[PATCH v3 0/7] btrfs-progs: Support for BG_TREE feature

[PATCH v3 0/7] btrfs-progs: Support for BG_TREE feature

[Qu Wenruo]

This patchset can be fetched from github:
https://github.com/adam900710/btrfs-progs/tree/bg_tree
Which is based on v5.2.2 tag.

This patchset provides the needed user space infrastructure for BG_TREE
feature.

Since it's an new incompatible feature, unlike SKINNY_METADATA, btrfs-progs
is needed to convert existing fs (unmounted) to new format.

Now btrfstune can convert regular extent tree fs to bg tree fs to
improve mount time.

For the performance improvement, please check the kernel patchset cover
letter or the last patch.
(SPOILER ALERT: It's super fast)

Changelog:
v2:
- Rebase to v5.2.2 tag
- Add btrfstune ability to convert existing fs to BG_TREE feature

v3:
- Fix a bug that temp chunks are not cleaned up properly
  This is caused by wrong timing btrfs_convert_to_bg_tree() is called.
  It should be called after temp chunks cleaned up.

- Fix a bug that an extent buffer get leaked
  This is caused by newly created bg tree not added to dirty list.

Qu Wenruo (7):
  btrfs-progs: Refactor excluded extent functions to use fs_info
  btrfs-progs: Refactor btrfs_read_block_groups()
  btrfs-progs: Enable read-write ability for 'bg_tree' feature
  btrfs-progs: mkfs: Introduce -O bg-tree
  btrfs-progs: dump-tree/dump-super: Introduce support for bg tree
  btrfs-progs: check: Introduce support for bg-tree feature
  btrfs-progs: btrfstune: Allow to enable bg-tree feature offline

 Documentation/btrfstune.asciidoc |   6 +
 btrfstune.c                      |  44 +++-
 check/main.c                     |   7 +-
 check/mode-lowmem.c              |   9 +-
 cmds/inspect-dump-super.c        |   3 +-
 cmds/inspect-dump-tree.c         |   5 +
 common/fsfeatures.c              |   6 +
 ctree.h                          |  18 +-
 disk-io.c                        |  29 ++-
 extent-tree.c                    | 365 +++++++++++++++++++++++--------
 mkfs/common.c                    |   6 +-
 mkfs/main.c                      |  25 +++
 print-tree.c                     |   3 +
 transaction.c                    |   1 +
 14 files changed, 419 insertions(+), 108 deletions(-)

-- 
2.23.0

[PATCH v3 1/7] btrfs-progs: Refactor excluded extent functions to use fs_info

[Qu Wenruo]

The following functions are just using @root to reach fs_info:
- exclude_super_stripes
- free_excluded_extents
- add_excluded_extent

Refactor them to use fs_info directly.

Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
---
 check/main.c  |  4 ++--
 ctree.h       |  4 ++--
 extent-tree.c | 20 ++++++++++----------
 3 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/check/main.c b/check/main.c
index c2d0f3949c5e..94ffab46cb70 100644
--- a/check/main.c
+++ b/check/main.c
@@ -5552,7 +5552,7 @@ static int check_space_cache(struct btrfs_root *root)
 		}
 
 		if (btrfs_fs_compat_ro(root->fs_info, FREE_SPACE_TREE)) {
-			ret = exclude_super_stripes(root, cache);
+			ret = exclude_super_stripes(root->fs_info, cache);
 			if (ret) {
 				errno = -ret;
 				fprintf(stderr,
@@ -5561,7 +5561,7 @@ static int check_space_cache(struct btrfs_root *root)
 				continue;
 			}
 			ret = load_free_space_tree(root->fs_info, cache);
-			free_excluded_extents(root, cache);
+			free_excluded_extents(root->fs_info, cache);
 			if (ret < 0) {
 				errno = -ret;
 				fprintf(stderr,
diff --git a/ctree.h b/ctree.h
index 0d12563b7261..8c7b3cb40151 100644
--- a/ctree.h
+++ b/ctree.h
@@ -2568,9 +2568,9 @@ int btrfs_record_file_extent(struct btrfs_trans_handle *trans,
 			      u64 num_bytes);
 int btrfs_free_block_group(struct btrfs_trans_handle *trans,
 			   struct btrfs_fs_info *fs_info, u64 bytenr, u64 len);
-void free_excluded_extents(struct btrfs_root *root,
+void free_excluded_extents(struct btrfs_fs_info *fs_info,
 			   struct btrfs_block_group_cache *cache);
-int exclude_super_stripes(struct btrfs_root *root,
+int exclude_super_stripes(struct btrfs_fs_info *fs_info,
 			  struct btrfs_block_group_cache *cache);
 u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
 		       struct btrfs_fs_info *info, u64 start, u64 end);
diff --git a/extent-tree.c b/extent-tree.c
index 932af2c644bd..19d1ea0df570 100644
--- a/extent-tree.c
+++ b/extent-tree.c
@@ -2716,7 +2716,7 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 		if (btrfs_chunk_readonly(info, cache->key.objectid))
 			cache->ro = 1;
 
-		exclude_super_stripes(root, cache);
+		exclude_super_stripes(info, cache);
 
 		ret = update_space_info(info, cache->flags, found_key.offset,
 					btrfs_block_group_used(&cache->item),
@@ -2760,7 +2760,7 @@ btrfs_add_block_group(struct btrfs_fs_info *fs_info, u64 bytes_used, u64 type,
 	cache->flags = type;
 	btrfs_set_block_group_flags(&cache->item, type);
 
-	exclude_super_stripes(fs_info->extent_root, cache);
+	exclude_super_stripes(fs_info, cache);
 	ret = update_space_info(fs_info, cache->flags, size, bytes_used,
 				&cache->space_info);
 	BUG_ON(ret);
@@ -3551,16 +3551,16 @@ int btrfs_record_file_extent(struct btrfs_trans_handle *trans,
 }
 
 
-static int add_excluded_extent(struct btrfs_root *root,
+static int add_excluded_extent(struct btrfs_fs_info *fs_info,
 			       u64 start, u64 num_bytes)
 {
 	u64 end = start + num_bytes - 1;
-	set_extent_bits(&root->fs_info->pinned_extents,
+	set_extent_bits(&fs_info->pinned_extents,
 			start, end, EXTENT_UPTODATE);
 	return 0;
 }
 
-void free_excluded_extents(struct btrfs_root *root,
+void free_excluded_extents(struct btrfs_fs_info *fs_info,
 			   struct btrfs_block_group_cache *cache)
 {
 	u64 start, end;
@@ -3568,11 +3568,11 @@ void free_excluded_extents(struct btrfs_root *root,
 	start = cache->key.objectid;
 	end = start + cache->key.offset - 1;
 
