      struct inode *inode, u64 start, u64 end);
 int btrfs_release_file(struct inode *inode, struct file *file);
@@ -3378,6 +3444,11 @@ static inline void __btrfs_set_fs_incompat(struct btrfs_fs_info *fs_info,
 	}
 }
 
+/*
+ * Call btrfs_abort_transaction as early as possible when an error condition is
+ * detected, that way the exact line number is reported.
+ */
+
 #define btrfs_abort_transaction(trans, root, errno)		\
 do {								\
 	__btrfs_abort_transaction(trans, root, __func__,	\

fs/btrfs/delayed-inode.c

@@ -29,7 +29,7 @@ static struct kmem_cache *delayed_node_cache;
 
 int __init btrfs_delayed_inode_init(void)
 {
-	delayed_node_cache = kmem_cache_create("delayed_node",
+	delayed_node_cache = kmem_cache_create("btrfs_delayed_node",
 					sizeof(struct btrfs_delayed_node),
 					0,
 					SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
@@ -650,7 +650,7 @@ static int btrfs_delayed_inode_reserve_metadata(
 	 * we're accounted for.
 	 */
 	if (!src_rsv || (!trans->bytes_reserved &&
-	    src_rsv != &root->fs_info->delalloc_block_rsv)) {
+			 src_rsv->type != BTRFS_BLOCK_RSV_DELALLOC)) {
 		ret = btrfs_block_rsv_add_noflush(root, dst_rsv, num_bytes);
 		/*
 		 * Since we're under a transaction reserve_metadata_bytes could
@@ -668,7 +668,7 @@ static int btrfs_delayed_inode_reserve_metadata(
 						      num_bytes, 1);
 		}
 		return ret;
-	} else if (src_rsv == &root->fs_info->delalloc_block_rsv) {
+	} else if (src_rsv->type == BTRFS_BLOCK_RSV_DELALLOC) {
 		spin_lock(&BTRFS_I(inode)->lock);
 		if (test_and_clear_bit(BTRFS_INODE_DELALLOC_META_RESERVED,
 				       &BTRFS_I(inode)->runtime_flags)) {

fs/btrfs/disk-io.c

@@ -46,6 +46,10 @@
 #include "check-integrity.h"
 #include "rcu-string.h"
 
+#ifdef CONFIG_X86
+#include <asm/cpufeature.h>
+#endif
+
 static struct extent_io_ops btree_extent_io_ops;
 static void end_workqueue_fn(struct btrfs_work *work);
 static void free_fs_root(struct btrfs_root *root);
@@ -217,26 +221,16 @@ static struct extent_map *btree_get_extent(struct inode *inode,
 	write_lock(&em_tree->lock);
 	ret = add_extent_mapping(em_tree, em);
 	if (ret == -EEXIST) {
-		u64 failed_start = em->start;
-		u64 failed_len = em->len;
-
 		free_extent_map(em);
 		em = lookup_extent_mapping(em_tree, start, len);
-		if (em) {
-			ret = 0;
-		} else {
-			em = lookup_extent_mapping(em_tree, failed_start,
-						   failed_len);
-			ret = -EIO;
-		}
+		if (!em)
+			em = ERR_PTR(-EIO);
 	} else if (ret) {
 		free_extent_map(em);
-		em = NULL;
+		em = ERR_PTR(ret);
 	}
 	write_unlock(&em_tree->lock);
 
-	if (ret)
-		em = ERR_PTR(ret);
 out:
 	return em;
 }
@@ -439,10 +433,6 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page)
 		WARN_ON(1);
 		return 0;
 	}
-	if (eb->pages[0] != page) {
-		WARN_ON(1);
-		return 0;
-	}
 	if (!PageUptodate(page)) {
 		WARN_ON(1);
 		return 0;
@@ -869,10 +859,22 @@ static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
 	return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 1);
 }
 
+static int check_async_write(struct inode *inode, unsigned long bio_flags)
+{
+	if (bio_flags & EXTENT_BIO_TREE_LOG)
+		return 0;
+#ifdef CONFIG_X86
+	if (cpu_has_xmm4_2)
+		return 0;
+#endif
+	return 1;
+}
+
 static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
 				 int mirror_num, unsigned long bio_flags,
 				 u64 bio_offset)
 {
+	int async = check_async_write(inode, bio_flags);
 	int ret;
 
 	if (!(rw & REQ_WRITE)) {
@@ -887,6 +889,12 @@ static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
 			return ret;
 		return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
 				     mirror_num, 0);
+	} else if (!async) {
+		ret = btree_csum_one_bio(bio);
+		if (ret)
+			return ret;
+		return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
+				     mirror_num, 0);
 	}
 
 	/*
@@ -1168,8 +1176,8 @@ static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
 	atomic_set(&root->log_commit[0], 0);
 	atomic_set(&root->log_commit[1], 0);
 	atomic_set(&root->log_writers, 0);
+	atomic_set(&root->log_batch, 0);
 	atomic_set(&root->orphan_inodes, 0);
-	root->log_batch = 0;
 	root->log_transid = 0;
 	root->last_log_commit = 0;
 	extent_io_tree_init(&root->dirty_log_pages,
@@ -1667,9 +1675,10 @@ static int transaction_kthread(void *arg)
 		spin_unlock(&root->fs_info->trans_lock);
 
 		/* If the file system is aborted, this will always fail. */
-		trans = btrfs_join_transaction(root);
+		trans = btrfs_attach_transaction(root);
 		if (IS_ERR(trans)) {
-			cannot_commit = true;
+			if (PTR_ERR(trans) != -ENOENT)
+				cannot_commit = true;
 			goto sleep;
 		}
 		if (transid == trans->transid) {
@@ -1994,13 +2003,11 @@ int open_ctree(struct super_block *sb,
 	INIT_LIST_HEAD(&fs_info->trans_list);
 	INIT_LIST_HEAD(&fs_info->dead_roots);
 	INIT_LIST_HEAD(&fs_info->delayed_iputs);
-	INIT_LIST_HEAD(&fs_info->hashers);
 	INIT_LIST_HEAD(&fs_info->delalloc_inodes);
 	INIT_LIST_HEAD(&fs_info->ordered_operations);
 	INIT_LIST_HEAD(&fs_info->caching_block_groups);
 	spin_lock_init(&fs_info->delalloc_lock);
 	spin_lock_init(&fs_info->trans_lock);
-	spin_lock_init(&fs_info->ref_cache_lock);
 	spin_lock_init(&fs_info->fs_roots_radix_lock);
 	spin_lock_init(&fs_info->delayed_iput_lock);
 	spin_lock_init(&fs_info->defrag_inodes_lock);
@@ -2014,12 +2021,15 @@ int open_ctree(struct super_block *sb,
 	INIT_LIST_HEAD(&fs_info->space_info);
 	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
 	btrfs_mapping_init(&fs_info->mapping_tree);
-	btrfs_init_block_rsv(&fs_info->global_block_rsv);
-	btrfs_init_block_rsv(&fs_info->delalloc_block_rsv);
-	btrfs_init_block_rsv(&fs_info->trans_block_rsv);
-	btrfs_init_block_rsv(&fs_info->chunk_block_rsv);
-	btrfs_init_block_rsv(&fs_info->empty_block_rsv);
-	btrfs_init_block_rsv(&fs_info->delayed_block_rsv);
+	btrfs_init_block_rsv(&fs_info->global_block_rsv,
+			     BTRFS_BLOCK_RSV_GLOBAL);
+	btrfs_init_block_rsv(&fs_info->delalloc_block_rsv,
+			     BTRFS_BLOCK_RSV_DELALLOC);
+	btrfs_init_block_rsv(&fs_info->trans_block_rsv, BTRFS_BLOCK_RSV_TRANS);
+	btrfs_init_block_rsv(&fs_info->chunk_block_rsv, BTRFS_BLOCK_RSV_CHUNK);
+	btrfs_init_block_rsv(&fs_info->empty_block_rsv, BTRFS_BLOCK_RSV_EMPTY);
+	btrfs_init_block_rsv(&fs_info->delayed_block_rsv,
+			     BTRFS_BLOCK_RSV_DELOPS);
 	atomic_set(&fs_info->nr_async_submits, 0);
 	atomic_set(&fs_info->async_delalloc_pages, 0);
 	atomic_set(&fs_info->async_submit_draining, 0);
@@ -2491,6 +2501,8 @@ retry_root_backup:
 		printk(KERN_ERR "Failed to read block groups: %d\n", ret);
 		goto fail_block_groups;
 	}
+	fs_info->num_tolerated_disk_barrier_failures =
+		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
 
 	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
 					       "btrfs-cleaner");
@@ -2874,12 +2886,10 @@ static int write_dev_flush(struct btrfs_device *device, int wait)
 			printk_in_rcu("btrfs: disabling barriers on dev %s\n",
 				      rcu_str_deref(device->name));
 			device->nobarriers = 1;
-		}
-		if (!bio_flagged(bio, BIO_UPTODATE)) {
+		} else if (!bio_flagged(bio, BIO_UPTODATE)) {
 			ret = -EIO;
-			if (!bio_flagged(bio, BIO_EOPNOTSUPP))
-				btrfs_dev_stat_inc_and_print(device,
-					BTRFS_DEV_STAT_FLUSH_ERRS);
+			btrfs_dev_stat_inc_and_print(device,
+				BTRFS_DEV_STAT_FLUSH_ERRS);
 		}
 
 		/* drop the reference from the wait == 0 run */
@@ -2918,14 +2928,15 @@ static int barrier_all_devices(struct btrfs_fs_info *info)
 {
 	struct list_head *head;
 	struct btrfs_device *dev;
-	int errors = 0;
+	int errors_send = 0;
+	int errors_wait = 0;
 	int ret;
 
 	/* send down all the barriers */
 	head = &info->fs_devices->devices;
 	list_for_each_entry_rcu(dev, head, dev_list) {
 		if (!dev->bdev) {
-			errors++;
+			errors_send++;
 			continue;
 		}
 		if (!dev->in_fs_metadata || !dev->writeable)
@@ -2933,13 +2944,13 @@ static int barrier_all_devices(struct btrfs_fs_info *info)
 
 		ret = write_dev_flush(dev, 0);
 		if (ret)
-			errors++;
+			errors_send++;
 	}
 
 	/* wait for all the barriers */
 	list_for_each_entry_rcu(dev, head, dev_list) {
 		if (!dev->bdev) {
-			errors++;
+			errors_wait++;
 			continue;
 		}
 		if (!dev->in_fs_metadata || !dev->writeable)
@@ -2947,13 +2958,87 @@ static int barrier_all_devices(struct btrfs_fs_info *info)
 
 		ret = write_dev_flush(dev, 1);
 		if (ret)
-			errors++;
+			errors_wait++;
 	}
-	if (errors)
+	if (errors_send > info->num_tolerated_disk_barrier_failures ||
+	    errors_wait > info->num_tolerated_disk_barrier_failures)
 		return -EIO;
 	return 0;
 }
 
+int btrfs_calc_num_tolerated_disk_barrier_failures(
+	struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_ioctl_space_info space;
+	struct btrfs_space_info *sinfo;
+	u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
+		       BTRFS_BLOCK_GROUP_SYSTEM,
+		       BTRFS_BLOCK_GROUP_METADATA,
+		       BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
+	int num_types = 4;
+	int i;
+	int c;
+	int num_tolerated_disk_barrier_failures =
+		(int)fs_info->fs_devices->num_devices;
+
+	for (i = 0; i < num_types; i++) {
+		struct btrfs_space_info *tmp;
+
+		sinfo = NULL;
+		rcu_read_lock();
+		list_for_each_entry_rcu(tmp, &fs_info->space_info, list) {
+			if (tmp->flags == types[i]) {
+				sinfo = tmp;
+				break;
+			}
+		}
+		rcu_read_unlock();
+
+		if (!sinfo)
+			continue;
+
+		down_read(&sinfo->groups_sem);
+		for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
+			if (!list_empty(&sinfo->block_groups[c])) {
+				u64 flags;
+
+				btrfs_get_block_group_info(
+					&sinfo->block_groups[c], &space);
+				if (space.total_bytes == 0 ||
+				    space.used_bytes == 0)
+					continue;
+				flags = space.flags;
+				/*
+				 * return
+				 * 0: if dup, single or RAID0 is configured for
+				 *    any of metadata, system or data, else
+				 * 1: if RAID5 is configured, or if RAID1 or
+				 *    RAID10 is configured and only two mirrors
+				 *    are used, else
+				 * 2: if RAID6 is configured, else
+				 * num_mirrors - 1: if RAID1 or RAID10 is
+				 *                  configured and more than
+				 *                  2 mirrors are used.
+				 */
+				if (num_tolerated_disk_barrier_failures > 0 &&
+				    ((flags & (BTRFS_BLOCK_GROUP_DUP |
+					       BTRFS_BLOCK_GROUP_RAID0)) ||
+				     ((flags & BTRFS_BLOCK_GROUP_PROFILE_MASK)
+				      == 0)))
+					num_tolerated_disk_barrier_failures = 0;
+				else if (num_tolerated_disk_barrier_failures > 1
+					 &&
+					 (flags & (BTRFS_BLOCK_GROUP_RAID1 |
+						   BTRFS_BLOCK_GROUP_RAID10)))
+					num_tolerated_disk_barrier_failures = 1;
+			}
+		}
+		up_read(&sinfo->groups_sem);
+	}
+
+	return num_tolerated_disk_barrier_failures;
+}
+
 int write_all_supers(struct btrfs_root *root, int max_mirrors)
 {
 	struct list_head *head;
@@ -2976,8 +3061,16 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors)
 	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
 	head = &root->fs_info->fs_devices->devices;
 
-	if (do_barriers)
-		barrier_all_devices(root->fs_info);
+	if (do_barriers) {
+		ret = barrier_all_devices(root->fs_info);
+		if (ret) {
+			mutex_unlock(
+				&root->fs_info->fs_devices->device_list_mutex);
+			btrfs_error(root->fs_info, ret,
+				    "errors while submitting device barriers.");
+			return ret;
+		}
+	}
 
 	list_for_each_entry_rcu(dev, head, dev_list) {
 		if (!dev->bdev) {
@@ -3211,10 +3304,6 @@ int close_ctree(struct btrfs_root *root)
 		printk(KERN_INFO "btrfs: at unmount delalloc count %llu\n",
 		       (unsigned long long)fs_info->delalloc_bytes);
 	}
-	if (fs_info->total_ref_cache_size) {
-		printk(KERN_INFO "btrfs: at umount reference cache size %llu\n",
-		       (unsigned long long)fs_info->total_ref_cache_size);
-	}
 
 	free_extent_buffer(fs_info->extent_root->node);
 	free_extent_buffer(fs_info->extent_root->commit_root);
@@ -3360,52 +3449,6 @@ int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
 	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
 }
 
-int btree_lock_page_hook(struct page *page, void *data,
-				void (*flush_fn)(void *))
-{
-	struct inode *inode = page->mapping->host;
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct extent_buffer *eb;
-
-	/*
-	 * We culled this eb but the page is still hanging out on the mapping,
-	 * carry on.
-	 */
-	if (!PagePrivate(page))
-		goto out;
-
-	eb = (struct extent_buffer *)page->private;
-	if (!eb) {
-		WARN_ON(1);
-		goto out;
-	}
-	if (page != eb->pages[0])
-		goto out;
-
-	if (!btrfs_try_tree_write_lock(eb)) {
-		flush_fn(data);
-		btrfs_tree_lock(eb);
-	}
-	btrfs_set_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN);
-
-	if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
-		spin_lock(&root->fs_info->delalloc_lock);
-		if (root->fs_info->dirty_metadata_bytes >= eb->len)
-			root->fs_info->dirty_metadata_bytes -= eb->len;
-		else
-			WARN_ON(1);
-		spin_unlock(&root->fs_info->delalloc_lock);
-	}
-
-	btrfs_tree_unlock(eb);
-out:
-	if (!trylock_page(page)) {
-		flush_fn(data);
-		lock_page(page);
-	}
-	return 0;
-}
-
 static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
 			      int read_only)
 {
@@ -3608,7 +3651,7 @@ static int btrfs_destroy_marked_extents(struct btrfs_root *root,
 
 	while (1) {
 		ret = find_first_extent_bit(dirty_pages, start, &start, &end,
-					    mark);
+					    mark, NULL);
 		if (ret)
 			break;
 
@@ -3663,7 +3706,7 @@ static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
 again:
 	while (1) {
 		ret = find_first_extent_bit(unpin, 0, &start, &end,
-					    EXTENT_DIRTY);
+					    EXTENT_DIRTY, NULL);
 		if (ret)
 			break;
 
@@ -3800,7 +3843,6 @@ int btrfs_cleanup_transaction(struct btrfs_root *root)
 }
 
 static struct extent_io_ops btree_extent_io_ops = {
-	.write_cache_pages_lock_hook = btree_lock_page_hook,
 	.readpage_end_io_hook = btree_readpage_end_io_hook,
 	.readpage_io_failed_hook = btree_io_failed_hook,
 	.submit_bio_hook = btree_submit_bio_hook,

fs/btrfs/disk-io.h

@@ -95,6 +95,8 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
 				     u64 objectid);
 int btree_lock_page_hook(struct page *page, void *data,
 				void (*flush_fn)(void *));
+int btrfs_calc_num_tolerated_disk_barrier_failures(
+	struct btrfs_fs_info *fs_info);
 
 #ifdef CONFIG_DEBUG_LOCK_ALLOC
 void btrfs_init_lockdep(void);

fs/btrfs/extent-tree.c

@@ -94,8 +94,8 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 				     u64 flags, struct btrfs_disk_key *key,
 				     int level, struct btrfs_key *ins);
 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
-			  struct btrfs_root *extent_root, u64 alloc_bytes,
-			  u64 flags, int force);
+			  struct btrfs_root *extent_root, u64 flags,
+			  int force);
 static int find_next_key(struct btrfs_path *path, int level,
 			 struct btrfs_key *key);
 static void dump_space_info(struct btrfs_space_info *info, u64 bytes,
@@ -312,7 +312,8 @@ static u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
 	while (start < end) {
 		ret = find_first_extent_bit(info->pinned_extents, start,
 					    &extent_start, &extent_end,
-					    EXTENT_DIRTY | EXTENT_UPTODATE);
+					    EXTENT_DIRTY | EXTENT_UPTODATE,
+					    NULL);
 		if (ret)
 			break;
 
@@ -2361,10 +2362,6 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans,
 		}
 
 next:
-		do_chunk_alloc(trans, fs_info->extent_root,
-			       2 * 1024 * 1024,
-			       btrfs_get_alloc_profile(root, 0),
-			       CHUNK_ALLOC_NO_FORCE);
 		cond_resched();
 		spin_lock(&delayed_refs->lock);
 	}
@@ -2478,10 +2475,6 @@ int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 	if (root == root->fs_info->extent_root)
 		root = root->fs_info->tree_root;
 
-	do_chunk_alloc(trans, root->fs_info->extent_root,
-		       2 * 1024 * 1024, btrfs_get_alloc_profile(root, 0),
-		       CHUNK_ALLOC_NO_FORCE);
-
 	btrfs_delayed_refs_qgroup_accounting(trans, root->fs_info);
 
 	delayed_refs = &trans->transaction->delayed_refs;
@@ -2551,6 +2544,12 @@ again:
 	}
 
 	if (run_all) {
+		if (!list_empty(&trans->new_bgs)) {
+			spin_unlock(&delayed_refs->lock);
+			btrfs_create_pending_block_groups(trans, root);
+			spin_lock(&delayed_refs->lock);
+		}
+
 		node = rb_first(&delayed_refs->root);
 		if (!node)
 			goto out;
@@ -3406,7 +3405,6 @@ alloc:
 				return PTR_ERR(trans);
 
 			ret = do_chunk_alloc(trans, root->fs_info->extent_root,
-					     bytes + 2 * 1024 * 1024,
 					     alloc_target,
 					     CHUNK_ALLOC_NO_FORCE);
 			btrfs_end_transaction(trans, root);
@@ -3488,8 +3486,7 @@ static void force_metadata_allocation(struct btrfs_fs_info *info)
 }
 
 static int should_alloc_chunk(struct btrfs_root *root,
-			      struct btrfs_space_info *sinfo, u64 alloc_bytes,
-			      int force)
+			      struct btrfs_space_info *sinfo, int force)
 {
 	struct btrfs_block_rsv *global_rsv = &root->fs_info->global_block_rsv;
 	u64 num_bytes = sinfo->total_bytes - sinfo->bytes_readonly;
@@ -3504,7 +3501,8 @@ static int should_alloc_chunk(struct btrfs_root *root,
 	 * and purposes it's used space.  Don't worry about locking the
 	 * global_rsv, it doesn't change except when the transaction commits.
 	 */
-	num_allocated += global_rsv->size;
+	if (sinfo->flags & BTRFS_BLOCK_GROUP_METADATA)
+		num_allocated += global_rsv->size;
 
 	/*
 	 * in limited mode, we want to have some free space up to
@@ -3518,15 +3516,8 @@ static int should_alloc_chunk(struct btrfs_root *root,
 		if (num_bytes - num_allocated < thresh)
 			return 1;
 	}
-	thresh = btrfs_super_total_bytes(root->fs_info->super_copy);
 
-	/* 256MB or 2% of the FS */
-	thresh = max_t(u64, 256 * 1024 * 1024, div_factor_fine(thresh, 2));
-	/* system chunks need a much small threshold */
-	if (sinfo->flags & BTRFS_BLOCK_GROUP_SYSTEM)
-		thresh = 32 * 1024 * 1024;
-
-	if (num_bytes > thresh && sinfo->bytes_used < div_factor(num_bytes, 8))
+	if (num_allocated + 2 * 1024 * 1024 < div_factor(num_bytes, 8))
 		return 0;
 	return 1;
 }
@@ -3576,8 +3567,7 @@ static void check_system_chunk(struct btrfs_trans_handle *trans,
 }
 
 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
-			  struct btrfs_root *extent_root, u64 alloc_bytes,
-			  u64 flags, int force)
+			  struct btrfs_root *extent_root, u64 flags, int force)
 {
 	struct btrfs_space_info *space_info;
 	struct btrfs_fs_info *fs_info = extent_root->fs_info;
@@ -3601,7 +3591,7 @@ again:
 		return 0;
 	}
 
-	if (!should_alloc_chunk(extent_root, space_info, alloc_bytes, force)) {
+	if (!should_alloc_chunk(extent_root, space_info, force)) {
 		spin_unlock(&space_info->lock);
 		return 0;
 	} else if (space_info->chunk_alloc) {
@@ -3669,6 +3659,46 @@ out:
 	return ret;
 }
 
+static int can_overcommit(struct btrfs_root *root,
+			  struct btrfs_space_info *space_info, u64 bytes,
+			  int flush)
+{
+	u64 profile = btrfs_get_alloc_profile(root, 0);
+	u64 avail;
+	u64 used;
+
+	used = space_info->bytes_used + space_info->bytes_reserved +
+		space_info->bytes_pinned + space_info->bytes_readonly +
+		space_info->bytes_may_use;
+
+	spin_lock(&root->fs_info->free_chunk_lock);
+	avail = root->fs_info->free_chunk_space;
+	spin_unlock(&root->fs_info->free_chunk_lock);
+
+	/*
+	 * If we have dup, raid1 or raid10 then only half of the free
+	 * space is actually useable.
+	 */
+	if (profile & (BTRFS_BLOCK_GROUP_DUP |
+		       BTRFS_BLOCK_GROUP_RAID1 |
+		       BTRFS_BLOCK_GROUP_RAID10))
+		avail >>= 1;
+
+	/*
+	 * If we aren't flushing don't let us overcommit too much, say
+	 * 1/8th of the space.  If we can flush, let it overcommit up to
+	 * 1/2 of the space.
+	 */
+	if (flush)
+		avail >>= 3;
+	else
+		avail >>= 1;
+
+	if (used + bytes < space_info->total_bytes + avail)
+		return 1;
+	return 0;
+}
+
 /*
  * shrink metadata reservation for delalloc
  */
@@ -3693,7 +3723,7 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig,
 	if (delalloc_bytes == 0) {
 		if (trans)
 			return;
-		btrfs_wait_ordered_extents(root, 0, 0);
+		btrfs_wait_ordered_extents(root, 0);
 		return;
 	}
 
@@ -3703,11 +3733,15 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig,
 		writeback_inodes_sb_nr_if_idle(root->fs_info->sb, nr_pages,
 					       WB_REASON_FS_FREE_SPACE);
 
+		/*
+		 * We need to wait for the async pages to actually start before
+		 * we do anything.
+		 */
+		wait_event(root->fs_info->async_submit_wait,
+			   !atomic_read(&root->fs_info->async_delalloc_pages));
+
 		spin_lock(&space_info->lock);
-		if (space_info->bytes_used + space_info->bytes_reserved +
-		    space_info->bytes_pinned + space_info->bytes_readonly +
-		    space_info->bytes_may_use + orig <=
-		    space_info->total_bytes) {
+		if (can_overcommit(root, space_info, orig, !trans)) {
 			spin_unlock(&space_info->lock);
 			break;
 		}
@@ -3715,7 +3749,7 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig,
 
 		loops++;
 		if (wait_ordered && !trans) {
-			btrfs_wait_ordered_extents(root, 0, 0);
+			btrfs_wait_ordered_extents(root, 0);
 		} else {
 			time_left = schedule_timeout_killable(1);
 			if (time_left)
@@ -3784,11 +3818,12 @@ commit:
 }
 
 enum flush_state {
-	FLUSH_DELALLOC		=	1,
-	FLUSH_DELALLOC_WAIT	=	2,
-	FLUSH_DELAYED_ITEMS_NR	=	3,
-	FLUSH_DELAYED_ITEMS	=	4,
-	COMMIT_TRANS		=	5,
+	FLUSH_DELAYED_ITEMS_NR	=	1,
+	FLUSH_DELAYED_ITEMS	=	2,
+	FLUSH_DELALLOC		=	3,
+	FLUSH_DELALLOC_WAIT	=	4,
+	ALLOC_CHUNK		=	5,
+	COMMIT_TRANS		=	6,
 };
 
 static int flush_space(struct btrfs_root *root,
@@ -3800,11 +3835,6 @@ static int flush_space(struct btrfs_root *root,
 	int ret = 0;
 
 	switch (state) {
-	case FLUSH_DELALLOC:
-	case FLUSH_DELALLOC_WAIT:
-		shrink_delalloc(root, num_bytes, orig_bytes,
-				state == FLUSH_DELALLOC_WAIT);
-		break;
 	case FLUSH_DELAYED_ITEMS_NR:
 	case FLUSH_DELAYED_ITEMS:
 		if (state == FLUSH_DELAYED_ITEMS_NR) {
@@ -3825,6 +3855,24 @@ static int flush_space(struct btrfs_root *root,
 		ret = btrfs_run_delayed_items_nr(trans, root, nr);
 		btrfs_end_transaction(trans, root);
 		break;
+	case FLUSH_DELALLOC:
+	case FLUSH_DELALLOC_WAIT:
+		shrink_delalloc(root, num_bytes, orig_bytes,
+				state == FLUSH_DELALLOC_WAIT);
+		break;
+	case ALLOC_CHUNK:
+		trans = btrfs_join_transaction(root);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			break;
+		}
+		ret = do_chunk_alloc(trans, root->fs_info->extent_root,
+				     btrfs_get_alloc_profile(root, 0),
+				     CHUNK_ALLOC_NO_FORCE);
+		btrfs_end_transaction(trans, root);
+		if (ret == -ENOSPC)
+			ret = 0;
+		break;
 	case COMMIT_TRANS:
 		ret = may_commit_transaction(root, space_info, orig_bytes, 0);
 		break;
@@ -3856,10 +3904,9 @@ static int reserve_metadata_bytes(struct btrfs_root *root,
 	struct btrfs_space_info *space_info = block_rsv->space_info;
 	u64 used;
 	u64 num_bytes = orig_bytes;
-	int flush_state = FLUSH_DELALLOC;
+	int flush_state = FLUSH_DELAYED_ITEMS_NR;
 	int ret = 0;
 	bool flushing = false;
-	bool committed = false;
 
 again:
 	ret = 0;
@@ -3922,57 +3969,12 @@ again:
 			(orig_bytes * 2);
 	}
 
-	if (ret) {
-		u64 profile = btrfs_get_alloc_profile(root, 0);
-		u64 avail;
-
-		/*
-		 * If we have a lot of space that's pinned, don't bother doing
-		 * the overcommit dance yet and just commit the transaction.
-		 */
-		avail = (space_info->total_bytes - space_info->bytes_used) * 8;
-		do_div(avail, 10);
-		if (space_info->bytes_pinned >= avail && flush && !committed) {
-			space_info->flush = 1;
-			flushing = true;
-			spin_unlock(&space_info->lock);
-			ret = may_commit_transaction(root, space_info,
-						     orig_bytes, 1);
-			if (ret)
-				goto out;
-			committed = true;
-			goto again;
-		}
-
-		spin_lock(&root->fs_info->free_chunk_lock);
-		avail = root->fs_info->free_chunk_space;
-
-		/*
-		 * If we have dup, raid1 or raid10 then only half of the free
-		 * space is actually useable.
-		 */
-		if (profile & (BTRFS_BLOCK_GROUP_DUP |
-			       BTRFS_BLOCK_GROUP_RAID1 |
-			       BTRFS_BLOCK_GROUP_RAID10))
-			avail >>= 1;
-
-		/*
-		 * If we aren't flushing don't let us overcommit too much, say
-		 * 1/8th of the space.  If we can flush, let it overcommit up to
-		 * 1/2 of the space.
-		 */
-		if (flush)
-			avail >>= 3;
-		else
-			avail >>= 1;
-		 spin_unlock(&root->fs_info->free_chunk_lock);
-
-		if (used + num_bytes < space_info->total_bytes + avail) {
-			space_info->bytes_may_use += orig_bytes;
-			trace_btrfs_space_reservation(root->fs_info,
-				"space_info", space_info->flags, orig_bytes, 1);
-			ret = 0;
-		}
+	if (ret && can_overcommit(root, space_info, orig_bytes, flush)) {
+		space_info->bytes_may_use += orig_bytes;
+		trace_btrfs_space_reservation(root->fs_info, "space_info",
+					      space_info->flags, orig_bytes,
+					      1);
+		ret = 0;
 	}
 
 	/*
@@ -4114,13 +4116,15 @@ static int block_rsv_migrate_bytes(struct btrfs_block_rsv *src,
 	return 0;
 }
 
-void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv)
+void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv, unsigned short type)
 {
 	memset(rsv, 0, sizeof(*rsv));
 	spin_lock_init(&rsv->lock);
+	rsv->type = type;
 }
 
-struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_root *root)
+struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_root *root,
+					      unsigned short type)
 {
 	struct btrfs_block_rsv *block_rsv;
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -4129,7 +4133,7 @@ struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_root *root)
 	if (!block_rsv)
 		return NULL;
 
-	btrfs_init_block_rsv(block_rsv);
+	btrfs_init_block_rsv(block_rsv, type);
 	block_rsv->space_info = __find_space_info(fs_info,
 						  BTRFS_BLOCK_GROUP_METADATA);
 	return block_rsv;
@@ -4138,6 +4142,8 @@ struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_root *root)
 void btrfs_free_block_rsv(struct btrfs_root *root,
 			  struct btrfs_block_rsv *rsv)
 {
+	if (!rsv)
+		return;
 	btrfs_block_rsv_release(root, rsv, (u64)-1);
 	kfree(rsv);
 }
@@ -4416,10 +4422,10 @@ int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans,
 	struct btrfs_block_rsv *src_rsv = get_block_rsv(trans, root);
 	struct btrfs_block_rsv *dst_rsv = &pending->block_rsv;
 	/*
-	 * two for root back/forward refs, two for directory entries
-	 * and one for root of the snapshot.
+	 * two for root back/forward refs, two for directory entries,
+	 * one for root of the snapshot and one for parent inode.
 	 */
-	u64 num_bytes = btrfs_calc_trans_metadata_size(root, 5);
+	u64 num_bytes = btrfs_calc_trans_metadata_size(root, 6);
 	dst_rsv->space_info = src_rsv->space_info;
 	return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes);
 }
@@ -5018,7 +5024,7 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 
 	while (1) {
 		ret = find_first_extent_bit(unpin, 0, &start, &end,
-					    EXTENT_DIRTY);
+					    EXTENT_DIRTY, NULL);
 		if (ret)
 			break;
 
@@ -5096,8 +5102,10 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 			ret = remove_extent_backref(trans, extent_root, path,
 						    NULL, refs_to_drop,
 						    is_data);
-			if (ret)
-				goto abort;
+			if (ret) {
+				btrfs_abort_transaction(trans, extent_root, ret);
+				goto out;
+			}
 			btrfs_release_path(path);
 			path->leave_spinning = 1;
 
@@ -5115,8 +5123,10 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 					btrfs_print_leaf(extent_root,
 							 path->nodes[0]);
 			}
-			if (ret < 0)
-				goto abort;
+			if (ret < 0) {
+				btrfs_abort_transaction(trans, extent_root, ret);
+				goto out;
+			}
 			extent_slot = path->slots[0];
 		}
 	} else if (ret == -ENOENT) {
@@ -5130,7 +5140,8 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 		       (unsigned long long)owner_objectid,
 		       (unsigned long long)owner_offset);
 	} else {
-		goto abort;
+		btrfs_abort_transaction(trans, extent_root, ret);
+		goto out;
 	}
 
 	leaf = path->nodes[0];
@@ -5140,8 +5151,10 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 		BUG_ON(found_extent || extent_slot != path->slots[0]);
 		ret = convert_extent_item_v0(trans, extent_root, path,
 					     owner_objectid, 0);
-		if (ret < 0)
-			goto abort;
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, extent_root, ret);
+			goto out;
+		}
 
 		btrfs_release_path(path);
 		path->leave_spinning = 1;
@@ -5158,8 +5171,11 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 			       (unsigned long long)bytenr);
 			btrfs_print_leaf(extent_root, path->nodes[0]);
 		}
-		if (ret < 0)
-			goto abort;
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, extent_root, ret);
+			goto out;
+		}
+
 		extent_slot = path->slots[0];
 		leaf = path->nodes[0];
 		item_size = btrfs_item_size_nr(leaf, extent_slot);
@@ -5196,8 +5212,10 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 			ret = remove_extent_backref(trans, extent_root, path,
 						    iref, refs_to_drop,
 						    is_data);
-			if (ret)
-				goto abort;
+			if (ret) {
+				btrfs_abort_transaction(trans, extent_root, ret);
+				goto out;
+			}
 		}
 	} else {
 		if (found_extent) {
@@ -5214,27 +5232,29 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 
 		ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
 				      num_to_del);
-		if (ret)
-			goto abort;
+		if (ret) {
+			btrfs_abort_transaction(trans, extent_root, ret);
+			goto out;
+		}
 		btrfs_release_path(path);
 
 		if (is_data) {
 			ret = btrfs_del_csums(trans, root, bytenr, num_bytes);
-			if (ret)
-				goto abort;
+			if (ret) {
+				btrfs_abort_transaction(trans, extent_root, ret);
+				goto out;
+			}
 		}
 
 		ret = update_block_group(trans, root, bytenr, num_bytes, 0);
-		if (ret)
-			goto abort;
+		if (ret) {
+			btrfs_abort_transaction(trans, extent_root, ret);
+			goto out;
+		}
 	}
 out:
 	btrfs_free_path(path);
 	return ret;
-
-abort:
-	btrfs_abort_transaction(trans, extent_root, ret);
-	goto out;
 }
 
 /*
@@ -5497,8 +5517,6 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 	struct btrfs_block_group_cache *used_block_group;
 	u64 search_start = 0;
 	int empty_cluster = 2 * 1024 * 1024;
-	int allowed_chunk_alloc = 0;
-	int done_chunk_alloc = 0;
 	struct btrfs_space_info *space_info;
 	int loop = 0;
 	int index = 0;
@@ -5530,9 +5548,6 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 	if (btrfs_mixed_space_info(space_info))
 		use_cluster = false;
 
-	if (orig_root->ref_cows || empty_size)
-		allowed_chunk_alloc = 1;
-
 	if (data & BTRFS_BLOCK_GROUP_METADATA && use_cluster) {
 		last_ptr = &root->fs_info->meta_alloc_cluster;
 		if (!btrfs_test_opt(root, SSD))
@@ -5806,10 +5821,6 @@ checks:
 
 		trace_btrfs_reserve_extent(orig_root, block_group,
 					   search_start, num_bytes);
-		if (offset < search_start)
-			btrfs_add_free_space(used_block_group, offset,
-					     search_start - offset);
-		BUG_ON(offset > search_start);
 		if (used_block_group != block_group)
 			btrfs_put_block_group(used_block_group);
 		btrfs_put_block_group(block_group);
@@ -5842,34 +5853,17 @@ loop:
 		index = 0;
 		loop++;
 		if (loop == LOOP_ALLOC_CHUNK) {
-		       if (allowed_chunk_alloc) {
-				ret = do_chunk_alloc(trans, root, num_bytes +
-						     2 * 1024 * 1024, data,
-						     CHUNK_ALLOC_LIMITED);
-				/*
-				 * Do not bail out on ENOSPC since we
-				 * can do more things.
-				 */
-				if (ret < 0 && ret != -ENOSPC) {
-					btrfs_abort_transaction(trans,
-								root, ret);
-					goto out;
-				}
-				allowed_chunk_alloc = 0;
-				if (ret == 1)
-					done_chunk_alloc = 1;
-			} else if (!done_chunk_alloc &&
-				   space_info->force_alloc ==
-				   CHUNK_ALLOC_NO_FORCE) {
-				space_info->force_alloc = CHUNK_ALLOC_LIMITED;
+			ret = do_chunk_alloc(trans, root, data,
+					     CHUNK_ALLOC_FORCE);
+			/*
+			 * Do not bail out on ENOSPC since we
+			 * can do more things.
+			 */
+			if (ret < 0 && ret != -ENOSPC) {
+				btrfs_abort_transaction(trans,
+							root, ret);
+				goto out;
 			}
-
-		       /*
-			* We didn't allocate a chunk, go ahead and drop the
-			* empty size and loop again.
-			*/
-		       if (!done_chunk_alloc)
-			       loop = LOOP_NO_EMPTY_SIZE;
 		}
 
 		if (loop == LOOP_NO_EMPTY_SIZE) {
@@ -5944,20 +5938,6 @@ int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
 
 	data = btrfs_get_alloc_profile(root, data);
 again:
-	/*
-	 * the only place that sets empty_size is btrfs_realloc_node, which
-	 * is not called recursively on allocations
-	 */
-	if (empty_size || root->ref_cows) {
-		ret = do_chunk_alloc(trans, root->fs_info->extent_root,
-				     num_bytes + 2 * 1024 * 1024, data,
-				     CHUNK_ALLOC_NO_FORCE);
-		if (ret < 0 && ret != -ENOSPC) {
-			btrfs_abort_transaction(trans, root, ret);
-			return ret;
-		}
-	}
-
 	WARN_ON(num_bytes < root->sectorsize);
 	ret = find_free_extent(trans, root, num_bytes, empty_size,
 			       hint_byte, ins, data);
@@ -5967,12 +5947,6 @@ again:
 			num_bytes = num_bytes >> 1;
 			num_bytes = num_bytes & ~(root->sectorsize - 1);
 			num_bytes = max(num_bytes, min_alloc_size);
-			ret = do_chunk_alloc(trans, root->fs_info->extent_root,
-				       num_bytes, data, CHUNK_ALLOC_FORCE);
-			if (ret < 0 && ret != -ENOSPC) {
-				btrfs_abort_transaction(trans, root, ret);
-				return ret;
-			}
 			if (num_bytes == min_alloc_size)
 				final_tried = true;
 			goto again;
@@ -6314,7 +6288,7 @@ use_block_rsv(struct btrfs_trans_handle *trans,
 	ret = block_rsv_use_bytes(block_rsv, blocksize);
 	if (!ret)
 		return block_rsv;
-	if (ret) {
+	if (ret && !block_rsv->failfast) {
 		static DEFINE_RATELIMIT_STATE(_rs,
 				DEFAULT_RATELIMIT_INTERVAL,
 				/*DEFAULT_RATELIMIT_BURST*/ 2);
@@ -7279,7 +7253,7 @@ int btrfs_set_block_group_ro(struct btrfs_root *root,
 
 	alloc_flags = update_block_group_flags(root, cache->flags);
 	if (alloc_flags != cache->flags) {
-		ret = do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags,
+		ret = do_chunk_alloc(trans, root, alloc_flags,
 				     CHUNK_ALLOC_FORCE);
 		if (ret < 0)
 			goto out;
@@ -7289,7 +7263,7 @@ int btrfs_set_block_group_ro(struct btrfs_root *root,
 	if (!ret)
 		goto out;
 	alloc_flags = get_alloc_profile(root, cache->space_info->flags);
-	ret = do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags,
+	ret = do_chunk_alloc(trans, root, alloc_flags,
 			     CHUNK_ALLOC_FORCE);
 	if (ret < 0)
 		goto out;
@@ -7303,7 +7277,7 @@ int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans,
 			    struct btrfs_root *root, u64 type)
 {
 	u64 alloc_flags = get_alloc_profile(root, type);
-	return do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags,
+	return do_chunk_alloc(trans, root, alloc_flags,
 			      CHUNK_ALLOC_FORCE);
 }
 
@@ -7810,6 +7784,34 @@ error:
 	return ret;
 }
 
+void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans,
+				       struct btrfs_root *root)
+{
+	struct btrfs_block_group_cache *block_group, *tmp;
+	struct btrfs_root *extent_root = root->fs_info->extent_root;
+	struct btrfs_block_group_item item;
+	struct btrfs_key key;
+	int ret = 0;
+
+	list_for_each_entry_safe(block_group, tmp, &trans->new_bgs,
+				 new_bg_list) {
+		list_del_init(&block_group->new_bg_list);
+
+		if (ret)
+			continue;
+
+		spin_lock(&block_group->lock);
+		memcpy(&item, &block_group->item, sizeof(item));
+		memcpy(&key, &block_group->key, sizeof(key));
+		spin_unlock(&block_group->lock);
+
+		ret = btrfs_insert_item(trans, extent_root, &key, &item,
+					sizeof(item));
+		if (ret)
+			btrfs_abort_transaction(trans, extent_root, ret);
+	}
+}
+
 int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root, u64 bytes_used,
 			   u64 type, u64 chunk_objectid, u64 chunk_offset,
@@ -7843,6 +7845,7 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 	spin_lock_init(&cache->lock);
 	INIT_LIST_HEAD(&cache->list);
 	INIT_LIST_HEAD(&cache->cluster_list);
+	INIT_LIST_HEAD(&cache->new_bg_list);
 
 	btrfs_init_free_space_ctl(cache);
 
@@ -7874,12 +7877,7 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 	ret = btrfs_add_block_group_cache(root->fs_info, cache);
 	BUG_ON(ret); /* Logic error */
 
-	ret = btrfs_insert_item(trans, extent_root, &cache->key, &cache->item,
-				sizeof(cache->item));
-	if (ret) {
-		btrfs_abort_transaction(trans, extent_root, ret);
-		return ret;
-	}
+	list_add_tail(&cache->new_bg_list, &trans->new_bgs);
 
 	set_avail_alloc_bits(extent_root->fs_info, type);
 

fs/btrfs/extent_io.c

@@ -45,6 +45,7 @@ struct extent_page_data {
 	struct bio *bio;
 	struct extent_io_tree *tree;
 	get_extent_t *get_extent;
+	unsigned long bio_flags;
 
 	/* tells writepage not to lock the state bits for this range
 	 * it still does the unlocking
@@ -64,13 +65,13 @@ tree_fs_info(struct extent_io_tree *tree)
 
 int __init extent_io_init(void)
 {
-	extent_state_cache = kmem_cache_create("extent_state",
+	extent_state_cache = kmem_cache_create("btrfs_extent_state",
 			sizeof(struct extent_state), 0,
 			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!extent_state_cache)
 		return -ENOMEM;
 
-	extent_buffer_cache = kmem_cache_create("extent_buffers",
+	extent_buffer_cache = kmem_cache_create("btrfs_extent_buffer",
 			sizeof(struct extent_buffer), 0,
 			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!extent_buffer_cache)
@@ -942,6 +943,7 @@ int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits,
  * @end:	the end offset in bytes (inclusive)
  * @bits:	the bits to set in this range
  * @clear_bits:	the bits to clear in this range
+ * @cached_state:	state that we're going to cache
  * @mask:	the allocation mask
  *
  * This will go through and set bits for the given range.  If any states exist
@@ -951,7 +953,8 @@ int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits,
  * boundary bits like LOCK.
  */
 int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		       int bits, int clear_bits, gfp_t mask)
+		       int bits, int clear_bits,
+		       struct extent_state **cached_state, gfp_t mask)
 {
 	struct extent_state *state;
 	struct extent_state *prealloc = NULL;
@@ -968,6 +971,15 @@ again:
 	}
 
 	spin_lock(&tree->lock);
+	if (cached_state && *cached_state) {
+		state = *cached_state;
+		if (state->start <= start && state->end > start &&
+		    state->tree) {
+			node = &state->rb_node;
+			goto hit_next;
+		}
+	}
+
 	/*
 	 * this search will find all the extents that end after
 	 * our range starts.
@@ -998,6 +1010,7 @@ hit_next:
 	 */
 	if (state->start == start && state->end <= end) {
 		set_state_bits(tree, state, &bits);
+		cache_state(state, cached_state);
 		state = clear_state_bit(tree, state, &clear_bits, 0);
 		if (last_end == (u64)-1)
 			goto out;
@@ -1038,6 +1051,7 @@ hit_next:
 			goto out;
 		if (state->end <= end) {
 			set_state_bits(tree, state, &bits);
+			cache_state(state, cached_state);
 			state = clear_state_bit(tree, state, &clear_bits, 0);
 			if (last_end == (u64)-1)
 				goto out;
@@ -1076,6 +1090,7 @@ hit_next:
 				   &bits);
 		if (err)
 			extent_io_tree_panic(tree, err);
+		cache_state(prealloc, cached_state);
 		prealloc = NULL;
 		start = this_end + 1;
 		goto search_again;
@@ -1098,6 +1113,7 @@ hit_next:
 			extent_io_tree_panic(tree, err);
 
 		set_state_bits(tree, prealloc, &bits);
+		cache_state(prealloc, cached_state);
 		clear_state_bit(tree, prealloc, &clear_bits, 0);
 		prealloc = NULL;
 		goto out;
@@ -1150,6 +1166,14 @@ int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
 			      NULL, cached_state, mask);
 }
 
+int set_extent_defrag(struct extent_io_tree *tree, u64 start, u64 end,
+		      struct extent_state **cached_state, gfp_t mask)
+{
+	return set_extent_bit(tree, start, end,
+			      EXTENT_DELALLOC | EXTENT_UPTODATE | EXTENT_DEFRAG,
+			      NULL, cached_state, mask);
+}
+
 int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 		       gfp_t mask)
 {
@@ -1294,18 +1318,42 @@ out:
  * If nothing was found, 1 is returned. If found something, return 0.
  */
 int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
-			  u64 *start_ret, u64 *end_ret, int bits)
+			  u64 *start_ret, u64 *end_ret, int bits,
+			  struct extent_state **cached_state)
 {
 	struct extent_state *state;
+	struct rb_node *n;
 	int ret = 1;
 
 	spin_lock(&tree->lock);
+	if (cached_state && *cached_state) {
+		state = *cached_state;
+		if (state->end == start - 1 && state->tree) {
+			n = rb_next(&state->rb_node);
+			while (n) {
+				state = rb_entry(n, struct extent_state,
+						 rb_node);
+				if (state->state & bits)
+					goto got_it;
+				n = rb_next(n);
+			}
+			free_extent_state(*cached_state);
+			*cached_state = NULL;
+			goto out;
+		}
+		free_extent_state(*cached_state);
+		*cached_state = NULL;
+	}
+
 	state = find_first_extent_bit_state(tree, start, bits);
+got_it:
 	if (state) {
+		cache_state(state, cached_state);
 		*start_ret = state->start;
 		*end_ret = state->end;
 		ret = 0;
 	}
+out:
 	spin_unlock(&tree->lock);
 	return ret;
 }
@@ -2068,7 +2116,7 @@ static int bio_readpage_error(struct bio *failed_bio, struct page *page,
 		}
 		read_unlock(&em_tree->lock);
 
-		if (!em || IS_ERR(em)) {
+		if (!em) {
 			kfree(failrec);
 			return -EIO;
 		}
@@ -2304,8 +2352,8 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 		struct extent_state *cached = NULL;
 		struct extent_state *state;
 
-		pr_debug("end_bio_extent_readpage: bi_vcnt=%d, idx=%d, err=%d, "
-			 "mirror=%ld\n", bio->bi_vcnt, bio->bi_idx, err,
+		pr_debug("end_bio_extent_readpage: bi_sector=%llu, err=%d, "
+			 "mirror=%ld\n", (u64)bio->bi_sector, err,
 			 (long int)bio->bi_bdev);
 		tree = &BTRFS_I(page->mapping->host)->io_tree;
 
@@ -2709,12 +2757,15 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 					 end_bio_extent_readpage, mirror_num,
 					 *bio_flags,
 					 this_bio_flag);
-			BUG_ON(ret == -ENOMEM);
-			nr++;
-			*bio_flags = this_bio_flag;
+			if (!ret) {
+				nr++;
+				*bio_flags = this_bio_flag;
+			}
 		}
-		if (ret)
+		if (ret) {
 			SetPageError(page);
+			unlock_extent(tree, cur, cur + iosize - 1);
+		}
 		cur = cur + iosize;
 		pg_offset += iosize;
 	}
@@ -3161,12 +3212,16 @@ static int write_one_eb(struct extent_buffer *eb,
 	struct block_device *bdev = fs_info->fs_devices->latest_bdev;
 	u64 offset = eb->start;
 	unsigned long i, num_pages;
+	unsigned long bio_flags = 0;
 	int rw = (epd->sync_io ? WRITE_SYNC : WRITE);
 	int ret = 0;
 
 	clear_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
 	num_pages = num_extent_pages(eb->start, eb->len);
 	atomic_set(&eb->io_pages, num_pages);
+	if (btrfs_header_owner(eb) == BTRFS_TREE_LOG_OBJECTID)
+		bio_flags = EXTENT_BIO_TREE_LOG;
+
 	for (i = 0; i < num_pages; i++) {
 		struct page *p = extent_buffer_page(eb, i);
 
@@ -3175,7 +3230,8 @@ static int write_one_eb(struct extent_buffer *eb,
 		ret = submit_extent_page(rw, eb->tree, p, offset >> 9,
 					 PAGE_CACHE_SIZE, 0, bdev, &epd->bio,
 					 -1, end_bio_extent_buffer_writepage,
-					 0, 0, 0);
+					 0, epd->bio_flags, bio_flags);
+		epd->bio_flags = bio_flags;
 		if (ret) {
 			set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
 			SetPageError(p);
@@ -3210,6 +3266,7 @@ int btree_write_cache_pages(struct address_space *mapping,
 		.tree = tree,
 		.extent_locked = 0,
 		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
+		.bio_flags = 0,
 	};
 	int ret = 0;
 	int done = 0;
@@ -3254,19 +3311,34 @@ retry:
 				break;
 			}
 
+			spin_lock(&mapping->private_lock);
+			if (!PagePrivate(page)) {
+				spin_unlock(&mapping->private_lock);
+				continue;
+			}
+
 			eb = (struct extent_buffer *)page->private;
+
+			/*
+			 * Shouldn't happen and normally this would be a BUG_ON
+			 * but no sense in crashing the users box for something
+			 * we can survive anyway.
+			 */
 			if (!eb) {
+				spin_unlock(&mapping->private_lock);
 				WARN_ON(1);
 				continue;
 			}
 
-			if (eb == prev_eb)
+			if (eb == prev_eb) {
+				spin_unlock(&mapping->private_lock);
 				continue;
+			}
 
-			if (!atomic_inc_not_zero(&eb->refs)) {
-				WARN_ON(1);
+			ret = atomic_inc_not_zero(&eb->refs);
+			spin_unlock(&mapping->private_lock);
+			if (!ret)
 				continue;
-			}
 
 			prev_eb = eb;
 			ret = lock_extent_buffer_for_io(eb, fs_info, &epd);
@@ -3457,7 +3529,7 @@ static void flush_epd_write_bio(struct extent_page_data *epd)
 		if (epd->sync_io)
 			rw = WRITE_SYNC;
 
-		ret = submit_one_bio(rw, epd->bio, 0, 0);
+		ret = submit_one_bio(rw, epd->bio, 0, epd->bio_flags);
 		BUG_ON(ret < 0); /* -ENOMEM */
 		epd->bio = NULL;
 	}
@@ -3480,6 +3552,7 @@ int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
 		.get_extent = get_extent,
 		.extent_locked = 0,
 		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
+		.bio_flags = 0,
 	};
 
 	ret = __extent_writepage(page, wbc, &epd);
@@ -3504,6 +3577,7 @@ int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode,
 		.get_extent = get_extent,
 		.extent_locked = 1,
 		.sync_io = mode == WB_SYNC_ALL,
+		.bio_flags = 0,
 	};
 	struct writeback_control wbc_writepages = {
 		.sync_mode	= mode,
@@ -3543,6 +3617,7 @@ int extent_writepages(struct extent_io_tree *tree,
 		.get_extent = get_extent,
 		.extent_locked = 0,
 		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
+		.bio_flags = 0,
 	};
 
 	ret = extent_write_cache_pages(tree, mapping, wbc,
@@ -3920,18 +3995,6 @@ out:
 	return ret;
 }
 
-inline struct page *extent_buffer_page(struct extent_buffer *eb,
-					      unsigned long i)
-{
-	return eb->pages[i];
-}
-
-inline unsigned long num_extent_pages(u64 start, u64 len)
-{
-	return ((start + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) -
-		(start >> PAGE_CACHE_SHIFT);
-}
-
 static void __free_extent_buffer(struct extent_buffer *eb)
 {
 #if LEAK_DEBUG
@@ -4047,7 +4110,7 @@ struct extent_buffer *alloc_dummy_extent_buffer(u64 start, unsigned long len)
 
 	return eb;
 err:
-	for (i--; i > 0; i--)
+	for (i--; i >= 0; i--)
 		__free_page(eb->pages[i]);
 	__free_extent_buffer(eb);
 	return NULL;
@@ -4192,10 +4255,8 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 
 	for (i = 0; i < num_pages; i++, index++) {
 		p = find_or_create_page(mapping, index, GFP_NOFS);
-		if (!p) {
-			WARN_ON(1);
+		if (!p)
 			goto free_eb;
-		}
 
 		spin_lock(&mapping->private_lock);
 		if (PagePrivate(p)) {
@@ -4338,7 +4399,6 @@ static int release_extent_buffer(struct extent_buffer *eb, gfp_t mask)
 
 		/* Should be safe to release our pages at this point */
 		btrfs_release_extent_buffer_page(eb, 0);
-
 		call_rcu(&eb->rcu_head, btrfs_release_extent_buffer_rcu);
 		return 1;
 	}

fs/btrfs/extent_io.h

@@ -27,6 +27,7 @@
  * type for this bio
  */
 #define EXTENT_BIO_COMPRESSED 1
+#define EXTENT_BIO_TREE_LOG 2
 #define EXTENT_BIO_FLAG_SHIFT 16
 
 /* these are bit numbers for test/set bit */
@@ -232,11 +233,15 @@ int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 		       gfp_t mask);
 int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		       int bits, int clear_bits, gfp_t mask);
+		       int bits, int clear_bits,
+		       struct extent_state **cached_state, gfp_t mask);
 int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
 			struct extent_state **cached_state, gfp_t mask);
+int set_extent_defrag(struct extent_io_tree *tree, u64 start, u64 end,
+		      struct extent_state **cached_state, gfp_t mask);
 int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
-			  u64 *start_ret, u64 *end_ret, int bits);
+			  u64 *start_ret, u64 *end_ret, int bits,
+			  struct extent_state **cached_state);
 struct extent_state *find_first_extent_bit_state(struct extent_io_tree *tree,
 						 u64 start, int bits);
 int extent_invalidatepage(struct extent_io_tree *tree,
@@ -277,8 +282,18 @@ void free_extent_buffer_stale(struct extent_buffer *eb);
 int read_extent_buffer_pages(struct extent_io_tree *tree,
 			     struct extent_buffer *eb, u64 start, int wait,
 			     get_extent_t *get_extent, int mirror_num);
-unsigned long num_extent_pages(u64 start, u64 len);
-struct page *extent_buffer_page(struct extent_buffer *eb, unsigned long i);
+
+static inline unsigned long num_extent_pages(u64 start, u64 len)
+{
+	return ((start + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) -
+		(start >> PAGE_CACHE_SHIFT);
+}
+
+static inline struct page *extent_buffer_page(struct extent_buffer *eb,
+					      unsigned long i)
+{
+	return eb->pages[i];
+}
 
 static inline void extent_buffer_get(struct extent_buffer *eb)
 {

fs/btrfs/extent_map.c

@@ -11,7 +11,7 @@ static struct kmem_cache *extent_map_cache;
 
 int __init extent_map_init(void)
 {
-	extent_map_cache = kmem_cache_create("extent_map",
+	extent_map_cache = kmem_cache_create("btrfs_extent_map",
 			sizeof(struct extent_map), 0,
 			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!extent_map_cache)
@@ -35,6 +35,7 @@ void extent_map_exit(void)
 void extent_map_tree_init(struct extent_map_tree *tree)
 {
 	tree->map = RB_ROOT;
+	INIT_LIST_HEAD(&tree->modified_extents);
 	rwlock_init(&tree->lock);
 }
 
@@ -54,7 +55,9 @@ struct extent_map *alloc_extent_map(void)
 	em->in_tree = 0;
 	em->flags = 0;
 	em->compress_type = BTRFS_COMPRESS_NONE;
+	em->generation = 0;
 	atomic_set(&em->refs, 1);
+	INIT_LIST_HEAD(&em->list);
 	return em;
 }
 
@@ -72,6 +75,7 @@ void free_extent_map(struct extent_map *em)
 	WARN_ON(atomic_read(&em->refs) == 0);
 	if (atomic_dec_and_test(&em->refs)) {
 		WARN_ON(em->in_tree);
+		WARN_ON(!list_empty(&em->list));
 		kmem_cache_free(extent_map_cache, em);
 	}
 }
@@ -198,6 +202,14 @@ static void try_merge_map(struct extent_map_tree *tree, struct extent_map *em)
 			em->block_len += merge->block_len;
 			em->block_start = merge->block_start;
 			merge->in_tree = 0;
+			if (merge->generation > em->generation) {
+				em->mod_start = em->start;
+				em->mod_len = em->len;
+				em->generation = merge->generation;
+				list_move(&em->list, &tree->modified_extents);
+			}
+
+			list_del_init(&merge->list);
 			rb_erase(&merge->rb_node, &tree->map);
 			free_extent_map(merge);
 		}
@@ -211,14 +223,34 @@ static void try_merge_map(struct extent_map_tree *tree, struct extent_map *em)
 		em->block_len += merge->len;
 		rb_erase(&merge->rb_node, &tree->map);
 		merge->in_tree = 0;
+		if (merge->generation > em->generation) {
+			em->mod_len = em->len;
+			em->generation = merge->generation;
+			list_move(&em->list, &tree->modified_extents);
+		}
+		list_del_init(&merge->list);
 		free_extent_map(merge);
 	}
 }
 
-int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len)
+/**
+ * unpint_extent_cache - unpin an extent from the cache
+ * @tree:	tree to unpin the extent in
+ * @start:	logical offset in the file
+ * @len:	length of the extent
+ * @gen:	generation that this extent has been modified in
+ * @prealloc:	if this is set we need to clear the prealloc flag
+ *
+ * Called after an extent has been written to disk properly.  Set the generation
+ * to the generation that actually added the file item to the inode so we know
+ * we need to sync this extent when we call fsync().
+ */
+int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len,
+		       u64 gen)
 {
 	int ret = 0;
 	struct extent_map *em;
+	bool prealloc = false;
 
 	write_lock(&tree->lock);
 	em = lookup_extent_mapping(tree, start, len);
@@ -228,10 +260,24 @@ int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len)
 	if (!em)
 		goto out;
 
+	list_move(&em->list, &tree->modified_extents);
+	em->generation = gen;
 	clear_bit(EXTENT_FLAG_PINNED, &em->flags);
+	em->mod_start = em->start;
+	em->mod_len = em->len;
+
+	if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
+		prealloc = true;
+		clear_bit(EXTENT_FLAG_PREALLOC, &em->flags);
+	}
 
 	try_merge_map(tree, em);
 
+	if (prealloc) {
+		em->mod_start = em->start;
+		em->mod_len = em->len;
+	}
+
 	free_extent_map(em);
 out:
 	write_unlock(&tree->lock);
@@ -269,6 +315,9 @@ int add_extent_mapping(struct extent_map_tree *tree,
 	}
 	atomic_inc(&em->refs);
 
+	em->mod_start = em->start;
+	em->mod_len = em->len;
+
 	try_merge_map(tree, em);
 out:
 	return ret;
@@ -358,6 +407,8 @@ int remove_extent_mapping(struct extent_map_tree *tree, struct extent_map *em)
 
 	WARN_ON(test_bit(EXTENT_FLAG_PINNED, &em->flags));
 	rb_erase(&em->rb_node, &tree->map);
+	if (!test_bit(EXTENT_FLAG_LOGGING, &em->flags))
+		list_del_init(&em->list);
 	em->in_tree = 0;
 	return ret;
 }

fs/btrfs/extent_map.h

@@ -13,6 +13,7 @@
 #define EXTENT_FLAG_COMPRESSED 1
 #define EXTENT_FLAG_VACANCY 2 /* no file extent item found */
 #define EXTENT_FLAG_PREALLOC 3 /* pre-allocated extent */
+#define EXTENT_FLAG_LOGGING 4 /* Logging this extent */
 
 struct extent_map {
 	struct rb_node rb_node;
@@ -20,18 +21,23 @@ struct extent_map {
 	/* all of these are in bytes */
 	u64 start;
 	u64 len;
+	u64 mod_start;
+	u64 mod_len;
 	u64 orig_start;
 	u64 block_start;
 	u64 block_len;
+	u64 generation;
 	unsigned long flags;
 	struct block_device *bdev;
 	atomic_t refs;
 	unsigned int in_tree;
 	unsigned int compress_type;
+	struct list_head list;
 };
 
 struct extent_map_tree {
 	struct rb_root map;
+	struct list_head modified_extents;
 	rwlock_t lock;
 };
 
@@ -60,7 +66,7 @@ struct extent_map *alloc_extent_map(void);
 void free_extent_map(struct extent_map *em);
 int __init extent_map_init(void);
 void extent_map_exit(void);
-int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len);
+int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len, u64 gen);
 struct extent_map *search_extent_mapping(struct extent_map_tree *tree,
 					 u64 start, u64 len);
 #endif

fs/btrfs/file-item.c

@@ -25,11 +25,12 @@
 #include "transaction.h"
 #include "print-tree.h"
 
-#define __MAX_CSUM_ITEMS(r, size) ((((BTRFS_LEAF_DATA_SIZE(r) - \
+#define __MAX_CSUM_ITEMS(r, size) ((unsigned long)(((BTRFS_LEAF_DATA_SIZE(r) - \
 				   sizeof(struct btrfs_item) * 2) / \
 				  size) - 1))
 
-#define MAX_CSUM_ITEMS(r, size) (min(__MAX_CSUM_ITEMS(r, size), PAGE_CACHE_SIZE))
+#define MAX_CSUM_ITEMS(r, size) (min_t(u32, __MAX_CSUM_ITEMS(r, size), \
+				       PAGE_CACHE_SIZE))
 
 #define MAX_ORDERED_SUM_BYTES(r) ((PAGE_SIZE - \
 				   sizeof(struct btrfs_ordered_sum)) / \

fs/btrfs/file.c

@@ -39,6 +39,7 @@
 #include "tree-log.h"
 #include "locking.h"
 #include "compat.h"
+#include "volumes.h"
 
 /*
  * when auto defrag is enabled we
@@ -458,14 +459,15 @@ int btrfs_dirty_pages(struct btrfs_root *root, struct inode *inode,
  * this drops all the extents in the cache that intersect the range
  * [start, end].  Existing extents are split as required.
  */
-int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
-			    int skip_pinned)
+void btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
+			     int skip_pinned)
 {
 	struct extent_map *em;
 	struct extent_map *split = NULL;
 	struct extent_map *split2 = NULL;
 	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	u64 len = end - start + 1;
+	u64 gen;
 	int ret;
 	int testend = 1;
 	unsigned long flags;
@@ -477,11 +479,14 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 		testend = 0;
 	}
 	while (1) {
+		int no_splits = 0;
+
 		if (!split)
 			split = alloc_extent_map();
 		if (!split2)
 			split2 = alloc_extent_map();
-		BUG_ON(!split || !split2); /* -ENOMEM */
+		if (!split || !split2)
+			no_splits = 1;
 
 		write_lock(&em_tree->lock);
 		em = lookup_extent_mapping(em_tree, start, len);
@@ -490,6 +495,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 			break;
 		}
 		flags = em->flags;
+		gen = em->generation;
 		if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
 			if (testend && em->start + em->len >= start + len) {
 				free_extent_map(em);
@@ -506,6 +512,8 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 		compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
 		clear_bit(EXTENT_FLAG_PINNED, &em->flags);
 		remove_extent_mapping(em_tree, em);
+		if (no_splits)
+			goto next;
 
 		if (em->block_start < EXTENT_MAP_LAST_BYTE &&
 		    em->start < start) {
@@ -518,12 +526,13 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 				split->block_len = em->block_len;
 			else
 				split->block_len = split->len;
-
+			split->generation = gen;
 			split->bdev = em->bdev;
 			split->flags = flags;
 			split->compress_type = em->compress_type;
 			ret = add_extent_mapping(em_tree, split);
 			BUG_ON(ret); /* Logic error */
+			list_move(&split->list, &em_tree->modified_extents);
 			free_extent_map(split);
 			split = split2;
 			split2 = NULL;
@@ -537,6 +546,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 			split->bdev = em->bdev;
 			split->flags = flags;
 			split->compress_type = em->compress_type;
+			split->generation = gen;
 
 			if (compressed) {
 				split->block_len = em->block_len;
@@ -550,9 +560,11 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 
 			ret = add_extent_mapping(em_tree, split);
 			BUG_ON(ret); /* Logic error */
+			list_move(&split->list, &em_tree->modified_extents);
 			free_extent_map(split);
 			split = NULL;
 		}
+next:
 		write_unlock(&em_tree->lock);
 
 		/* once for us */
@@ -564,7 +576,6 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 		free_extent_map(split);
 	if (split2)
 		free_extent_map(split2);
-	return 0;
 }
 
 /*
@@ -576,13 +587,13 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
  * it is either truncated or split.  Anything entirely inside the range
  * is deleted from the tree.
  */
-int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode,
-		       u64 start, u64 end, u64 *hint_byte, int drop_cache)
+int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
+			 struct btrfs_root *root, struct inode *inode,
+			 struct btrfs_path *path, u64 start, u64 end,
+			 u64 *drop_end, int drop_cache)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct extent_buffer *leaf;
 	struct btrfs_file_extent_item *fi;
-	struct btrfs_path *path;
 	struct btrfs_key key;
 	struct btrfs_key new_key;
 	u64 ino = btrfs_ino(inode);
@@ -597,14 +608,12 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode,
 	int recow;
 	int ret;
 	int modify_tree = -1;
+	int update_refs = (root->ref_cows || root == root->fs_info->tree_root);
+	int found = 0;
 
 	if (drop_cache)
 		btrfs_drop_extent_cache(inode, start, end - 1, 0);
 
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
 	if (start >= BTRFS_I(inode)->disk_i_size)
 		modify_tree = 0;
 
@@ -666,6 +675,7 @@ next_slot:
 			goto next_slot;
 		}
 
+		found = 1;
 		search_start = max(key.offset, start);
 		if (recow || !modify_tree) {
 			modify_tree = -1;
@@ -707,14 +717,13 @@ next_slot:
 							extent_end - start);
 			btrfs_mark_buffer_dirty(leaf);
 
-			if (disk_bytenr > 0) {
+			if (update_refs && disk_bytenr > 0) {
 				ret = btrfs_inc_extent_ref(trans, root,
 						disk_bytenr, num_bytes, 0,
 						root->root_key.objectid,
 						new_key.objectid,
 						start - extent_offset, 0);
 				BUG_ON(ret); /* -ENOMEM */
-				*hint_byte = disk_bytenr;
 			}
 			key.offset = start;
 		}
@@ -734,10 +743,8 @@ next_slot:
 			btrfs_set_file_extent_num_bytes(leaf, fi,
 							extent_end - end);
 			btrfs_mark_buffer_dirty(leaf);
-			if (disk_bytenr > 0) {
+			if (update_refs && disk_bytenr > 0)
 				inode_sub_bytes(inode, end - key.offset);
-				*hint_byte = disk_bytenr;
-			}
 			break;
 		}
 
@@ -753,10 +760,8 @@ next_slot:
 			btrfs_set_file_extent_num_bytes(leaf, fi,
 							start - key.offset);
 			btrfs_mark_buffer_dirty(leaf);
-			if (disk_bytenr > 0) {
+			if (update_refs && disk_bytenr > 0)
 				inode_sub_bytes(inode, extent_end - start);
-				*hint_byte = disk_bytenr;
-			}
 			if (end == extent_end)
 				break;
 
@@ -777,12 +782,13 @@ next_slot:
 				del_nr++;
 			}
 
-			if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
+			if (update_refs &&
+			    extent_type == BTRFS_FILE_EXTENT_INLINE) {
 				inode_sub_bytes(inode,
 						extent_end - key.offset);
 				extent_end = ALIGN(extent_end,
 						   root->sectorsize);
-			} else if (disk_bytenr > 0) {
+			} else if (update_refs && disk_bytenr > 0) {
 				ret = btrfs_free_extent(trans, root,
 						disk_bytenr, num_bytes, 0,
 						root->root_key.objectid,
@@ -791,7 +797,6 @@ next_slot:
 				BUG_ON(ret); /* -ENOMEM */
 				inode_sub_bytes(inode,
 						extent_end - key.offset);
-				*hint_byte = disk_bytenr;
 			}
 
 			if (end == extent_end)
@@ -806,7 +811,7 @@ next_slot:
 					      del_nr);
 			if (ret) {
 				btrfs_abort_transaction(trans, root, ret);
-				goto out;
+				break;
 			}
 
 			del_nr = 0;
@@ -825,7 +830,24 @@ next_slot:
 			btrfs_abort_transaction(trans, root, ret);
 	}
 
-out:
+	if (drop_end)
+		*drop_end = found ? min(end, extent_end) : end;
+	btrfs_release_path(path);
+	return ret;
+}
+
+int btrfs_drop_extents(struct btrfs_trans_handle *trans,
+		       struct btrfs_root *root, struct inode *inode, u64 start,
+		       u64 end, int drop_cache)
+{
+	struct btrfs_path *path;
+	int ret;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+	ret = __btrfs_drop_extents(trans, root, inode, path, start, end, NULL,
+				   drop_cache);
 	btrfs_free_path(path);
 	return ret;
 }
@@ -892,8 +914,6 @@ int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
 	int ret;
 	u64 ino = btrfs_ino(inode);
 
-	btrfs_drop_extent_cache(inode, start, end - 1, 0);
-
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
@@ -935,12 +955,16 @@ again:
 			btrfs_set_item_key_safe(trans, root, path, &new_key);
 			fi = btrfs_item_ptr(leaf, path->slots[0],
 					    struct btrfs_file_extent_item);
+			btrfs_set_file_extent_generation(leaf, fi,
+							 trans->transid);
 			btrfs_set_file_extent_num_bytes(leaf, fi,
 							extent_end - end);
 			btrfs_set_file_extent_offset(leaf, fi,
 						     end - orig_offset);
 			fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
 					    struct btrfs_file_extent_item);
+			btrfs_set_file_extent_generation(leaf, fi,
+							 trans->transid);
 			btrfs_set_file_extent_num_bytes(leaf, fi,
 							end - other_start);
 			btrfs_mark_buffer_dirty(leaf);
@@ -958,12 +982,16 @@ again:
 					    struct btrfs_file_extent_item);
 			btrfs_set_file_extent_num_bytes(leaf, fi,
 							start - key.offset);
+			btrfs_set_file_extent_generation(leaf, fi,
+							 trans->transid);
 			path->slots[0]++;
 			new_key.offset = start;
 			btrfs_set_item_key_safe(trans, root, path, &new_key);
 
 			fi = btrfs_item_ptr(leaf, path->slots[0],
 					    struct btrfs_file_extent_item);
+			btrfs_set_file_extent_generation(leaf, fi,
+							 trans->transid);
 			btrfs_set_file_extent_num_bytes(leaf, fi,
 							other_end - start);
 			btrfs_set_file_extent_offset(leaf, fi,
@@ -991,12 +1019,14 @@ again:
 		leaf = path->nodes[0];
 		fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
 				    struct btrfs_file_extent_item);
+		btrfs_set_file_extent_generation(leaf, fi, trans->transid);
 		btrfs_set_file_extent_num_bytes(leaf, fi,
 						split - key.offset);
 
 		fi = btrfs_item_ptr(leaf, path->slots[0],
 				    struct btrfs_file_extent_item);
 
+		btrfs_set_file_extent_generation(leaf, fi, trans->transid);
 		btrfs_set_file_extent_offset(leaf, fi, split - orig_offset);
 		btrfs_set_file_extent_num_bytes(leaf, fi,
 						extent_end - split);
@@ -1056,12 +1086,14 @@ again:
 			   struct btrfs_file_extent_item);
 		btrfs_set_file_extent_type(leaf, fi,
 					   BTRFS_FILE_EXTENT_REG);
+		btrfs_set_file_extent_generation(leaf, fi, trans->transid);
 		btrfs_mark_buffer_dirty(leaf);
 	} else {
 		fi = btrfs_item_ptr(leaf, del_slot - 1,
 			   struct btrfs_file_extent_item);
 		btrfs_set_file_extent_type(leaf, fi,
 					   BTRFS_FILE_EXTENT_REG);
+		btrfs_set_file_extent_generation(leaf, fi, trans->transid);
 		btrfs_set_file_extent_num_bytes(leaf, fi,
 						extent_end - key.offset);
 		btrfs_mark_buffer_dirty(leaf);
@@ -1173,8 +1205,8 @@ again:
 
 		clear_extent_bit(&BTRFS_I(inode)->io_tree, start_pos,
 				  last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
-				  EXTENT_DO_ACCOUNTING, 0, 0, &cached_state,
-				  GFP_NOFS);
+				  EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
+				  0, 0, &cached_state, GFP_NOFS);
 		unlock_extent_cached(&BTRFS_I(inode)->io_tree,
 				     start_pos, last_pos - 1, &cached_state,
 				     GFP_NOFS);
@@ -1514,16 +1546,24 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 
 	trace_btrfs_sync_file(file, datasync);
 
+	/*
+	 * We write the dirty pages in the range and wait until they complete
+	 * out of the ->i_mutex. If so, we can flush the dirty pages by
+	 * multi-task, and make the performance up.
+	 */
+	ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
+	if (ret)
+		return ret;
+
 	mutex_lock(&inode->i_mutex);
 