-	clear_extent_bits(&root->fs_info->pinned_extents,
+	clear_extent_bits(&fs_info->pinned_extents,
 			  start, end, EXTENT_UPTODATE);
 }
 
-int exclude_super_stripes(struct btrfs_root *root,
+int exclude_super_stripes(struct btrfs_fs_info *fs_info,
 			  struct btrfs_block_group_cache *cache)
 {
 	u64 bytenr;
@@ -3583,7 +3583,7 @@ int exclude_super_stripes(struct btrfs_root *root,
 	if (cache->key.objectid < BTRFS_SUPER_INFO_OFFSET) {
 		stripe_len = BTRFS_SUPER_INFO_OFFSET - cache->key.objectid;
 		cache->bytes_super += stripe_len;
-		ret = add_excluded_extent(root, cache->key.objectid,
+		ret = add_excluded_extent(fs_info, cache->key.objectid,
 					  stripe_len);
 		if (ret)
 			return ret;
@@ -3591,7 +3591,7 @@ int exclude_super_stripes(struct btrfs_root *root,
 
 	for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
 		bytenr = btrfs_sb_offset(i);
-		ret = btrfs_rmap_block(root->fs_info,
+		ret = btrfs_rmap_block(fs_info,
 				       cache->key.objectid, bytenr,
 				       &logical, &nr, &stripe_len);
 		if (ret)
@@ -3618,7 +3618,7 @@ int exclude_super_stripes(struct btrfs_root *root,
 			}
 
 			cache->bytes_super += len;
-			ret = add_excluded_extent(root, start, len);
+			ret = add_excluded_extent(fs_info, start, len);
 			if (ret) {
 				kfree(logical);
 				return ret;
-- 
2.23.0

[PATCH v3 2/7] btrfs-progs: Refactor btrfs_read_block_groups()

[Qu Wenruo]

This patch does the following refactor:
- Refactor parameter from @root to @fs_info

- Refactor the large loop body into another function
  Now we have a helper function, read_one_block_group(), to handle
  block group cache and space info related routine.

- Refactor the return value
  Even we have the code handling ret > 0 from find_first_block_group(),
  it never works, as when there is no more block group,
  find_first_block_group() just return -ENOENT other than 1.

  This is super confusing, it's almost a mircle it even works.

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 ctree.h       |   2 +-
 disk-io.c     |   9 ++-
 extent-tree.c | 160 ++++++++++++++++++++++++++++----------------------
 3 files changed, 97 insertions(+), 74 deletions(-)

diff --git a/ctree.h b/ctree.h
index 8c7b3cb40151..2899de358613 100644
--- a/ctree.h
+++ b/ctree.h
@@ -2550,7 +2550,7 @@ int update_space_info(struct btrfs_fs_info *info, u64 flags,
 		      u64 total_bytes, u64 bytes_used,
 		      struct btrfs_space_info **space_info);
 int btrfs_free_block_groups(struct btrfs_fs_info *info);
-int btrfs_read_block_groups(struct btrfs_root *root);
+int btrfs_read_block_groups(struct btrfs_fs_info *info);
 struct btrfs_block_group_cache *
 btrfs_add_block_group(struct btrfs_fs_info *fs_info, u64 bytes_used, u64 type,
 		      u64 chunk_offset, u64 size);
diff --git a/disk-io.c b/disk-io.c
index be44eead5cef..8978f0cb60c7 100644
--- a/disk-io.c
+++ b/disk-io.c
@@ -983,14 +983,17 @@ int btrfs_setup_all_roots(struct btrfs_fs_info *fs_info, u64 root_tree_bytenr,
 	fs_info->last_trans_committed = generation;
 	if (extent_buffer_uptodate(fs_info->extent_root->node) &&
 	    !(flags & OPEN_CTREE_NO_BLOCK_GROUPS)) {
-		ret = btrfs_read_block_groups(fs_info->tree_root);
+		ret = btrfs_read_block_groups(fs_info);
 		/*
 		 * If we don't find any blockgroups (ENOENT) we're either
 		 * restoring or creating the filesystem, where it's expected,
 		 * anything else is error
 		 */
-		if (ret != -ENOENT)
-			return -EIO;
+		if (ret < 0 && ret != -ENOENT) {
+			errno = -ret;
+			error("failed to read block groups: %m");
+			return ret;
+		}
 	}
 
 	key.objectid = BTRFS_FS_TREE_OBJECTID;
diff --git a/extent-tree.c b/extent-tree.c
index 19d1ea0df570..9713d627764c 100644
--- a/extent-tree.c
+++ b/extent-tree.c
@@ -2607,6 +2607,13 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 	return 0;
 }
 
+/*
+ * Find a block group which starts >= @key->objectid in extent tree.
+ *
+ * Return 0 for found
+ * Retrun >0 for not found
+ * Return <0 for error
+ */
 static int find_first_block_group(struct btrfs_root *root,
 		struct btrfs_path *path, struct btrfs_key *key)
 {
@@ -2636,36 +2643,95 @@ static int find_first_block_group(struct btrfs_root *root,
 			return 0;
 		path->slots[0]++;
 	}
-	ret = -ENOENT;
+	ret = 1;
 error:
 	return ret;
 }
 
-int btrfs_read_block_groups(struct btrfs_root *root)
+/*
+ * Helper function to read out one BLOCK_GROUP_ITEM and insert it into
+ * block group cache.
+ *
+ * Return 0 if nothing wrong (either insert the bg cache or skip 0 sized bg)
+ * Return <0 for error.
+ */
+static int read_one_block_group(struct btrfs_fs_info *fs_info,
+				 struct btrfs_path *path)
 {
-	struct btrfs_path *path;
-	int ret;
-	int bit;
-	struct btrfs_block_group_cache *cache;
-	struct btrfs_fs_info *info = root->fs_info;
+	struct extent_io_tree *block_group_cache = &fs_info->block_group_cache;
+	struct extent_buffer *leaf = path->nodes[0];
 	struct btrfs_space_info *space_info;
-	struct extent_io_tree *block_group_cache;
+	struct btrfs_block_group_cache *cache;
 	struct btrfs_key key;
-	struct btrfs_key found_key;
-	struct extent_buffer *leaf;
+	int slot = path->slots[0];
+	int bit = 0;
+	int ret;
 