 	/*
-	 * we wait first, since the writeback may change the inode, also wait
-	 * ordered range does a filemape_write_and_wait_range which is why we
-	 * don't do it above like other file systems.
+	 * We flush the dirty pages again to avoid some dirty pages in the
+	 * range being left.
 	 */
-	root->log_batch++;
+	atomic_inc(&root->log_batch);
 	btrfs_wait_ordered_range(inode, start, end);
-	root->log_batch++;
+	atomic_inc(&root->log_batch);
 
 	/*
 	 * check the transaction that last modified this inode
@@ -1544,6 +1584,14 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	    BTRFS_I(inode)->last_trans <=
 	    root->fs_info->last_trans_committed) {
 		BTRFS_I(inode)->last_trans = 0;
+
+		/*
+		 * We'v had everything committed since the last time we were
+		 * modified so clear this flag in case it was set for whatever
+		 * reason, it's no longer relevant.
+		 */
+		clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+			  &BTRFS_I(inode)->runtime_flags);
 		mutex_unlock(&inode->i_mutex);
 		goto out;
 	}
@@ -1615,6 +1663,324 @@ static int btrfs_file_mmap(struct file	*filp, struct vm_area_struct *vma)
 	return 0;
 }
 
+static int hole_mergeable(struct inode *inode, struct extent_buffer *leaf,
+			  int slot, u64 start, u64 end)
+{
+	struct btrfs_file_extent_item *fi;
+	struct btrfs_key key;
+
+	if (slot < 0 || slot >= btrfs_header_nritems(leaf))
+		return 0;
+
+	btrfs_item_key_to_cpu(leaf, &key, slot);
+	if (key.objectid != btrfs_ino(inode) ||
+	    key.type != BTRFS_EXTENT_DATA_KEY)
+		return 0;
+
+	fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
+
+	if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG)
+		return 0;
+
+	if (btrfs_file_extent_disk_bytenr(leaf, fi))
+		return 0;
+
+	if (key.offset == end)
+		return 1;
+	if (key.offset + btrfs_file_extent_num_bytes(leaf, fi) == start)
+		return 1;
+	return 0;
+}
+
+static int fill_holes(struct btrfs_trans_handle *trans, struct inode *inode,
+		      struct btrfs_path *path, u64 offset, u64 end)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct extent_buffer *leaf;
+	struct btrfs_file_extent_item *fi;
+	struct extent_map *hole_em;
+	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	struct btrfs_key key;
+	int ret;
+
+	key.objectid = btrfs_ino(inode);
+	key.type = BTRFS_EXTENT_DATA_KEY;
+	key.offset = offset;
+
+
+	ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
+	if (ret < 0)
+		return ret;
+	BUG_ON(!ret);
+
+	leaf = path->nodes[0];
+	if (hole_mergeable(inode, leaf, path->slots[0]-1, offset, end)) {
+		u64 num_bytes;
+
+		path->slots[0]--;
+		fi = btrfs_item_ptr(leaf, path->slots[0],
+				    struct btrfs_file_extent_item);
+		num_bytes = btrfs_file_extent_num_bytes(leaf, fi) +
+			end - offset;
+		btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
+		btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
+		btrfs_set_file_extent_offset(leaf, fi, 0);
+		btrfs_mark_buffer_dirty(leaf);
+		goto out;
+	}
+
+	if (hole_mergeable(inode, leaf, path->slots[0]+1, offset, end)) {
+		u64 num_bytes;
+
+		path->slots[0]++;
+		key.offset = offset;
+		btrfs_set_item_key_safe(trans, root, path, &key);
+		fi = btrfs_item_ptr(leaf, path->slots[0],
+				    struct btrfs_file_extent_item);
+		num_bytes = btrfs_file_extent_num_bytes(leaf, fi) + end -
+			offset;
+		btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
+		btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
+		btrfs_set_file_extent_offset(leaf, fi, 0);
+		btrfs_mark_buffer_dirty(leaf);
+		goto out;
+	}
+	btrfs_release_path(path);
+
+	ret = btrfs_insert_file_extent(trans, root, btrfs_ino(inode), offset,
+				       0, 0, end - offset, 0, end - offset,
+				       0, 0, 0);
+	if (ret)
+		return ret;
+
+out:
+	btrfs_release_path(path);
+
+	hole_em = alloc_extent_map();
+	if (!hole_em) {
+		btrfs_drop_extent_cache(inode, offset, end - 1, 0);
+		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+			&BTRFS_I(inode)->runtime_flags);
+	} else {
+		hole_em->start = offset;
+		hole_em->len = end - offset;
+		hole_em->orig_start = offset;
+
+		hole_em->block_start = EXTENT_MAP_HOLE;
+		hole_em->block_len = 0;
+		hole_em->bdev = root->fs_info->fs_devices->latest_bdev;
+		hole_em->compress_type = BTRFS_COMPRESS_NONE;
+		hole_em->generation = trans->transid;
+
+		do {
+			btrfs_drop_extent_cache(inode, offset, end - 1, 0);
+			write_lock(&em_tree->lock);
+			ret = add_extent_mapping(em_tree, hole_em);
+			if (!ret)
+				list_move(&hole_em->list,
+					  &em_tree->modified_extents);
+			write_unlock(&em_tree->lock);
+		} while (ret == -EEXIST);
+		free_extent_map(hole_em);
+		if (ret)
+			set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+				&BTRFS_I(inode)->runtime_flags);
+	}
+
+	return 0;
+}
+
+static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct extent_state *cached_state = NULL;
+	struct btrfs_path *path;
+	struct btrfs_block_rsv *rsv;
+	struct btrfs_trans_handle *trans;
+	u64 mask = BTRFS_I(inode)->root->sectorsize - 1;
+	u64 lockstart = (offset + mask) & ~mask;
+	u64 lockend = ((offset + len) & ~mask) - 1;
+	u64 cur_offset = lockstart;
+	u64 min_size = btrfs_calc_trunc_metadata_size(root, 1);
+	u64 drop_end;
+	unsigned long nr;
+	int ret = 0;
+	int err = 0;
+	bool same_page = (offset >> PAGE_CACHE_SHIFT) ==
+		((offset + len) >> PAGE_CACHE_SHIFT);
+
+	btrfs_wait_ordered_range(inode, offset, len);
+
+	mutex_lock(&inode->i_mutex);
+	if (offset >= inode->i_size) {
+		mutex_unlock(&inode->i_mutex);
+		return 0;
+	}
+
+	/*
+	 * Only do this if we are in the same page and we aren't doing the
+	 * entire page.
+	 */
+	if (same_page && len < PAGE_CACHE_SIZE) {
+		ret = btrfs_truncate_page(inode, offset, len, 0);
+		mutex_unlock(&inode->i_mutex);
+		return ret;
+	}
+
+	/* zero back part of the first page */
+	ret = btrfs_truncate_page(inode, offset, 0, 0);
+	if (ret) {
+		mutex_unlock(&inode->i_mutex);
+		return ret;
+	}
+
+	/* zero the front end of the last page */
+	ret = btrfs_truncate_page(inode, offset + len, 0, 1);
+	if (ret) {
+		mutex_unlock(&inode->i_mutex);
+		return ret;
+	}
+
+	if (lockend < lockstart) {
+		mutex_unlock(&inode->i_mutex);
+		return 0;
+	}
+
+	while (1) {
+		struct btrfs_ordered_extent *ordered;
+
+		truncate_pagecache_range(inode, lockstart, lockend);
+
+		lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+				 0, &cached_state);
+		ordered = btrfs_lookup_first_ordered_extent(inode, lockend);
+
+		/*
+		 * We need to make sure we have no ordered extents in this range
+		 * and nobody raced in and read a page in this range, if we did
+		 * we need to try again.
+		 */
+		if ((!ordered ||
+		    (ordered->file_offset + ordered->len < lockstart ||
+		     ordered->file_offset > lockend)) &&
+		     !test_range_bit(&BTRFS_I(inode)->io_tree, lockstart,
+				     lockend, EXTENT_UPTODATE, 0,
+				     cached_state)) {
+			if (ordered)
+				btrfs_put_ordered_extent(ordered);
+			break;
+		}
+		if (ordered)
+			btrfs_put_ordered_extent(ordered);
+		unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart,
+				     lockend, &cached_state, GFP_NOFS);
+		btrfs_wait_ordered_range(inode, lockstart,
+					 lockend - lockstart + 1);
+	}
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP);
+	if (!rsv) {
+		ret = -ENOMEM;
+		goto out_free;
+	}
+	rsv->size = btrfs_calc_trunc_metadata_size(root, 1);
+	rsv->failfast = 1;
+
+	/*
+	 * 1 - update the inode
+	 * 1 - removing the extents in the range
+	 * 1 - adding the hole extent
+	 */
+	trans = btrfs_start_transaction(root, 3);
+	if (IS_ERR(trans)) {
+		err = PTR_ERR(trans);
+		goto out_free;
+	}
+
+	ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv, rsv,
+				      min_size);
+	BUG_ON(ret);
+	trans->block_rsv = rsv;
+
+	while (cur_offset < lockend) {
+		ret = __btrfs_drop_extents(trans, root, inode, path,
+					   cur_offset, lockend + 1,
+					   &drop_end, 1);
+		if (ret != -ENOSPC)
+			break;
+
+		trans->block_rsv = &root->fs_info->trans_block_rsv;
+
+		ret = fill_holes(trans, inode, path, cur_offset, drop_end);
+		if (ret) {
+			err = ret;
+			break;
+		}
+
+		cur_offset = drop_end;
+
+		ret = btrfs_update_inode(trans, root, inode);
+		if (ret) {
+			err = ret;
+			break;
+		}
+
+		nr = trans->blocks_used;
+		btrfs_end_transaction(trans, root);
+		btrfs_btree_balance_dirty(root, nr);
+
+		trans = btrfs_start_transaction(root, 3);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			trans = NULL;
+			break;
+		}
+
+		ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv,
+					      rsv, min_size);
+		BUG_ON(ret);	/* shouldn't happen */
+		trans->block_rsv = rsv;
+	}
+
+	if (ret) {
+		err = ret;
+		goto out_trans;
+	}
+
+	trans->block_rsv = &root->fs_info->trans_block_rsv;
+	ret = fill_holes(trans, inode, path, cur_offset, drop_end);
+	if (ret) {
+		err = ret;
+		goto out_trans;
+	}
+
+out_trans:
+	if (!trans)
+		goto out_free;
+
+	trans->block_rsv = &root->fs_info->trans_block_rsv;
+	ret = btrfs_update_inode(trans, root, inode);
+	nr = trans->blocks_used;
+	btrfs_end_transaction(trans, root);
+	btrfs_btree_balance_dirty(root, nr);
+out_free:
+	btrfs_free_path(path);
+	btrfs_free_block_rsv(root, rsv);
+out:
+	unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+			     &cached_state, GFP_NOFS);
+	mutex_unlock(&inode->i_mutex);
+	if (ret && !err)
+		err = ret;
+	return err;
+}
+
 static long btrfs_fallocate(struct file *file, int mode,
 			    loff_t offset, loff_t len)
 {
@@ -1633,15 +1999,18 @@ static long btrfs_fallocate(struct file *file, int mode,
 	alloc_start = offset & ~mask;
 	alloc_end =  (offset + len + mask) & ~mask;
 
-	/* We only support the FALLOC_FL_KEEP_SIZE mode */
-	if (mode & ~FALLOC_FL_KEEP_SIZE)
+	/* Make sure we aren't being give some crap mode */
+	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
 		return -EOPNOTSUPP;
 
+	if (mode & FALLOC_FL_PUNCH_HOLE)
+		return btrfs_punch_hole(inode, offset, len);
+
 	/*
 	 * Make sure we have enough space before we do the
 	 * allocation.
 	 */
-	ret = btrfs_check_data_free_space(inode, len);
+	ret = btrfs_check_data_free_space(inode, alloc_end - alloc_start + 1);
 	if (ret)
 		return ret;
 
@@ -1748,7 +2117,7 @@ static long btrfs_fallocate(struct file *file, int mode,
 out:
 	mutex_unlock(&inode->i_mutex);
 	/* Let go of our reservation. */
-	btrfs_free_reserved_data_space(inode, len);
+	btrfs_free_reserved_data_space(inode, alloc_end - alloc_start + 1);
 	return ret;
 }
 

fs/btrfs/free-space-cache.c

@@ -966,7 +966,7 @@ int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 			       block_group->key.offset)) {
 		ret = find_first_extent_bit(unpin, start,
 					    &extent_start, &extent_end,
-					    EXTENT_DIRTY);
+					    EXTENT_DIRTY, NULL);
 		if (ret) {
 			ret = 0;
 			break;
@@ -1454,9 +1454,7 @@ static int search_bitmap(struct btrfs_free_space_ctl *ctl,
 			  max_t(u64, *offset, bitmap_info->offset));
 	bits = bytes_to_bits(*bytes, ctl->unit);
 
-	for (i = find_next_bit(bitmap_info->bitmap, BITS_PER_BITMAP, i);
-	     i < BITS_PER_BITMAP;
-	     i = find_next_bit(bitmap_info->bitmap, BITS_PER_BITMAP, i + 1)) {
+	for_each_set_bit_from(i, bitmap_info->bitmap, BITS_PER_BITMAP) {
 		next_zero = find_next_zero_bit(bitmap_info->bitmap,
 					       BITS_PER_BITMAP, i);
 		if ((next_zero - i) >= bits) {
@@ -2307,9 +2305,7 @@ static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
 
 again:
 	found_bits = 0;
-	for (i = find_next_bit(entry->bitmap, BITS_PER_BITMAP, i);
-	     i < BITS_PER_BITMAP;
-	     i = find_next_bit(entry->bitmap, BITS_PER_BITMAP, i + 1)) {
+	for_each_set_bit_from(i, entry->bitmap, BITS_PER_BITMAP) {
 		next_zero = find_next_zero_bit(entry->bitmap,
 					       BITS_PER_BITMAP, i);
 		if (next_zero - i >= min_bits) {

fs/btrfs/hash.h

@@ -24,4 +24,14 @@ static inline u64 btrfs_name_hash(const char *name, int len)
 {
 	return crc32c((u32)~1, name, len);
 }
+
+/*
+ * Figure the key offset of an extended inode ref
+ */
+static inline u64 btrfs_extref_hash(u64 parent_objectid, const char *name,
+				    int len)
+{
+	return (u64) crc32c(parent_objectid, name, len);
+}
+
 #endif

fs/btrfs/inode-item.c

@@ -18,6 +18,7 @@
 
 #include "ctree.h"
 #include "disk-io.h"
+#include "hash.h"
 #include "transaction.h"
 #include "print-tree.h"
 
@@ -50,18 +51,57 @@ static int find_name_in_backref(struct btrfs_path *path, const char *name,
 	return 0;
 }
 
-struct btrfs_inode_ref *
+int btrfs_find_name_in_ext_backref(struct btrfs_path *path, u64 ref_objectid,
+				   const char *name, int name_len,
+				   struct btrfs_inode_extref **extref_ret)
+{
+	struct extent_buffer *leaf;
+	struct btrfs_inode_extref *extref;
+	unsigned long ptr;
+	unsigned long name_ptr;
+	u32 item_size;
+	u32 cur_offset = 0;
+	int ref_name_len;
+
+	leaf = path->nodes[0];
+	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+	ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
+
+	/*
+	 * Search all extended backrefs in this item. We're only
+	 * looking through any collisions so most of the time this is
+	 * just going to compare against one buffer. If all is well,
+	 * we'll return success and the inode ref object.
+	 */
+	while (cur_offset < item_size) {
+		extref = (struct btrfs_inode_extref *) (ptr + cur_offset);
+		name_ptr = (unsigned long)(&extref->name);
+		ref_name_len = btrfs_inode_extref_name_len(leaf, extref);
+
+		if (ref_name_len == name_len &&
+		    btrfs_inode_extref_parent(leaf, extref) == ref_objectid &&
+		    (memcmp_extent_buffer(leaf, name, name_ptr, name_len) == 0)) {
+			if (extref_ret)
+				*extref_ret = extref;
+			return 1;
+		}
+
+		cur_offset += ref_name_len + sizeof(*extref);
+	}
+	return 0;
+}
+
+static struct btrfs_inode_ref *
 btrfs_lookup_inode_ref(struct btrfs_trans_handle *trans,
-			struct btrfs_root *root,
-			struct btrfs_path *path,
-			const char *name, int name_len,
-			u64 inode_objectid, u64 ref_objectid, int mod)
+		       struct btrfs_root *root,
+		       struct btrfs_path *path,
+		       const char *name, int name_len,
+		       u64 inode_objectid, u64 ref_objectid, int ins_len,
+		       int cow)
 {
+	int ret;
 	struct btrfs_key key;
 	struct btrfs_inode_ref *ref;
-	int ins_len = mod < 0 ? -1 : 0;
-	int cow = mod != 0;
-	int ret;
 
 	key.objectid = inode_objectid;
 	key.type = BTRFS_INODE_REF_KEY;
@@ -77,10 +117,147 @@ btrfs_lookup_inode_ref(struct btrfs_trans_handle *trans,
 	return ref;
 }
 
-int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
+/* Returns NULL if no extref found */
+struct btrfs_inode_extref *
+btrfs_lookup_inode_extref(struct btrfs_trans_handle *trans,
+			  struct btrfs_root *root,
+			  struct btrfs_path *path,
+			  const char *name, int name_len,
+			  u64 inode_objectid, u64 ref_objectid, int ins_len,
+			  int cow)
+{
+	int ret;
+	struct btrfs_key key;
+	struct btrfs_inode_extref *extref;
+
+	key.objectid = inode_objectid;
+	key.type = BTRFS_INODE_EXTREF_KEY;
+	key.offset = btrfs_extref_hash(ref_objectid, name, name_len);
+
+	ret = btrfs_search_slot(trans, root, &key, path, ins_len, cow);
+	if (ret < 0)
+		return ERR_PTR(ret);
+	if (ret > 0)
+		return NULL;
+	if (!btrfs_find_name_in_ext_backref(path, ref_objectid, name, name_len, &extref))
+		return NULL;
+	return extref;
+}
+
+int btrfs_get_inode_ref_index(struct btrfs_trans_handle *trans,
+			      struct btrfs_root *root,
+			      struct btrfs_path *path,
+			      const char *name, int name_len,
+			      u64 inode_objectid, u64 ref_objectid, int mod,
+			      u64 *ret_index)
+{
+	struct btrfs_inode_ref *ref;
+	struct btrfs_inode_extref *extref;
+	int ins_len = mod < 0 ? -1 : 0;
+	int cow = mod != 0;
+
+	ref = btrfs_lookup_inode_ref(trans, root, path, name, name_len,
+				     inode_objectid, ref_objectid, ins_len,
+				     cow);
+	if (IS_ERR(ref))
+		return PTR_ERR(ref);
+
+	if (ref != NULL) {
+		*ret_index = btrfs_inode_ref_index(path->nodes[0], ref);
+		return 0;
+	}
+
+	btrfs_release_path(path);
+
+	extref = btrfs_lookup_inode_extref(trans, root, path, name,
+					   name_len, inode_objectid,
+					   ref_objectid, ins_len, cow);
+	if (IS_ERR(extref))
+		return PTR_ERR(extref);
+
+	if (extref) {
+		*ret_index = btrfs_inode_extref_index(path->nodes[0], extref);
+		return 0;
+	}
+
+	return -ENOENT;
+}
+
+int btrfs_del_inode_extref(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root,
 			   const char *name, int name_len,
 			   u64 inode_objectid, u64 ref_objectid, u64 *index)
+{
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	struct btrfs_inode_extref *extref;
+	struct extent_buffer *leaf;
+	int ret;
+	int del_len = name_len + sizeof(*extref);
+	unsigned long ptr;
+	unsigned long item_start;
+	u32 item_size;
+
+	key.objectid = inode_objectid;
+	btrfs_set_key_type(&key, BTRFS_INODE_EXTREF_KEY);
+	key.offset = btrfs_extref_hash(ref_objectid, name, name_len);
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	path->leave_spinning = 1;
+
+	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+	if (ret > 0)
+		ret = -ENOENT;
+	if (ret < 0)
+		goto out;
+
+	/*
+	 * Sanity check - did we find the right item for this name?
+	 * This should always succeed so error here will make the FS
+	 * readonly.
+	 */
+	if (!btrfs_find_name_in_ext_backref(path, ref_objectid,
+					    name, name_len, &extref)) {
+		btrfs_std_error(root->fs_info, -ENOENT);
+		ret = -EROFS;
+		goto out;
+	}
+
+	leaf = path->nodes[0];
+	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+	if (index)
+		*index = btrfs_inode_extref_index(leaf, extref);
+
+	if (del_len == item_size) {
+		/*
+		 * Common case only one ref in the item, remove the
+		 * whole item.
+		 */
+		ret = btrfs_del_item(trans, root, path);
+		goto out;
+	}
+
+	ptr = (unsigned long)extref;
+	item_start = btrfs_item_ptr_offset(leaf, path->slots[0]);
+
+	memmove_extent_buffer(leaf, ptr, ptr + del_len,
+			      item_size - (ptr + del_len - item_start));
+
+	btrfs_truncate_item(trans, root, path, item_size - del_len, 1);
+
+out:
+	btrfs_free_path(path);
+
+	return ret;
+}
+
+int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
+			struct btrfs_root *root,
+			const char *name, int name_len,
+			u64 inode_objectid, u64 ref_objectid, u64 *index)
 {
 	struct btrfs_path *path;
 	struct btrfs_key key;
@@ -91,6 +268,7 @@ int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
 	u32 item_size;
 	u32 sub_item_len;
 	int ret;
+	int search_ext_refs = 0;
 	int del_len = name_len + sizeof(*ref);
 
 	key.objectid = inode_objectid;
@@ -106,12 +284,14 @@ int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
 	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 	if (ret > 0) {
 		ret = -ENOENT;
+		search_ext_refs = 1;
 		goto out;
 	} else if (ret < 0) {
 		goto out;
 	}
 	if (!find_name_in_backref(path, name, name_len, &ref)) {
 		ret = -ENOENT;
+		search_ext_refs = 1;
 		goto out;
 	}
 	leaf = path->nodes[0];
@@ -129,8 +309,78 @@ int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
 	item_start = btrfs_item_ptr_offset(leaf, path->slots[0]);
 	memmove_extent_buffer(leaf, ptr, ptr + sub_item_len,
 			      item_size - (ptr + sub_item_len - item_start));
-	btrfs_truncate_item(trans, root, path,
-				  item_size - sub_item_len, 1);
+	btrfs_truncate_item(trans, root, path, item_size - sub_item_len, 1);
+out:
+	btrfs_free_path(path);
+
+	if (search_ext_refs) {
+		/*
+		 * No refs were found, or we could not find the
+		 * name in our ref array. Find and remove the extended
+		 * inode ref then.
+		 */
+		return btrfs_del_inode_extref(trans, root, name, name_len,
+					      inode_objectid, ref_objectid, index);
+	}
+
+	return ret;
+}
+
+/*
+ * btrfs_insert_inode_extref() - Inserts an extended inode ref into a tree.
+ *
+ * The caller must have checked against BTRFS_LINK_MAX already.
+ */
+static int btrfs_insert_inode_extref(struct btrfs_trans_handle *trans,
+				     struct btrfs_root *root,
+				     const char *name, int name_len,
+				     u64 inode_objectid, u64 ref_objectid, u64 index)
+{
+	struct btrfs_inode_extref *extref;
+	int ret;
+	int ins_len = name_len + sizeof(*extref);
+	unsigned long ptr;
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	struct extent_buffer *leaf;
+	struct btrfs_item *item;
+
+	key.objectid = inode_objectid;
+	key.type = BTRFS_INODE_EXTREF_KEY;
+	key.offset = btrfs_extref_hash(ref_objectid, name, name_len);
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	path->leave_spinning = 1;
+	ret = btrfs_insert_empty_item(trans, root, path, &key,
+				      ins_len);
+	if (ret == -EEXIST) {
+		if (btrfs_find_name_in_ext_backref(path, ref_objectid,
+						   name, name_len, NULL))
+			goto out;
+
+		btrfs_extend_item(trans, root, path, ins_len);
+		ret = 0;
+	}
+	if (ret < 0)
+		goto out;
+
+	leaf = path->nodes[0];
+	item = btrfs_item_nr(leaf, path->slots[0]);
+	ptr = (unsigned long)btrfs_item_ptr(leaf, path->slots[0], char);
+	ptr += btrfs_item_size(leaf, item) - ins_len;
+	extref = (struct btrfs_inode_extref *)ptr;
+
+	btrfs_set_inode_extref_name_len(path->nodes[0], extref, name_len);
+	btrfs_set_inode_extref_index(path->nodes[0], extref, index);
+	btrfs_set_inode_extref_parent(path->nodes[0], extref, ref_objectid);
+
+	ptr = (unsigned long)&extref->name;
+	write_extent_buffer(path->nodes[0], name, ptr, name_len);
+	btrfs_mark_buffer_dirty(path->nodes[0]);
+
 out:
 	btrfs_free_path(path);
 	return ret;
@@ -191,6 +441,19 @@ int btrfs_insert_inode_ref(struct btrfs_trans_handle *trans,
 
 out:
 	btrfs_free_path(path);
+
+	if (ret == -EMLINK) {
+		struct btrfs_super_block *disk_super = root->fs_info->super_copy;
+		/* We ran out of space in the ref array. Need to
+		 * add an extended ref. */
+		if (btrfs_super_incompat_flags(disk_super)
+		    & BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF)
+			ret = btrfs_insert_inode_extref(trans, root, name,
+							name_len,
+							inode_objectid,
+							ref_objectid, index);
+	}
+
 	return ret;
 }
 

fs/btrfs/inode.c

@@ -230,7 +230,6 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans,
 	u64 inline_len = actual_end - start;
 	u64 aligned_end = (end + root->sectorsize - 1) &
 			~((u64)root->sectorsize - 1);
-	u64 hint_byte;
 	u64 data_len = inline_len;
 	int ret;
 
@@ -247,8 +246,7 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans,
 		return 1;
 	}
 
-	ret = btrfs_drop_extents(trans, inode, start, aligned_end,
-				 &hint_byte, 1);
+	ret = btrfs_drop_extents(trans, root, inode, start, aligned_end, 1);
 	if (ret)
 		return ret;
 
@@ -664,7 +662,7 @@ retry:
 					   async_extent->compressed_size,
 					   async_extent->compressed_size,
 					   0, alloc_hint, &ins, 1);
-			if (ret)
+			if (ret && ret != -ENOSPC)
 				btrfs_abort_transaction(trans, root, ret);
 			btrfs_end_transaction(trans, root);
 		}
@@ -1308,6 +1306,7 @@ out_check:
 			em->block_start = disk_bytenr;
 			em->bdev = root->fs_info->fs_devices->latest_bdev;
 			set_bit(EXTENT_FLAG_PINNED, &em->flags);
+			set_bit(EXTENT_FLAG_PREALLOC, &em->flags);
 			while (1) {
 				write_lock(&em_tree->lock);
 				ret = add_extent_mapping(em_tree, em);
@@ -1364,11 +1363,7 @@ out_check:
 	}
 
 error:
-	if (nolock) {
-		err = btrfs_end_transaction_nolock(trans, root);
-	} else {
-		err = btrfs_end_transaction(trans, root);
-	}
+	err = btrfs_end_transaction(trans, root);
 	if (!ret)
 		ret = err;
 
@@ -1785,7 +1780,6 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	struct btrfs_path *path;
 	struct extent_buffer *leaf;
 	struct btrfs_key ins;
-	u64 hint;
 	int ret;
 
 	path = btrfs_alloc_path();
@@ -1803,8 +1797,8 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	 * the caller is expected to unpin it and allow it to be merged
 	 * with the others.
 	 */
-	ret = btrfs_drop_extents(trans, inode, file_pos, file_pos + num_bytes,
-				 &hint, 0);
+	ret = btrfs_drop_extents(trans, root, inode, file_pos,
+				 file_pos + num_bytes, 0);
 	if (ret)
 		goto out;
 
@@ -1828,10 +1822,8 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	btrfs_set_file_extent_encryption(leaf, fi, encryption);
 	btrfs_set_file_extent_other_encoding(leaf, fi, other_encoding);
 
-	btrfs_unlock_up_safe(path, 1);
-	btrfs_set_lock_blocking(leaf);
-
 	btrfs_mark_buffer_dirty(leaf);
+	btrfs_release_path(path);
 
 	inode_add_bytes(inode, num_bytes);
 
@@ -1929,11 +1921,10 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 						ordered_extent->len,
 						compress_type, 0, 0,
 						BTRFS_FILE_EXTENT_REG);
-		unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
-				   ordered_extent->file_offset,
-				   ordered_extent->len);
 	}
-
+	unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
+			   ordered_extent->file_offset, ordered_extent->len,
+			   trans->transid);
 	if (ret < 0) {
 		btrfs_abort_transaction(trans, root, ret);
 		goto out_unlock;
@@ -1949,6 +1940,8 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 			btrfs_abort_transaction(trans, root, ret);
 			goto out_unlock;
 		}
+	} else {
+		btrfs_set_inode_last_trans(trans, inode);
 	}
 	ret = 0;
 out_unlock:
@@ -1958,12 +1951,8 @@ out_unlock:
 out:
 	if (root != root->fs_info->tree_root)
 		btrfs_delalloc_release_metadata(inode, ordered_extent->len);
-	if (trans) {
-		if (nolock)
-			btrfs_end_transaction_nolock(trans, root);
-		else
-			btrfs_end_transaction(trans, root);
-	}
+	if (trans)
+		btrfs_end_transaction(trans, root);
 
 	if (ret)
 		clear_extent_uptodate(io_tree, ordered_extent->file_offset,
@@ -2119,7 +2108,6 @@ void btrfs_run_delayed_iputs(struct btrfs_root *root)
 	if (empty)
 		return;
 
-	down_read(&root->fs_info->cleanup_work_sem);
 	spin_lock(&fs_info->delayed_iput_lock);
 	list_splice_init(&fs_info->delayed_iputs, &list);
 	spin_unlock(&fs_info->delayed_iput_lock);
@@ -2130,7 +2118,6 @@ void btrfs_run_delayed_iputs(struct btrfs_root *root)
 		iput(delayed->inode);
 		kfree(delayed);
 	}
-	up_read(&root->fs_info->cleanup_work_sem);
 }
 
 enum btrfs_orphan_cleanup_state {
@@ -2198,7 +2185,7 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode)
 	int ret;
 
 	if (!root->orphan_block_rsv) {
-		block_rsv = btrfs_alloc_block_rsv(root);
+		block_rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP);
 		if (!block_rsv)
 			return -ENOMEM;
 	}
@@ -2225,7 +2212,7 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode)
 			insert = 1;
 #endif
 		insert = 1;
-		atomic_dec(&root->orphan_inodes);
+		atomic_inc(&root->orphan_inodes);
 	}
 
 	if (!test_and_set_bit(BTRFS_INODE_ORPHAN_META_RESERVED,
@@ -2590,6 +2577,18 @@ static void btrfs_read_locked_inode(struct inode *inode)
 
 	inode_set_bytes(inode, btrfs_inode_nbytes(leaf, inode_item));
 	BTRFS_I(inode)->generation = btrfs_inode_generation(leaf, inode_item);
+	BTRFS_I(inode)->last_trans = btrfs_inode_transid(leaf, inode_item);
+
+	/*
+	 * If we were modified in the current generation and evicted from memory
+	 * and then re-read we need to do a full sync since we don't have any
+	 * idea about which extents were modified before we were evicted from
+	 * cache.
+	 */
+	if (BTRFS_I(inode)->last_trans == root->fs_info->generation)
+		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+			&BTRFS_I(inode)->runtime_flags);
+
 	inode->i_version = btrfs_inode_sequence(leaf, inode_item);
 	inode->i_generation = BTRFS_I(inode)->generation;
 	inode->i_rdev = 0;
@@ -2894,7 +2893,6 @@ static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir,
 	struct btrfs_trans_handle *trans;
 	struct btrfs_root *root = BTRFS_I(dir)->root;
 	struct btrfs_path *path;
-	struct btrfs_inode_ref *ref;
 	struct btrfs_dir_item *di;
 	struct inode *inode = dentry->d_inode;
 	u64 index;
@@ -3008,17 +3006,17 @@ static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir,
 	}
 	btrfs_release_path(path);
 
-	ref = btrfs_lookup_inode_ref(trans, root, path,
-				dentry->d_name.name, dentry->d_name.len,
-				ino, dir_ino, 0);
-	if (IS_ERR(ref)) {
-		err = PTR_ERR(ref);
+	ret = btrfs_get_inode_ref_index(trans, root, path, dentry->d_name.name,
+					dentry->d_name.len, ino, dir_ino, 0,
+					&index);
+	if (ret) {
+		err = ret;
 		goto out;
 	}
-	BUG_ON(!ref); /* Logic error */
+
 	if (check_path_shared(root, path))
 		goto out;
-	index = btrfs_inode_ref_index(path->nodes[0], ref);
+
 	btrfs_release_path(path);
 
 	/*
@@ -3061,7 +3059,7 @@ out:
 static void __unlink_end_trans(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root)
 {
-	if (trans->block_rsv == &root->fs_info->global_block_rsv) {
+	if (trans->block_rsv->type == BTRFS_BLOCK_RSV_GLOBAL) {
 		btrfs_block_rsv_release(root, trans->block_rsv,
 					trans->bytes_reserved);
 		trans->block_rsv = &root->fs_info->trans_block_rsv;
@@ -3191,9 +3189,10 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 	struct btrfs_trans_handle *trans;
 	unsigned long nr = 0;
 
-	if (inode->i_size > BTRFS_EMPTY_DIR_SIZE ||
-	    btrfs_ino(inode) == BTRFS_FIRST_FREE_OBJECTID)
+	if (inode->i_size > BTRFS_EMPTY_DIR_SIZE)
 		return -ENOTEMPTY;
+	if (btrfs_ino(inode) == BTRFS_FIRST_FREE_OBJECTID)
+		return -EPERM;
 
 	trans = __unlink_start_trans(dir, dentry);
 	if (IS_ERR(trans))
@@ -3267,8 +3266,13 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
 		return -ENOMEM;
 	path->reada = -1;
 
+	/*
+	 * We want to drop from the next block forward in case this new size is
+	 * not block aligned since we will be keeping the last block of the
+	 * extent just the way it is.
+	 */
 	if (root->ref_cows || root == root->fs_info->tree_root)
-		btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0);
+		btrfs_drop_extent_cache(inode, (new_size + mask) & (~mask), (u64)-1, 0);
 
 	/*
 	 * This function is also used to drop the items in the log tree before
@@ -3429,12 +3433,6 @@ delete:
 
 		if (path->slots[0] == 0 ||
 		    path->slots[0] != pending_del_slot) {
-			if (root->ref_cows &&
-			    BTRFS_I(inode)->location.objectid !=
-						BTRFS_FREE_INO_OBJECTID) {
-				err = -EAGAIN;
-				goto out;
-			}
 			if (pending_del_nr) {
 				ret = btrfs_del_items(trans, root, path,
 						pending_del_slot,
@@ -3465,12 +3463,20 @@ error:
 }
 
 /*
- * taken from block_truncate_page, but does cow as it zeros out
- * any bytes left in the last page in the file.
+ * btrfs_truncate_page - read, zero a chunk and write a page
+ * @inode - inode that we're zeroing
+ * @from - the offset to start zeroing
+ * @len - the length to zero, 0 to zero the entire range respective to the
+ *	offset
+ * @front - zero up to the offset instead of from the offset on
+ *
+ * This will find the page for the "from" offset and cow the page and zero the
+ * part we want to zero.  This is used with truncate and hole punching.
  */
-static int btrfs_truncate_page(struct address_space *mapping, loff_t from)
+int btrfs_truncate_page(struct inode *inode, loff_t from, loff_t len,
+			int front)
 {
-	struct inode *inode = mapping->host;
+	struct address_space *mapping = inode->i_mapping;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct btrfs_ordered_extent *ordered;
@@ -3485,7 +3491,8 @@ static int btrfs_truncate_page(struct address_space *mapping, loff_t from)
 	u64 page_start;
 	u64 page_end;
 
-	if ((offset & (blocksize - 1)) == 0)
+	if ((offset & (blocksize - 1)) == 0 &&
+	    (!len || ((len & (blocksize - 1)) == 0)))
 		goto out;
 	ret = btrfs_delalloc_reserve_space(inode, PAGE_CACHE_SIZE);
 	if (ret)
@@ -3532,7 +3539,8 @@ again:
 	}
 
 	clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, page_end,
-			  EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING,
+			  EXTENT_DIRTY | EXTENT_DELALLOC |
+			  EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
 			  0, 0, &cached_state, GFP_NOFS);
 
 	ret = btrfs_set_extent_delalloc(inode, page_start, page_end,
@@ -3545,8 +3553,13 @@ again:
 
 	ret = 0;
 	if (offset != PAGE_CACHE_SIZE) {
+		if (!len)
+			len = PAGE_CACHE_SIZE - offset;
 		kaddr = kmap(page);
-		memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset);
+		if (front)
+			memset(kaddr, 0, offset);
+		else
+			memset(kaddr + offset, 0, len);
 		flush_dcache_page(page);
 		kunmap(page);
 	}
@@ -3577,6 +3590,7 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct extent_map *em = NULL;
 	struct extent_state *cached_state = NULL;
+	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	u64 mask = root->sectorsize - 1;
 	u64 hole_start = (oldsize + mask) & ~mask;
 	u64 block_end = (size + mask) & ~mask;
@@ -3613,7 +3627,7 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 		last_byte = min(extent_map_end(em), block_end);
 		last_byte = (last_byte + mask) & ~mask;
 		if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
-			u64 hint_byte = 0;
+			struct extent_map *hole_em;
 			hole_size = last_byte - cur_offset;
 
 			trans = btrfs_start_transaction(root, 3);
@@ -3622,9 +3636,9 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 				break;
 			}
 
-			err = btrfs_drop_extents(trans, inode, cur_offset,
-						 cur_offset + hole_size,
-						 &hint_byte, 1);
+			err = btrfs_drop_extents(trans, root, inode,
+						 cur_offset,
+						 cur_offset + hole_size, 1);
 			if (err) {
 				btrfs_abort_transaction(trans, root, err);
 				btrfs_end_transaction(trans, root);
@@ -3641,9 +3655,39 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 				break;
 			}
 
-			btrfs_drop_extent_cache(inode, hole_start,
-					last_byte - 1, 0);
+			btrfs_drop_extent_cache(inode, cur_offset,
+						cur_offset + hole_size - 1, 0);
+			hole_em = alloc_extent_map();
+			if (!hole_em) {
+				set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+					&BTRFS_I(inode)->runtime_flags);
+				goto next;
+			}
+			hole_em->start = cur_offset;
+			hole_em->len = hole_size;
+			hole_em->orig_start = cur_offset;
 
+			hole_em->block_start = EXTENT_MAP_HOLE;
+			hole_em->block_len = 0;
+			hole_em->bdev = root->fs_info->fs_devices->latest_bdev;
+			hole_em->compress_type = BTRFS_COMPRESS_NONE;
+			hole_em->generation = trans->transid;
+
+			while (1) {
+				write_lock(&em_tree->lock);
+				err = add_extent_mapping(em_tree, hole_em);
+				if (!err)
+					list_move(&hole_em->list,
+						  &em_tree->modified_extents);
+				write_unlock(&em_tree->lock);
+				if (err != -EEXIST)
+					break;
+				btrfs_drop_extent_cache(inode, cur_offset,
+							cur_offset +
+							hole_size - 1, 0);
+			}
+			free_extent_map(hole_em);
+next:
 			btrfs_update_inode(trans, root, inode);
 			btrfs_end_transaction(trans, root);
 		}
@@ -3768,26 +3812,22 @@ void btrfs_evict_inode(struct inode *inode)
 		goto no_delete;
 	}
 
-	rsv = btrfs_alloc_block_rsv(root);
+	rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP);
 	if (!rsv) {
 		btrfs_orphan_del(NULL, inode);
 		goto no_delete;
 	}
 	rsv->size = min_size;
+	rsv->failfast = 1;
 	global_rsv = &root->fs_info->global_block_rsv;
 
 	btrfs_i_size_write(inode, 0);
 
 	/*
-	 * This is a bit simpler than btrfs_truncate since
-	 *
-	 * 1) We've already reserved our space for our orphan item in the
-	 *    unlink.
-	 * 2) We're going to delete the inode item, so we don't need to update
-	 *    it at all.
-	 *
-	 * So we just need to reserve some slack space in case we add bytes when
-	 * doing the truncate.
+	 * This is a bit simpler than btrfs_truncate since we've already
+	 * reserved our space for our orphan item in the unlink, so we just
+	 * need to reserve some slack space in case we add bytes and update
+	 * inode item when doing the truncate.
 	 */
 	while (1) {
 		ret = btrfs_block_rsv_refill_noflush(root, rsv, min_size);
@@ -3808,7 +3848,7 @@ void btrfs_evict_inode(struct inode *inode)
 			goto no_delete;
 		}
 
-		trans = btrfs_start_transaction(root, 0);
+		trans = btrfs_start_transaction_noflush(root, 1);
 		if (IS_ERR(trans)) {
 			btrfs_orphan_del(NULL, inode);
 			btrfs_free_block_rsv(root, rsv);
@@ -3818,9 +3858,13 @@ void btrfs_evict_inode(struct inode *inode)
 		trans->block_rsv = rsv;
 
 		ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0);
-		if (ret != -EAGAIN)
+		if (ret != -ENOSPC)
 			break;
 
+		trans->block_rsv = &root->fs_info->trans_block_rsv;
+		ret = btrfs_update_inode(trans, root, inode);
+		BUG_ON(ret);
+
 		nr = trans->blocks_used;
 		btrfs_end_transaction(trans, root);
 		trans = NULL;
@@ -4470,10 +4514,7 @@ int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc)
 			trans = btrfs_join_transaction(root);
 		if (IS_ERR(trans))
 			return PTR_ERR(trans);
-		if (nolock)
-			ret = btrfs_end_transaction_nolock(trans, root);
-		else
-			ret = btrfs_commit_transaction(trans, root);
+		ret = btrfs_commit_transaction(trans, root);
 	}
 	return ret;
 }
@@ -4671,6 +4712,14 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
 	BTRFS_I(inode)->generation = trans->transid;
 	inode->i_generation = BTRFS_I(inode)->generation;
 
+	/*
+	 * We could have gotten an inode number from somebody who was fsynced
+	 * and then removed in this same transaction, so let's just set full
+	 * sync since it will be a full sync anyway and this will blow away the
+	 * old info in the log.
+	 */
+	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags);
+
 	if (S_ISDIR(mode))
 		owner = 0;
 	else
@@ -4680,6 +4729,12 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
 	btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY);
 	key[0].offset = 0;
 
+	/*
+	 * Start new inodes with an inode_ref. This is slightly more
+	 * efficient for small numbers of hard links since they will
+	 * be packed into one item. Extended refs will kick in if we
+	 * add more hard links than can fit in the ref item.
+	 */
 	key[1].objectid = objectid;
 	btrfs_set_key_type(&key[1], BTRFS_INODE_REF_KEY);
 	key[1].offset = ref_objectid;
@@ -4986,7 +5041,7 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
 	if (root->objectid != BTRFS_I(inode)->root->objectid)
 		return -EXDEV;
 
-	if (inode->i_nlink == ~0U)
+	if (inode->i_nlink >= BTRFS_LINK_MAX)
 		return -EMLINK;
 
 	err = btrfs_set_inode_index(dir, &index);
@@ -5450,7 +5505,8 @@ insert:
 	write_unlock(&em_tree->lock);
 out:
 
-	trace_btrfs_get_extent(root, em);
+	if (em)
+		trace_btrfs_get_extent(root, em);
 
 	if (path)
 		btrfs_free_path(path);
@@ -5836,6 +5892,48 @@ static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend,
 	return ret;
 }
 
+static struct extent_map *create_pinned_em(struct inode *inode, u64 start,
+					   u64 len, u64 orig_start,
+					   u64 block_start, u64 block_len,
+					   int type)
+{
+	struct extent_map_tree *em_tree;
+	struct extent_map *em;
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	int ret;
+
+	em_tree = &BTRFS_I(inode)->extent_tree;
+	em = alloc_extent_map();
+	if (!em)
+		return ERR_PTR(-ENOMEM);
+
+	em->start = start;
+	em->orig_start = orig_start;
+	em->len = len;
+	em->block_len = block_len;
+	em->block_start = block_start;
+	em->bdev = root->fs_info->fs_devices->latest_bdev;
+	set_bit(EXTENT_FLAG_PINNED, &em->flags);
+	if (type == BTRFS_ORDERED_PREALLOC)
+		set_bit(EXTENT_FLAG_PREALLOC, &em->flags);
+
+	do {
+		btrfs_drop_extent_cache(inode, em->start,
+				em->start + em->len - 1, 0);
+		write_lock(&em_tree->lock);
+		ret = add_extent_mapping(em_tree, em);
+		write_unlock(&em_tree->lock);
+	} while (ret == -EEXIST);
+
+	if (ret) {
+		free_extent_map(em);
+		return ERR_PTR(ret);
+	}
+
+	return em;
+}
+
+
 static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 				   struct buffer_head *bh_result, int create)
 {
@@ -5950,6 +6048,19 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 			goto must_cow;
 
 		if (can_nocow_odirect(trans, inode, start, len) == 1) {
+			u64 orig_start = em->start;
+
+			if (type == BTRFS_ORDERED_PREALLOC) {
+				free_extent_map(em);
+				em = create_pinned_em(inode, start, len,
+						       orig_start,
+						       block_start, len, type);
+				if (IS_ERR(em)) {
+					btrfs_end_transaction(trans, root);
+					goto unlock_err;
+				}
+			}
+
 			ret = btrfs_add_ordered_extent_dio(inode, start,
 					   block_start, len, len, type);
 			btrfs_end_transaction(trans, root);
@@ -5999,7 +6110,8 @@ unlock:
 	if (lockstart < lockend) {
 		if (create && len < lockend - lockstart) {
 			clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
-					 lockstart + len - 1, unlock_bits, 1, 0,
+					 lockstart + len - 1,
+					 unlock_bits | EXTENT_DEFRAG, 1, 0,
 					 &cached_state, GFP_NOFS);
 			/*
 			 * Beside unlock, we also need to cleanup reserved space
@@ -6007,8 +6119,8 @@ unlock:
 			 */
 			clear_extent_bit(&BTRFS_I(inode)->io_tree,
 					 lockstart + len, lockend,
-					 unlock_bits | EXTENT_DO_ACCOUNTING,
-					 1, 0, NULL, GFP_NOFS);
+					 unlock_bits | EXTENT_DO_ACCOUNTING |
+					 EXTENT_DEFRAG, 1, 0, NULL, GFP_NOFS);
 		} else {
 			clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
 					 lockend, unlock_bits, 1, 0,
@@ -6573,8 +6685,8 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset)
 		 */
 		clear_extent_bit(tree, page_start, page_end,
 				 EXTENT_DIRTY | EXTENT_DELALLOC |
-				 EXTENT_LOCKED | EXTENT_DO_ACCOUNTING, 1, 0,
-				 &cached_state, GFP_NOFS);
+				 EXTENT_LOCKED | EXTENT_DO_ACCOUNTING |
+				 EXTENT_DEFRAG, 1, 0, &cached_state, GFP_NOFS);
 		/*
 		 * whoever cleared the private bit is responsible
 		 * for the finish_ordered_io
@@ -6590,7 +6702,8 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset)
 	}
 	clear_extent_bit(tree, page_start, page_end,
 		 EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
-		 EXTENT_DO_ACCOUNTING, 1, 1, &cached_state, GFP_NOFS);
+		 EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 1, 1,
+		 &cached_state, GFP_NOFS);
 	__btrfs_releasepage(page, GFP_NOFS);
 
 	ClearPageChecked(page);
@@ -6687,7 +6800,8 @@ again:
 	 * prepare_pages in the normal write path.
 	 */
 	clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, page_end,
-			  EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING,
+			  EXTENT_DIRTY | EXTENT_DELALLOC |
+			  EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
 			  0, 0, &cached_state, GFP_NOFS);
 
 	ret = btrfs_set_extent_delalloc(inode, page_start, page_end,
@@ -6718,6 +6832,7 @@ again:
 
 	BTRFS_I(inode)->last_trans = root->fs_info->generation;
 	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;
 
 	unlock_extent_cached(io_tree, page_start, page_end, &cached_state, GFP_NOFS);
 
@@ -6745,7 +6860,7 @@ static int btrfs_truncate(struct inode *inode)
 	u64 mask = root->sectorsize - 1;
 	u64 min_size = btrfs_calc_trunc_metadata_size(root, 1);
 
-	ret = btrfs_truncate_page(inode->i_mapping, inode->i_size);
+	ret = btrfs_truncate_page(inode, inode->i_size, 0, 0);
 	if (ret)
 		return ret;
 
@@ -6788,10 +6903,11 @@ static int btrfs_truncate(struct inode *inode)
 	 * 3) fs_info->trans_block_rsv - this will have 1 items worth left for
 	 * updating the inode.
 	 */
-	rsv = btrfs_alloc_block_rsv(root);
+	rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP);
 	if (!rsv)
 		return -ENOMEM;
 	rsv->size = min_size;
+	rsv->failfast = 1;
 
 	/*
 	 * 1 for the truncate slack space
@@ -6837,36 +6953,21 @@ static int btrfs_truncate(struct inode *inode)
 					   &BTRFS_I(inode)->runtime_flags))
 		btrfs_add_ordered_operation(trans, root, inode);
 
-	while (1) {
-		ret = btrfs_block_rsv_refill(root, rsv, min_size);
-		if (ret) {
-			/*
-			 * This can only happen with the original transaction we
-			 * started above, every other time we shouldn't have a
-			 * transaction started yet.
-			 */
-			if (ret == -EAGAIN)
-				goto end_trans;
-			err = ret;
-			break;
-		}
-
-		if (!trans) {
-			/* Just need the 1 for updating the inode */
-			trans = btrfs_start_transaction(root, 1);
-			if (IS_ERR(trans)) {
-				ret = err = PTR_ERR(trans);
-				trans = NULL;
-				break;
-			}
-		}
-
-		trans->block_rsv = rsv;
+	/*
+	 * So if we truncate and then write and fsync we normally would just
+	 * write the extents that changed, which is a problem if we need to
+	 * first truncate that entire inode.  So set this flag so we write out
+	 * all of the extents in the inode to the sync log so we're completely
+	 * safe.
+	 */
+	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags);
+	trans->block_rsv = rsv;
 
+	while (1) {
 		ret = btrfs_truncate_inode_items(trans, root, inode,
 						 inode->i_size,
 						 BTRFS_EXTENT_DATA_KEY);
-		if (ret != -EAGAIN) {
+		if (ret != -ENOSPC) {
 			err = ret;
 			break;
 		}
@@ -6877,11 +6978,22 @@ static int btrfs_truncate(struct inode *inode)
 			err = ret;
 			break;
 		}
-end_trans:
+
 		nr = trans->blocks_used;
 		btrfs_end_transaction(trans, root);
-		trans = NULL;
 		btrfs_btree_balance_dirty(root, nr);
+
+		trans = btrfs_start_transaction(root, 2);
+		if (IS_ERR(trans)) {
+			ret = err = PTR_ERR(trans);
+			trans = NULL;
+			break;
+		}
+
+		ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv,
+					      rsv, min_size);
+		BUG_ON(ret);	/* shouldn't happen */
+		trans->block_rsv = rsv;
 	}
 
 	if (ret == 0 && inode->i_nlink > 0) {
@@ -6965,6 +7077,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	ei->csum_bytes = 0;
 	ei->index_cnt = (u64)-1;
 	ei->last_unlink_trans = 0;
+	ei->last_log_commit = 0;
 
 	spin_lock_init(&ei->lock);
 	ei->outstanding_extents = 0;
@@ -7095,31 +7208,31 @@ void btrfs_destroy_cachep(void)
 
 int btrfs_init_cachep(void)
 {
-	btrfs_inode_cachep = kmem_cache_create("btrfs_inode_cache",
+	btrfs_inode_cachep = kmem_cache_create("btrfs_inode",
 			sizeof(struct btrfs_inode), 0,
 			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, init_once);
 	if (!btrfs_inode_cachep)
 		goto fail;
 
-	btrfs_trans_handle_cachep = kmem_cache_create("btrfs_trans_handle_cache",
+	btrfs_trans_handle_cachep = kmem_cache_create("btrfs_trans_handle",
 			sizeof(struct btrfs_trans_handle), 0,
 			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!btrfs_trans_handle_cachep)
 		goto fail;
 
-	btrfs_transaction_cachep = kmem_cache_create("btrfs_transaction_cache",
+	btrfs_transaction_cachep = kmem_cache_create("btrfs_transaction",
 			sizeof(struct btrfs_transaction), 0,
 			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!btrfs_transaction_cachep)
 		goto fail;
 
-	btrfs_path_cachep = kmem_cache_create("btrfs_path_cache",
+	btrfs_path_cachep = kmem_cache_create("btrfs_path",
 			sizeof(struct btrfs_path), 0,
 			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!btrfs_path_cachep)
 		goto fail;
 
-	btrfs_free_space_cachep = kmem_cache_create("btrfs_free_space_cache",
+	btrfs_free_space_cachep = kmem_cache_create("btrfs_free_space",
 			sizeof(struct btrfs_free_space), 0,
 			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!btrfs_free_space_cachep)
@@ -7513,6 +7626,8 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 				       loff_t actual_len, u64 *alloc_hint,
 				       struct btrfs_trans_handle *trans)
 {
+	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	struct extent_map *em;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_key ins;
 	u64 cur_offset = start;
@@ -7553,6 +7668,37 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 		btrfs_drop_extent_cache(inode, cur_offset,
 					cur_offset + ins.offset -1, 0);
 
+		em = alloc_extent_map();
+		if (!em) {
+			set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+				&BTRFS_I(inode)->runtime_flags);
+			goto next;
+		}
+
+		em->start = cur_offset;
+		em->orig_start = cur_offset;
+		em->len = ins.offset;
+		em->block_start = ins.objectid;
+		em->block_len = ins.offset;
+		em->bdev = root->fs_info->fs_devices->latest_bdev;
+		set_bit(EXTENT_FLAG_PREALLOC, &em->flags);
+		em->generation = trans->transid;
+
+		while (1) {
+			write_lock(&em_tree->lock);
+			ret = add_extent_mapping(em_tree, em);
+			if (!ret)
+				list_move(&em->list,
+					  &em_tree->modified_extents);
+			write_unlock(&em_tree->lock);
+			if (ret != -EEXIST)
+				break;
+			btrfs_drop_extent_cache(inode, cur_offset,
+						cur_offset + ins.offset - 1,
+						0);
+		}
+		free_extent_map(em);
+next:
 		num_bytes -= ins.offset;
 		cur_offset += ins.offset;
 		*alloc_hint = ins.objectid + ins.offset;

fs/btrfs/ioctl.c

@@ -181,6 +181,7 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
 	int ret;
 	u64 ip_oldflags;
 	unsigned int i_oldflags;
+	umode_t mode;
 
 	if (btrfs_root_readonly(root))
 		return -EROFS;
@@ -203,6 +204,7 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
 
 	ip_oldflags = ip->flags;
 	i_oldflags = inode->i_flags;
+	mode = inode->i_mode;
 
 	flags = btrfs_mask_flags(inode->i_mode, flags);
 	oldflags = btrfs_flags_to_ioctl(ip->flags);
@@ -237,10 +239,31 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
 		ip->flags |= BTRFS_INODE_DIRSYNC;
 	else
 		ip->flags &= ~BTRFS_INODE_DIRSYNC;
-	if (flags & FS_NOCOW_FL)
-		ip->flags |= BTRFS_INODE_NODATACOW;
-	else
-		ip->flags &= ~BTRFS_INODE_NODATACOW;
+	if (flags & FS_NOCOW_FL) {
+		if (S_ISREG(mode)) {
+			/*
+			 * It's safe to turn csums off here, no extents exist.
+			 * Otherwise we want the flag to reflect the real COW
+			 * status of the file and will not set it.
+			 */
+			if (inode->i_size == 0)
+				ip->flags |= BTRFS_INODE_NODATACOW
+					   | BTRFS_INODE_NODATASUM;
+		} else {
+			ip->flags |= BTRFS_INODE_NODATACOW;
+		}
+	} else {
+		/*
+		 * Revert back under same assuptions as above
+		 */
+		if (S_ISREG(mode)) {
+			if (inode->i_size == 0)
+				ip->flags &= ~(BTRFS_INODE_NODATACOW
+				             | BTRFS_INODE_NODATASUM);
+		} else {
+			ip->flags &= ~BTRFS_INODE_NODATACOW;
+		}
+	}
 
 	/*
 	 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
@@ -516,7 +539,8 @@ static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
 	if (!pending_snapshot)
 		return -ENOMEM;
 
-	btrfs_init_block_rsv(&pending_snapshot->block_rsv);
+	btrfs_init_block_rsv(&pending_snapshot->block_rsv,
+			     BTRFS_BLOCK_RSV_TEMP);
 	pending_snapshot->dentry = dentry;
 	pending_snapshot->root = root;
 	pending_snapshot->readonly = readonly;
@@ -525,7 +549,7 @@ static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
 		*inherit = NULL;	/* take responsibility to free it */
 	}
 
-	trans = btrfs_start_transaction(root->fs_info->extent_root, 5);
+	trans = btrfs_start_transaction(root->fs_info->extent_root, 6);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
 		goto fail;
@@ -1022,8 +1046,8 @@ again:
 			 page_start, page_end - 1, 0, &cached_state);
 	clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
 			  page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
-			  EXTENT_DO_ACCOUNTING, 0, 0, &cached_state,
-			  GFP_NOFS);
+			  EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0,
+			  &cached_state, GFP_NOFS);
 
 	if (i_done != page_cnt) {
 		spin_lock(&BTRFS_I(inode)->lock);
@@ -1034,8 +1058,8 @@ again:
 	}
 
 
-	btrfs_set_extent_delalloc(inode, page_start, page_end - 1,
-				  &cached_state);
+	set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1,
+			  &cached_state, GFP_NOFS);
 
 	unlock_extent_cached(&BTRFS_I(inode)->io_tree,
 			     page_start, page_end - 1, &cached_state,
@@ -2351,7 +2375,6 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 	int ret;
 	u64 len = olen;
 	u64 bs = root->fs_info->sb->s_blocksize;
-	u64 hint_byte;
 
 	/*
 	 * TODO:
@@ -2456,13 +2479,13 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 	   another, and lock file content */
 	while (1) {
 		struct btrfs_ordered_extent *ordered;
-		lock_extent(&BTRFS_I(src)->io_tree, off, off+len);
-		ordered = btrfs_lookup_first_ordered_extent(src, off+len);
+		lock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
+		ordered = btrfs_lookup_first_ordered_extent(src, off + len - 1);
 		if (!ordered &&
-		    !test_range_bit(&BTRFS_I(src)->io_tree, off, off+len,
-				   EXTENT_DELALLOC, 0, NULL))
+		    !test_range_bit(&BTRFS_I(src)->io_tree, off, off + len - 1,
+				    EXTENT_DELALLOC, 0, NULL))
 			break;
-		unlock_extent(&BTRFS_I(src)->io_tree, off, off+len);
+		unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
 		if (ordered)
 			btrfs_put_ordered_extent(ordered);
 		btrfs_wait_ordered_range(src, off, len);
@@ -2536,7 +2559,7 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 			btrfs_release_path(path);
 
 			if (key.offset + datal <= off ||
-			    key.offset >= off+len)
+			    key.offset >= off + len - 1)
 				goto next;
 
 			memcpy(&new_key, &key, sizeof(new_key));
@@ -2574,10 +2597,10 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 					datal -= off - key.offset;
 				}
 
-				ret = btrfs_drop_extents(trans, inode,
+				ret = btrfs_drop_extents(trans, root, inode,
 							 new_key.offset,
 							 new_key.offset + datal,
-							 &hint_byte, 1);
+							 1);
 				if (ret) {
 					btrfs_abort_transaction(trans, root,
 								ret);
@@ -2637,8 +2660,8 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 					new_key.offset += skip;
 				}
 
-				if (key.offset + datal > off+len)
-					trim = key.offset + datal - (off+len);
+				if (key.offset + datal > off + len)
+					trim = key.offset + datal - (off + len);
 
 				if (comp && (skip || trim)) {
 					ret = -EINVAL;
@@ -2648,10 +2671,10 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 				size -= skip + trim;
 				datal -= skip + trim;
 
-				ret = btrfs_drop_extents(trans, inode,
+				ret = btrfs_drop_extents(trans, root, inode,
 							 new_key.offset,
 							 new_key.offset + datal,
-							 &hint_byte, 1);
+							 1);
 				if (ret) {
 					btrfs_abort_transaction(trans, root,
 								ret);
@@ -2715,7 +2738,7 @@ next:
 	ret = 0;
 out:
 	btrfs_release_path(path);
-	unlock_extent(&BTRFS_I(src)->io_tree, off, off+len);
+	unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
 out_unlock:
 	mutex_unlock(&src->i_mutex);
 	mutex_unlock(&inode->i_mutex);
@@ -2850,8 +2873,8 @@ static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
 	return 0;
 }
 
-static void get_block_group_info(struct list_head *groups_list,
-				 struct btrfs_ioctl_space_info *space)
+void btrfs_get_block_group_info(struct list_head *groups_list,
+				struct btrfs_ioctl_space_info *space)
 {
 	struct btrfs_block_group_cache *block_group;
 
@@ -2959,8 +2982,8 @@ long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg)
 		down_read(&info->groups_sem);
 		for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
 			if (!list_empty(&info->block_groups[c])) {
-				get_block_group_info(&info->block_groups[c],
-						     &space);
+				btrfs_get_block_group_info(
+					&info->block_groups[c], &space);
 				memcpy(dest, &space, sizeof(space));
 				dest++;
 				space_args.total_spaces++;
@@ -3208,11 +3231,9 @@ static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
 {
 	int ret = 0;
 	int size;
-	u64 extent_item_pos;
 	struct btrfs_ioctl_logical_ino_args *loi;
 	struct btrfs_data_container *inodes = NULL;
 	struct btrfs_path *path = NULL;
-	struct btrfs_key key;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
@@ -3230,7 +3251,7 @@ static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
 		goto out;
 	}
 
-	size = min_t(u32, loi->size, 4096);
+	size = min_t(u32, loi->size, 64 * 1024);
 	inodes = init_data_container(size);
 	if (IS_ERR(inodes)) {
 		ret = PTR_ERR(inodes);
@@ -3238,22 +3259,13 @@ static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
 		goto out;
 	}
 
-	ret = extent_from_logical(root->fs_info, loi->logical, path, &key);
-	btrfs_release_path(path);
-
-	if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+	ret = iterate_inodes_from_logical(loi->logical, root->fs_info, path,
+					  build_ino_list, inodes);
+	if (ret == -EINVAL)
 		ret = -ENOENT;
 	if (ret < 0)
 		goto out;
 
-	extent_item_pos = loi->logical - key.objectid;
-	ret = iterate_extent_inodes(root->fs_info, key.objectid,
-					extent_item_pos, 0, build_ino_list,
-					inodes);
-
-	if (ret < 0)
-		goto out;
-
 	ret = copy_to_user((void *)(unsigned long)loi->inodes,
 			   (void *)(unsigned long)inodes, size);
 	if (ret)
@@ -3261,7 +3273,7 @@ static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
 
 out:
 	btrfs_free_path(path);
-	kfree(inodes);
+	vfree(inodes);
 	kfree(loi);
 
 	return ret;

fs/btrfs/ordered-data.c

@@ -25,6 +25,8 @@
 #include "btrfs_inode.h"
 #include "extent_io.h"
 
+static struct kmem_cache *btrfs_ordered_extent_cache;
+
 static u64 entry_end(struct btrfs_ordered_extent *entry)
 {
 	if (entry->file_offset + entry->len < entry->file_offset)
@@ -187,7 +189,7 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 	struct btrfs_ordered_extent *entry;
 
 	tree = &BTRFS_I(inode)->ordered_tree;
-	entry = kzalloc(sizeof(*entry), GFP_NOFS);
+	entry = kmem_cache_zalloc(btrfs_ordered_extent_cache, GFP_NOFS);
 	if (!entry)
 		return -ENOMEM;
 
@@ -421,7 +423,7 @@ void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 			list_del(&sum->list);
 			kfree(sum);
 		}
-		kfree(entry);
+		kmem_cache_free(btrfs_ordered_extent_cache, entry);
 	}
 }
 
@@ -466,8 +468,7 @@ void btrfs_remove_ordered_extent(struct inode *inode,
  * wait for all the ordered extents in a root.  This is done when balancing
  * space between drives.
  */
-void btrfs_wait_ordered_extents(struct btrfs_root *root,
-				int nocow_only, int delay_iput)
+void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput)
 {
 	struct list_head splice;
 	struct list_head *cur;
@@ -482,15 +483,6 @@ void btrfs_wait_ordered_extents(struct btrfs_root *root,
 		cur = splice.next;
 		ordered = list_entry(cur, struct btrfs_ordered_extent,
 				     root_extent_list);
-		if (nocow_only &&
-		    !test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags) &&
-		    !test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags)) {
-			list_move(&ordered->root_extent_list,
-				  &root->fs_info->ordered_extents);
-			cond_resched_lock(&root->fs_info->ordered_extent_lock);
-			continue;
-		}
-
 		list_del_init(&ordered->root_extent_list);
 		atomic_inc(&ordered->refs);
 
@@ -775,7 +767,6 @@ int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
 	struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
 	u64 disk_i_size;
 	u64 new_i_size;
-	u64 i_size_test;
 	u64 i_size = i_size_read(inode);
 	struct rb_node *node;
 	struct rb_node *prev = NULL;
@@ -835,55 +826,30 @@ int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
 			break;
 		if (test->file_offset >= i_size)
 			break;
-		if (test->file_offset >= disk_i_size)
+		if (test->file_offset >= disk_i_size) {
+			/*
+			 * we don't update disk_i_size now, so record this
+			 * undealt i_size. Or we will not know the real
+			 * i_size.
+			 */
+			if (test->outstanding_isize < offset)
+				test->outstanding_isize = offset;
+			if (ordered &&
+			    ordered->outstanding_isize >
+			    test->outstanding_isize)
+				test->outstanding_isize =
+						ordered->outstanding_isize;
 			goto out;
-	}
-	new_i_size = min_t(u64, offset, i_size);
-
-	/*
-	 * at this point, we know we can safely update i_size to at least
-	 * the offset from this ordered extent.  But, we need to
-	 * walk forward and see if ios from higher up in the file have
-	 * finished.
-	 */
-	if (ordered) {
-		node = rb_next(&ordered->rb_node);
-	} else {
-		if (prev)
-			node = rb_next(prev);
-		else
-			node = rb_first(&tree->tree);
-	}
-
-	/*
-	 * We are looking for an area between our current extent and the next
-	 * ordered extent to update the i_size to.  There are 3 cases here
-	 *
-	 * 1) We don't actually have anything and we can update to i_size.
-	 * 2) We have stuff but they already did their i_size update so again we
-	 * can just update to i_size.
-	 * 3) We have an outstanding ordered extent so the most we can update
-	 * our disk_i_size to is the start of the next offset.
-	 */
-	i_size_test = i_size;
-	for (; node; node = rb_next(node)) {
-		test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
-
-		if (test_bit(BTRFS_ORDERED_UPDATED_ISIZE, &test->flags))
-			continue;
-		if (test->file_offset > offset) {
-			i_size_test = test->file_offset;
-			break;
 		}
 	}
+	new_i_size = min_t(u64, offset, i_size);
 
 	/*
-	 * i_size_test is the end of a region after this ordered
-	 * extent where there are no ordered extents, we can safely set
-	 * disk_i_size to this.
+	 * Some ordered extents may completed before the current one, and
+	 * we hold the real i_size in ->outstanding_isize.
 	 */
-	if (i_size_test > offset)
-		new_i_size = min_t(u64, i_size_test, i_size);
+	if (ordered && ordered->outstanding_isize > new_i_size)
+		new_i_size = min_t(u64, ordered->outstanding_isize, i_size);
 	BTRFS_I(inode)->disk_i_size = new_i_size;
 	ret = 0;
 out:
@@ -984,3 +950,20 @@ void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
 	}
 	spin_unlock(&root->fs_info->ordered_extent_lock);
 }
+
+int __init ordered_data_init(void)
+{
+	btrfs_ordered_extent_cache = kmem_cache_create("btrfs_ordered_extent",
+				     sizeof(struct btrfs_ordered_extent), 0,
+				     SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
+				     NULL);
+	if (!btrfs_ordered_extent_cache)
+		return -ENOMEM;
+	return 0;
+}
+
+void ordered_data_exit(void)
+{
+	if (btrfs_ordered_extent_cache)
+		kmem_cache_destroy(btrfs_ordered_extent_cache);
+}

fs/btrfs/ordered-data.h

@@ -96,6 +96,13 @@ struct btrfs_ordered_extent {
 	/* number of bytes that still need writing */
 	u64 bytes_left;
 
+	/*
+	 * the end of the ordered extent which is behind it but
+	 * didn't update disk_i_size. Please see the comment of
+	 * btrfs_ordered_update_i_size();
+	 */
+	u64 outstanding_isize;
+
 	/* flags (described above) */
 	unsigned long flags;
 
@@ -183,6 +190,7 @@ void btrfs_run_ordered_operations(struct btrfs_root *root, int wait);
 void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
 				 struct btrfs_root *root,
 				 struct inode *inode);
-void btrfs_wait_ordered_extents(struct btrfs_root *root,
-				int nocow_only, int delay_iput);
+void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput);
+int __init ordered_data_init(void);
+void ordered_data_exit(void);
 #endif

fs/btrfs/qgroup.c

@@ -1145,12 +1145,12 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
 
 		ulist_reinit(tmp);
 						/* XXX id not needed */
-		ulist_add(tmp, qg->qgroupid, (unsigned long)qg, GFP_ATOMIC);
+		ulist_add(tmp, qg->qgroupid, (u64)(uintptr_t)qg, GFP_ATOMIC);
 		ULIST_ITER_INIT(&tmp_uiter);
 		while ((tmp_unode = ulist_next(tmp, &tmp_uiter))) {
 			struct btrfs_qgroup_list *glist;
 
-			qg = (struct btrfs_qgroup *)tmp_unode->aux;
+			qg = (struct btrfs_qgroup *)(uintptr_t)tmp_unode->aux;
 			if (qg->refcnt < seq)
 				qg->refcnt = seq + 1;
 			else
@@ -1158,7 +1158,7 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
 
 			list_for_each_entry(glist, &qg->groups, next_group) {
 				ulist_add(tmp, glist->group->qgroupid,
-					  (unsigned long)glist->group,
+					  (u64)(uintptr_t)glist->group,
 					  GFP_ATOMIC);
 			}
 		}
@@ -1168,13 +1168,13 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
 	 * step 2: walk from the new root
 	 */
 	ulist_reinit(tmp);
-	ulist_add(tmp, qgroup->qgroupid, (unsigned long)qgroup, GFP_ATOMIC);
+	ulist_add(tmp, qgroup->qgroupid, (uintptr_t)qgroup, GFP_ATOMIC);
 	ULIST_ITER_INIT(&uiter);
 	while ((unode = ulist_next(tmp, &uiter))) {
 		struct btrfs_qgroup *qg;
 		struct btrfs_qgroup_list *glist;
 
-		qg = (struct btrfs_qgroup *)unode->aux;
+		qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux;
 		if (qg->refcnt < seq) {
 			/* not visited by step 1 */
 			qg->rfer += sgn * node->num_bytes;
@@ -1190,7 +1190,7 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
 
 		list_for_each_entry(glist, &qg->groups, next_group) {
 			ulist_add(tmp, glist->group->qgroupid,
-				  (unsigned long)glist->group, GFP_ATOMIC);
+				  (uintptr_t)glist->group, GFP_ATOMIC);
 		}
 	}
 
@@ -1208,12 +1208,12 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
 			continue;
 
 		ulist_reinit(tmp);
-		ulist_add(tmp, qg->qgroupid, (unsigned long)qg, GFP_ATOMIC);
+		ulist_add(tmp, qg->qgroupid, (uintptr_t)qg, GFP_ATOMIC);
 		ULIST_ITER_INIT(&tmp_uiter);
 		while ((tmp_unode = ulist_next(tmp, &tmp_uiter))) {
 			struct btrfs_qgroup_list *glist;
 
-			qg = (struct btrfs_qgroup *)tmp_unode->aux;
+			qg = (struct btrfs_qgroup *)(uintptr_t)tmp_unode->aux;
 			if (qg->tag == seq)
 				continue;
 
@@ -1225,7 +1225,7 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
 
 			list_for_each_entry(glist, &qg->groups, next_group) {
 				ulist_add(tmp, glist->group->qgroupid,
-					  (unsigned long)glist->group,
+					  (uintptr_t)glist->group,
 					  GFP_ATOMIC);
 			}
 		}
@@ -1469,13 +1469,17 @@ int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes)
 	 * be exceeded
 	 */
 	ulist = ulist_alloc(GFP_ATOMIC);
-	ulist_add(ulist, qgroup->qgroupid, (unsigned long)qgroup, GFP_ATOMIC);
+	if (!ulist) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	ulist_add(ulist, qgroup->qgroupid, (uintptr_t)qgroup, GFP_ATOMIC);
 	ULIST_ITER_INIT(&uiter);
 	while ((unode = ulist_next(ulist, &uiter))) {
 		struct btrfs_qgroup *qg;
 		struct btrfs_qgroup_list *glist;
 
-		qg = (struct btrfs_qgroup *)unode->aux;
+		qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux;
 
 		if ((qg->lim_flags & BTRFS_QGROUP_LIMIT_MAX_RFER) &&
 		    qg->reserved + qg->rfer + num_bytes >
@@ -1489,7 +1493,7 @@ int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes)
 
 		list_for_each_entry(glist, &qg->groups, next_group) {
 			ulist_add(ulist, glist->group->qgroupid,
-				  (unsigned long)glist->group, GFP_ATOMIC);
+				  (uintptr_t)glist->group, GFP_ATOMIC);
 		}
 	}
 	if (ret)
@@ -1502,7 +1506,7 @@ int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes)
 	while ((unode = ulist_next(ulist, &uiter))) {
 		struct btrfs_qgroup *qg;
 
-		qg = (struct btrfs_qgroup *)unode->aux;
+		qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux;
 
 		qg->reserved += num_bytes;
 	}
@@ -1541,19 +1545,23 @@ void btrfs_qgroup_free(struct btrfs_root *root, u64 num_bytes)
 		goto out;
 
 	ulist = ulist_alloc(GFP_ATOMIC);
-	ulist_add(ulist, qgroup->qgroupid, (unsigned long)qgroup, GFP_ATOMIC);
+	if (!ulist) {
+		btrfs_std_error(fs_info, -ENOMEM);
+		goto out;
+	}
+	ulist_add(ulist, qgroup->qgroupid, (uintptr_t)qgroup, GFP_ATOMIC);
 	ULIST_ITER_INIT(&uiter);
 	while ((unode = ulist_next(ulist, &uiter))) {
 		struct btrfs_qgroup *qg;
 		struct btrfs_qgroup_list *glist;
 