-	block_group_cache = &info->block_group_cache;
+	btrfs_item_key_to_cpu(leaf, &key, slot);
+	ASSERT(key.type == BTRFS_BLOCK_GROUP_ITEM_KEY);
+
+	/*
+	 * Skip 0 sized block group, don't insert them into block
+	 * group cache tree, as its length is 0, it won't get
+	 * freed at close_ctree() time.
+	 */
+	if (key.offset == 0)
+		return 0;
+
+	cache = kzalloc(sizeof(*cache), GFP_NOFS);
+	if (!cache)
+		return -ENOMEM;
+	read_extent_buffer(leaf, &cache->item,
+			   btrfs_item_ptr_offset(leaf, slot),
+			   sizeof(cache->item));
+	memcpy(&cache->key, &key, sizeof(key));
+	cache->cached = 0;
+	cache->pinned = 0;
+	cache->flags = btrfs_block_group_flags(&cache->item);
+	if (cache->flags & BTRFS_BLOCK_GROUP_DATA) {
+		bit = BLOCK_GROUP_DATA;
+	} else if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
+		bit = BLOCK_GROUP_SYSTEM;
+	} else if (cache->flags & BTRFS_BLOCK_GROUP_METADATA) {
+		bit = BLOCK_GROUP_METADATA;
+	}
+	set_avail_alloc_bits(fs_info, cache->flags);
+	if (btrfs_chunk_readonly(fs_info, cache->key.objectid))
+		cache->ro = 1;
+	exclude_super_stripes(fs_info, cache);
+
+	ret = update_space_info(fs_info, cache->flags, cache->key.offset,
+				btrfs_block_group_used(&cache->item),
+				&space_info);
+	if (ret < 0) {
+		free(cache);
+		return ret;
+	}
+	cache->space_info = space_info;
+
+	set_extent_bits(block_group_cache, cache->key.objectid,
+			cache->key.objectid + cache->key.offset - 1,
+			bit | EXTENT_LOCKED);
+	set_state_private(block_group_cache, cache->key.objectid,
+			  (unsigned long)cache);
+	return 0;
+}
 
-	root = info->extent_root;
+int btrfs_read_block_groups(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_path path;
+	struct btrfs_root *root;
+	int ret;
+	struct btrfs_key key;
+
+	root = fs_info->extent_root;
 	key.objectid = 0;
 	key.offset = 0;
 	key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
+	btrfs_init_path(&path);
 
 	while(1) {
-		ret = find_first_block_group(root, path, &key);
+		ret = find_first_block_group(root, &path, &key);
 		if (ret > 0) {
 			ret = 0;
 			goto error;
@@ -2673,67 +2739,21 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 		if (ret != 0) {
 			goto error;
 		}
-		leaf = path->nodes[0];
-		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+		btrfs_item_key_to_cpu(path.nodes[0], &key, path.slots[0]);
 
-		cache = kzalloc(sizeof(*cache), GFP_NOFS);
-		if (!cache) {
-			ret = -ENOMEM;
+		ret = read_one_block_group(fs_info, &path);
+		if (ret < 0)
 			goto error;
-		}
 
-		read_extent_buffer(leaf, &cache->item,
-				   btrfs_item_ptr_offset(leaf, path->slots[0]),
-				   sizeof(cache->item));
-		memcpy(&cache->key, &found_key, sizeof(found_key));
-		cache->cached = 0;
-		cache->pinned = 0;
-		key.objectid = found_key.objectid + found_key.offset;
-		if (found_key.offset == 0)
+		if (key.offset == 0)
 			key.objectid++;
-		btrfs_release_path(path);
-
-		/*
-		 * Skip 0 sized block group, don't insert them into block
-		 * group cache tree, as its length is 0, it won't get
-		 * freed at close_ctree() time.
-		 */
-		if (found_key.offset == 0) {
-			free(cache);
-			continue;
-		}
-
-		cache->flags = btrfs_block_group_flags(&cache->item);
-		bit = 0;
-		if (cache->flags & BTRFS_BLOCK_GROUP_DATA) {
-			bit = BLOCK_GROUP_DATA;
-		} else if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
-			bit = BLOCK_GROUP_SYSTEM;
-		} else if (cache->flags & BTRFS_BLOCK_GROUP_METADATA) {
-			bit = BLOCK_GROUP_METADATA;
-		}
-		set_avail_alloc_bits(info, cache->flags);
-		if (btrfs_chunk_readonly(info, cache->key.objectid))
-			cache->ro = 1;
-
-		exclude_super_stripes(info, cache);
-
-		ret = update_space_info(info, cache->flags, found_key.offset,
-					btrfs_block_group_used(&cache->item),
-					&space_info);
-		BUG_ON(ret);
-		cache->space_info = space_info;
-
-		/* use EXTENT_LOCKED to prevent merging */
-		set_extent_bits(block_group_cache, found_key.objectid,
-				found_key.objectid + found_key.offset - 1,
-				bit | EXTENT_LOCKED);
-		set_state_private(block_group_cache, found_key.objectid,
-				  (unsigned long)cache);
+		else
+			key.objectid = key.objectid + key.offset;
+		btrfs_release_path(&path);
 	}
 	ret = 0;
 error:
-	btrfs_free_path(path);
+	btrfs_release_path(&path);
 	return ret;
 }
 
-- 
2.23.0

[PATCH v3 3/7] btrfs-progs: Enable read-write ability for 'bg_tree' feature

[Qu Wenruo]

Allow btrfs-progs to open, read and write 'bg_tree' enabled fs.

The modification itself is not large, as block groups items are only
used at 4 timing:
1) open_ctree()
   We only need to populate fs_info->bg_root and read block group items
   from fs_info->bg_root.
   The obvious change is, we don't need to do btrfs_search_slot() for
   each block group item, but btrfs_next_item() is enough.

   This should hugely reduce open_ctree() execution duration.

2) btrfs_commit_transaction()
   We need to write back dirty block group items back to bg_root.

   The modification here is to insert new block group item if we can't
   find one existing in bg_root, and delete the old one in extent tree
   if we're converting to bg_tree feature.

3) btrfs_make_block_group()
   Just change @root for btrfs_insert_item() from @extent_root to
   @bg_root for fs with that feature.

   This modification needs extra handling for converting case, where
   block group items can be either in extent tree or bg tree.

4) free_block_group_item()
   Just delete the block group item in extent tree.