-		qg = (struct btrfs_qgroup *)unode->aux;
+		qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux;
 
 		qg->reserved -= num_bytes;
 
 		list_for_each_entry(glist, &qg->groups, next_group) {
 			ulist_add(ulist, glist->group->qgroupid,
-				  (unsigned long)glist->group, GFP_ATOMIC);
+				  (uintptr_t)glist->group, GFP_ATOMIC);
 		}
 	}
 

fs/btrfs/relocation.c

@@ -3270,8 +3270,8 @@ static int delete_block_group_cache(struct btrfs_fs_info *fs_info,
 	key.offset = 0;
 
 	inode = btrfs_iget(fs_info->sb, &key, root, NULL);
-	if (IS_ERR_OR_NULL(inode) || is_bad_inode(inode)) {
-		if (inode && !IS_ERR(inode))
+	if (IS_ERR(inode) || is_bad_inode(inode)) {
+		if (!IS_ERR(inode))
 			iput(inode);
 		return -ENOENT;
 	}
@@ -3621,7 +3621,7 @@ next:
 
 		ret = find_first_extent_bit(&rc->processed_blocks,
 					    key.objectid, &start, &end,
-					    EXTENT_DIRTY);
+					    EXTENT_DIRTY, NULL);
 
 		if (ret == 0 && start <= key.objectid) {
 			btrfs_release_path(path);
@@ -3674,7 +3674,8 @@ int prepare_to_relocate(struct reloc_control *rc)
 	struct btrfs_trans_handle *trans;
 	int ret;
 
-	rc->block_rsv = btrfs_alloc_block_rsv(rc->extent_root);
+	rc->block_rsv = btrfs_alloc_block_rsv(rc->extent_root,
+					      BTRFS_BLOCK_RSV_TEMP);
 	if (!rc->block_rsv)
 		return -ENOMEM;
 
@@ -4057,7 +4058,7 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start)
 	       (unsigned long long)rc->block_group->flags);
 
 	btrfs_start_delalloc_inodes(fs_info->tree_root, 0);
-	btrfs_wait_ordered_extents(fs_info->tree_root, 0, 0);
+	btrfs_wait_ordered_extents(fs_info->tree_root, 0);
 
 	while (1) {
 		mutex_lock(&fs_info->cleaner_mutex);

fs/btrfs/root-tree.c

@@ -141,8 +141,10 @@ int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
 		return -ENOMEM;
 
 	ret = btrfs_search_slot(trans, root, key, path, 0, 1);
-	if (ret < 0)
-		goto out_abort;
+	if (ret < 0) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out;
+	}
 
 	if (ret != 0) {
 		btrfs_print_leaf(root, path->nodes[0]);
@@ -166,16 +168,23 @@ int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
 		btrfs_release_path(path);
 		ret = btrfs_search_slot(trans, root, key, path,
 				-1, 1);
-		if (ret < 0)
-			goto out_abort;
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out;
+		}
+
 		ret = btrfs_del_item(trans, root, path);
-		if (ret < 0)
-			goto out_abort;
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out;
+		}
 		btrfs_release_path(path);
 		ret = btrfs_insert_empty_item(trans, root, path,
 				key, sizeof(*item));
-		if (ret < 0)
-			goto out_abort;
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out;
+		}
 		l = path->nodes[0];
 		slot = path->slots[0];
 		ptr = btrfs_item_ptr_offset(l, slot);
@@ -192,10 +201,6 @@ int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
 out:
 	btrfs_free_path(path);
 	return ret;
-
-out_abort:
-	btrfs_abort_transaction(trans, root, ret);
-	goto out;
 }
 
 int btrfs_insert_root(struct btrfs_trans_handle *trans, struct btrfs_root *root,

fs/btrfs/scrub.c

@@ -352,13 +352,14 @@ static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
 	struct extent_buffer *eb;
 	struct btrfs_extent_item *ei;
 	struct scrub_warning swarn;
-	u32 item_size;
-	int ret;
+	unsigned long ptr = 0;
+	u64 extent_item_pos;
+	u64 flags = 0;
 	u64 ref_root;
+	u32 item_size;
 	u8 ref_level;
-	unsigned long ptr = 0;
 	const int bufsize = 4096;
-	u64 extent_item_pos;
+	int ret;
 
 	path = btrfs_alloc_path();
 
@@ -375,7 +376,8 @@ static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
 	if (!path || !swarn.scratch_buf || !swarn.msg_buf)
 		goto out;
 
-	ret = extent_from_logical(fs_info, swarn.logical, path, &found_key);
+	ret = extent_from_logical(fs_info, swarn.logical, path, &found_key,
+				  &flags);
 	if (ret < 0)
 		goto out;
 
@@ -387,7 +389,7 @@ static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
 	item_size = btrfs_item_size_nr(eb, path->slots[0]);
 	btrfs_release_path(path);
 
-	if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+	if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
 		do {
 			ret = tree_backref_for_extent(&ptr, eb, ei, item_size,
 							&ref_root, &ref_level);
@@ -1029,6 +1031,7 @@ static int scrub_setup_recheck_block(struct scrub_dev *sdev,
 				spin_lock(&sdev->stat_lock);
 				sdev->stat.malloc_errors++;
 				spin_unlock(&sdev->stat_lock);
+				kfree(bbio);
 				return -ENOMEM;
 			}
 			sblock->page_count++;
@@ -1666,21 +1669,6 @@ static void scrub_bio_end_io_worker(struct btrfs_work *work)
 		scrub_block_put(sblock);
 	}
 
-	if (sbio->err) {
-		/* what is this good for??? */
-		sbio->bio->bi_flags &= ~(BIO_POOL_MASK - 1);
-		sbio->bio->bi_flags |= 1 << BIO_UPTODATE;
-		sbio->bio->bi_phys_segments = 0;
-		sbio->bio->bi_idx = 0;
-
-		for (i = 0; i < sbio->page_count; i++) {
-			struct bio_vec *bi;
-			bi = &sbio->bio->bi_io_vec[i];
-			bi->bv_offset = 0;
-			bi->bv_len = PAGE_SIZE;
-		}
-	}
-
 	bio_put(sbio->bio);
 	sbio->bio = NULL;
 	spin_lock(&sdev->list_lock);

fs/btrfs/send.c

@@ -107,7 +107,6 @@ struct send_ctx {
 	int cur_inode_new;
 	int cur_inode_new_gen;
 	int cur_inode_deleted;
-	int cur_inode_first_ref_orphan;
 	u64 cur_inode_size;
 	u64 cur_inode_mode;
 
@@ -126,7 +125,15 @@ struct send_ctx {
 
 struct name_cache_entry {
 	struct list_head list;
-	struct list_head use_list;
+	/*
+	 * radix_tree has only 32bit entries but we need to handle 64bit inums.
+	 * We use the lower 32bit of the 64bit inum to store it in the tree. If
+	 * more then one inum would fall into the same entry, we use radix_list
+	 * to store the additional entries. radix_list is also used to store
+	 * entries where two entries have the same inum but different
+	 * generations.
+	 */
+	struct list_head radix_list;
 	u64 ino;
 	u64 gen;
 	u64 parent_ino;
@@ -328,6 +335,7 @@ out:
 	return ret;
 }
 
+#if 0
 static void fs_path_remove(struct fs_path *p)
 {
 	BUG_ON(p->reversed);
@@ -335,6 +343,7 @@ static void fs_path_remove(struct fs_path *p)
 		p->end--;
 	*p->end = 0;
 }
+#endif
 
 static int fs_path_copy(struct fs_path *p, struct fs_path *from)
 {
@@ -377,7 +386,7 @@ static struct btrfs_path *alloc_path_for_send(void)
 	return path;
 }
 
-static int write_buf(struct send_ctx *sctx, const void *buf, u32 len)
+int write_buf(struct file *filp, const void *buf, u32 len, loff_t *off)
 {
 	int ret;
 	mm_segment_t old_fs;
@@ -387,8 +396,7 @@ static int write_buf(struct send_ctx *sctx, const void *buf, u32 len)
 	set_fs(KERNEL_DS);
 
 	while (pos < len) {
-		ret = vfs_write(sctx->send_filp, (char *)buf + pos, len - pos,
-				&sctx->send_off);
+		ret = vfs_write(filp, (char *)buf + pos, len - pos, off);
 		/* TODO handle that correctly */
 		/*if (ret == -ERESTARTSYS) {
 			continue;
@@ -544,7 +552,8 @@ static int send_header(struct send_ctx *sctx)
 	strcpy(hdr.magic, BTRFS_SEND_STREAM_MAGIC);
 	hdr.version = cpu_to_le32(BTRFS_SEND_STREAM_VERSION);
 
-	return write_buf(sctx, &hdr, sizeof(hdr));
+	return write_buf(sctx->send_filp, &hdr, sizeof(hdr),
+					&sctx->send_off);
 }
 
 /*
@@ -581,7 +590,8 @@ static int send_cmd(struct send_ctx *sctx)
 	crc = crc32c(0, (unsigned char *)sctx->send_buf, sctx->send_size);
 	hdr->crc = cpu_to_le32(crc);
 
-	ret = write_buf(sctx, sctx->send_buf, sctx->send_size);
+	ret = write_buf(sctx->send_filp, sctx->send_buf, sctx->send_size,
+					&sctx->send_off);
 
 	sctx->total_send_size += sctx->send_size;
 	sctx->cmd_send_size[le16_to_cpu(hdr->cmd)] += sctx->send_size;
@@ -687,7 +697,8 @@ out:
  */
 static int get_inode_info(struct btrfs_root *root,
 			  u64 ino, u64 *size, u64 *gen,
-			  u64 *mode, u64 *uid, u64 *gid)
+			  u64 *mode, u64 *uid, u64 *gid,
+			  u64 *rdev)
 {
 	int ret;
 	struct btrfs_inode_item *ii;
@@ -721,6 +732,8 @@ static int get_inode_info(struct btrfs_root *root,
 		*uid = btrfs_inode_uid(path->nodes[0], ii);
 	if (gid)
 		*gid = btrfs_inode_gid(path->nodes[0], ii);
+	if (rdev)
+		*rdev = btrfs_inode_rdev(path->nodes[0], ii);
 
 out:
 	btrfs_free_path(path);
@@ -852,7 +865,6 @@ static int iterate_dir_item(struct send_ctx *sctx,
 	struct extent_buffer *eb;
 	struct btrfs_item *item;
 	struct btrfs_dir_item *di;
-	struct btrfs_path *tmp_path = NULL;
 	struct btrfs_key di_key;
 	char *buf = NULL;
 	char *buf2 = NULL;
@@ -874,12 +886,6 @@ static int iterate_dir_item(struct send_ctx *sctx,
 		goto out;
 	}
 
-	tmp_path = alloc_path_for_send();
-	if (!tmp_path) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
 	eb = path->nodes[0];
 	slot = path->slots[0];
 	item = btrfs_item_nr(eb, slot);
@@ -941,7 +947,6 @@ static int iterate_dir_item(struct send_ctx *sctx,
 	}
 
 out:
-	btrfs_free_path(tmp_path);
 	if (buf_virtual)
 		vfree(buf);
 	else
@@ -1026,12 +1031,12 @@ struct backref_ctx {
 	u64 extent_len;
 
 	/* Just to check for bugs in backref resolving */
-	int found_in_send_root;
+	int found_itself;
 };
 
 static int __clone_root_cmp_bsearch(const void *key, const void *elt)
 {
-	u64 root = (u64)key;
+	u64 root = (u64)(uintptr_t)key;
 	struct clone_root *cr = (struct clone_root *)elt;
 
 	if (root < cr->root->objectid)
@@ -1055,6 +1060,7 @@ static int __clone_root_cmp_sort(const void *e1, const void *e2)
 
 /*
  * Called for every backref that is found for the current extent.
+ * Results are collected in sctx->clone_roots->ino/offset/found_refs
  */
 static int __iterate_backrefs(u64 ino, u64 offset, u64 root, void *ctx_)
 {
@@ -1064,7 +1070,7 @@ static int __iterate_backrefs(u64 ino, u64 offset, u64 root, void *ctx_)
 	u64 i_size;
 
 	/* First check if the root is in the list of accepted clone sources */
-	found = bsearch((void *)root, bctx->sctx->clone_roots,
+	found = bsearch((void *)(uintptr_t)root, bctx->sctx->clone_roots,
 			bctx->sctx->clone_roots_cnt,
 			sizeof(struct clone_root),
 			__clone_root_cmp_bsearch);
@@ -1074,14 +1080,15 @@ static int __iterate_backrefs(u64 ino, u64 offset, u64 root, void *ctx_)
 	if (found->root == bctx->sctx->send_root &&
 	    ino == bctx->cur_objectid &&
 	    offset == bctx->cur_offset) {
-		bctx->found_in_send_root = 1;
+		bctx->found_itself = 1;
 	}
 
 	/*
-	 * There are inodes that have extents that lie behind it's i_size. Don't
+	 * There are inodes that have extents that lie behind its i_size. Don't
 	 * accept clones from these extents.
 	 */
-	ret = get_inode_info(found->root, ino, &i_size, NULL, NULL, NULL, NULL);
+	ret = get_inode_info(found->root, ino, &i_size, NULL, NULL, NULL, NULL,
+			NULL);
 	if (ret < 0)
 		return ret;
 
@@ -1101,16 +1108,12 @@ static int __iterate_backrefs(u64 ino, u64 offset, u64 root, void *ctx_)
 		 */
 		if (ino >= bctx->cur_objectid)
 			return 0;
-		/*if (ino > ctx->cur_objectid)
+#if 0
+		if (ino > bctx->cur_objectid)
 			return 0;
-		if (offset + ctx->extent_len > ctx->cur_offset)
-			return 0;*/
-
-		bctx->found++;
-		found->found_refs++;
-		found->ino = ino;
-		found->offset = offset;
-		return 0;
+		if (offset + bctx->extent_len > bctx->cur_offset)
+			return 0;
+#endif
 	}
 
 	bctx->found++;
@@ -1130,6 +1133,12 @@ static int __iterate_backrefs(u64 ino, u64 offset, u64 root, void *ctx_)
 }
 
 /*
+ * Given an inode, offset and extent item, it finds a good clone for a clone
+ * instruction. Returns -ENOENT when none could be found. The function makes
+ * sure that the returned clone is usable at the point where sending is at the
+ * moment. This means, that no clones are accepted which lie behind the current
+ * inode+offset.
+ *
  * path must point to the extent item when called.
  */
 static int find_extent_clone(struct send_ctx *sctx,
@@ -1141,20 +1150,29 @@ static int find_extent_clone(struct send_ctx *sctx,
 	int ret;
 	int extent_type;
 	u64 logical;
+	u64 disk_byte;
 	u64 num_bytes;
 	u64 extent_item_pos;
+	u64 flags = 0;
 	struct btrfs_file_extent_item *fi;
 	struct extent_buffer *eb = path->nodes[0];
-	struct backref_ctx backref_ctx;
+	struct backref_ctx *backref_ctx = NULL;
 	struct clone_root *cur_clone_root;
 	struct btrfs_key found_key;
 	struct btrfs_path *tmp_path;
+	int compressed;
 	u32 i;
 
 	tmp_path = alloc_path_for_send();
 	if (!tmp_path)
 		return -ENOMEM;
 
+	backref_ctx = kmalloc(sizeof(*backref_ctx), GFP_NOFS);
+	if (!backref_ctx) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
 	if (data_offset >= ino_size) {
 		/*
 		 * There may be extents that lie behind the file's size.
@@ -1172,22 +1190,23 @@ static int find_extent_clone(struct send_ctx *sctx,
 		ret = -ENOENT;
 		goto out;
 	}
+	compressed = btrfs_file_extent_compression(eb, fi);
 
 	num_bytes = btrfs_file_extent_num_bytes(eb, fi);
-	logical = btrfs_file_extent_disk_bytenr(eb, fi);
-	if (logical == 0) {
+	disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
+	if (disk_byte == 0) {
 		ret = -ENOENT;
 		goto out;
 	}
-	logical += btrfs_file_extent_offset(eb, fi);
+	logical = disk_byte + btrfs_file_extent_offset(eb, fi);
 
-	ret = extent_from_logical(sctx->send_root->fs_info,
-			logical, tmp_path, &found_key);
+	ret = extent_from_logical(sctx->send_root->fs_info, disk_byte, tmp_path,
+				  &found_key, &flags);
 	btrfs_release_path(tmp_path);
 
 	if (ret < 0)
 		goto out;
-	if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+	if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
 		ret = -EIO;
 		goto out;
 	}
@@ -1202,12 +1221,12 @@ static int find_extent_clone(struct send_ctx *sctx,
 		cur_clone_root->found_refs = 0;
 	}
 
-	backref_ctx.sctx = sctx;
-	backref_ctx.found = 0;
-	backref_ctx.cur_objectid = ino;
-	backref_ctx.cur_offset = data_offset;
-	backref_ctx.found_in_send_root = 0;
-	backref_ctx.extent_len = num_bytes;
+	backref_ctx->sctx = sctx;
+	backref_ctx->found = 0;
+	backref_ctx->cur_objectid = ino;
+	backref_ctx->cur_offset = data_offset;
+	backref_ctx->found_itself = 0;
+	backref_ctx->extent_len = num_bytes;
 
 	/*
 	 * The last extent of a file may be too large due to page alignment.
@@ -1215,25 +1234,31 @@ static int find_extent_clone(struct send_ctx *sctx,
 	 * __iterate_backrefs work.
 	 */
 	if (data_offset + num_bytes >= ino_size)
-		backref_ctx.extent_len = ino_size - data_offset;
+		backref_ctx->extent_len = ino_size - data_offset;
 
 	/*
 	 * Now collect all backrefs.
 	 */
+	if (compressed == BTRFS_COMPRESS_NONE)
+		extent_item_pos = logical - found_key.objectid;
+	else
+		extent_item_pos = 0;
+
 	extent_item_pos = logical - found_key.objectid;
 	ret = iterate_extent_inodes(sctx->send_root->fs_info,
 					found_key.objectid, extent_item_pos, 1,
-					__iterate_backrefs, &backref_ctx);
+					__iterate_backrefs, backref_ctx);
+
 	if (ret < 0)
 		goto out;
 
-	if (!backref_ctx.found_in_send_root) {
+	if (!backref_ctx->found_itself) {
 		/* found a bug in backref code? */
 		ret = -EIO;
 		printk(KERN_ERR "btrfs: ERROR did not find backref in "
 				"send_root. inode=%llu, offset=%llu, "
-				"logical=%llu\n",
-				ino, data_offset, logical);
+				"disk_byte=%llu found extent=%llu\n",
+				ino, data_offset, disk_byte, found_key.objectid);
 		goto out;
 	}
 
@@ -1242,7 +1267,7 @@ verbose_printk(KERN_DEBUG "btrfs: find_extent_clone: data_offset=%llu, "
 		"num_bytes=%llu, logical=%llu\n",
 		data_offset, ino, num_bytes, logical);
 
-	if (!backref_ctx.found)
+	if (!backref_ctx->found)
 		verbose_printk("btrfs:    no clones found\n");
 
 	cur_clone_root = NULL;
@@ -1253,7 +1278,6 @@ verbose_printk(KERN_DEBUG "btrfs: find_extent_clone: data_offset=%llu, "
 			else if (sctx->clone_roots[i].root == sctx->send_root)
 				/* prefer clones from send_root over others */
 				cur_clone_root = sctx->clone_roots + i;
-			break;
 		}
 
 	}
@@ -1267,6 +1291,7 @@ verbose_printk(KERN_DEBUG "btrfs: find_extent_clone: data_offset=%llu, "
 
 out:
 	btrfs_free_path(tmp_path);
+	kfree(backref_ctx);
 	return ret;
 }
 
@@ -1307,8 +1332,6 @@ static int read_symlink(struct send_ctx *sctx,
 	len = btrfs_file_extent_inline_len(path->nodes[0], ei);
 
 	ret = fs_path_add_from_extent_buffer(dest, path->nodes[0], off, len);
-	if (ret < 0)
-		goto out;
 
 out:
 	btrfs_free_path(path);
@@ -1404,7 +1427,7 @@ static int get_cur_inode_state(struct send_ctx *sctx, u64 ino, u64 gen)
 	u64 right_gen;
 
 	ret = get_inode_info(sctx->send_root, ino, NULL, &left_gen, NULL, NULL,
-			NULL);
+			NULL, NULL);
 	if (ret < 0 && ret != -ENOENT)
 		goto out;
 	left_ret = ret;
@@ -1413,16 +1436,16 @@ static int get_cur_inode_state(struct send_ctx *sctx, u64 ino, u64 gen)
 		right_ret = -ENOENT;
 	} else {
 		ret = get_inode_info(sctx->parent_root, ino, NULL, &right_gen,
-				NULL, NULL, NULL);
+				NULL, NULL, NULL, NULL);
 		if (ret < 0 && ret != -ENOENT)
 			goto out;
 		right_ret = ret;
 	}
 
 	if (!left_ret && !right_ret) {
-		if (left_gen == gen && right_gen == gen)
+		if (left_gen == gen && right_gen == gen) {
 			ret = inode_state_no_change;
-		else if (left_gen == gen) {
+		} else if (left_gen == gen) {
 			if (ino < sctx->send_progress)
 				ret = inode_state_did_create;
 			else
@@ -1516,6 +1539,10 @@ out:
 	return ret;
 }
 
+/*
+ * Looks up the first btrfs_inode_ref of a given ino. It returns the parent dir,
+ * generation of the parent dir and the name of the dir entry.
+ */
 static int get_first_ref(struct send_ctx *sctx,
 			 struct btrfs_root *root, u64 ino,
 			 u64 *dir, u64 *dir_gen, struct fs_path *name)
@@ -1557,7 +1584,7 @@ static int get_first_ref(struct send_ctx *sctx,
 	btrfs_release_path(path);
 
 	ret = get_inode_info(root, found_key.offset, NULL, dir_gen, NULL, NULL,
-			NULL);
+			NULL, NULL);
 	if (ret < 0)
 		goto out;
 
@@ -1586,22 +1613,28 @@ static int is_first_ref(struct send_ctx *sctx,
 	if (ret < 0)
 		goto out;
 
-	if (name_len != fs_path_len(tmp_name)) {
+	if (dir != tmp_dir || name_len != fs_path_len(tmp_name)) {
 		ret = 0;
 		goto out;
 	}
 
-	ret = memcmp(tmp_name->start, name, name_len);
-	if (ret)
-		ret = 0;
-	else
-		ret = 1;
+	ret = !memcmp(tmp_name->start, name, name_len);
 
 out:
 	fs_path_free(sctx, tmp_name);
 	return ret;
 }
 
+/*
+ * Used by process_recorded_refs to determine if a new ref would overwrite an
+ * already existing ref. In case it detects an overwrite, it returns the
+ * inode/gen in who_ino/who_gen.
+ * When an overwrite is detected, process_recorded_refs does proper orphanizing
+ * to make sure later references to the overwritten inode are possible.
+ * Orphanizing is however only required for the first ref of an inode.
+ * process_recorded_refs does an additional is_first_ref check to see if
+ * orphanizing is really required.
+ */
 static int will_overwrite_ref(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 			      const char *name, int name_len,
 			      u64 *who_ino, u64 *who_gen)
@@ -1626,9 +1659,14 @@ static int will_overwrite_ref(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 		goto out;
 	}
 
+	/*
+	 * Check if the overwritten ref was already processed. If yes, the ref
+	 * was already unlinked/moved, so we can safely assume that we will not
+	 * overwrite anything at this point in time.
+	 */
 	if (other_inode > sctx->send_progress) {
 		ret = get_inode_info(sctx->parent_root, other_inode, NULL,
-				who_gen, NULL, NULL, NULL);
+				who_gen, NULL, NULL, NULL, NULL);
 		if (ret < 0)
 			goto out;
 
@@ -1642,6 +1680,13 @@ out:
 	return ret;
 }
 
+/*
+ * Checks if the ref was overwritten by an already processed inode. This is
+ * used by __get_cur_name_and_parent to find out if the ref was orphanized and
+ * thus the orphan name needs be used.
+ * process_recorded_refs also uses it to avoid unlinking of refs that were
+ * overwritten.
+ */
 static int did_overwrite_ref(struct send_ctx *sctx,
 			    u64 dir, u64 dir_gen,
 			    u64 ino, u64 ino_gen,
@@ -1671,7 +1716,7 @@ static int did_overwrite_ref(struct send_ctx *sctx,
 	}
 
 	ret = get_inode_info(sctx->send_root, ow_inode, NULL, &gen, NULL, NULL,
-			NULL);
+			NULL, NULL);
 	if (ret < 0)
 		goto out;
 
@@ -1690,6 +1735,11 @@ out:
 	return ret;
 }
 
+/*
+ * Same as did_overwrite_ref, but also checks if it is the first ref of an inode
+ * that got overwritten. This is used by process_recorded_refs to determine
+ * if it has to use the path as returned by get_cur_path or the orphan name.
+ */
 static int did_overwrite_first_ref(struct send_ctx *sctx, u64 ino, u64 gen)
 {
 	int ret = 0;
@@ -1710,39 +1760,40 @@ static int did_overwrite_first_ref(struct send_ctx *sctx, u64 ino, u64 gen)
 
 	ret = did_overwrite_ref(sctx, dir, dir_gen, ino, gen,
 			name->start, fs_path_len(name));
-	if (ret < 0)
-		goto out;
 
 out:
 	fs_path_free(sctx, name);
 	return ret;
 }
 
+/*
+ * Insert a name cache entry. On 32bit kernels the radix tree index is 32bit,
+ * so we need to do some special handling in case we have clashes. This function
+ * takes care of this with the help of name_cache_entry::radix_list.
+ * In case of error, nce is kfreed.
+ */
 static int name_cache_insert(struct send_ctx *sctx,
 			     struct name_cache_entry *nce)
 {
 	int ret = 0;
-	struct name_cache_entry **ncea;
-
-	ncea = radix_tree_lookup(&sctx->name_cache, nce->ino);
-	if (ncea) {
-		if (!ncea[0])
-			ncea[0] = nce;
-		else if (!ncea[1])
-			ncea[1] = nce;
-		else
-			BUG();
-	} else {
-		ncea = kmalloc(sizeof(void *) * 2, GFP_NOFS);
-		if (!ncea)
+	struct list_head *nce_head;
+
+	nce_head = radix_tree_lookup(&sctx->name_cache,
+			(unsigned long)nce->ino);
+	if (!nce_head) {
+		nce_head = kmalloc(sizeof(*nce_head), GFP_NOFS);
+		if (!nce_head)
 			return -ENOMEM;
+		INIT_LIST_HEAD(nce_head);
 
-		ncea[0] = nce;
-		ncea[1] = NULL;
-		ret = radix_tree_insert(&sctx->name_cache, nce->ino, ncea);
-		if (ret < 0)
+		ret = radix_tree_insert(&sctx->name_cache, nce->ino, nce_head);
+		if (ret < 0) {
+			kfree(nce_head);
+			kfree(nce);
 			return ret;
+		}
 	}
+	list_add_tail(&nce->radix_list, nce_head);
 	list_add_tail(&nce->list, &sctx->name_cache_list);
 	sctx->name_cache_size++;
 
@@ -1752,50 +1803,52 @@ static int name_cache_insert(struct send_ctx *sctx,
 static void name_cache_delete(struct send_ctx *sctx,
 			      struct name_cache_entry *nce)
 {
-	struct name_cache_entry **ncea;
-
-	ncea = radix_tree_lookup(&sctx->name_cache, nce->ino);
-	BUG_ON(!ncea);
-
-	if (ncea[0] == nce)
-		ncea[0] = NULL;
-	else if (ncea[1] == nce)
-		ncea[1] = NULL;
-	else
-		BUG();
+	struct list_head *nce_head;
 
-	if (!ncea[0] && !ncea[1]) {
-		radix_tree_delete(&sctx->name_cache, nce->ino);
-		kfree(ncea);
-	}
+	nce_head = radix_tree_lookup(&sctx->name_cache,
+			(unsigned long)nce->ino);
+	BUG_ON(!nce_head);
 
+	list_del(&nce->radix_list);
 	list_del(&nce->list);
-
 	sctx->name_cache_size--;
+
+	if (list_empty(nce_head)) {
+		radix_tree_delete(&sctx->name_cache, (unsigned long)nce->ino);
+		kfree(nce_head);
+	}
 }
 
 static struct name_cache_entry *name_cache_search(struct send_ctx *sctx,
 						    u64 ino, u64 gen)
 {
-	struct name_cache_entry **ncea;
+	struct list_head *nce_head;
+	struct name_cache_entry *cur;
 
-	ncea = radix_tree_lookup(&sctx->name_cache, ino);
-	if (!ncea)
+	nce_head = radix_tree_lookup(&sctx->name_cache, (unsigned long)ino);
+	if (!nce_head)
 		return NULL;
 
-	if (ncea[0] && ncea[0]->gen == gen)
-		return ncea[0];
-	else if (ncea[1] && ncea[1]->gen == gen)
-		return ncea[1];
+	list_for_each_entry(cur, nce_head, radix_list) {
+		if (cur->ino == ino && cur->gen == gen)
+			return cur;
+	}
 	return NULL;
 }
 
+/*
+ * Removes the entry from the list and adds it back to the end. This marks the
+ * entry as recently used so that name_cache_clean_unused does not remove it.
+ */
 static void name_cache_used(struct send_ctx *sctx, struct name_cache_entry *nce)
 {
 	list_del(&nce->list);
 	list_add_tail(&nce->list, &sctx->name_cache_list);
 }
 
+/*
+ * Remove some entries from the beginning of name_cache_list.
+ */
 static void name_cache_clean_unused(struct send_ctx *sctx)
 {
 	struct name_cache_entry *nce;
@@ -1814,13 +1867,23 @@ static void name_cache_clean_unused(struct send_ctx *sctx)
 static void name_cache_free(struct send_ctx *sctx)
 {
 	struct name_cache_entry *nce;
-	struct name_cache_entry *tmp;
 
-	list_for_each_entry_safe(nce, tmp, &sctx->name_cache_list, list) {
+	while (!list_empty(&sctx->name_cache_list)) {
+		nce = list_entry(sctx->name_cache_list.next,
+				struct name_cache_entry, list);
 		name_cache_delete(sctx, nce);
+		kfree(nce);
 	}
 }
 
+/*
+ * Used by get_cur_path for each ref up to the root.
+ * Returns 0 if it succeeded.
+ * Returns 1 if the inode is not existent or got overwritten. In that case, the
+ * name is an orphan name. This instructs get_cur_path to stop iterating. If 1
+ * is returned, parent_ino/parent_gen are not guaranteed to be valid.
+ * Returns <0 in case of error.
+ */
 static int __get_cur_name_and_parent(struct send_ctx *sctx,
 				     u64 ino, u64 gen,
 				     u64 *parent_ino,
@@ -1832,6 +1895,11 @@ static int __get_cur_name_and_parent(struct send_ctx *sctx,
 	struct btrfs_path *path = NULL;
 	struct name_cache_entry *nce = NULL;
 
+	/*
+	 * First check if we already did a call to this function with the same
+	 * ino/gen. If yes, check if the cache entry is still up-to-date. If yes
+	 * return the cached result.
+	 */
 	nce = name_cache_search(sctx, ino, gen);
 	if (nce) {
 		if (ino < sctx->send_progress && nce->need_later_update) {
@@ -1854,6 +1922,11 @@ static int __get_cur_name_and_parent(struct send_ctx *sctx,
 	if (!path)
 		return -ENOMEM;
 
+	/*
+	 * If the inode is not existent yet, add the orphan name and return 1.
+	 * This should only happen for the parent dir that we determine in
+	 * __record_new_ref
+	 */
 	ret = is_inode_existent(sctx, ino, gen);
 	if (ret < 0)
 		goto out;
@@ -1866,6 +1939,10 @@ static int __get_cur_name_and_parent(struct send_ctx *sctx,
 		goto out_cache;
 	}
 
+	/*
+	 * Depending on whether the inode was already processed or not, use
+	 * send_root or parent_root for ref lookup.
+	 */
 	if (ino < sctx->send_progress)
 		ret = get_first_ref(sctx, sctx->send_root, ino,
 				parent_ino, parent_gen, dest);
@@ -1875,6 +1952,10 @@ static int __get_cur_name_and_parent(struct send_ctx *sctx,
 	if (ret < 0)
 		goto out;
 
+	/*
+	 * Check if the ref was overwritten by an inode's ref that was processed
+	 * earlier. If yes, treat as orphan and return 1.
+	 */
 	ret = did_overwrite_ref(sctx, *parent_ino, *parent_gen, ino, gen,
 			dest->start, dest->end - dest->start);
 	if (ret < 0)
@@ -1888,6 +1969,9 @@ static int __get_cur_name_and_parent(struct send_ctx *sctx,
 	}
 
 out_cache:
+	/*
+	 * Store the result of the lookup in the name cache.
+	 */
 	nce = kmalloc(sizeof(*nce) + fs_path_len(dest) + 1, GFP_NOFS);
 	if (!nce) {
 		ret = -ENOMEM;
@@ -1901,7 +1985,6 @@ out_cache:
 	nce->name_len = fs_path_len(dest);
 	nce->ret = ret;
 	strcpy(nce->name, dest->start);
-	memset(&nce->use_list, 0, sizeof(nce->use_list));
 
 	if (ino < sctx->send_progress)
 		nce->need_later_update = 0;
@@ -2107,9 +2190,6 @@ static int send_subvol_begin(struct send_ctx *sctx)
 	read_extent_buffer(leaf, name, (unsigned long)(ref + 1), namelen);
 	btrfs_release_path(path);
 
-	if (ret < 0)
-		goto out;
-
 	if (parent_root) {
 		ret = begin_cmd(sctx, BTRFS_SEND_C_SNAPSHOT);
 		if (ret < 0)
@@ -2276,7 +2356,7 @@ verbose_printk("btrfs: send_utimes %llu\n", ino);
 			btrfs_inode_mtime(ii));
 	TLV_PUT_BTRFS_TIMESPEC(sctx, BTRFS_SEND_A_CTIME, eb,
 			btrfs_inode_ctime(ii));
-	/* TODO otime? */
+	/* TODO Add otime support when the otime patches get into upstream */
 
 	ret = send_cmd(sctx);
 
@@ -2292,39 +2372,39 @@ out:
  * a valid path yet because we did not process the refs yet. So, the inode
  * is created as orphan.
  */
-static int send_create_inode(struct send_ctx *sctx, struct btrfs_path *path,
-			     struct btrfs_key *key)
+static int send_create_inode(struct send_ctx *sctx, u64 ino)
 {
 	int ret = 0;
-	struct extent_buffer *eb = path->nodes[0];
-	struct btrfs_inode_item *ii;
 	struct fs_path *p;
-	int slot = path->slots[0];
 	int cmd;
+	u64 gen;
 	u64 mode;
+	u64 rdev;
 
-verbose_printk("btrfs: send_create_inode %llu\n", sctx->cur_ino);
+verbose_printk("btrfs: send_create_inode %llu\n", ino);
 
 	p = fs_path_alloc(sctx);
 	if (!p)
 		return -ENOMEM;
 
-	ii = btrfs_item_ptr(eb, slot, struct btrfs_inode_item);
-	mode = btrfs_inode_mode(eb, ii);
+	ret = get_inode_info(sctx->send_root, ino, NULL, &gen, &mode, NULL,
+			NULL, &rdev);
+	if (ret < 0)
+		goto out;
 
-	if (S_ISREG(mode))
+	if (S_ISREG(mode)) {
 		cmd = BTRFS_SEND_C_MKFILE;
-	else if (S_ISDIR(mode))
+	} else if (S_ISDIR(mode)) {
 		cmd = BTRFS_SEND_C_MKDIR;
-	else if (S_ISLNK(mode))
+	} else if (S_ISLNK(mode)) {
 		cmd = BTRFS_SEND_C_SYMLINK;
-	else if (S_ISCHR(mode) || S_ISBLK(mode))
+	} else if (S_ISCHR(mode) || S_ISBLK(mode)) {
 		cmd = BTRFS_SEND_C_MKNOD;
-	else if (S_ISFIFO(mode))
+	} else if (S_ISFIFO(mode)) {
 		cmd = BTRFS_SEND_C_MKFIFO;
-	else if (S_ISSOCK(mode))
+	} else if (S_ISSOCK(mode)) {
 		cmd = BTRFS_SEND_C_MKSOCK;
-	else {
+	} else {
 		printk(KERN_WARNING "btrfs: unexpected inode type %o",
 				(int)(mode & S_IFMT));
 		ret = -ENOTSUPP;
@@ -2335,22 +2415,22 @@ verbose_printk("btrfs: send_create_inode %llu\n", sctx->cur_ino);
 	if (ret < 0)
 		goto out;
 
-	ret = gen_unique_name(sctx, sctx->cur_ino, sctx->cur_inode_gen, p);
+	ret = gen_unique_name(sctx, ino, gen, p);
 	if (ret < 0)
 		goto out;
 
 	TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
-	TLV_PUT_U64(sctx, BTRFS_SEND_A_INO, sctx->cur_ino);
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_INO, ino);
 
 	if (S_ISLNK(mode)) {
 		fs_path_reset(p);
-		ret = read_symlink(sctx, sctx->send_root, sctx->cur_ino, p);
+		ret = read_symlink(sctx, sctx->send_root, ino, p);
 		if (ret < 0)
 			goto out;
 		TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH_LINK, p);
 	} else if (S_ISCHR(mode) || S_ISBLK(mode) ||
 		   S_ISFIFO(mode) || S_ISSOCK(mode)) {
-		TLV_PUT_U64(sctx, BTRFS_SEND_A_RDEV, btrfs_inode_rdev(eb, ii));
+		TLV_PUT_U64(sctx, BTRFS_SEND_A_RDEV, rdev);
 	}
 
 	ret = send_cmd(sctx);
@@ -2364,6 +2444,92 @@ out:
 	return ret;
 }
 
+/*
+ * We need some special handling for inodes that get processed before the parent
+ * directory got created. See process_recorded_refs for details.
+ * This function does the check if we already created the dir out of order.
+ */
+static int did_create_dir(struct send_ctx *sctx, u64 dir)
+{
+	int ret = 0;
+	struct btrfs_path *path = NULL;
+	struct btrfs_key key;
+	struct btrfs_key found_key;
+	struct btrfs_key di_key;
+	struct extent_buffer *eb;
+	struct btrfs_dir_item *di;
+	int slot;
+
+	path = alloc_path_for_send();
+	if (!path) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	key.objectid = dir;
+	key.type = BTRFS_DIR_INDEX_KEY;
+	key.offset = 0;
+	while (1) {
+		ret = btrfs_search_slot_for_read(sctx->send_root, &key, path,
+				1, 0);
+		if (ret < 0)
+			goto out;
+		if (!ret) {
+			eb = path->nodes[0];
+			slot = path->slots[0];
+			btrfs_item_key_to_cpu(eb, &found_key, slot);
+		}
+		if (ret || found_key.objectid != key.objectid ||
+		    found_key.type != key.type) {
+			ret = 0;
+			goto out;
+		}
+
+		di = btrfs_item_ptr(eb, slot, struct btrfs_dir_item);
+		btrfs_dir_item_key_to_cpu(eb, di, &di_key);
+
+		if (di_key.objectid < sctx->send_progress) {
+			ret = 1;
+			goto out;
+		}
+
+		key.offset = found_key.offset + 1;
+		btrfs_release_path(path);
+	}
+
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
+/*
+ * Only creates the inode if it is:
+ * 1. Not a directory
+ * 2. Or a directory which was not created already due to out of order
+ *    directories. See did_create_dir and process_recorded_refs for details.
+ */
+static int send_create_inode_if_needed(struct send_ctx *sctx)
+{
+	int ret;
+
+	if (S_ISDIR(sctx->cur_inode_mode)) {
+		ret = did_create_dir(sctx, sctx->cur_ino);
+		if (ret < 0)
+			goto out;
+		if (ret) {
+			ret = 0;
+			goto out;
+		}
+	}
+
+	ret = send_create_inode(sctx, sctx->cur_ino);
+	if (ret < 0)
+		goto out;
+
+out:
+	return ret;
+}
+
 struct recorded_ref {
 	struct list_head list;
 	char *dir_path;
@@ -2416,13 +2582,13 @@ static int record_ref(struct list_head *head, u64 dir,
 static void __free_recorded_refs(struct send_ctx *sctx, struct list_head *head)
 {
 	struct recorded_ref *cur;
-	struct recorded_ref *tmp;
 
-	list_for_each_entry_safe(cur, tmp, head, list) {
+	while (!list_empty(head)) {
+		cur = list_entry(head->next, struct recorded_ref, list);
 		fs_path_free(sctx, cur->full_path);
+		list_del(&cur->list);
 		kfree(cur);
 	}
-	INIT_LIST_HEAD(head);
 }
 
 static void free_recorded_refs(struct send_ctx *sctx)
@@ -2432,7 +2598,7 @@ static void free_recorded_refs(struct send_ctx *sctx)
 }
 
 /*
- * Renames/moves a file/dir to it's orphan name. Used when the first
+ * Renames/moves a file/dir to its orphan name. Used when the first
  * ref of an unprocessed inode gets overwritten and for all non empty
  * directories.
  */
@@ -2472,6 +2638,12 @@ static int can_rmdir(struct send_ctx *sctx, u64 dir, u64 send_progress)
 	struct btrfs_key loc;
 	struct btrfs_dir_item *di;
 
+	/*
+	 * Don't try to rmdir the top/root subvolume dir.
+	 */
+	if (dir == BTRFS_FIRST_FREE_OBJECTID)
+		return 0;
+
 	path = alloc_path_for_send();
 	if (!path)
 		return -ENOMEM;
@@ -2513,160 +2685,6 @@ out:
 	return ret;
 }
 
-struct finish_unordered_dir_ctx {
-	struct send_ctx *sctx;
-	struct fs_path *cur_path;
-	struct fs_path *dir_path;
-	u64 dir_ino;
-	int need_delete;
-	int delete_pass;
-};
-
-int __finish_unordered_dir(int num, struct btrfs_key *di_key,
-			   const char *name, int name_len,
-			   const char *data, int data_len,
-			   u8 type, void *ctx)
-{
-	int ret = 0;
-	struct finish_unordered_dir_ctx *fctx = ctx;
-	struct send_ctx *sctx = fctx->sctx;
-	u64 di_gen;
-	u64 di_mode;
-	int is_orphan = 0;
-
-	if (di_key->objectid >= fctx->dir_ino)
-		goto out;
-
-	fs_path_reset(fctx->cur_path);
-
-	ret = get_inode_info(sctx->send_root, di_key->objectid,
-			NULL, &di_gen, &di_mode, NULL, NULL);
-	if (ret < 0)
-		goto out;
-
-	ret = is_first_ref(sctx, sctx->send_root, di_key->objectid,
-			fctx->dir_ino, name, name_len);
-	if (ret < 0)
-		goto out;
-	if (ret) {
-		is_orphan = 1;
-		ret = gen_unique_name(sctx, di_key->objectid, di_gen,
-				fctx->cur_path);
-	} else {
-		ret = get_cur_path(sctx, di_key->objectid, di_gen,
-				fctx->cur_path);
-	}
-	if (ret < 0)
-		goto out;
-
-	ret = fs_path_add(fctx->dir_path, name, name_len);
-	if (ret < 0)
-		goto out;
-
-	if (!fctx->delete_pass) {
-		if (S_ISDIR(di_mode)) {
-			ret = send_rename(sctx, fctx->cur_path,
-					fctx->dir_path);
-		} else {
-			ret = send_link(sctx, fctx->dir_path,
-					fctx->cur_path);
-			if (is_orphan)
-				fctx->need_delete = 1;
-		}
-	} else if (!S_ISDIR(di_mode)) {
-		ret = send_unlink(sctx, fctx->cur_path);
-	} else {
-		ret = 0;
-	}
-
-	fs_path_remove(fctx->dir_path);
-
-out:
-	return ret;
-}
-
-/*
- * Go through all dir items and see if we find refs which could not be created
- * in the past because the dir did not exist at that time.
- */
-static int finish_outoforder_dir(struct send_ctx *sctx, u64 dir, u64 dir_gen)
-{
-	int ret = 0;
-	struct btrfs_path *path = NULL;
-	struct btrfs_key key;
-	struct btrfs_key found_key;
-	struct extent_buffer *eb;
-	struct finish_unordered_dir_ctx fctx;
-	int slot;
-
-	path = alloc_path_for_send();
-	if (!path) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
-	memset(&fctx, 0, sizeof(fctx));
-	fctx.sctx = sctx;
-	fctx.cur_path = fs_path_alloc(sctx);
-	fctx.dir_path = fs_path_alloc(sctx);
-	if (!fctx.cur_path || !fctx.dir_path) {
-		ret = -ENOMEM;
-		goto out;
-	}
-	fctx.dir_ino = dir;
-
-	ret = get_cur_path(sctx, dir, dir_gen, fctx.dir_path);
-	if (ret < 0)
-		goto out;
-
-	/*
-	 * We do two passes. The first links in the new refs and the second
-	 * deletes orphans if required. Deletion of orphans is not required for
-	 * directory inodes, as we always have only one ref and use rename
-	 * instead of link for those.
-	 */
-
-again:
-	key.objectid = dir;
-	key.type = BTRFS_DIR_ITEM_KEY;
-	key.offset = 0;
-	while (1) {
-		ret = btrfs_search_slot_for_read(sctx->send_root, &key, path,
-				1, 0);
-		if (ret < 0)
-			goto out;
-		eb = path->nodes[0];
-		slot = path->slots[0];
-		btrfs_item_key_to_cpu(eb, &found_key, slot);
-
-		if (found_key.objectid != key.objectid ||
-		    found_key.type != key.type) {
-			btrfs_release_path(path);
-			break;
-		}
-
-		ret = iterate_dir_item(sctx, sctx->send_root, path,
-				&found_key, __finish_unordered_dir,
-				&fctx);
-		if (ret < 0)
-			goto out;
-
-		key.offset = found_key.offset + 1;
-		btrfs_release_path(path);
-	}
-
-	if (!fctx.delete_pass && fctx.need_delete) {
-		fctx.delete_pass = 1;
-		goto again;
-	}
-
-out:
-	btrfs_free_path(path);
-	fs_path_free(sctx, fctx.cur_path);
-	fs_path_free(sctx, fctx.dir_path);
-	return ret;
-}
-
 /*
  * This does all the move/link/unlink/rmdir magic.
  */
@@ -2674,6 +2692,7 @@ static int process_recorded_refs(struct send_ctx *sctx)
 {
 	int ret = 0;
 	struct recorded_ref *cur;
+	struct recorded_ref *cur2;
 	struct ulist *check_dirs = NULL;
 	struct ulist_iterator uit;
 	struct ulist_node *un;
@@ -2685,6 +2704,12 @@ static int process_recorded_refs(struct send_ctx *sctx)
 
 verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 
+	/*
+	 * This should never happen as the root dir always has the same ref
+	 * which is always '..'
+	 */
+	BUG_ON(sctx->cur_ino <= BTRFS_FIRST_FREE_OBJECTID);
+
 	valid_path = fs_path_alloc(sctx);
 	if (!valid_path) {
 		ret = -ENOMEM;
@@ -2730,6 +2755,46 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 	}
 
 	list_for_each_entry(cur, &sctx->new_refs, list) {
+		/*
+		 * We may have refs where the parent directory does not exist
+		 * yet. This happens if the parent directories inum is higher
+		 * the the current inum. To handle this case, we create the
+		 * parent directory out of order. But we need to check if this
+		 * did already happen before due to other refs in the same dir.
+		 */
+		ret = get_cur_inode_state(sctx, cur->dir, cur->dir_gen);
+		if (ret < 0)
+			goto out;
+		if (ret == inode_state_will_create) {
+			ret = 0;
+			/*
+			 * First check if any of the current inodes refs did
+			 * already create the dir.
+			 */
+			list_for_each_entry(cur2, &sctx->new_refs, list) {
+				if (cur == cur2)
+					break;
+				if (cur2->dir == cur->dir) {
+					ret = 1;
+					break;
+				}
+			}
+
+			/*
+			 * If that did not happen, check if a previous inode
+			 * did already create the dir.
+			 */
+			if (!ret)
+				ret = did_create_dir(sctx, cur->dir);
+			if (ret < 0)
+				goto out;
+			if (!ret) {
+				ret = send_create_inode(sctx, cur->dir);
+				if (ret < 0)
+					goto out;
+			}
+		}
+
 		/*
 		 * Check if this new ref would overwrite the first ref of
 		 * another unprocessed inode. If yes, orphanize the
@@ -2764,7 +2829,7 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 		 * inode, move it and update valid_path. If not, link or move
 		 * it depending on the inode mode.
 		 */
-		if (is_orphan && !sctx->cur_inode_first_ref_orphan) {
+		if (is_orphan) {
 			ret = send_rename(sctx, valid_path, cur->full_path);
 			if (ret < 0)
 				goto out;
@@ -2827,6 +2892,17 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 			if (ret < 0)
 				goto out;
 		}
+	} else if (S_ISDIR(sctx->cur_inode_mode) &&
+		   !list_empty(&sctx->deleted_refs)) {
+		/*
+		 * We have a moved dir. Add the old parent to check_dirs
+		 */
+		cur = list_entry(sctx->deleted_refs.next, struct recorded_ref,
+				list);
+		ret = ulist_add(check_dirs, cur->dir, cur->dir_gen,
+				GFP_NOFS);
+		if (ret < 0)
+			goto out;
 	} else if (!S_ISDIR(sctx->cur_inode_mode)) {
 		/*
 		 * We have a non dir inode. Go through all deleted refs and
@@ -2840,35 +2916,9 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 			if (ret < 0)
 				goto out;
 			if (!ret) {
-				/*
-				 * In case the inode was moved to a directory
-				 * that was not created yet (see
-				 * __record_new_ref), we can not unlink the ref
-				 * as it will be needed later when the parent
-				 * directory is created, so that we can move in
-				 * the inode to the new dir.
-				 */
-				if (!is_orphan &&
-				    sctx->cur_inode_first_ref_orphan) {
-					ret = orphanize_inode(sctx,
-							sctx->cur_ino,
-							sctx->cur_inode_gen,
-							cur->full_path);
-					if (ret < 0)
-						goto out;
-					ret = gen_unique_name(sctx,
-							sctx->cur_ino,
-							sctx->cur_inode_gen,
-							valid_path);
-					if (ret < 0)
-						goto out;
-					is_orphan = 1;
-
-				} else {
-					ret = send_unlink(sctx, cur->full_path);
-					if (ret < 0)
-						goto out;
-				}
+				ret = send_unlink(sctx, cur->full_path);
+				if (ret < 0)
+					goto out;
 			}
 			ret = ulist_add(check_dirs, cur->dir, cur->dir_gen,
 					GFP_NOFS);
@@ -2880,12 +2930,11 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 		 * If the inode is still orphan, unlink the orphan. This may
 		 * happen when a previous inode did overwrite the first ref
 		 * of this inode and no new refs were added for the current
-		 * inode.
-		 * We can however not delete the orphan in case the inode relies
-		 * in a directory that was not created yet (see
-		 * __record_new_ref)
+		 * inode. Unlinking does not mean that the inode is deleted in
+		 * all cases. There may still be links to this inode in other
+		 * places.
 		 */
-		if (is_orphan && !sctx->cur_inode_first_ref_orphan) {
+		if (is_orphan) {
 			ret = send_unlink(sctx, valid_path);
 			if (ret < 0)
 				goto out;
@@ -2900,6 +2949,11 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 	 */
 	ULIST_ITER_INIT(&uit);
 	while ((un = ulist_next(check_dirs, &uit))) {
+		/*
+		 * In case we had refs into dirs that were not processed yet,
+		 * we don't need to do the utime and rmdir logic for these dirs.
+		 * The dir will be processed later.
+		 */
 		if (un->val > sctx->cur_ino)
 			continue;
 
@@ -2929,25 +2983,6 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 		}
 	}
 
-	/*
-	 * Current inode is now at it's new position, so we must increase
-	 * send_progress
-	 */
-	sctx->send_progress = sctx->cur_ino + 1;
-
-	/*
-	 * We may have a directory here that has pending refs which could not
-	 * be created before (because the dir did not exist before, see
-	 * __record_new_ref). finish_outoforder_dir will link/move the pending
-	 * refs.
-	 */
-	if (S_ISDIR(sctx->cur_inode_mode) && sctx->cur_inode_new) {
-		ret = finish_outoforder_dir(sctx, sctx->cur_ino,
-				sctx->cur_inode_gen);
-		if (ret < 0)
-			goto out;
-	}
-
 	ret = 0;
 
 out:
@@ -2971,34 +3006,9 @@ static int __record_new_ref(int num, u64 dir, int index,
 		return -ENOMEM;
 
 	ret = get_inode_info(sctx->send_root, dir, NULL, &gen, NULL, NULL,
-			NULL);
-	if (ret < 0)
-		goto out;
-
-	/*
-	 * The parent may be non-existent at this point in time. This happens
-	 * if the ino of the parent dir is higher then the current ino. In this
-	 * case, we can not process this ref until the parent dir is finally
-	 * created. If we reach the parent dir later, process_recorded_refs
-	 * will go through all dir items and process the refs that could not be
-	 * processed before. In case this is the first ref, we set
-	 * cur_inode_first_ref_orphan to 1 to inform process_recorded_refs to
-	 * keep an orphan of the inode so that it later can be used for
-	 * link/move
-	 */
-	ret = is_inode_existent(sctx, dir, gen);
+			NULL, NULL);
 	if (ret < 0)
 		goto out;
-	if (!ret) {
-		ret = is_first_ref(sctx, sctx->send_root, sctx->cur_ino, dir,
-				name->start, fs_path_len(name));
-		if (ret < 0)
-			goto out;
-		if (ret)
-			sctx->cur_inode_first_ref_orphan = 1;
-		ret = 0;
-		goto out;
-	}
 
 	ret = get_cur_path(sctx, dir, gen, p);
 	if (ret < 0)
@@ -3029,7 +3039,7 @@ static int __record_deleted_ref(int num, u64 dir, int index,
 		return -ENOMEM;
 
 	ret = get_inode_info(sctx->parent_root, dir, NULL, &gen, NULL, NULL,
-			NULL);
+			NULL, NULL);
 	if (ret < 0)
 		goto out;
 
@@ -3206,33 +3216,28 @@ static int process_all_refs(struct send_ctx *sctx,
 	key.offset = 0;
 	while (1) {
 		ret = btrfs_search_slot_for_read(root, &key, path, 1, 0);
-		if (ret < 0) {
-			btrfs_release_path(path);
+		if (ret < 0)
 			goto out;
-		}
-		if (ret) {
-			btrfs_release_path(path);
+		if (ret)
 			break;
-		}
 
 		eb = path->nodes[0];
 		slot = path->slots[0];
 		btrfs_item_key_to_cpu(eb, &found_key, slot);
 
 		if (found_key.objectid != key.objectid ||
-		    found_key.type != key.type) {
-			btrfs_release_path(path);
+		    found_key.type != key.type)
 			break;
-		}
 
-		ret = iterate_inode_ref(sctx, sctx->parent_root, path,
-				&found_key, 0, cb, sctx);
+		ret = iterate_inode_ref(sctx, root, path, &found_key, 0, cb,
+				sctx);
 		btrfs_release_path(path);
 		if (ret < 0)
 			goto out;
 
 		key.offset = found_key.offset + 1;
 	}
+	btrfs_release_path(path);
 
 	ret = process_recorded_refs(sctx);
 
@@ -3555,7 +3560,7 @@ static int send_write(struct send_ctx *sctx, u64 offset, u32 len)
 	int ret = 0;
 	struct fs_path *p;
 	loff_t pos = offset;
-	int readed = 0;
+	int num_read = 0;
 	mm_segment_t old_fs;
 
 	p = fs_path_alloc(sctx);
@@ -3580,8 +3585,8 @@ verbose_printk("btrfs: send_write offset=%llu, len=%d\n", offset, len);
 	ret = vfs_read(sctx->cur_inode_filp, sctx->read_buf, len, &pos);
 	if (ret < 0)
 		goto out;
-	readed = ret;
-	if (!readed)
+	num_read = ret;
+	if (!num_read)
 		goto out;
 
 	ret = begin_cmd(sctx, BTRFS_SEND_C_WRITE);
@@ -3594,7 +3599,7 @@ verbose_printk("btrfs: send_write offset=%llu, len=%d\n", offset, len);
 
 	TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_FILE_OFFSET, offset);
-	TLV_PUT(sctx, BTRFS_SEND_A_DATA, sctx->read_buf, readed);
+	TLV_PUT(sctx, BTRFS_SEND_A_DATA, sctx->read_buf, num_read);
 
 	ret = send_cmd(sctx);
 
@@ -3604,7 +3609,7 @@ out:
 	set_fs(old_fs);
 	if (ret < 0)
 		return ret;
-	return readed;
+	return num_read;
 }
 
 /*
@@ -3615,7 +3620,6 @@ static int send_clone(struct send_ctx *sctx,
 		      struct clone_root *clone_root)
 {
 	int ret = 0;
-	struct btrfs_root *clone_root2 = clone_root->root;
 	struct fs_path *p;
 	u64 gen;
 
@@ -3640,22 +3644,23 @@ verbose_printk("btrfs: send_clone offset=%llu, len=%d, clone_root=%llu, "
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_LEN, len);
 	TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
 
-	if (clone_root2 == sctx->send_root) {
+	if (clone_root->root == sctx->send_root) {
 		ret = get_inode_info(sctx->send_root, clone_root->ino, NULL,
-				&gen, NULL, NULL, NULL);
+				&gen, NULL, NULL, NULL, NULL);
 		if (ret < 0)
 			goto out;
 		ret = get_cur_path(sctx, clone_root->ino, gen, p);
 	} else {
-		ret = get_inode_path(sctx, clone_root2, clone_root->ino, p);
+		ret = get_inode_path(sctx, clone_root->root,
+				clone_root->ino, p);
 	}
 	if (ret < 0)
 		goto out;
 
 	TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
-			clone_root2->root_item.uuid);
+			clone_root->root->root_item.uuid);
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_CTRANSID,
-			clone_root2->root_item.ctransid);
+			clone_root->root->root_item.ctransid);
 	TLV_PUT_PATH(sctx, BTRFS_SEND_A_CLONE_PATH, p);
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_OFFSET,
 			clone_root->offset);
@@ -3684,10 +3689,17 @@ static int send_write_or_clone(struct send_ctx *sctx,
 	ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
 			struct btrfs_file_extent_item);
 	type = btrfs_file_extent_type(path->nodes[0], ei);
-	if (type == BTRFS_FILE_EXTENT_INLINE)
+	if (type == BTRFS_FILE_EXTENT_INLINE) {
 		len = btrfs_file_extent_inline_len(path->nodes[0], ei);
-	else
+		/*
+		 * it is possible the inline item won't cover the whole page,
+		 * but there may be items after this page.  Make
+		 * sure to send the whole thing
+		 */
+		len = PAGE_CACHE_ALIGN(len);
+	} else {
 		len = btrfs_file_extent_num_bytes(path->nodes[0], ei);
+	}
 
 	if (offset + len > sctx->cur_inode_size)
 		len = sctx->cur_inode_size - offset;
@@ -3735,6 +3747,8 @@ static int is_extent_unchanged(struct send_ctx *sctx,
 	u64 left_offset_fixed;
 	u64 left_len;
 	u64 right_len;
+	u64 left_gen;
+	u64 right_gen;
 	u8 left_type;
 	u8 right_type;
 
@@ -3744,17 +3758,17 @@ static int is_extent_unchanged(struct send_ctx *sctx,
 
 	eb = left_path->nodes[0];
 	slot = left_path->slots[0];
-
 	ei = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
 	left_type = btrfs_file_extent_type(eb, ei);
-	left_disknr = btrfs_file_extent_disk_bytenr(eb, ei);
-	left_len = btrfs_file_extent_num_bytes(eb, ei);
-	left_offset = btrfs_file_extent_offset(eb, ei);
 
 	if (left_type != BTRFS_FILE_EXTENT_REG) {
 		ret = 0;
 		goto out;
 	}
+	left_disknr = btrfs_file_extent_disk_bytenr(eb, ei);
+	left_len = btrfs_file_extent_num_bytes(eb, ei);
+	left_offset = btrfs_file_extent_offset(eb, ei);
+	left_gen = btrfs_file_extent_generation(eb, ei);
 
 	/*
 	 * Following comments will refer to these graphics. L is the left
@@ -3810,6 +3824,7 @@ static int is_extent_unchanged(struct send_ctx *sctx,
 		right_disknr = btrfs_file_extent_disk_bytenr(eb, ei);
 		right_len = btrfs_file_extent_num_bytes(eb, ei);
 		right_offset = btrfs_file_extent_offset(eb, ei);
+		right_gen = btrfs_file_extent_generation(eb, ei);
 
 		if (right_type != BTRFS_FILE_EXTENT_REG) {
 			ret = 0;
@@ -3820,7 +3835,7 @@ static int is_extent_unchanged(struct send_ctx *sctx,
 		 * Are we at extent 8? If yes, we know the extent is changed.
 		 * This may only happen on the first iteration.
 		 */
-		if (found_key.offset + right_len < ekey->offset) {
+		if (found_key.offset + right_len <= ekey->offset) {
 			ret = 0;
 			goto out;
 		}
@@ -3837,8 +3852,9 @@ static int is_extent_unchanged(struct send_ctx *sctx,
 		/*
 		 * Check if we have the same extent.
 		 */
-		if (left_disknr + left_offset_fixed !=
-				right_disknr + right_offset) {
+		if (left_disknr != right_disknr ||
+		    left_offset_fixed != right_offset ||
+		    left_gen != right_gen) {
 			ret = 0;
 			goto out;
 		}
@@ -3977,6 +3993,15 @@ static int process_recorded_refs_if_needed(struct send_ctx *sctx, int at_end)
 		goto out;
 
 	ret = process_recorded_refs(sctx);
+	if (ret < 0)
+		goto out;
+
+	/*
+	 * We have processed the refs and thus need to advance send_progress.
+	 * Now, calls to get_cur_xxx will take the updated refs of the current
+	 * inode into account.
+	 */
+	sctx->send_progress = sctx->cur_ino + 1;
 
 out:
 	return ret;
@@ -4004,7 +4029,7 @@ static int finish_inode_if_needed(struct send_ctx *sctx, int at_end)
 		goto out;
 
 	ret = get_inode_info(sctx->send_root, sctx->cur_ino, NULL, NULL,
-			&left_mode, &left_uid, &left_gid);
+			&left_mode, &left_uid, &left_gid, NULL);
 	if (ret < 0)
 		goto out;
 
@@ -4015,7 +4040,7 @@ static int finish_inode_if_needed(struct send_ctx *sctx, int at_end)
 		} else {
 			ret = get_inode_info(sctx->parent_root, sctx->cur_ino,
 					NULL, NULL, &right_mode, &right_uid,
-					&right_gid);
+					&right_gid, NULL);
 			if (ret < 0)
 				goto out;
 
@@ -4074,7 +4099,12 @@ static int changed_inode(struct send_ctx *sctx,
 
 	sctx->cur_ino = key->objectid;
 	sctx->cur_inode_new_gen = 0;
-	sctx->cur_inode_first_ref_orphan = 0;
+
+	/*
+	 * Set send_progress to current inode. This will tell all get_cur_xxx
+	 * functions that the current inode's refs are not updated yet. Later,
+	 * when process_recorded_refs is finished, it is set to cur_ino + 1.
+	 */
 	sctx->send_progress = sctx->cur_ino;
 
 	if (result == BTRFS_COMPARE_TREE_NEW ||
@@ -4098,7 +4128,14 @@ static int changed_inode(struct send_ctx *sctx,
 
 		right_gen = btrfs_inode_generation(sctx->right_path->nodes[0],
 				right_ii);
-		if (left_gen != right_gen)
+
+		/*
+		 * The cur_ino = root dir case is special here. We can't treat
+		 * the inode as deleted+reused because it would generate a
+		 * stream that tries to delete/mkdir the root dir.
+		 */
+		if (left_gen != right_gen &&
+		    sctx->cur_ino != BTRFS_FIRST_FREE_OBJECTID)
 			sctx->cur_inode_new_gen = 1;
 	}
 
@@ -4111,8 +4148,7 @@ static int changed_inode(struct send_ctx *sctx,
 		sctx->cur_inode_mode = btrfs_inode_mode(
 				sctx->left_path->nodes[0], left_ii);
 		if (sctx->cur_ino != BTRFS_FIRST_FREE_OBJECTID)
-			ret = send_create_inode(sctx, sctx->left_path,
-					sctx->cmp_key);
+			ret = send_create_inode_if_needed(sctx);
 	} else if (result == BTRFS_COMPARE_TREE_DELETED) {
 		sctx->cur_inode_gen = right_gen;
 		sctx->cur_inode_new = 0;
@@ -4122,7 +4158,17 @@ static int changed_inode(struct send_ctx *sctx,
 		sctx->cur_inode_mode = btrfs_inode_mode(
 				sctx->right_path->nodes[0], right_ii);
 	} else if (result == BTRFS_COMPARE_TREE_CHANGED) {
+		/*
+		 * We need to do some special handling in case the inode was
+		 * reported as changed with a changed generation number. This
+		 * means that the original inode was deleted and new inode
+		 * reused the same inum. So we have to treat the old inode as
+		 * deleted and the new one as new.
+		 */
 		if (sctx->cur_inode_new_gen) {
+			/*
+			 * First, process the inode as if it was deleted.
+			 */
 			sctx->cur_inode_gen = right_gen;
 			sctx->cur_inode_new = 0;
 			sctx->cur_inode_deleted = 1;
@@ -4135,6 +4181,9 @@ static int changed_inode(struct send_ctx *sctx,
 			if (ret < 0)
 				goto out;
 
+			/*
+			 * Now process the inode as if it was new.
+			 */
 			sctx->cur_inode_gen = left_gen;
 			sctx->cur_inode_new = 1;
 			sctx->cur_inode_deleted = 0;
@@ -4142,14 +4191,23 @@ static int changed_inode(struct send_ctx *sctx,
 					sctx->left_path->nodes[0], left_ii);
 			sctx->cur_inode_mode = btrfs_inode_mode(
 					sctx->left_path->nodes[0], left_ii);
-			ret = send_create_inode(sctx, sctx->left_path,
-					sctx->cmp_key);
+			ret = send_create_inode_if_needed(sctx);
 			if (ret < 0)
 				goto out;
 
 			ret = process_all_refs(sctx, BTRFS_COMPARE_TREE_NEW);
 			if (ret < 0)
 				goto out;
+			/*
+			 * Advance send_progress now as we did not get into
+			 * process_recorded_refs_if_needed in the new_gen case.
+			 */
+			sctx->send_progress = sctx->cur_ino + 1;
+
+			/*
+			 * Now process all extents and xattrs of the inode as if
+			 * they were all new.
+			 */
 			ret = process_all_extents(sctx);
 			if (ret < 0)
 				goto out;
@@ -4172,6 +4230,16 @@ out:
 	return ret;
 }
 
+/*
+ * We have to process new refs before deleted refs, but compare_trees gives us
+ * the new and deleted refs mixed. To fix this, we record the new/deleted refs
+ * first and later process them in process_recorded_refs.
+ * For the cur_inode_new_gen case, we skip recording completely because
+ * changed_inode did already initiate processing of refs. The reason for this is
+ * that in this case, compare_tree actually compares the refs of 2 different
+ * inodes. To fix this, process_all_refs is used in changed_inode to handle all
+ * refs of the right tree as deleted and all refs of the left tree as new.
+ */
 static int changed_ref(struct send_ctx *sctx,
 		       enum btrfs_compare_tree_result result)
 {
@@ -4192,6 +4260,11 @@ static int changed_ref(struct send_ctx *sctx,
 	return ret;
 }
 
+/*
+ * Process new/deleted/changed xattrs. We skip processing in the
+ * cur_inode_new_gen case because changed_inode did already initiate processing
+ * of xattrs. The reason is the same as in changed_ref
+ */
 static int changed_xattr(struct send_ctx *sctx,
 			 enum btrfs_compare_tree_result result)
 {
@@ -4211,6 +4284,11 @@ static int changed_xattr(struct send_ctx *sctx,
 	return ret;
 }
 
+/*
+ * Process new/deleted/changed extents. We skip processing in the
+ * cur_inode_new_gen case because changed_inode did already initiate processing
+ * of extents. The reason is the same as in changed_ref
+ */
 static int changed_extent(struct send_ctx *sctx,
 			  enum btrfs_compare_tree_result result)
 {
@@ -4227,7 +4305,10 @@ static int changed_extent(struct send_ctx *sctx,
 	return ret;
 }
 
-
+/*
+ * Updates compare related fields in sctx and simply forwards to the actual
+ * changed_xxx functions.
+ */
 static int changed_cb(struct btrfs_root *left_root,
 		      struct btrfs_root *right_root,
 		      struct btrfs_path *left_path,
@@ -4247,6 +4328,11 @@ static int changed_cb(struct btrfs_root *left_root,
 	if (ret < 0)
 		goto out;
 
+	/* Ignore non-FS objects */
+	if (key->objectid == BTRFS_FREE_INO_OBJECTID ||
+	    key->objectid == BTRFS_FREE_SPACE_OBJECTID)
+		goto out;
+
 	if (key->type == BTRFS_INODE_ITEM_KEY)
 		ret = changed_inode(sctx, result);
 	else if (key->type == BTRFS_INODE_REF_KEY)
@@ -4299,7 +4385,8 @@ join_trans:
 	}
 
 	/*
-	 * Make sure the tree has not changed
+	 * Make sure the tree has not changed after re-joining. We detect this
+	 * by comparing start_ctransid and ctransid. They should always match.
 	 */
 	spin_lock(&send_root->root_times_lock);
 	ctransid = btrfs_root_ctransid(&send_root->root_item);

fs/btrfs/send.h

@@ -130,4 +130,5 @@ enum {
 
 #ifdef __KERNEL__
 long btrfs_ioctl_send(struct file *mnt_file, void __user *arg);
+int write_buf(struct file *filp, const void *buf, u32 len, loff_t *off);
 #endif

fs/btrfs/super.c

@@ -243,12 +243,18 @@ void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root, const char *function,
 			       unsigned int line, int errno)
 {
-	WARN_ONCE(1, KERN_DEBUG "btrfs: Transaction aborted");
+	WARN_ONCE(1, KERN_DEBUG "btrfs: Transaction aborted\n");
 	trans->aborted = errno;
 	/* Nothing used. The other threads that have joined this
 	 * transaction may be able to continue. */
 	if (!trans->blocks_used) {
-		btrfs_printk(root->fs_info, "Aborting unused transaction.\n");
+		char nbuf[16];
+		const char *errstr;
+
+		errstr = btrfs_decode_error(root->fs_info, errno, nbuf);
+		btrfs_printk(root->fs_info,
+			     "%s:%d: Aborting unused transaction(%s).\n",
+			     function, line, errstr);
 		return;
 	}
 	trans->transaction->aborted = errno;
@@ -407,7 +413,15 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
 			btrfs_set_opt(info->mount_opt, NODATASUM);
 			break;
 		case Opt_nodatacow:
-			printk(KERN_INFO "btrfs: setting nodatacow\n");
+			if (!btrfs_test_opt(root, COMPRESS) ||
+				!btrfs_test_opt(root, FORCE_COMPRESS)) {
+					printk(KERN_INFO "btrfs: setting nodatacow, compression disabled\n");
+			} else {
+				printk(KERN_INFO "btrfs: setting nodatacow\n");
+			}
+			info->compress_type = BTRFS_COMPRESS_NONE;
+			btrfs_clear_opt(info->mount_opt, COMPRESS);
+			btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
 			btrfs_set_opt(info->mount_opt, NODATACOW);
 			btrfs_set_opt(info->mount_opt, NODATASUM);
 			break;
@@ -422,10 +436,14 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
 				compress_type = "zlib";
 				info->compress_type = BTRFS_COMPRESS_ZLIB;
 				btrfs_set_opt(info->mount_opt, COMPRESS);
+				btrfs_clear_opt(info->mount_opt, NODATACOW);
+				btrfs_clear_opt(info->mount_opt, NODATASUM);
 			} else if (strcmp(args[0].from, "lzo") == 0) {
 				compress_type = "lzo";
 				info->compress_type = BTRFS_COMPRESS_LZO;
 				btrfs_set_opt(info->mount_opt, COMPRESS);
+				btrfs_clear_opt(info->mount_opt, NODATACOW);
+				btrfs_clear_opt(info->mount_opt, NODATASUM);
 				btrfs_set_fs_incompat(info, COMPRESS_LZO);
 			} else if (strncmp(args[0].from, "no", 2) == 0) {
 				compress_type = "no";
@@ -543,11 +561,11 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
 			btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG);
 			break;
 		case Opt_defrag:
-			printk(KERN_INFO "btrfs: enabling auto defrag");
+			printk(KERN_INFO "btrfs: enabling auto defrag\n");
 			btrfs_set_opt(info->mount_opt, AUTO_DEFRAG);
 			break;
 		case Opt_recovery:
-			printk(KERN_INFO "btrfs: enabling auto recovery");
+			printk(KERN_INFO "btrfs: enabling auto recovery\n");
 			btrfs_set_opt(info->mount_opt, RECOVERY);
 			break;
 		case Opt_skip_balance:
@@ -846,18 +864,15 @@ int btrfs_sync_fs(struct super_block *sb, int wait)
 		return 0;
 	}
 
-	btrfs_wait_ordered_extents(root, 0, 0);
-
-	spin_lock(&fs_info->trans_lock);
-	if (!fs_info->running_transaction) {
-		spin_unlock(&fs_info->trans_lock);
-		return 0;
-	}
-	spin_unlock(&fs_info->trans_lock);
+	btrfs_wait_ordered_extents(root, 0);
 
-	trans = btrfs_join_transaction(root);
-	if (IS_ERR(trans))
+	trans = btrfs_attach_transaction(root);
+	if (IS_ERR(trans)) {
+		/* no transaction, don't bother */
+		if (PTR_ERR(trans) == -ENOENT)
+			return 0;
 		return PTR_ERR(trans);
+	}
 	return btrfs_commit_transaction(trans, root);
 }
 
@@ -1508,17 +1523,21 @@ static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
 
 static int btrfs_freeze(struct super_block *sb)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
-	mutex_lock(&fs_info->transaction_kthread_mutex);
-	mutex_lock(&fs_info->cleaner_mutex);
-	return 0;
+	struct btrfs_trans_handle *trans;
+	struct btrfs_root *root = btrfs_sb(sb)->tree_root;
+
+	trans = btrfs_attach_transaction(root);
+	if (IS_ERR(trans)) {
+		/* no transaction, don't bother */
+		if (PTR_ERR(trans) == -ENOENT)
+			return 0;
+		return PTR_ERR(trans);
+	}
+	return btrfs_commit_transaction(trans, root);
 }
 
 static int btrfs_unfreeze(struct super_block *sb)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
-	mutex_unlock(&fs_info->cleaner_mutex);
-	mutex_unlock(&fs_info->transaction_kthread_mutex);
 	return 0;
 }
 
@@ -1595,7 +1614,7 @@ static int btrfs_interface_init(void)
 static void btrfs_interface_exit(void)
 {
 	if (misc_deregister(&btrfs_misc) < 0)
-		printk(KERN_INFO "misc_deregister failed for control device");
+		printk(KERN_INFO "btrfs: misc_deregister failed for control device\n");
 }
 
 static int __init init_btrfs_fs(void)
@@ -1620,10 +1639,14 @@ static int __init init_btrfs_fs(void)
 	if (err)
 		goto free_extent_io;
 