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 ctree.h       |  11 +++-
 disk-io.c     |  20 +++++++
 extent-tree.c | 152 +++++++++++++++++++++++++++++++++++++++++++++-----
 3 files changed, 168 insertions(+), 15 deletions(-)

diff --git a/ctree.h b/ctree.h
index 2899de358613..c2a18c8ab72f 100644
--- a/ctree.h
+++ b/ctree.h
@@ -89,6 +89,9 @@ struct btrfs_free_space_ctl;
 /* tracks free space in block groups. */
 #define BTRFS_FREE_SPACE_TREE_OBJECTID 10ULL
 
+/* store BLOCK_GROUP_ITEMS in a seperate tree */
+#define BTRFS_BLOCK_GROUP_TREE_OBJECTID 11ULL
+
 /* device stats in the device tree */
 #define BTRFS_DEV_STATS_OBJECTID 0ULL
 
@@ -490,6 +493,7 @@ struct btrfs_super_block {
 #define BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA	(1ULL << 8)
 #define BTRFS_FEATURE_INCOMPAT_NO_HOLES		(1ULL << 9)
 #define BTRFS_FEATURE_INCOMPAT_METADATA_UUID    (1ULL << 10)
+#define BTRFS_FEATURE_INCOMPAT_BG_TREE		(1ULL << 11)
 
 #define BTRFS_FEATURE_COMPAT_SUPP		0ULL
 
@@ -513,7 +517,8 @@ struct btrfs_super_block {
 	 BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS |		\
 	 BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA |	\
 	 BTRFS_FEATURE_INCOMPAT_NO_HOLES |		\
-	 BTRFS_FEATURE_INCOMPAT_METADATA_UUID)
+	 BTRFS_FEATURE_INCOMPAT_METADATA_UUID |		\
+	 BTRFS_FEATURE_INCOMPAT_BG_TREE)
 
 /*
  * A leaf is full of items. offset and size tell us where to find
@@ -1123,6 +1128,7 @@ struct btrfs_fs_info {
 	struct btrfs_root *quota_root;
 	struct btrfs_root *free_space_root;
 	struct btrfs_root *uuid_root;
+	struct btrfs_root *bg_root;
 
 	struct rb_root fs_root_tree;
 
@@ -1175,6 +1181,9 @@ struct btrfs_fs_info {
 	unsigned int avoid_sys_chunk_alloc:1;
 	unsigned int finalize_on_close:1;
 
+	/* Converting from bg in extent tree to bg tree */
+	unsigned int convert_to_bg_tree:1;
+
 	int transaction_aborted;
 
 	int (*free_extent_hook)(struct btrfs_fs_info *fs_info,
diff --git a/disk-io.c b/disk-io.c
index 8978f0cb60c7..38248aa895b8 100644
--- a/disk-io.c
+++ b/disk-io.c
@@ -716,6 +716,8 @@ struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info,
 	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
 		return fs_info->free_space_root ? fs_info->free_space_root :
 						ERR_PTR(-ENOENT);
+	if (location->objectid == BTRFS_BLOCK_GROUP_TREE_OBJECTID)
+		return fs_info->bg_root ? fs_info->bg_root : ERR_PTR(-ENOENT);
 
 	BUG_ON(location->objectid == BTRFS_TREE_RELOC_OBJECTID ||
 	       location->offset != (u64)-1);
@@ -768,6 +770,7 @@ struct btrfs_fs_info *btrfs_new_fs_info(int writable, u64 sb_bytenr)
 	fs_info->quota_root = calloc(1, sizeof(struct btrfs_root));
 	fs_info->free_space_root = calloc(1, sizeof(struct btrfs_root));
 	fs_info->uuid_root = calloc(1, sizeof(struct btrfs_root));
+	fs_info->bg_root = calloc(1, sizeof(struct btrfs_root));
 	fs_info->super_copy = calloc(1, BTRFS_SUPER_INFO_SIZE);
 
 	if (!fs_info->tree_root || !fs_info->extent_root ||
@@ -930,6 +933,21 @@ int btrfs_setup_all_roots(struct btrfs_fs_info *fs_info, u64 root_tree_bytenr,
 		return ret;
 	fs_info->extent_root->track_dirty = 1;
 
+	if (btrfs_fs_incompat(fs_info, BG_TREE)) {
+		ret = setup_root_or_create_block(fs_info, flags,
+					fs_info->bg_root,
+					BTRFS_BLOCK_GROUP_TREE_OBJECTID, "bg");
+		if (ret < 0) {
+			error("Couldn't setup bg tree");
+			return ret;
+		}
+		fs_info->bg_root->track_dirty = 1;
+		fs_info->bg_root->ref_cows = 0;
+	} else {
+		free(fs_info->bg_root);
+		fs_info->bg_root = NULL;
+	}
+
 	ret = find_and_setup_root(root, fs_info, BTRFS_DEV_TREE_OBJECTID,
 				  fs_info->dev_root);
 	if (ret) {
@@ -1012,6 +1030,8 @@ void btrfs_release_all_roots(struct btrfs_fs_info *fs_info)
 		free_extent_buffer(fs_info->free_space_root->node);
 	if (fs_info->quota_root)
 		free_extent_buffer(fs_info->quota_root->node);
+	if (fs_info->bg_root)
+		free_extent_buffer(fs_info->bg_root->node);
 	if (fs_info->csum_root)
 		free_extent_buffer(fs_info->csum_root->node);
 	if (fs_info->dev_root)
diff --git a/extent-tree.c b/extent-tree.c
index 9713d627764c..cb3d7a1add0f 100644
--- a/extent-tree.c
+++ b/extent-tree.c
@@ -1528,22 +1528,68 @@ static int write_one_cache_group(struct btrfs_trans_handle *trans,
 				 struct btrfs_path *path,
 				 struct btrfs_block_group_cache *cache)
 {
+	bool is_bg_tree = btrfs_fs_incompat(trans->fs_info, BG_TREE);
 	int ret;
-	struct btrfs_root *extent_root = trans->fs_info->extent_root;
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_root *root;
 	unsigned long bi;
 	struct extent_buffer *leaf;
 
-	ret = btrfs_search_slot(trans, extent_root, &cache->key, path, 0, 1);
+	if (is_bg_tree)
+		root = fs_info->bg_root;
+	else
+		root = fs_info->extent_root;
+
+	ret = btrfs_search_slot(trans, root, &cache->key, path, 0, 1);
 	if (ret < 0)
-		goto fail;
-	BUG_ON(ret);
+		goto out;
 