-	err = btrfs_delayed_inode_init();
+	err = ordered_data_init();
 	if (err)
 		goto free_extent_map;
 
+	err = btrfs_delayed_inode_init();
+	if (err)
+		goto free_ordered_data;
+
 	err = btrfs_interface_init();
 	if (err)
 		goto free_delayed_inode;
@@ -1641,6 +1664,8 @@ unregister_ioctl:
 	btrfs_interface_exit();
 free_delayed_inode:
 	btrfs_delayed_inode_exit();
+free_ordered_data:
+	ordered_data_exit();
 free_extent_map:
 	extent_map_exit();
 free_extent_io:
@@ -1657,6 +1682,7 @@ static void __exit exit_btrfs_fs(void)
 {
 	btrfs_destroy_cachep();
 	btrfs_delayed_inode_exit();
+	ordered_data_exit();
 	extent_map_exit();
 	extent_io_exit();
 	btrfs_interface_exit();

fs/btrfs/transaction.c

@@ -53,7 +53,7 @@ static noinline void switch_commit_root(struct btrfs_root *root)
 /*
  * either allocate a new transaction or hop into the existing one
  */
-static noinline int join_transaction(struct btrfs_root *root, int nofail)
+static noinline int join_transaction(struct btrfs_root *root, int type)
 {
 	struct btrfs_transaction *cur_trans;
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -67,7 +67,13 @@ loop:
 	}
 
 	if (fs_info->trans_no_join) {
-		if (!nofail) {
+		/* 
+		 * If we are JOIN_NOLOCK we're already committing a current
+		 * transaction, we just need a handle to deal with something
+		 * when committing the transaction, such as inode cache and
+		 * space cache. It is a special case.
+		 */
+		if (type != TRANS_JOIN_NOLOCK) {
 			spin_unlock(&fs_info->trans_lock);
 			return -EBUSY;
 		}
@@ -87,6 +93,13 @@ loop:
 	}
 	spin_unlock(&fs_info->trans_lock);
 
+	/*
+	 * If we are ATTACH, we just want to catch the current transaction,
+	 * and commit it. If there is no transaction, just return ENOENT.
+	 */
+	if (type == TRANS_ATTACH)
+		return -ENOENT;
+
 	cur_trans = kmem_cache_alloc(btrfs_transaction_cachep, GFP_NOFS);
 	if (!cur_trans)
 		return -ENOMEM;
@@ -267,13 +280,6 @@ static void wait_current_trans(struct btrfs_root *root)
 	}
 }
 
-enum btrfs_trans_type {
-	TRANS_START,
-	TRANS_JOIN,
-	TRANS_USERSPACE,
-	TRANS_JOIN_NOLOCK,
-};
-
 static int may_wait_transaction(struct btrfs_root *root, int type)
 {
 	if (root->fs_info->log_root_recovering)
@@ -290,7 +296,8 @@ static int may_wait_transaction(struct btrfs_root *root, int type)
 }
 
 static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
-						    u64 num_items, int type)
+						    u64 num_items, int type,
+						    int noflush)
 {
 	struct btrfs_trans_handle *h;
 	struct btrfs_transaction *cur_trans;
@@ -324,9 +331,14 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
 		}
 
 		num_bytes = btrfs_calc_trans_metadata_size(root, num_items);
-		ret = btrfs_block_rsv_add(root,
-					  &root->fs_info->trans_block_rsv,
-					  num_bytes);
+		if (noflush)
+			ret = btrfs_block_rsv_add_noflush(root,
+						&root->fs_info->trans_block_rsv,
+						num_bytes);
+		else
+			ret = btrfs_block_rsv_add(root,
+						&root->fs_info->trans_block_rsv,
+						num_bytes);
 		if (ret)
 			return ERR_PTR(ret);
 	}
@@ -335,19 +347,34 @@ again:
 	if (!h)
 		return ERR_PTR(-ENOMEM);
 
-	sb_start_intwrite(root->fs_info->sb);
+	/*
+	 * If we are JOIN_NOLOCK we're already committing a transaction and
+	 * waiting on this guy, so we don't need to do the sb_start_intwrite
+	 * because we're already holding a ref.  We need this because we could
+	 * have raced in and did an fsync() on a file which can kick a commit
+	 * and then we deadlock with somebody doing a freeze.
+	 *
+	 * If we are ATTACH, it means we just want to catch the current
+	 * transaction and commit it, so we needn't do sb_start_intwrite(). 
+	 */
+	if (type < TRANS_JOIN_NOLOCK)
+		sb_start_intwrite(root->fs_info->sb);
 
 	if (may_wait_transaction(root, type))
 		wait_current_trans(root);
 
 	do {
-		ret = join_transaction(root, type == TRANS_JOIN_NOLOCK);
+		ret = join_transaction(root, type);
 		if (ret == -EBUSY)
 			wait_current_trans(root);
 	} while (ret == -EBUSY);
 
 	if (ret < 0) {
-		sb_end_intwrite(root->fs_info->sb);
+		/* We must get the transaction if we are JOIN_NOLOCK. */
+		BUG_ON(type == TRANS_JOIN_NOLOCK);
+
+		if (type < TRANS_JOIN_NOLOCK)
+			sb_end_intwrite(root->fs_info->sb);
 		kmem_cache_free(btrfs_trans_handle_cachep, h);
 		return ERR_PTR(ret);
 	}
@@ -367,7 +394,9 @@ again:
 	h->aborted = 0;
 	h->qgroup_reserved = qgroup_reserved;
 	h->delayed_ref_elem.seq = 0;
+	h->type = type;
 	INIT_LIST_HEAD(&h->qgroup_ref_list);
+	INIT_LIST_HEAD(&h->new_bgs);
 
 	smp_mb();
 	if (cur_trans->blocked && may_wait_transaction(root, type)) {
@@ -393,21 +422,33 @@ got_it:
 struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
 						   int num_items)
 {
-	return start_transaction(root, num_items, TRANS_START);
+	return start_transaction(root, num_items, TRANS_START, 0);
+}
+
+struct btrfs_trans_handle *btrfs_start_transaction_noflush(
+					struct btrfs_root *root, int num_items)
+{
+	return start_transaction(root, num_items, TRANS_START, 1);
 }
+
 struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root)
 {
-	return start_transaction(root, 0, TRANS_JOIN);
+	return start_transaction(root, 0, TRANS_JOIN, 0);
 }
 
 struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root)
 {
-	return start_transaction(root, 0, TRANS_JOIN_NOLOCK);
+	return start_transaction(root, 0, TRANS_JOIN_NOLOCK, 0);
 }
 
 struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *root)
 {
-	return start_transaction(root, 0, TRANS_USERSPACE);
+	return start_transaction(root, 0, TRANS_USERSPACE, 0);
+}
+
+struct btrfs_trans_handle *btrfs_attach_transaction(struct btrfs_root *root)
+{
+	return start_transaction(root, 0, TRANS_ATTACH, 0);
 }
 
 /* wait for a transaction commit to be fully complete */
@@ -506,11 +547,12 @@ int btrfs_should_end_transaction(struct btrfs_trans_handle *trans,
 }
 
 static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
-			  struct btrfs_root *root, int throttle, int lock)
+			  struct btrfs_root *root, int throttle)
 {
 	struct btrfs_transaction *cur_trans = trans->transaction;
 	struct btrfs_fs_info *info = root->fs_info;
 	int count = 0;
+	int lock = (trans->type != TRANS_JOIN_NOLOCK);
 	int err = 0;
 
 	if (--trans->use_count) {
@@ -536,6 +578,9 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 		trans->qgroup_reserved = 0;
 	}
 
+	if (!list_empty(&trans->new_bgs))
+		btrfs_create_pending_block_groups(trans, root);
+
 	while (count < 2) {
 		unsigned long cur = trans->delayed_ref_updates;
 		trans->delayed_ref_updates = 0;
@@ -551,7 +596,8 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	btrfs_trans_release_metadata(trans, root);
 	trans->block_rsv = NULL;
 
-	sb_end_intwrite(root->fs_info->sb);
+	if (!list_empty(&trans->new_bgs))
+		btrfs_create_pending_block_groups(trans, root);
 
 	if (lock && !atomic_read(&root->fs_info->open_ioctl_trans) &&
 	    should_end_transaction(trans, root)) {
@@ -573,6 +619,9 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 		}
 	}
 
+	if (trans->type < TRANS_JOIN_NOLOCK)
+		sb_end_intwrite(root->fs_info->sb);
+
 	WARN_ON(cur_trans != info->running_transaction);
 	WARN_ON(atomic_read(&cur_trans->num_writers) < 1);
 	atomic_dec(&cur_trans->num_writers);
@@ -604,7 +653,7 @@ int btrfs_end_transaction(struct btrfs_trans_handle *trans,
 {
 	int ret;
 
-	ret = __btrfs_end_transaction(trans, root, 0, 1);
+	ret = __btrfs_end_transaction(trans, root, 0);
 	if (ret)
 		return ret;
 	return 0;
@@ -615,18 +664,7 @@ int btrfs_end_transaction_throttle(struct btrfs_trans_handle *trans,
 {
 	int ret;
 
-	ret = __btrfs_end_transaction(trans, root, 1, 1);
-	if (ret)
-		return ret;
-	return 0;
-}
-
-int btrfs_end_transaction_nolock(struct btrfs_trans_handle *trans,
-				 struct btrfs_root *root)
-{
-	int ret;
-
-	ret = __btrfs_end_transaction(trans, root, 0, 0);
+	ret = __btrfs_end_transaction(trans, root, 1);
 	if (ret)
 		return ret;
 	return 0;
@@ -635,7 +673,7 @@ int btrfs_end_transaction_nolock(struct btrfs_trans_handle *trans,
 int btrfs_end_transaction_dmeta(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root)
 {
-	return __btrfs_end_transaction(trans, root, 1, 1);
+	return __btrfs_end_transaction(trans, root, 1);
 }
 
 /*
@@ -649,13 +687,15 @@ int btrfs_write_marked_extents(struct btrfs_root *root,
 	int err = 0;
 	int werr = 0;
 	struct address_space *mapping = root->fs_info->btree_inode->i_mapping;
+	struct extent_state *cached_state = NULL;
 	u64 start = 0;
 	u64 end;
 
 	while (!find_first_extent_bit(dirty_pages, start, &start, &end,
-				      mark)) {
-		convert_extent_bit(dirty_pages, start, end, EXTENT_NEED_WAIT, mark,
-				   GFP_NOFS);
+				      mark, &cached_state)) {
+		convert_extent_bit(dirty_pages, start, end, EXTENT_NEED_WAIT,
+				   mark, &cached_state, GFP_NOFS);
+		cached_state = NULL;
 		err = filemap_fdatawrite_range(mapping, start, end);
 		if (err)
 			werr = err;
@@ -679,12 +719,14 @@ int btrfs_wait_marked_extents(struct btrfs_root *root,
 	int err = 0;
 	int werr = 0;
 	struct address_space *mapping = root->fs_info->btree_inode->i_mapping;
+	struct extent_state *cached_state = NULL;
 	u64 start = 0;
 	u64 end;
 
 	while (!find_first_extent_bit(dirty_pages, start, &start, &end,
-				      EXTENT_NEED_WAIT)) {
-		clear_extent_bits(dirty_pages, start, end, EXTENT_NEED_WAIT, GFP_NOFS);
+				      EXTENT_NEED_WAIT, &cached_state)) {
+		clear_extent_bit(dirty_pages, start, end, EXTENT_NEED_WAIT,
+				 0, 0, &cached_state, GFP_NOFS);
 		err = filemap_fdatawait_range(mapping, start, end);
 		if (err)
 			werr = err;
@@ -955,6 +997,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	struct btrfs_root *parent_root;
 	struct btrfs_block_rsv *rsv;
 	struct inode *parent_inode;
+	struct btrfs_path *path;
+	struct btrfs_dir_item *dir_item;
 	struct dentry *parent;
 	struct dentry *dentry;
 	struct extent_buffer *tmp;
@@ -967,18 +1011,22 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	u64 root_flags;
 	uuid_le new_uuid;
 
-	rsv = trans->block_rsv;
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = pending->error = -ENOMEM;
+		goto path_alloc_fail;
+	}
 
 	new_root_item = kmalloc(sizeof(*new_root_item), GFP_NOFS);
 	if (!new_root_item) {
 		ret = pending->error = -ENOMEM;
-		goto fail;
+		goto root_item_alloc_fail;
 	}
 
 	ret = btrfs_find_free_objectid(tree_root, &objectid);
 	if (ret) {
 		pending->error = ret;
-		goto fail;
+		goto no_free_objectid;
 	}
 
 	btrfs_reloc_pre_snapshot(trans, pending, &to_reserve);
@@ -988,22 +1036,22 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 						  to_reserve);
 		if (ret) {
 			pending->error = ret;
-			goto fail;
+			goto no_free_objectid;
 		}
 	}
 
 	ret = btrfs_qgroup_inherit(trans, fs_info, root->root_key.objectid,
 				   objectid, pending->inherit);
-	kfree(pending->inherit);
 	if (ret) {
 		pending->error = ret;
-		goto fail;
+		goto no_free_objectid;
 	}
 
 	key.objectid = objectid;
 	key.offset = (u64)-1;
 	key.type = BTRFS_ROOT_ITEM_KEY;
 
+	rsv = trans->block_rsv;
 	trans->block_rsv = &pending->block_rsv;
 
 	dentry = pending->dentry;
@@ -1017,24 +1065,21 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	 */
 	ret = btrfs_set_inode_index(parent_inode, &index);
 	BUG_ON(ret); /* -ENOMEM */
-	ret = btrfs_insert_dir_item(trans, parent_root,
-				dentry->d_name.name, dentry->d_name.len,
-				parent_inode, &key,
-				BTRFS_FT_DIR, index);
-	if (ret == -EEXIST) {
+
+	/* check if there is a file/dir which has the same name. */
+	dir_item = btrfs_lookup_dir_item(NULL, parent_root, path,
+					 btrfs_ino(parent_inode),
+					 dentry->d_name.name,
+					 dentry->d_name.len, 0);
+	if (dir_item != NULL && !IS_ERR(dir_item)) {
 		pending->error = -EEXIST;
-		dput(parent);
 		goto fail;
-	} else if (ret) {
-		goto abort_trans_dput;
+	} else if (IS_ERR(dir_item)) {
+		ret = PTR_ERR(dir_item);
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
 	}
-
-	btrfs_i_size_write(parent_inode, parent_inode->i_size +
-					 dentry->d_name.len * 2);
-	parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
-	ret = btrfs_update_inode(trans, parent_root, parent_inode);
-	if (ret)
-		goto abort_trans_dput;
+	btrfs_release_path(path);
 
 	/*
 	 * pull in the delayed directory update
@@ -1043,8 +1088,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	 * snapshot
 	 */
 	ret = btrfs_run_delayed_items(trans, root);
-	if (ret) { /* Transaction aborted */
-		dput(parent);
+	if (ret) {	/* Transaction aborted */
+		btrfs_abort_transaction(trans, root, ret);
 		goto fail;
 	}
 
@@ -1079,7 +1124,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	if (ret) {
 		btrfs_tree_unlock(old);
 		free_extent_buffer(old);
-		goto abort_trans_dput;
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
 	}
 
 	btrfs_set_lock_blocking(old);
@@ -1088,8 +1134,10 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	/* clean up in any case */
 	btrfs_tree_unlock(old);
 	free_extent_buffer(old);
-	if (ret)
-		goto abort_trans_dput;
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
 
 	/* see comments in should_cow_block() */
 	root->force_cow = 1;
@@ -1101,8 +1149,10 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	ret = btrfs_insert_root(trans, tree_root, &key, new_root_item);
 	btrfs_tree_unlock(tmp);
 	free_extent_buffer(tmp);
-	if (ret)
-		goto abort_trans_dput;
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
 
 	/*
 	 * insert root back/forward references
@@ -1111,32 +1161,58 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 				 parent_root->root_key.objectid,
 				 btrfs_ino(parent_inode), index,
 				 dentry->d_name.name, dentry->d_name.len);
-	dput(parent);
-	if (ret)
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
 		goto fail;
+	}
 
 	key.offset = (u64)-1;
 	pending->snap = btrfs_read_fs_root_no_name(root->fs_info, &key);
 	if (IS_ERR(pending->snap)) {
 		ret = PTR_ERR(pending->snap);
-		goto abort_trans;
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
 	}
 
 	ret = btrfs_reloc_post_snapshot(trans, pending);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
+
+	ret = btrfs_run_delayed_refs(trans, root, (unsigned long)-1);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
+
+	ret = btrfs_insert_dir_item(trans, parent_root,
+				    dentry->d_name.name, dentry->d_name.len,
+				    parent_inode, &key,
+				    BTRFS_FT_DIR, index);
+	/* We have check then name at the beginning, so it is impossible. */
+	BUG_ON(ret == -EEXIST);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
+
+	btrfs_i_size_write(parent_inode, parent_inode->i_size +
+					 dentry->d_name.len * 2);
+	parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
+	ret = btrfs_update_inode(trans, parent_root, parent_inode);
 	if (ret)
-		goto abort_trans;
-	ret = 0;
+		btrfs_abort_transaction(trans, root, ret);
 fail:
-	kfree(new_root_item);
+	dput(parent);
 	trans->block_rsv = rsv;
+no_free_objectid:
+	kfree(new_root_item);
+root_item_alloc_fail:
+	btrfs_free_path(path);
+path_alloc_fail:
 	btrfs_block_rsv_release(root, &pending->block_rsv, (u64)-1);
 	return ret;
-
-abort_trans_dput:
-	dput(parent);
-abort_trans:
-	btrfs_abort_transaction(trans, root, ret);
-	goto fail;
 }
 
 /*
@@ -1229,6 +1305,16 @@ static void do_async_commit(struct work_struct *work)
 	struct btrfs_async_commit *ac =
 		container_of(work, struct btrfs_async_commit, work.work);
 
+	/*
+	 * We've got freeze protection passed with the transaction.
+	 * Tell lockdep about it.
+	 */
+	rwsem_acquire_read(
+		&ac->root->fs_info->sb->s_writers.lock_map[SB_FREEZE_FS-1],
+		0, 1, _THIS_IP_);
+
+	current->journal_info = ac->newtrans;
+
 	btrfs_commit_transaction(ac->newtrans, ac->root);
 	kfree(ac);
 }
@@ -1258,6 +1344,14 @@ int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
 	atomic_inc(&cur_trans->use_count);
 
 	btrfs_end_transaction(trans, root);
+
+	/*
+	 * Tell lockdep we've released the freeze rwsem, since the
+	 * async commit thread will be the one to unlock it.
+	 */
+	rwsem_release(&root->fs_info->sb->s_writers.lock_map[SB_FREEZE_FS-1],
+		      1, _THIS_IP_);
+
 	schedule_delayed_work(&ac->work, 0);
 
 	/* wait for transaction to start and unblock */
@@ -1348,6 +1442,9 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	 */
 	cur_trans->delayed_refs.flushing = 1;
 
+	if (!list_empty(&trans->new_bgs))
+		btrfs_create_pending_block_groups(trans, root);
+
 	ret = btrfs_run_delayed_refs(trans, root, 0);
 	if (ret)
 		goto cleanup_transaction;
@@ -1403,7 +1500,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
 		if (flush_on_commit || snap_pending) {
 			btrfs_start_delalloc_inodes(root, 1);
-			btrfs_wait_ordered_extents(root, 0, 1);
+			btrfs_wait_ordered_extents(root, 1);
 		}
 
 		ret = btrfs_run_delayed_items(trans, root);
@@ -1456,13 +1553,28 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	 */
 	mutex_lock(&root->fs_info->reloc_mutex);
 
-	ret = btrfs_run_delayed_items(trans, root);
+	/*
+	 * We needn't worry about the delayed items because we will
+	 * deal with them in create_pending_snapshot(), which is the
+	 * core function of the snapshot creation.
+	 */
+	ret = create_pending_snapshots(trans, root->fs_info);
 	if (ret) {
 		mutex_unlock(&root->fs_info->reloc_mutex);
 		goto cleanup_transaction;
 	}
 
-	ret = create_pending_snapshots(trans, root->fs_info);
+	/*
+	 * We insert the dir indexes of the snapshots and update the inode
+	 * of the snapshots' parents after the snapshot creation, so there
+	 * are some delayed items which are not dealt with. Now deal with
+	 * them.
+	 *
+	 * We needn't worry that this operation will corrupt the snapshots,
+	 * because all the tree which are snapshoted will be forced to COW
+	 * the nodes and leaves.
+	 */
+	ret = btrfs_run_delayed_items(trans, root);
 	if (ret) {
 		mutex_unlock(&root->fs_info->reloc_mutex);
 		goto cleanup_transaction;
@@ -1584,7 +1696,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	put_transaction(cur_trans);
 	put_transaction(cur_trans);
 
-	sb_end_intwrite(root->fs_info->sb);
+	if (trans->type < TRANS_JOIN_NOLOCK)
+		sb_end_intwrite(root->fs_info->sb);
 
 	trace_btrfs_transaction_commit(root);
 

fs/btrfs/transaction.h

@@ -47,6 +47,14 @@ struct btrfs_transaction {
 	int aborted;
 };
 
+enum btrfs_trans_type {
+	TRANS_START,
+	TRANS_JOIN,
+	TRANS_USERSPACE,
+	TRANS_JOIN_NOLOCK,
+	TRANS_ATTACH,
+};
+
 struct btrfs_trans_handle {
 	u64 transid;
 	u64 bytes_reserved;
@@ -58,8 +66,9 @@ struct btrfs_trans_handle {
 	struct btrfs_transaction *transaction;
 	struct btrfs_block_rsv *block_rsv;
 	struct btrfs_block_rsv *orig_rsv;
-	int aborted;
-	int adding_csums;
+	short aborted;
+	short adding_csums;
+	enum btrfs_trans_type type;
 	/*
 	 * this root is only needed to validate that the root passed to
 	 * start_transaction is the same as the one passed to end_transaction.
@@ -68,6 +77,7 @@ struct btrfs_trans_handle {
 	struct btrfs_root *root;
 	struct seq_list delayed_ref_elem;
 	struct list_head qgroup_ref_list;
+	struct list_head new_bgs;
 };
 
 struct btrfs_pending_snapshot {
@@ -88,16 +98,18 @@ static inline void btrfs_set_inode_last_trans(struct btrfs_trans_handle *trans,
 {
 	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;
 }
 
 int btrfs_end_transaction(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root);
-int btrfs_end_transaction_nolock(struct btrfs_trans_handle *trans,
-				 struct btrfs_root *root);
 struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
 						   int num_items);
+struct btrfs_trans_handle *btrfs_start_transaction_noflush(
+					struct btrfs_root *root, int num_items);
 struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root);
 struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root);
+struct btrfs_trans_handle *btrfs_attach_transaction(struct btrfs_root *root);
 struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *root);
 int btrfs_wait_for_commit(struct btrfs_root *root, u64 transid);
 int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,

fs/btrfs/tree-log.c

@@ -18,13 +18,16 @@
 
 #include <linux/sched.h>
 #include <linux/slab.h>
+#include <linux/list_sort.h>
 #include "ctree.h"
 #include "transaction.h"
 #include "disk-io.h"
 #include "locking.h"
 #include "print-tree.h"
+#include "backref.h"
 #include "compat.h"
 #include "tree-log.h"
+#include "hash.h"
 
 /* magic values for the inode_only field in btrfs_log_inode:
  *
@@ -146,7 +149,7 @@ static int start_log_trans(struct btrfs_trans_handle *trans,
 			root->log_multiple_pids = true;
 		}
 
-		root->log_batch++;
+		atomic_inc(&root->log_batch);
 		atomic_inc(&root->log_writers);
 		mutex_unlock(&root->log_mutex);
 		return 0;
@@ -165,7 +168,7 @@ static int start_log_trans(struct btrfs_trans_handle *trans,
 			err = ret;
 	}
 	mutex_unlock(&root->fs_info->tree_log_mutex);
-	root->log_batch++;
+	atomic_inc(&root->log_batch);
 	atomic_inc(&root->log_writers);
 	mutex_unlock(&root->log_mutex);
 	return err;
@@ -484,7 +487,6 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 	int found_type;
 	u64 mask = root->sectorsize - 1;
 	u64 extent_end;
-	u64 alloc_hint;
 	u64 start = key->offset;
 	u64 saved_nbytes;
 	struct btrfs_file_extent_item *item;
@@ -550,8 +552,7 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 
 	saved_nbytes = inode_get_bytes(inode);
 	/* drop any overlapping extents */
-	ret = btrfs_drop_extents(trans, inode, start, extent_end,
-				 &alloc_hint, 1);
+	ret = btrfs_drop_extents(trans, root, inode, start, extent_end, 1);
 	BUG_ON(ret);
 
 	if (found_type == BTRFS_FILE_EXTENT_REG ||
@@ -744,6 +745,7 @@ out:
  */
 static noinline int backref_in_log(struct btrfs_root *log,
 				   struct btrfs_key *key,
+				   u64 ref_objectid,
 				   char *name, int namelen)
 {
 	struct btrfs_path *path;
@@ -764,8 +766,17 @@ static noinline int backref_in_log(struct btrfs_root *log,
 	if (ret != 0)
 		goto out;
 
-	item_size = btrfs_item_size_nr(path->nodes[0], path->slots[0]);
 	ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]);
+
+	if (key->type == BTRFS_INODE_EXTREF_KEY) {
+		if (btrfs_find_name_in_ext_backref(path, ref_objectid,
+						   name, namelen, NULL))
+			match = 1;
+
+		goto out;
+	}
+
+	item_size = btrfs_item_size_nr(path->nodes[0], path->slots[0]);
 	ptr_end = ptr + item_size;
 	while (ptr < ptr_end) {
 		ref = (struct btrfs_inode_ref *)ptr;
@@ -786,91 +797,42 @@ out:
 	return match;
 }
 
-
-/*
- * replay one inode back reference item found in the log tree.
- * eb, slot and key refer to the buffer and key found in the log tree.
- * root is the destination we are replaying into, and path is for temp
- * use by this function.  (it should be released on return).
- */
-static noinline int add_inode_ref(struct btrfs_trans_handle *trans,
+static inline int __add_inode_ref(struct btrfs_trans_handle *trans,
 				  struct btrfs_root *root,
-				  struct btrfs_root *log,
 				  struct btrfs_path *path,
-				  struct extent_buffer *eb, int slot,
-				  struct btrfs_key *key)
+				  struct btrfs_root *log_root,
+				  struct inode *dir, struct inode *inode,
+				  struct extent_buffer *eb,
+				  u64 inode_objectid, u64 parent_objectid,
+				  u64 ref_index, char *name, int namelen,
+				  int *search_done)
 {
-	struct btrfs_inode_ref *ref;
-	struct btrfs_dir_item *di;
-	struct inode *dir;
-	struct inode *inode;
-	unsigned long ref_ptr;
-	unsigned long ref_end;
-	char *name;
-	int namelen;
 	int ret;
-	int search_done = 0;
-
-	/*
-	 * it is possible that we didn't log all the parent directories
-	 * for a given inode.  If we don't find the dir, just don't
-	 * copy the back ref in.  The link count fixup code will take
-	 * care of the rest
-	 */
-	dir = read_one_inode(root, key->offset);
-	if (!dir)
-		return -ENOENT;
-
-	inode = read_one_inode(root, key->objectid);
-	if (!inode) {
-		iput(dir);
-		return -EIO;
-	}
-
-	ref_ptr = btrfs_item_ptr_offset(eb, slot);
-	ref_end = ref_ptr + btrfs_item_size_nr(eb, slot);
+	char *victim_name;
+	int victim_name_len;
+	struct extent_buffer *leaf;
+	struct btrfs_dir_item *di;
+	struct btrfs_key search_key;
+	struct btrfs_inode_extref *extref;
 
 again:
-	ref = (struct btrfs_inode_ref *)ref_ptr;
-
-	namelen = btrfs_inode_ref_name_len(eb, ref);
-	name = kmalloc(namelen, GFP_NOFS);
-	BUG_ON(!name);
-
-	read_extent_buffer(eb, name, (unsigned long)(ref + 1), namelen);
-
-	/* if we already have a perfect match, we're done */
-	if (inode_in_dir(root, path, btrfs_ino(dir), btrfs_ino(inode),
-			 btrfs_inode_ref_index(eb, ref),
-			 name, namelen)) {
-		goto out;
-	}
-
-	/*
-	 * look for a conflicting back reference in the metadata.
-	 * if we find one we have to unlink that name of the file
-	 * before we add our new link.  Later on, we overwrite any
-	 * existing back reference, and we don't want to create
-	 * dangling pointers in the directory.
-	 */
-
-	if (search_done)
-		goto insert;
-
-	ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
+	/* Search old style refs */
+	search_key.objectid = inode_objectid;
+	search_key.type = BTRFS_INODE_REF_KEY;
+	search_key.offset = parent_objectid;
+	ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0);
 	if (ret == 0) {
-		char *victim_name;
-		int victim_name_len;
 		struct btrfs_inode_ref *victim_ref;
 		unsigned long ptr;
 		unsigned long ptr_end;
-		struct extent_buffer *leaf = path->nodes[0];
+
+		leaf = path->nodes[0];
 
 		/* are we trying to overwrite a back ref for the root directory
 		 * if so, just jump out, we're done
 		 */
-		if (key->objectid == key->offset)
-			goto out_nowrite;
+		if (search_key.objectid == search_key.offset)
+			return 1;
 
 		/* check all the names in this back reference to see
 		 * if they are in the log.  if so, we allow them to stay
@@ -889,7 +851,9 @@ again:
 					   (unsigned long)(victim_ref + 1),
 					   victim_name_len);
 
-			if (!backref_in_log(log, key, victim_name,
+			if (!backref_in_log(log_root, &search_key,
+					    parent_objectid,
+					    victim_name,
 					    victim_name_len)) {
 				btrfs_inc_nlink(inode);
 				btrfs_release_path(path);
@@ -897,9 +861,14 @@ again:
 				ret = btrfs_unlink_inode(trans, root, dir,
 							 inode, victim_name,
 							 victim_name_len);
+				BUG_ON(ret);
 				btrfs_run_delayed_items(trans, root);
+				kfree(victim_name);
+				*search_done = 1;
+				goto again;
 			}
 			kfree(victim_name);
+
 			ptr = (unsigned long)(victim_ref + 1) + victim_name_len;
 		}
 		BUG_ON(ret);
@@ -908,14 +877,78 @@ again:
 		 * NOTE: we have searched root tree and checked the
 		 * coresponding ref, it does not need to check again.
 		 */
-		search_done = 1;
+		*search_done = 1;
+	}
+	btrfs_release_path(path);
+
+	/* Same search but for extended refs */
+	extref = btrfs_lookup_inode_extref(NULL, root, path, name, namelen,
+					   inode_objectid, parent_objectid, 0,
+					   0);
+	if (!IS_ERR_OR_NULL(extref)) {
+		u32 item_size;
+		u32 cur_offset = 0;
+		unsigned long base;
+		struct inode *victim_parent;
+
+		leaf = path->nodes[0];
+
+		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+		base = btrfs_item_ptr_offset(leaf, path->slots[0]);
+
+		while (cur_offset < item_size) {
+			extref = (struct btrfs_inode_extref *)base + cur_offset;
+
+			victim_name_len = btrfs_inode_extref_name_len(leaf, extref);
+
+			if (btrfs_inode_extref_parent(leaf, extref) != parent_objectid)
+				goto next;
+
+			victim_name = kmalloc(victim_name_len, GFP_NOFS);
+			read_extent_buffer(leaf, victim_name, (unsigned long)&extref->name,
+					   victim_name_len);
+
+			search_key.objectid = inode_objectid;
+			search_key.type = BTRFS_INODE_EXTREF_KEY;
+			search_key.offset = btrfs_extref_hash(parent_objectid,
+							      victim_name,
+							      victim_name_len);
+			ret = 0;
+			if (!backref_in_log(log_root, &search_key,
+					    parent_objectid, victim_name,
+					    victim_name_len)) {
+				ret = -ENOENT;
+				victim_parent = read_one_inode(root,
+							       parent_objectid);
+				if (victim_parent) {
+					btrfs_inc_nlink(inode);
+					btrfs_release_path(path);
+
+					ret = btrfs_unlink_inode(trans, root,
+								 victim_parent,
+								 inode,
+								 victim_name,
+								 victim_name_len);
+					btrfs_run_delayed_items(trans, root);
+				}
+				BUG_ON(ret);
+				iput(victim_parent);
+				kfree(victim_name);
+				*search_done = 1;
+				goto again;
+			}
+			kfree(victim_name);
+			BUG_ON(ret);
+next:
+			cur_offset += victim_name_len + sizeof(*extref);
+		}
+		*search_done = 1;
 	}
 	btrfs_release_path(path);
 
 	/* look for a conflicting sequence number */
 	di = btrfs_lookup_dir_index_item(trans, root, path, btrfs_ino(dir),
-					 btrfs_inode_ref_index(eb, ref),
-					 name, namelen, 0);
+					 ref_index, name, namelen, 0);
 	if (di && !IS_ERR(di)) {
 		ret = drop_one_dir_item(trans, root, path, dir, di);
 		BUG_ON(ret);
@@ -931,25 +964,173 @@ again:
 	}
 	btrfs_release_path(path);
 
-insert:
-	/* insert our name */
-	ret = btrfs_add_link(trans, dir, inode, name, namelen, 0,
-			     btrfs_inode_ref_index(eb, ref));
-	BUG_ON(ret);
+	return 0;
+}
 
-	btrfs_update_inode(trans, root, inode);
+static int extref_get_fields(struct extent_buffer *eb, unsigned long ref_ptr,
+			     u32 *namelen, char **name, u64 *index,
+			     u64 *parent_objectid)
+{
+	struct btrfs_inode_extref *extref;
 
-out:
-	ref_ptr = (unsigned long)(ref + 1) + namelen;
-	kfree(name);
-	if (ref_ptr < ref_end)
-		goto again;
+	extref = (struct btrfs_inode_extref *)ref_ptr;
+
+	*namelen = btrfs_inode_extref_name_len(eb, extref);
+	*name = kmalloc(*namelen, GFP_NOFS);
+	if (*name == NULL)
+		return -ENOMEM;
+
+	read_extent_buffer(eb, *name, (unsigned long)&extref->name,
+			   *namelen);
+
+	*index = btrfs_inode_extref_index(eb, extref);
+	if (parent_objectid)
+		*parent_objectid = btrfs_inode_extref_parent(eb, extref);
+
+	return 0;
+}
+
+static int ref_get_fields(struct extent_buffer *eb, unsigned long ref_ptr,
+			  u32 *namelen, char **name, u64 *index)
+{
+	struct btrfs_inode_ref *ref;
+
+	ref = (struct btrfs_inode_ref *)ref_ptr;
+
+	*namelen = btrfs_inode_ref_name_len(eb, ref);
+	*name = kmalloc(*namelen, GFP_NOFS);
+	if (*name == NULL)
+		return -ENOMEM;
+
+	read_extent_buffer(eb, *name, (unsigned long)(ref + 1), *namelen);
+
+	*index = btrfs_inode_ref_index(eb, ref);
+
+	return 0;
+}
+
+/*
+ * replay one inode back reference item found in the log tree.
+ * eb, slot and key refer to the buffer and key found in the log tree.
+ * root is the destination we are replaying into, and path is for temp
+ * use by this function.  (it should be released on return).
+ */
+static noinline int add_inode_ref(struct btrfs_trans_handle *trans,
+				  struct btrfs_root *root,
+				  struct btrfs_root *log,
+				  struct btrfs_path *path,
+				  struct extent_buffer *eb, int slot,
+				  struct btrfs_key *key)
+{
+	struct inode *dir;
+	struct inode *inode;
+	unsigned long ref_ptr;
+	unsigned long ref_end;
+	char *name;
+	int namelen;
+	int ret;
+	int search_done = 0;
+	int log_ref_ver = 0;
+	u64 parent_objectid;
+	u64 inode_objectid;
+	u64 ref_index = 0;
+	int ref_struct_size;
+
+	ref_ptr = btrfs_item_ptr_offset(eb, slot);
+	ref_end = ref_ptr + btrfs_item_size_nr(eb, slot);
+
+	if (key->type == BTRFS_INODE_EXTREF_KEY) {
+		struct btrfs_inode_extref *r;
+
+		ref_struct_size = sizeof(struct btrfs_inode_extref);
+		log_ref_ver = 1;
+		r = (struct btrfs_inode_extref *)ref_ptr;
+		parent_objectid = btrfs_inode_extref_parent(eb, r);
+	} else {
+		ref_struct_size = sizeof(struct btrfs_inode_ref);
+		parent_objectid = key->offset;
+	}
+	inode_objectid = key->objectid;
+
+	/*
+	 * it is possible that we didn't log all the parent directories
+	 * for a given inode.  If we don't find the dir, just don't
+	 * copy the back ref in.  The link count fixup code will take
+	 * care of the rest
+	 */
+	dir = read_one_inode(root, parent_objectid);
+	if (!dir)
+		return -ENOENT;
+
+	inode = read_one_inode(root, inode_objectid);
+	if (!inode) {
+		iput(dir);
+		return -EIO;
+	}
+
+	while (ref_ptr < ref_end) {
+		if (log_ref_ver) {
+			ret = extref_get_fields(eb, ref_ptr, &namelen, &name,
+						&ref_index, &parent_objectid);
+			/*
+			 * parent object can change from one array
+			 * item to another.
+			 */
+			if (!dir)
+				dir = read_one_inode(root, parent_objectid);
+			if (!dir)
+				return -ENOENT;
+		} else {
+			ret = ref_get_fields(eb, ref_ptr, &namelen, &name,
+					     &ref_index);
+		}
+		if (ret)
+			return ret;
+
+		/* if we already have a perfect match, we're done */
+		if (!inode_in_dir(root, path, btrfs_ino(dir), btrfs_ino(inode),
+				  ref_index, name, namelen)) {
+			/*
+			 * look for a conflicting back reference in the
+			 * metadata. if we find one we have to unlink that name
+			 * of the file before we add our new link.  Later on, we
+			 * overwrite any existing back reference, and we don't
+			 * want to create dangling pointers in the directory.
+			 */
+
+			if (!search_done) {
+				ret = __add_inode_ref(trans, root, path, log,
+						      dir, inode, eb,
+						      inode_objectid,
+						      parent_objectid,
+						      ref_index, name, namelen,
+						      &search_done);
+				if (ret == 1)
+					goto out;
+				BUG_ON(ret);
+			}
+
+			/* insert our name */
+			ret = btrfs_add_link(trans, dir, inode, name, namelen,
+					     0, ref_index);
+			BUG_ON(ret);
+
+			btrfs_update_inode(trans, root, inode);
+		}
+
+		ref_ptr = (unsigned long)(ref_ptr + ref_struct_size) + namelen;
+		kfree(name);
+		if (log_ref_ver) {
+			iput(dir);
+			dir = NULL;
+		}
+	}
 