-	leaf = path->nodes[0];
-	bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
-	write_extent_buffer(leaf, &cache->item, bi, sizeof(cache->item));
-	btrfs_mark_buffer_dirty(leaf);
-	btrfs_release_path(path);
-fail:
+	if (ret == 0) {
+		/* Update existing bg */
+		leaf = path->nodes[0];
+		bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
+		write_extent_buffer(leaf, &cache->item, bi, sizeof(cache->item));
+		btrfs_mark_buffer_dirty(leaf);
+		btrfs_release_path(path);
+	} else {
+		btrfs_release_path(path);
+
+		/*
+		 * Insert new bg item
+		 *
+		 * This only happens for bg_tree feature
+		 */
+		if (!is_bg_tree) {
+			error("can't find block group item for bytenr %llu",
+			      cache->key.objectid);
+			ret = -ENOENT;
+			goto out;
+		}
+		ret = btrfs_insert_item(trans, root, &cache->key, &cache->item,
+					sizeof(cache->item));
+		if (ret < 0)
+			goto out;
+
+		/* Also delete the existing one in next tree if needed */
+		if (fs_info->convert_to_bg_tree) {
+			ret = btrfs_search_slot(trans, fs_info->extent_root,
+						&cache->key, path, -1, 1);
+			if (ret < 0) {
+				btrfs_release_path(path);
+				goto out;
+			}
+			/* Good, already converted */
+			if (ret > 0) {
+				ret = 0;
+				btrfs_release_path(path);
+				goto out;
+			}
+			/* Delete old block group item in extent tree */
+			ret = btrfs_del_item(trans, fs_info->extent_root, path);
+			btrfs_release_path(path);
+		}
+	}
+out:
 	if (ret)
 		return ret;
 	return 0;
@@ -2717,14 +2763,66 @@ static int read_one_block_group(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
+static int read_block_group_tree(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_root *root = fs_info->bg_root;
+	struct btrfs_key key = { 0 };
+	struct btrfs_path path;
+	int ret;
+
+	btrfs_init_path(&path);
+	ret = btrfs_search_slot(NULL, root, &key, &path, 0, 0);
+	if (ret < 0) {
+		errno = -ret;
+		error("failed to search block group tree: %m");
+		return ret;
+	}
+	if (ret == 0)
+		goto invalid_key;
+
+	while (1) {
+		btrfs_item_key_to_cpu(path.nodes[0], &key, path.slots[0]);
+		if (key.type != BTRFS_BLOCK_GROUP_ITEM_KEY)
+			goto invalid_key;
+
+		ret = read_one_block_group(fs_info, &path);
+		if (ret < 0) {
+			errno = -ret;
+			error("failed to read one block group: %m");
+			goto out;
+		}
+		ret = btrfs_next_item(root, &path);
+		if (ret < 0) {
+			errno = -ret;
+			error("failed to search block group tree: %m");
+			goto out;
+		}
+		if (ret > 0) {
+			ret = 0;
+			break;
+		}
+	}
+out:
+	btrfs_release_path(&path);
+	return ret;
+
+invalid_key:
+	error("invalid key (%llu, %u, %llu) found in block group tree",
+	      key.objectid, key.type, key.offset);
+	btrfs_release_path(&path);
+	return -EUCLEAN;
+}
+
 int btrfs_read_block_groups(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_path path;
-	struct btrfs_root *root;
+	struct btrfs_root *root = fs_info->extent_root;
 	int ret;
 	struct btrfs_key key;
 
-	root = fs_info->extent_root;
+	if (btrfs_fs_incompat(fs_info, BG_TREE))
+		return read_block_group_tree(fs_info);
+
 	key.objectid = 0;
 	key.offset = 0;
 	key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
@@ -2804,12 +2902,17 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 			   u64 type, u64 chunk_offset, u64 size)
 {
 	int ret;
-	struct btrfs_root *extent_root = fs_info->extent_root;
+	struct btrfs_root *root;
 	struct btrfs_block_group_cache *cache;
 
+	if (btrfs_fs_incompat(fs_info, BG_TREE))
+		root = fs_info->bg_root;
+	else
+		root = fs_info->extent_root;
+
 	cache = btrfs_add_block_group(fs_info, bytes_used, type, chunk_offset,
 				      size);
-	ret = btrfs_insert_item(trans, extent_root, &cache->key, &cache->item,
+	ret = btrfs_insert_item(trans, root, &cache->key, &cache->item,
 				sizeof(cache->item));
 	BUG_ON(ret);
 
@@ -2943,8 +3046,16 @@ static int free_block_group_item(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
+	/* Using bg tree only */
+	if (btrfs_fs_incompat(fs_info, BG_TREE) && !fs_info->convert_to_bg_tree)
+		goto bg_tree;
+
 	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 	if (ret > 0) {
+		if (btrfs_fs_incompat(fs_info, BG_TREE)) {
+			btrfs_release_path(path);
+			goto bg_tree;
+		}
 		ret = -ENOENT;
 		goto out;
 	}
@@ -2952,6 +3063,19 @@ static int free_block_group_item(struct btrfs_trans_handle *trans,
 		goto out;
 
 	ret = btrfs_del_item(trans, root, path);
+	goto out;
+
+bg_tree:
+	root = fs_info->bg_root;
+	ret = btrfs_search_slot(trans, fs_info->bg_root, &key, path, -1, 1);
+	if (ret < 0)
+		goto out;
+	if (ret > 0) {
+		ret = -ENOENT;
+		goto out;
+	}
+	ret = btrfs_del_item(trans, root, path);
+
 out:
 	btrfs_free_path(path);
 	return ret;
-- 
2.23.0

[PATCH v3 4/7] btrfs-progs: mkfs: Introduce -O bg-tree

[Qu Wenruo]

This allow mkfs.btrfs to create a btrfs with bg-tree feature.

This patch introduce a global function, btrfs_convert_to_bg_tree() in
extent-tree.c, to do the work.

The workflow is pretty simple:
- Create a new tree block for bg tree
- Set the BG_TREE feature for superblock
- Set the fs_info->convert_to_bg_tree flag
- Mark all block group items as dirty
- Commit transaction
  * With fs_info->convert_to_bg_tree set, we will try to delete the
    BLOCK_GROUP_ITEM in extent tree first, then write the new
    BLOCK_GROUP_ITEM into bg tree.

This btrfs_convert_to_bg_tree() will be used in mkfs after the basic fs
is created.

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 common/fsfeatures.c |  6 ++++++
 ctree.h             |  1 +
 extent-tree.c       | 39 +++++++++++++++++++++++++++++++++++++++
 mkfs/common.c       |  6 ++++--
 mkfs/main.c         | 25 +++++++++++++++++++++++++
 transaction.c       |  1 +
 6 files changed, 76 insertions(+), 2 deletions(-)

diff --git a/common/fsfeatures.c b/common/fsfeatures.c
index 50934bd161b0..b9bd70a4b3b6 100644
--- a/common/fsfeatures.c
+++ b/common/fsfeatures.c
@@ -86,6 +86,12 @@ static const struct btrfs_fs_feature {
 		VERSION_TO_STRING2(4,0),
 		NULL, 0,
 		"no explicit hole extents for files" },
+	{ "bg-tree", BTRFS_FEATURE_INCOMPAT_BG_TREE,
+		"bg_tree",
+		VERSION_TO_STRING2(5, 0),
+		NULL, 0,
+		NULL, 0,
+		"store block group items in dedicated tree" },
 	/* Keep this one last */
 	{ "list-all", BTRFS_FEATURE_LIST_ALL, NULL }
 };
diff --git a/ctree.h b/ctree.h
index c2a18c8ab72f..3d3992487a53 100644
--- a/ctree.h
+++ b/ctree.h
@@ -2856,5 +2856,6 @@ int btrfs_read_file(struct btrfs_root *root, u64 ino, u64 start, int len,
 
 /* extent-tree.c */
 int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans, unsigned long nr);
+int btrfs_convert_to_bg_tree(struct btrfs_trans_handle *trans);
 
 #endif
diff --git a/extent-tree.c b/extent-tree.c
index cb3d7a1add0f..7dda933956ce 100644
--- a/extent-tree.c
+++ b/extent-tree.c
@@ -1524,6 +1524,45 @@ int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 	return __btrfs_mod_ref(trans, root, buf, record_parent, 0);
 }
 
+int btrfs_convert_to_bg_tree(struct btrfs_trans_handle *trans)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_block_group_cache *bg;
+	struct btrfs_root *bg_root;
+	u64 features = btrfs_super_incompat_flags(fs_info->super_copy);
+	int ret;
+
+	/* create bg tree first */
+	bg_root = btrfs_create_tree(trans, fs_info, BTRFS_BLOCK_GROUP_TREE_OBJECTID);
+	if (IS_ERR(bg_root)) {
+		ret = PTR_ERR(bg_root);
+		errno = -ret;
+		error("failed to create bg tree: %m");
+		return ret;
+	}
+	fs_info->bg_root = bg_root;
+	fs_info->bg_root->track_dirty = 1;
+	fs_info->bg_root->ref_cows = 0;
+	add_root_to_dirty_list(bg_root);
+
+	/* set BG_TREE feature and mark the fs into bg_tree convert status */
+	btrfs_set_super_incompat_flags(fs_info->super_copy,
+			features | BTRFS_FEATURE_INCOMPAT_BG_TREE);
+	fs_info->convert_to_bg_tree = 1;
+
+	/*
+	 * Mark all block groups dirty so they will get converted to bg tree at
+	 * commit transaction time
+	 */
+	for (bg = btrfs_lookup_first_block_group(fs_info, 0); bg;
+	     bg = btrfs_lookup_first_block_group(fs_info,
+				bg->key.objectid + bg->key.offset))
+		set_extent_bits(&fs_info->block_group_cache, bg->key.objectid,
+				bg->key.objectid + bg->key.offset - 1,
+				BLOCK_GROUP_DIRTY);
+	return 0;
+}
+
 static int write_one_cache_group(struct btrfs_trans_handle *trans,
 				 struct btrfs_path *path,
 				 struct btrfs_block_group_cache *cache)
diff --git a/mkfs/common.c b/mkfs/common.c
index caca5e707233..876193838612 100644
--- a/mkfs/common.c
+++ b/mkfs/common.c
@@ -111,6 +111,9 @@ static int btrfs_create_tree_root(int fd, struct btrfs_mkfs_config *cfg,
 	return ret;
 }
 
+/* These features will not be set in the temporary fs */
+#define MASKED_FEATURES		(~(BTRFS_FEATURE_INCOMPAT_BG_TREE))
+
 /*
  * @fs_uuid - if NULL, generates a UUID, returns back the new filesystem UUID
  *
@@ -204,7 +207,7 @@ int make_btrfs(int fd, struct btrfs_mkfs_config *cfg)
 	btrfs_set_super_csum_type(&super, BTRFS_CSUM_TYPE_CRC32);
 	btrfs_set_super_chunk_root_generation(&super, 1);
 	btrfs_set_super_cache_generation(&super, -1);
-	btrfs_set_super_incompat_flags(&super, cfg->features);
+	btrfs_set_super_incompat_flags(&super, cfg->features & MASKED_FEATURES);
 	if (cfg->label)
 		__strncpy_null(super.label, cfg->label, BTRFS_LABEL_SIZE - 1);
 
@@ -824,4 +827,3 @@ int test_minimum_size(const char *file, u64 min_dev_size)
 	return 0;
 }
 
-
diff --git a/mkfs/main.c b/mkfs/main.c
index b752da13aba9..55bc4288dc08 100644
--- a/mkfs/main.c
+++ b/mkfs/main.c
@@ -1312,6 +1312,31 @@ raid_groups:
 		goto out;
 	}
 
+	/*
+	 * Bg tree are converted after temp chunks cleaned up, or we can
+	 * populate temp chunks.
+	 */
+	if (mkfs_cfg.features & BTRFS_FEATURE_INCOMPAT_BG_TREE) {
+		trans = btrfs_start_transaction(fs_info->tree_root, 1);
+		if (IS_ERR(trans)) {
+			error("failed to start transaction: %d", ret);
+			goto out;
+		}
+		ret = btrfs_convert_to_bg_tree(trans);
+		if (ret < 0) {
+			errno = -ret;
+			error(
+		"bg-tree feature will not be enabled, due to error: %m");
+			btrfs_abort_transaction(trans, ret);
+			goto out;
+		}
+		ret = btrfs_commit_transaction(trans, fs_info->tree_root);
+		if (ret < 0) {
+			error("failed to commit transaction: %d", ret);
+			goto out;
+		}
+	}
+
 	if (source_dir_set) {
 		ret = btrfs_mkfs_fill_dir(source_dir, root, verbose);
 		if (ret) {
diff --git a/transaction.c b/transaction.c
index 45bb9e1f9de6..5de967fb015f 100644
--- a/transaction.c
+++ b/transaction.c
@@ -225,6 +225,7 @@ commit_tree:
 	root->commit_root = NULL;
 	fs_info->running_transaction = NULL;
 	fs_info->last_trans_committed = transid;
+	fs_info->convert_to_bg_tree = 0;
 	list_for_each_entry(sinfo, &fs_info->space_info, list) {
 		if (sinfo->bytes_reserved) {
 			warning(
-- 
2.23.0

[PATCH v3 5/7] btrfs-progs: dump-tree/dump-super: Introduce support for bg tree

[Qu Wenruo]

Just a new tree called BLOCK_GROUP_TREE.