 	/* finally write the back reference in the inode */
 	ret = overwrite_item(trans, root, path, eb, slot, key);
 	BUG_ON(ret);
 
-out_nowrite:
+out:
 	btrfs_release_path(path);
 	iput(dir);
 	iput(inode);
@@ -966,25 +1147,55 @@ static int insert_orphan_item(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
+static int count_inode_extrefs(struct btrfs_root *root,
+			       struct inode *inode, struct btrfs_path *path)
+{
+	int ret = 0;
+	int name_len;
+	unsigned int nlink = 0;
+	u32 item_size;
+	u32 cur_offset = 0;
+	u64 inode_objectid = btrfs_ino(inode);
+	u64 offset = 0;
+	unsigned long ptr;
+	struct btrfs_inode_extref *extref;
+	struct extent_buffer *leaf;
 
-/*
- * There are a few corners where the link count of the file can't
- * be properly maintained during replay.  So, instead of adding
- * lots of complexity to the log code, we just scan the backrefs
- * for any file that has been through replay.
- *
- * The scan will update the link count on the inode to reflect the
- * number of back refs found.  If it goes down to zero, the iput
- * will free the inode.
- */
-static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
-					   struct btrfs_root *root,
-					   struct inode *inode)
+	while (1) {
+		ret = btrfs_find_one_extref(root, inode_objectid, offset, path,
+					    &extref, &offset);
+		if (ret)
+			break;
+
+		leaf = path->nodes[0];
+		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+		ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
+
+		while (cur_offset < item_size) {
+			extref = (struct btrfs_inode_extref *) (ptr + cur_offset);
+			name_len = btrfs_inode_extref_name_len(leaf, extref);
+
+			nlink++;
+
+			cur_offset += name_len + sizeof(*extref);
+		}
+
+		offset++;
+		btrfs_release_path(path);
+	}
+	btrfs_release_path(path);
+
+	if (ret < 0)
+		return ret;
+	return nlink;
+}
+
+static int count_inode_refs(struct btrfs_root *root,
+			       struct inode *inode, struct btrfs_path *path)
 {
-	struct btrfs_path *path;
 	int ret;
 	struct btrfs_key key;
-	u64 nlink = 0;
+	unsigned int nlink = 0;
 	unsigned long ptr;
 	unsigned long ptr_end;
 	int name_len;
@@ -994,10 +1205,6 @@ static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
 	key.type = BTRFS_INODE_REF_KEY;
 	key.offset = (u64)-1;
 
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
 	while (1) {
 		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 		if (ret < 0)
@@ -1031,6 +1238,50 @@ static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
 		btrfs_release_path(path);
 	}
 	btrfs_release_path(path);
+
+	return nlink;
+}
+
+/*
+ * There are a few corners where the link count of the file can't
+ * be properly maintained during replay.  So, instead of adding
+ * lots of complexity to the log code, we just scan the backrefs
+ * for any file that has been through replay.
+ *
+ * The scan will update the link count on the inode to reflect the
+ * number of back refs found.  If it goes down to zero, the iput
+ * will free the inode.
+ */
+static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
+					   struct btrfs_root *root,
+					   struct inode *inode)
+{
+	struct btrfs_path *path;
+	int ret;
+	u64 nlink = 0;
+	u64 ino = btrfs_ino(inode);
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	ret = count_inode_refs(root, inode, path);
+	if (ret < 0)
+		goto out;
+
+	nlink = ret;
+
+	ret = count_inode_extrefs(root, inode, path);
+	if (ret == -ENOENT)
+		ret = 0;
+
+	if (ret < 0)
+		goto out;
+
+	nlink += ret;
+
+	ret = 0;
+
 	if (nlink != inode->i_nlink) {
 		set_nlink(inode, nlink);
 		btrfs_update_inode(trans, root, inode);
@@ -1046,9 +1297,10 @@ static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
 		ret = insert_orphan_item(trans, root, ino);
 		BUG_ON(ret);
 	}
-	btrfs_free_path(path);
 
-	return 0;
+out:
+	btrfs_free_path(path);
+	return ret;
 }
 
 static noinline int fixup_inode_link_counts(struct btrfs_trans_handle *trans,
@@ -1695,6 +1947,10 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
 			ret = add_inode_ref(wc->trans, root, log, path,
 					    eb, i, &key);
 			BUG_ON(ret && ret != -ENOENT);
+		} else if (key.type == BTRFS_INODE_EXTREF_KEY) {
+			ret = add_inode_ref(wc->trans, root, log, path,
+					    eb, i, &key);
+			BUG_ON(ret && ret != -ENOENT);
 		} else if (key.type == BTRFS_EXTENT_DATA_KEY) {
 			ret = replay_one_extent(wc->trans, root, path,
 						eb, i, &key);
@@ -2037,7 +2293,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	if (atomic_read(&root->log_commit[(index1 + 1) % 2]))
 		wait_log_commit(trans, root, root->log_transid - 1);
 	while (1) {
-		unsigned long batch = root->log_batch;
+		int batch = atomic_read(&root->log_batch);
 		/* when we're on an ssd, just kick the log commit out */
 		if (!btrfs_test_opt(root, SSD) && root->log_multiple_pids) {
 			mutex_unlock(&root->log_mutex);
@@ -2045,7 +2301,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 			mutex_lock(&root->log_mutex);
 		}
 		wait_for_writer(trans, root);
-		if (batch == root->log_batch)
+		if (batch == atomic_read(&root->log_batch))
 			break;
 	}
 
@@ -2074,7 +2330,6 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 
 	btrfs_set_root_node(&log->root_item, log->node);
 
-	root->log_batch = 0;
 	root->log_transid++;
 	log->log_transid = root->log_transid;
 	root->log_start_pid = 0;
@@ -2087,7 +2342,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	mutex_unlock(&root->log_mutex);
 
 	mutex_lock(&log_root_tree->log_mutex);
-	log_root_tree->log_batch++;
+	atomic_inc(&log_root_tree->log_batch);
 	atomic_inc(&log_root_tree->log_writers);
 	mutex_unlock(&log_root_tree->log_mutex);
 
@@ -2157,7 +2412,6 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	btrfs_set_super_log_root_level(root->fs_info->super_for_commit,
 				btrfs_header_level(log_root_tree->node));
 
-	log_root_tree->log_batch = 0;
 	log_root_tree->log_transid++;
 	smp_mb();
 
@@ -2171,9 +2425,12 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	 * in and cause problems either.
 	 */
 	btrfs_scrub_pause_super(root);
-	write_ctree_super(trans, root->fs_info->tree_root, 1);
+	ret = write_ctree_super(trans, root->fs_info->tree_root, 1);
 	btrfs_scrub_continue_super(root);
-	ret = 0;
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out_wake_log_root;
+	}
 
 	mutex_lock(&root->log_mutex);
 	if (root->last_log_commit < log_transid)
@@ -2209,7 +2466,8 @@ static void free_log_tree(struct btrfs_trans_handle *trans,
 
 	while (1) {
 		ret = find_first_extent_bit(&log->dirty_log_pages,
-				0, &start, &end, EXTENT_DIRTY | EXTENT_NEW);
+				0, &start, &end, EXTENT_DIRTY | EXTENT_NEW,
+				NULL);
 		if (ret)
 			break;
 
@@ -2646,6 +2904,7 @@ static int drop_objectid_items(struct btrfs_trans_handle *trans,
 	int ret;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
+	int start_slot;
 
 	key.objectid = objectid;
 	key.type = max_key_type;
@@ -2667,8 +2926,18 @@ static int drop_objectid_items(struct btrfs_trans_handle *trans,
 		if (found_key.objectid != objectid)
 			break;
 
-		ret = btrfs_del_item(trans, log, path);
-		if (ret)
+		found_key.offset = 0;
+		found_key.type = 0;
+		ret = btrfs_bin_search(path->nodes[0], &found_key, 0,
+				       &start_slot);
+
+		ret = btrfs_del_items(trans, log, path, start_slot,
+				      path->slots[0] - start_slot + 1);
+		/*
+		 * If start slot isn't 0 then we don't need to re-search, we've
+		 * found the last guy with the objectid in this tree.
+		 */
+		if (ret || start_slot != 0)
 			break;
 		btrfs_release_path(path);
 	}
@@ -2678,14 +2947,64 @@ static int drop_objectid_items(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
+static void fill_inode_item(struct btrfs_trans_handle *trans,
+			    struct extent_buffer *leaf,
+			    struct btrfs_inode_item *item,
+			    struct inode *inode, int log_inode_only)
+{
+	btrfs_set_inode_uid(leaf, item, inode->i_uid);
+	btrfs_set_inode_gid(leaf, item, inode->i_gid);
+	btrfs_set_inode_mode(leaf, item, inode->i_mode);
+	btrfs_set_inode_nlink(leaf, item, inode->i_nlink);
+
+	btrfs_set_timespec_sec(leaf, btrfs_inode_atime(item),
+			       inode->i_atime.tv_sec);
+	btrfs_set_timespec_nsec(leaf, btrfs_inode_atime(item),
+				inode->i_atime.tv_nsec);
+
+	btrfs_set_timespec_sec(leaf, btrfs_inode_mtime(item),
+			       inode->i_mtime.tv_sec);
+	btrfs_set_timespec_nsec(leaf, btrfs_inode_mtime(item),
+				inode->i_mtime.tv_nsec);
+
+	btrfs_set_timespec_sec(leaf, btrfs_inode_ctime(item),
+			       inode->i_ctime.tv_sec);
+	btrfs_set_timespec_nsec(leaf, btrfs_inode_ctime(item),
+				inode->i_ctime.tv_nsec);
+
+	btrfs_set_inode_nbytes(leaf, item, inode_get_bytes(inode));
+
+	btrfs_set_inode_sequence(leaf, item, inode->i_version);
+	btrfs_set_inode_transid(leaf, item, trans->transid);
+	btrfs_set_inode_rdev(leaf, item, inode->i_rdev);
+	btrfs_set_inode_flags(leaf, item, BTRFS_I(inode)->flags);
+	btrfs_set_inode_block_group(leaf, item, 0);
+
+	if (log_inode_only) {
+		/* set the generation to zero so the recover code
+		 * can tell the difference between an logging
+		 * just to say 'this inode exists' and a logging
+		 * to say 'update this inode with these values'
+		 */
+		btrfs_set_inode_generation(leaf, item, 0);
+		btrfs_set_inode_size(leaf, item, 0);
+	} else {
+		btrfs_set_inode_generation(leaf, item,
+					   BTRFS_I(inode)->generation);
+		btrfs_set_inode_size(leaf, item, inode->i_size);
+	}
+
+}
+
 static noinline int copy_items(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *log,
+			       struct inode *inode,
 			       struct btrfs_path *dst_path,
 			       struct extent_buffer *src,
 			       int start_slot, int nr, int inode_only)
 {
 	unsigned long src_offset;
 	unsigned long dst_offset;
+	struct btrfs_root *log = BTRFS_I(inode)->root->log_root;
 	struct btrfs_file_extent_item *extent;
 	struct btrfs_inode_item *inode_item;
 	int ret;
@@ -2694,6 +3013,7 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 	char *ins_data;
 	int i;
 	struct list_head ordered_sums;
+	int skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
 
 	INIT_LIST_HEAD(&ordered_sums);
 
@@ -2722,29 +3042,23 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 
 		src_offset = btrfs_item_ptr_offset(src, start_slot + i);
 
-		copy_extent_buffer(dst_path->nodes[0], src, dst_offset,
-				   src_offset, ins_sizes[i]);
-
-		if (inode_only == LOG_INODE_EXISTS &&
-		    ins_keys[i].type == BTRFS_INODE_ITEM_KEY) {
+		if (ins_keys[i].type == BTRFS_INODE_ITEM_KEY) {
 			inode_item = btrfs_item_ptr(dst_path->nodes[0],
 						    dst_path->slots[0],
 						    struct btrfs_inode_item);
-			btrfs_set_inode_size(dst_path->nodes[0], inode_item, 0);
-
-			/* set the generation to zero so the recover code
-			 * can tell the difference between an logging
-			 * just to say 'this inode exists' and a logging
-			 * to say 'update this inode with these values'
-			 */
-			btrfs_set_inode_generation(dst_path->nodes[0],
-						   inode_item, 0);
+			fill_inode_item(trans, dst_path->nodes[0], inode_item,
+					inode, inode_only == LOG_INODE_EXISTS);
+		} else {
+			copy_extent_buffer(dst_path->nodes[0], src, dst_offset,
+					   src_offset, ins_sizes[i]);
 		}
+
 		/* take a reference on file data extents so that truncates
 		 * or deletes of this inode don't have to relog the inode
 		 * again
 		 */
-		if (btrfs_key_type(ins_keys + i) == BTRFS_EXTENT_DATA_KEY) {
+		if (btrfs_key_type(ins_keys + i) == BTRFS_EXTENT_DATA_KEY &&
+		    !skip_csum) {
 			int found_type;
 			extent = btrfs_item_ptr(src, start_slot + i,
 						struct btrfs_file_extent_item);
@@ -2753,8 +3067,7 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 				continue;
 
 			found_type = btrfs_file_extent_type(src, extent);
-			if (found_type == BTRFS_FILE_EXTENT_REG ||
-			    found_type == BTRFS_FILE_EXTENT_PREALLOC) {
+			if (found_type == BTRFS_FILE_EXTENT_REG) {
 				u64 ds, dl, cs, cl;
 				ds = btrfs_file_extent_disk_bytenr(src,
 								extent);
@@ -2803,6 +3116,239 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
+static int extent_cmp(void *priv, struct list_head *a, struct list_head *b)
+{
+	struct extent_map *em1, *em2;
+
+	em1 = list_entry(a, struct extent_map, list);
+	em2 = list_entry(b, struct extent_map, list);
+
+	if (em1->start < em2->start)
+		return -1;
+	else if (em1->start > em2->start)
+		return 1;
+	return 0;
+}
+
+struct log_args {
+	struct extent_buffer *src;
+	u64 next_offset;
+	int start_slot;
+	int nr;
+};
+
+static int log_one_extent(struct btrfs_trans_handle *trans,
+			  struct inode *inode, struct btrfs_root *root,
+			  struct extent_map *em, struct btrfs_path *path,
+			  struct btrfs_path *dst_path, struct log_args *args)
+{
+	struct btrfs_root *log = root->log_root;
+	struct btrfs_file_extent_item *fi;
+	struct btrfs_key key;
+	u64 start = em->mod_start;
+	u64 search_start = start;
+	u64 len = em->mod_len;
+	u64 num_bytes;
+	int nritems;
+	int ret;
+
+	if (BTRFS_I(inode)->logged_trans == trans->transid) {
+		ret = __btrfs_drop_extents(trans, log, inode, dst_path, start,
+					   start + len, NULL, 0);
+		if (ret)
+			return ret;
+	}
+
+	while (len) {
+		if (args->nr)
+			goto next_slot;
+again:
+		key.objectid = btrfs_ino(inode);
+		key.type = BTRFS_EXTENT_DATA_KEY;
+		key.offset = search_start;
+
+		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+		if (ret < 0)
+			return ret;
+
+		if (ret) {
+			/*
+			 * A rare case were we can have an em for a section of a
+			 * larger extent so we need to make sure that this em
+			 * falls within the extent we've found.  If not we just
+			 * bail and go back to ye-olde way of doing things but
+			 * it happens often enough in testing that we need to do
+			 * this dance to make sure.
+			 */
+			do {
+				if (path->slots[0] == 0) {
+					btrfs_release_path(path);
+					if (search_start == 0)
+						return -ENOENT;
+					search_start--;
+					goto again;
+				}
+
+				path->slots[0]--;
+				btrfs_item_key_to_cpu(path->nodes[0], &key,
+						      path->slots[0]);
+				if (key.objectid != btrfs_ino(inode) ||
+				    key.type != BTRFS_EXTENT_DATA_KEY) {
+					btrfs_release_path(path);
+					return -ENOENT;
+				}
+			} while (key.offset > start);
+
+			fi = btrfs_item_ptr(path->nodes[0], path->slots[0],
+					    struct btrfs_file_extent_item);
+			num_bytes = btrfs_file_extent_num_bytes(path->nodes[0],
+								fi);
+			if (key.offset + num_bytes <= start) {
+				btrfs_release_path(path);
+				return -ENOENT;
+			}
+		}
+		args->src = path->nodes[0];
+next_slot:
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+		fi = btrfs_item_ptr(args->src, path->slots[0],
+				    struct btrfs_file_extent_item);
+		if (args->nr &&
+		    args->start_slot + args->nr == path->slots[0]) {
+			args->nr++;
+		} else if (args->nr) {
+			ret = copy_items(trans, inode, dst_path, args->src,
+					 args->start_slot, args->nr,
+					 LOG_INODE_ALL);
+			if (ret)
+				return ret;
+			args->nr = 1;
+			args->start_slot = path->slots[0];
+		} else if (!args->nr) {
+			args->nr = 1;
+			args->start_slot = path->slots[0];
+		}
+		nritems = btrfs_header_nritems(path->nodes[0]);
+		path->slots[0]++;
+		num_bytes = btrfs_file_extent_num_bytes(args->src, fi);
+		if (len < num_bytes) {
+			/* I _think_ this is ok, envision we write to a
+			 * preallocated space that is adjacent to a previously
+			 * written preallocated space that gets merged when we
+			 * mark this preallocated space written.  If we do not
+			 * have the adjacent extent in cache then when we copy
+			 * this extent it could end up being larger than our EM
+			 * thinks it is, which is a-ok, so just set len to 0.
+			 */
+			len = 0;
+		} else {
+			len -= num_bytes;
+		}
+		start = key.offset + num_bytes;
+		args->next_offset = start;
+		search_start = start;
+
+		if (path->slots[0] < nritems) {
+			if (len)
+				goto next_slot;
+			break;
+		}
+
+		if (args->nr) {
+			ret = copy_items(trans, inode, dst_path, args->src,
+					 args->start_slot, args->nr,
+					 LOG_INODE_ALL);
+			if (ret)
+				return ret;
+			args->nr = 0;
+			btrfs_release_path(path);
+		}
+	}
+
+	return 0;
+}
+
+static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans,
+				     struct btrfs_root *root,
+				     struct inode *inode,
+				     struct btrfs_path *path,
+				     struct btrfs_path *dst_path)
+{
+	struct log_args args;
+	struct extent_map *em, *n;
+	struct list_head extents;
+	struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree;
+	u64 test_gen;
+	int ret = 0;
+
+	INIT_LIST_HEAD(&extents);
+
+	memset(&args, 0, sizeof(args));
+
+	write_lock(&tree->lock);
+	test_gen = root->fs_info->last_trans_committed;
+
+	list_for_each_entry_safe(em, n, &tree->modified_extents, list) {
+		list_del_init(&em->list);
+		if (em->generation <= test_gen)
+			continue;
+		/* Need a ref to keep it from getting evicted from cache */
+		atomic_inc(&em->refs);
+		set_bit(EXTENT_FLAG_LOGGING, &em->flags);
+		list_add_tail(&em->list, &extents);
+	}
+
+	list_sort(NULL, &extents, extent_cmp);
+
+	while (!list_empty(&extents)) {
+		em = list_entry(extents.next, struct extent_map, list);
+
+		list_del_init(&em->list);
+		clear_bit(EXTENT_FLAG_LOGGING, &em->flags);
+
+		/*
+		 * If we had an error we just need to delete everybody from our
+		 * private list.
+		 */
+		if (ret) {
+			free_extent_map(em);
+			continue;
+		}
+
+		write_unlock(&tree->lock);
+
+		/*
+		 * If the previous EM and the last extent we left off on aren't
+		 * sequential then we need to copy the items we have and redo
+		 * our search
+		 */
+		if (args.nr && em->mod_start != args.next_offset) {
+			ret = copy_items(trans, inode, dst_path, args.src,
+					 args.start_slot, args.nr,
+					 LOG_INODE_ALL);
+			if (ret) {
+				free_extent_map(em);
+				write_lock(&tree->lock);
+				continue;
+			}
+			btrfs_release_path(path);
+			args.nr = 0;
+		}
+
+		ret = log_one_extent(trans, inode, root, em, path, dst_path, &args);
+		free_extent_map(em);
+		write_lock(&tree->lock);
+	}
+	WARN_ON(!list_empty(&extents));
+	write_unlock(&tree->lock);
+
+	if (!ret && args.nr)
+		ret = copy_items(trans, inode, dst_path, args.src,
+				 args.start_slot, args.nr, LOG_INODE_ALL);
+	btrfs_release_path(path);
+	return ret;
+}
+
 /* log a single inode in the tree log.
  * At least one parent directory for this inode must exist in the tree
  * or be logged already.
@@ -2832,6 +3378,7 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 	int nritems;
 	int ins_start_slot = 0;
 	int ins_nr;
+	bool fast_search = false;
 	u64 ino = btrfs_ino(inode);
 
 	log = root->log_root;
@@ -2851,21 +3398,23 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 
 	max_key.objectid = ino;
 
-	/* today the code can only do partial logging of directories */
-	if (!S_ISDIR(inode->i_mode))
-	    inode_only = LOG_INODE_ALL;
 
+	/* today the code can only do partial logging of directories */
 	if (inode_only == LOG_INODE_EXISTS || S_ISDIR(inode->i_mode))
 		max_key.type = BTRFS_XATTR_ITEM_KEY;
 	else
 		max_key.type = (u8)-1;
 	max_key.offset = (u64)-1;
 
-	ret = btrfs_commit_inode_delayed_items(trans, inode);
-	if (ret) {
-		btrfs_free_path(path);
-		btrfs_free_path(dst_path);
-		return ret;
+	/* Only run delayed items if we are a dir or a new file */
+	if (S_ISDIR(inode->i_mode) ||
+	    BTRFS_I(inode)->generation > root->fs_info->last_trans_committed) {
+		ret = btrfs_commit_inode_delayed_items(trans, inode);
+		if (ret) {
+			btrfs_free_path(path);
+			btrfs_free_path(dst_path);
+			return ret;
+		}
 	}
 
 	mutex_lock(&BTRFS_I(inode)->log_mutex);
@@ -2881,7 +3430,16 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 			max_key_type = BTRFS_XATTR_ITEM_KEY;
 		ret = drop_objectid_items(trans, log, path, ino, max_key_type);
 	} else {
-		ret = btrfs_truncate_inode_items(trans, log, inode, 0, 0);
+		if (test_and_clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+				       &BTRFS_I(inode)->runtime_flags)) {
+			ret = btrfs_truncate_inode_items(trans, log,
+							 inode, 0, 0);
+		} else {
+			fast_search = true;
+			max_key.type = BTRFS_XATTR_ITEM_KEY;
+			ret = drop_objectid_items(trans, log, path, ino,
+						  BTRFS_XATTR_ITEM_KEY);
+		}
 	}
 	if (ret) {
 		err = ret;
@@ -2912,7 +3470,7 @@ again:
 			goto next_slot;
 		}
 
-		ret = copy_items(trans, log, dst_path, src, ins_start_slot,
+		ret = copy_items(trans, inode, dst_path, src, ins_start_slot,
 				 ins_nr, inode_only);
 		if (ret) {
 			err = ret;
@@ -2930,7 +3488,7 @@ next_slot:
 			goto again;
 		}
 		if (ins_nr) {
-			ret = copy_items(trans, log, dst_path, src,
+			ret = copy_items(trans, inode, dst_path, src,
 					 ins_start_slot,
 					 ins_nr, inode_only);
 			if (ret) {
@@ -2951,8 +3509,7 @@ next_slot:
 			break;
 	}
 	if (ins_nr) {
-		ret = copy_items(trans, log, dst_path, src,
-				 ins_start_slot,
+		ret = copy_items(trans, inode, dst_path, src, ins_start_slot,
 				 ins_nr, inode_only);
 		if (ret) {
 			err = ret;
@@ -2960,7 +3517,24 @@ next_slot:
 		}
 		ins_nr = 0;
 	}
-	WARN_ON(ins_nr);
+
+	if (fast_search) {
+		btrfs_release_path(path);
+		btrfs_release_path(dst_path);
+		ret = btrfs_log_changed_extents(trans, root, inode, path,
+						dst_path);
+		if (ret) {
+			err = ret;
+			goto out_unlock;
+		}
+	} else {
+		struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree;
+		struct extent_map *em, *n;
+
+		list_for_each_entry_safe(em, n, &tree->modified_extents, list)
+			list_del_init(&em->list);
+	}
+
 	if (inode_only == LOG_INODE_ALL && S_ISDIR(inode->i_mode)) {
 		btrfs_release_path(path);
 		btrfs_release_path(dst_path);
@@ -2971,6 +3545,7 @@ next_slot:
 		}
 	}
 	BTRFS_I(inode)->logged_trans = trans->transid;
+	BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->last_sub_trans;
 out_unlock:
 	mutex_unlock(&BTRFS_I(inode)->log_mutex);
 
@@ -3138,7 +3713,7 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 end_trans:
 	dput(old_parent);
 	if (ret < 0) {
-		BUG_ON(ret != -ENOSPC);
+		WARN_ON(ret != -ENOSPC);
 		root->fs_info->last_trans_log_full_commit = trans->transid;
 		ret = 1;
 	}

fs/btrfs/ulist.c

@@ -143,14 +143,13 @@ EXPORT_SYMBOL(ulist_free);
  * In case of allocation failure -ENOMEM is returned and the ulist stays
  * unaltered.
  */
-int ulist_add(struct ulist *ulist, u64 val, unsigned long aux,
-	      gfp_t gfp_mask)
+int ulist_add(struct ulist *ulist, u64 val, u64 aux, gfp_t gfp_mask)
 {
 	return ulist_add_merge(ulist, val, aux, NULL, gfp_mask);
 }
 
-int ulist_add_merge(struct ulist *ulist, u64 val, unsigned long aux,
-		    unsigned long *old_aux, gfp_t gfp_mask)
+int ulist_add_merge(struct ulist *ulist, u64 val, u64 aux,
+		    u64 *old_aux, gfp_t gfp_mask)
 {
 	int i;
 

fs/btrfs/ulist.h

@@ -33,7 +33,7 @@ struct ulist_iterator {
  */
 struct ulist_node {
 	u64 val;		/* value to store */
-	unsigned long aux;	/* auxiliary value saved along with the val */
+	u64 aux;		/* auxiliary value saved along with the val */
 };
 
 struct ulist {
@@ -65,10 +65,9 @@ void ulist_fini(struct ulist *ulist);
 void ulist_reinit(struct ulist *ulist);
 struct ulist *ulist_alloc(gfp_t gfp_mask);
 void ulist_free(struct ulist *ulist);
-int ulist_add(struct ulist *ulist, u64 val, unsigned long aux,
-	      gfp_t gfp_mask);
-int ulist_add_merge(struct ulist *ulist, u64 val, unsigned long aux,
-		    unsigned long *old_aux, gfp_t gfp_mask);
+int ulist_add(struct ulist *ulist, u64 val, u64 aux, gfp_t gfp_mask);
+int ulist_add_merge(struct ulist *ulist, u64 val, u64 aux,
+		    u64 *old_aux, gfp_t gfp_mask);
 struct ulist_node *ulist_next(struct ulist *ulist,
 			      struct ulist_iterator *uiter);
 

fs/btrfs/volumes.c

@@ -639,7 +639,7 @@ static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 
 		bdev = blkdev_get_by_path(device->name->str, flags, holder);
 		if (IS_ERR(bdev)) {
-			printk(KERN_INFO "open %s failed\n", device->name->str);
+			printk(KERN_INFO "btrfs: open %s failed\n", device->name->str);
 			goto error;
 		}
 		filemap_write_and_wait(bdev->bd_inode->i_mapping);
@@ -1475,6 +1475,9 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
 		free_fs_devices(cur_devices);
 	}
 
+	root->fs_info->num_tolerated_disk_barrier_failures =
+		btrfs_calc_num_tolerated_disk_barrier_failures(root->fs_info);
+
 	/*
 	 * at this point, the device is zero sized.  We want to
 	 * remove it from the devices list and zero out the old super
@@ -1775,15 +1778,21 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 
 	if (seeding_dev) {
 		ret = init_first_rw_device(trans, root, device);
-		if (ret)
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
 			goto error_trans;
+		}
 		ret = btrfs_finish_sprout(trans, root);
-		if (ret)
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
 			goto error_trans;
+		}
 	} else {
 		ret = btrfs_add_device(trans, root, device);
-		if (ret)
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
 			goto error_trans;
+		}
 	}
 
 	/*
@@ -1793,6 +1802,8 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 	btrfs_clear_space_info_full(root->fs_info);
 
 	unlock_chunks(root);
+	root->fs_info->num_tolerated_disk_barrier_failures =
+		btrfs_calc_num_tolerated_disk_barrier_failures(root->fs_info);
 	ret = btrfs_commit_transaction(trans, root);
 
 	if (seeding_dev) {
@@ -1814,7 +1825,6 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 
 error_trans:
 	unlock_chunks(root);
-	btrfs_abort_transaction(trans, root, ret);
 	btrfs_end_transaction(trans, root);
 	rcu_string_free(device->name);
 	kfree(device);
@@ -2804,6 +2814,26 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		}
 	}
 
+	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
+		int num_tolerated_disk_barrier_failures;
+		u64 target = bctl->sys.target;
+
+		num_tolerated_disk_barrier_failures =
+			btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
+		if (num_tolerated_disk_barrier_failures > 0 &&
+		    (target &
+		     (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID0 |
+		      BTRFS_AVAIL_ALLOC_BIT_SINGLE)))
+			num_tolerated_disk_barrier_failures = 0;
+		else if (num_tolerated_disk_barrier_failures > 1 &&
+			 (target &
+			  (BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10)))
+			num_tolerated_disk_barrier_failures = 1;
+
+		fs_info->num_tolerated_disk_barrier_failures =
+			num_tolerated_disk_barrier_failures;
+	}
+
 	ret = insert_balance_item(fs_info->tree_root, bctl);
 	if (ret && ret != -EEXIST)
 		goto out;
@@ -2836,6 +2866,11 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		__cancel_balance(fs_info);
 	}
 
+	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
+		fs_info->num_tolerated_disk_barrier_failures =
+			btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
+	}
+
 	wake_up(&fs_info->balance_wait_q);
 
 	return ret;
@@ -3608,12 +3643,16 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
 	ret = __btrfs_alloc_chunk(trans, extent_root, &sys_map,
 				  &sys_chunk_size, &sys_stripe_size,
 				  sys_chunk_offset, alloc_profile);
-	if (ret)
-		goto abort;
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out;
+	}
 
 	ret = btrfs_add_device(trans, fs_info->chunk_root, device);
-	if (ret)
-		goto abort;
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out;
+	}
 
 	/*
 	 * Modifying chunk tree needs allocating new blocks from both
@@ -3623,19 +3662,19 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
 	 */
 	ret = __finish_chunk_alloc(trans, extent_root, map, chunk_offset,
 				   chunk_size, stripe_size);
-	if (ret)
-		goto abort;
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out;
+	}
 
 	ret = __finish_chunk_alloc(trans, extent_root, sys_map,
 				   sys_chunk_offset, sys_chunk_size,
 				   sys_stripe_size);
 	if (ret)
-		goto abort;
+		btrfs_abort_transaction(trans, root, ret);
 
-	return 0;
+out:
 
-abort:
-	btrfs_abort_transaction(trans, root, ret);
 	return ret;
 }
 
@@ -3760,7 +3799,7 @@ static int __btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
 	read_unlock(&em_tree->lock);
 
 	if (!em) {
-		printk(KERN_CRIT "unable to find logical %llu len %llu\n",
+		printk(KERN_CRIT "btrfs: unable to find logical %llu len %llu\n",
 		       (unsigned long long)logical,
 		       (unsigned long long)*length);
 		BUG();
@@ -4217,7 +4256,7 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
 
 	total_devs = bbio->num_stripes;
 	if (map_length < length) {
-		printk(KERN_CRIT "mapping failed logical %llu bio len %llu "
+		printk(KERN_CRIT "btrfs: mapping failed logical %llu bio len %llu "
 		       "len %llu\n", (unsigned long long)logical,
 		       (unsigned long long)length,
 		       (unsigned long long)map_length);

fs/btrfs/zlib.c

@@ -97,7 +97,7 @@ static int zlib_compress_pages(struct list_head *ws,
 	*total_in = 0;
 
 	if (Z_OK != zlib_deflateInit(&workspace->def_strm, 3)) {
-		printk(KERN_WARNING "deflateInit failed\n");
+		printk(KERN_WARNING "btrfs: deflateInit failed\n");
 		ret = -1;
 		goto out;
 	}
@@ -125,7 +125,7 @@ static int zlib_compress_pages(struct list_head *ws,
 	while (workspace->def_strm.total_in < len) {
 		ret = zlib_deflate(&workspace->def_strm, Z_SYNC_FLUSH);
 		if (ret != Z_OK) {
-			printk(KERN_DEBUG "btrfs deflate in loop returned %d\n",
+			printk(KERN_DEBUG "btrfs: deflate in loop returned %d\n",
 			       ret);
 			zlib_deflateEnd(&workspace->def_strm);
 			ret = -1;
@@ -252,7 +252,7 @@ static int zlib_decompress_biovec(struct list_head *ws, struct page **pages_in,
 	}
 
 	if (Z_OK != zlib_inflateInit2(&workspace->inf_strm, wbits)) {
-		printk(KERN_WARNING "inflateInit failed\n");
+		printk(KERN_WARNING "btrfs: inflateInit failed\n");
 		return -1;
 	}
 	while (workspace->inf_strm.total_in < srclen) {
@@ -336,7 +336,7 @@ static int zlib_decompress(struct list_head *ws, unsigned char *data_in,
 	}
 
 	if (Z_OK != zlib_inflateInit2(&workspace->inf_strm, wbits)) {
-		printk(KERN_WARNING "inflateInit failed\n");
+		printk(KERN_WARNING "btrfs: inflateInit failed\n");
 		return -1;
 	}
 

include/trace/events/btrfs.h

@@ -445,6 +445,7 @@ TRACE_EVENT(btrfs_delayed_tree_ref,
 		__field(	u64,  ref_root		)
 		__field(	int,  level		)
 		__field(	int,  type		)
+		__field(	u64,  seq		)
 	),
 
 	TP_fast_assign(
@@ -455,17 +456,19 @@ TRACE_EVENT(btrfs_delayed_tree_ref,
 		__entry->ref_root	= full_ref->root;
 		__entry->level		= full_ref->level;
 		__entry->type		= ref->type;
+		__entry->seq		= ref->seq;
 	),
 
 	TP_printk("bytenr = %llu, num_bytes = %llu, action = %s, "
 		  "parent = %llu(%s), ref_root = %llu(%s), level = %d, "
-		  "type = %s",
+		  "type = %s, seq = %llu",
 		  (unsigned long long)__entry->bytenr,
 		  (unsigned long long)__entry->num_bytes,
 		  show_ref_action(__entry->action),
 		  show_root_type(__entry->parent),
 		  show_root_type(__entry->ref_root),
-		  __entry->level, show_ref_type(__entry->type))
+		  __entry->level, show_ref_type(__entry->type),
+		  (unsigned long long)__entry->seq)
 );
 
 TRACE_EVENT(btrfs_delayed_data_ref,
@@ -485,6 +488,7 @@ TRACE_EVENT(btrfs_delayed_data_ref,
 		__field(	u64,  owner		)
 		__field(	u64,  offset		)
 		__field(	int,  type		)
+		__field(	u64,  seq		)
 	),
 
 	TP_fast_assign(
@@ -496,11 +500,12 @@ TRACE_EVENT(btrfs_delayed_data_ref,
 		__entry->owner		= full_ref->objectid;
 		__entry->offset		= full_ref->offset;
 		__entry->type		= ref->type;
+		__entry->seq		= ref->seq;
 	),
 
 	TP_printk("bytenr = %llu, num_bytes = %llu, action = %s, "
 		  "parent = %llu(%s), ref_root = %llu(%s), owner = %llu, "
-		  "offset = %llu, type = %s",
+		  "offset = %llu, type = %s, seq = %llu",
 		  (unsigned long long)__entry->bytenr,
 		  (unsigned long long)__entry->num_bytes,
 		  show_ref_action(__entry->action),
@@ -508,7 +513,8 @@ TRACE_EVENT(btrfs_delayed_data_ref,
 		  show_root_type(__entry->ref_root),
 		  (unsigned long long)__entry->owner,
 		  (unsigned long long)__entry->offset,
-		  show_ref_type(__entry->type))
+		  show_ref_type(__entry->type),
+		  (unsigned long long)__entry->seq)
 );
 
 TRACE_EVENT(btrfs_delayed_ref_head,