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 cmds/inspect-dump-super.c | 3 ++-
 cmds/inspect-dump-tree.c  | 5 +++++
 print-tree.c              | 3 +++
 3 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/cmds/inspect-dump-super.c b/cmds/inspect-dump-super.c
index 65fb3506eac6..414d9c2317d8 100644
--- a/cmds/inspect-dump-super.c
+++ b/cmds/inspect-dump-super.c
@@ -229,7 +229,8 @@ static struct readable_flag_entry incompat_flags_array[] = {
 	DEF_INCOMPAT_FLAG_ENTRY(RAID56),
 	DEF_INCOMPAT_FLAG_ENTRY(SKINNY_METADATA),
 	DEF_INCOMPAT_FLAG_ENTRY(NO_HOLES),
-	DEF_INCOMPAT_FLAG_ENTRY(METADATA_UUID)
+	DEF_INCOMPAT_FLAG_ENTRY(METADATA_UUID),
+	DEF_INCOMPAT_FLAG_ENTRY(BG_TREE)
 };
 static const int incompat_flags_num = sizeof(incompat_flags_array) /
 				      sizeof(struct readable_flag_entry);
diff --git a/cmds/inspect-dump-tree.c b/cmds/inspect-dump-tree.c
index e50130a4a161..def5bea39a2b 100644
--- a/cmds/inspect-dump-tree.c
+++ b/cmds/inspect-dump-tree.c
@@ -150,6 +150,7 @@ static u64 treeid_from_string(const char *str, const char **end)
 		{ "CSUM", BTRFS_CSUM_TREE_OBJECTID },
 		{ "CHECKSUM", BTRFS_CSUM_TREE_OBJECTID },
 		{ "QUOTA", BTRFS_QUOTA_TREE_OBJECTID },
+		{ "BG", BTRFS_BLOCK_GROUP_TREE_OBJECTID },
 		{ "UUID", BTRFS_UUID_TREE_OBJECTID },
 		{ "FREE_SPACE", BTRFS_FREE_SPACE_TREE_OBJECTID },
 		{ "TREE_LOG_FIXUP", BTRFS_TREE_LOG_FIXUP_OBJECTID },
@@ -661,6 +662,10 @@ again:
 				if (!skip)
 					printf("free space");
 				break;
+			case BTRFS_BLOCK_GROUP_TREE_OBJECTID:
+				if (!skip)
+					printf("block group");
+				break;
 			case BTRFS_MULTIPLE_OBJECTIDS:
 				if (!skip) {
 					printf("multiple");
diff --git a/print-tree.c b/print-tree.c
index e079f1a971d3..e2a43226ff87 100644
--- a/print-tree.c
+++ b/print-tree.c
@@ -770,6 +770,9 @@ void print_objectid(FILE *stream, u64 objectid, u8 type)
 	case BTRFS_FREE_SPACE_TREE_OBJECTID:
 		fprintf(stream, "FREE_SPACE_TREE");
 		break;
+	case BTRFS_BLOCK_GROUP_TREE_OBJECTID:
+		fprintf(stream, "BLOCK_GROUP_TREE");
+		break;
 	case BTRFS_MULTIPLE_OBJECTIDS:
 		fprintf(stream, "MULTIPLE");
 		break;
-- 
2.23.0

[PATCH v3 6/7] btrfs-progs: check: Introduce support for bg-tree feature

[Qu Wenruo]

Just some minor modification.

- original mode:
  * Block group item can occur in extent tree and bg tree.
- lowmem mode:
  * search block group items in bg tree if BG_TREE feature is set.

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 check/main.c        | 3 ++-
 check/mode-lowmem.c | 9 +++++++--
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/check/main.c b/check/main.c
index 94ffab46cb70..3fd0bb2317bb 100644
--- a/check/main.c
+++ b/check/main.c
@@ -6034,7 +6034,8 @@ static int check_type_with_root(u64 rootid, u8 key_type)
 	case BTRFS_EXTENT_ITEM_KEY:
 	case BTRFS_METADATA_ITEM_KEY:
 	case BTRFS_BLOCK_GROUP_ITEM_KEY:
-		if (rootid != BTRFS_EXTENT_TREE_OBJECTID)
+		if (rootid != BTRFS_EXTENT_TREE_OBJECTID &&
+		    rootid != BTRFS_BLOCK_GROUP_TREE_OBJECTID)
 			goto err;
 		break;
 	case BTRFS_ROOT_ITEM_KEY:
diff --git a/check/mode-lowmem.c b/check/mode-lowmem.c
index 5f7f101daab1..fcb8210984eb 100644
--- a/check/mode-lowmem.c
+++ b/check/mode-lowmem.c
@@ -4365,7 +4365,7 @@ next:
 static int check_chunk_item(struct btrfs_fs_info *fs_info,
 			    struct extent_buffer *eb, int slot)
 {
-	struct btrfs_root *extent_root = fs_info->extent_root;
+	struct btrfs_root *root;
 	struct btrfs_root *dev_root = fs_info->dev_root;
 	struct btrfs_path path;
 	struct btrfs_key chunk_key;
@@ -4387,6 +4387,11 @@ static int check_chunk_item(struct btrfs_fs_info *fs_info,
 	int ret;
 	int err = 0;
 
+	if (btrfs_fs_incompat(fs_info, BG_TREE))
+		root = fs_info->bg_root;
+	else
+		root = fs_info->extent_root;
+
 	btrfs_item_key_to_cpu(eb, &chunk_key, slot);
 	chunk = btrfs_item_ptr(eb, slot, struct btrfs_chunk);
 	length = btrfs_chunk_length(eb, chunk);
@@ -4406,7 +4411,7 @@ static int check_chunk_item(struct btrfs_fs_info *fs_info,
 	bg_key.offset = length;
 
 	btrfs_init_path(&path);
-	ret = btrfs_search_slot(NULL, extent_root, &bg_key, &path, 0, 0);
+	ret = btrfs_search_slot(NULL, root, &bg_key, &path, 0, 0);
 	if (ret) {
 		error(
 		"chunk[%llu %llu) did not find the related block group item",
-- 
2.23.0

[PATCH v3 7/7] btrfs-progs: btrfstune: Allow to enable bg-tree feature offline

[Qu Wenruo]

Add a new option '-b' for btrfstune, to enable bg-tree feature for a
unmounted fs.

This feature will convert all BLOCK_GROUP_ITEMs in extent tree to bg
tree, by reusing the existing btrfs_convert_to_bg_tree() function.

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 Documentation/btrfstune.asciidoc |  6 +++++
 btrfstune.c                      | 44 ++++++++++++++++++++++++++++++--
 2 files changed, 48 insertions(+), 2 deletions(-)

diff --git a/Documentation/btrfstune.asciidoc b/Documentation/btrfstune.asciidoc
index 1d6bc98deed8..ed54c2e1597f 100644
--- a/Documentation/btrfstune.asciidoc
+++ b/Documentation/btrfstune.asciidoc
@@ -26,6 +26,12 @@ means.  Please refer to the 'FILESYSTEM FEATURES' in `btrfs`(5).
 OPTIONS
 -------
 
+-b::
+(since kernel: 5.x)
++
+enable bg-tree feature (faster mount time for large fs), enabled by mkfs
+feature 'bg-tree'.
+
 -f::
 Allow dangerous changes, e.g. clear the seeding flag or change fsid. Make sure
 that you are aware of the dangers.
diff --git a/btrfstune.c b/btrfstune.c
index afa3aae35412..aa1ac568aef0 100644
--- a/btrfstune.c
+++ b/btrfstune.c
@@ -476,11 +476,39 @@ static void print_usage(void)
 	printf("\t-m          change fsid in metadata_uuid to a random UUID\n");
 	printf("\t            (incompat change, more lightweight than -u|-U)\n");
 	printf("\t-M UUID     change fsid in metadata_uuid to UUID\n");
+	printf("\t-b          enable bg-tree feature (mkfs: bg-tree, for faster mount time)\n");
 	printf("  general:\n");
 	printf("\t-f          allow dangerous operations, make sure that you are aware of the dangers\n");
 	printf("\t--help      print this help\n");
 }
 
+static int convert_to_bg_tree(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_trans_handle *trans;
+	int ret;
+
+	trans = btrfs_start_transaction(fs_info->tree_root, 1);
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
+		errno = -ret;
+		error("failed to start transaction: %m");
+		return ret;
+	}
+	ret = btrfs_convert_to_bg_tree(trans);
+	if (ret < 0) {
+		errno = -ret;
+		error("failed to convert: %m");
+		btrfs_abort_transaction(trans, ret);
+		return ret;
+	}
+	ret = btrfs_commit_transaction(trans, fs_info->tree_root);
+	if (ret < 0) {
+		errno = -ret;
+		error("failed to commit transaction: %m");
+	}
+	return ret;
+}
+
 int BOX_MAIN(btrfstune)(int argc, char *argv[])
 {
 	struct btrfs_root *root;
@@ -491,6 +519,7 @@ int BOX_MAIN(btrfstune)(int argc, char *argv[])
 	u64 seeding_value = 0;
 	int random_fsid = 0;
 	int change_metadata_uuid = 0;
+	bool to_bg_tree = false;
 	char *new_fsid_str = NULL;
 	int ret;
 	u64 super_flags = 0;
@@ -501,7 +530,7 @@ int BOX_MAIN(btrfstune)(int argc, char *argv[])
 			{ "help", no_argument, NULL, GETOPT_VAL_HELP},
 			{ NULL, 0, NULL, 0 }
 		};
-		int c = getopt_long(argc, argv, "S:rxfuU:nmM:", long_options, NULL);
+		int c = getopt_long(argc, argv, "S:rxfuU:nmM:b", long_options, NULL);
 
 		if (c < 0)
 			break;
@@ -539,6 +568,9 @@ int BOX_MAIN(btrfstune)(int argc, char *argv[])
 			ctree_flags |= OPEN_CTREE_IGNORE_FSID_MISMATCH;
 			change_metadata_uuid = 1;
 			break;
+		case 'b':
+			to_bg_tree = true;
+			break;
 		case GETOPT_VAL_HELP:
 		default:
 			print_usage();
@@ -556,7 +588,7 @@ int BOX_MAIN(btrfstune)(int argc, char *argv[])
 		return 1;
 	}
 	if (!super_flags && !seeding_flag && !(random_fsid || new_fsid_str) &&
-	    !change_metadata_uuid) {
+	    !change_metadata_uuid && !to_bg_tree) {
 		error("at least one option should be specified");
 		print_usage();
 		return 1;
@@ -602,6 +634,14 @@ int BOX_MAIN(btrfstune)(int argc, char *argv[])
 		return 1;
 	}
 
+	if (to_bg_tree) {
+		ret = convert_to_bg_tree(root->fs_info);
+		if (ret < 0) {
+			errno = -ret;
+			error("failed to convert to bg-tree feature: %m");
+			goto out;
+		}
+	}
 	if (seeding_flag) {
 		if (btrfs_fs_incompat(root->fs_info, METADATA_UUID)) {
 			fprintf(stderr, "SEED flag cannot be changed on a metadata-uuid changed fs\n");
-- 
2.23.0

Re: [PATCH v3 0/7] btrfs-progs: Support for BG_TREE feature

[David Sterba]

On Thu, Oct 10, 2019 at 02:41:49PM +0800, Qu Wenruo wrote:
> This patchset can be fetched from github:
> https://github.com/adam900710/btrfs-progs/tree/bg_tree
> Which is based on v5.2.2 tag.
> 
> This patchset provides the needed user space infrastructure for BG_TREE
> feature.
> 
> Since it's an new incompatible feature, unlike SKINNY_METADATA, btrfs-progs
> is needed to convert existing fs (unmounted) to new format.
> 
> Now btrfstune can convert regular extent tree fs to bg tree fs to
> improve mount time.
> 
> For the performance improvement, please check the kernel patchset cover
> letter or the last patch.
> (SPOILER ALERT: It's super fast)
> 
> Changelog:
> v2:
> - Rebase to v5.2.2 tag
> - Add btrfstune ability to convert existing fs to BG_TREE feature
> 
> v3:
> - Fix a bug that temp chunks are not cleaned up properly
>   This is caused by wrong timing btrfs_convert_to_bg_tree() is called.
>   It should be called after temp chunks cleaned up.
> 
> - Fix a bug that an extent buffer get leaked
>   This is caused by newly created bg tree not added to dirty list.
> 
> Qu Wenruo (7):
>   btrfs-progs: Refactor excluded extent functions to use fs_info
>   btrfs-progs: Refactor btrfs_read_block_groups()

I'll add 1 and 2 to devel as they're independent. Thanks.