Btrfs: incompatible format change to remove hole extents
Btrfs has always had these filler extent data items for holes in inodes. This
has made somethings very easy, like logging hole punches and sending hole
punches. However for large holey files these extent data items are pure
overhead. So add an incompatible feature to no longer add hole extents to
reduce the amount of metadata used by these sort of files. This has a few
changes for logging and send obviously since they will need to detect holes and
log/send the holes if there are any. I've tested this thoroughly with xfstests
and it doesn't cause any issues with and without the incompat format set.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <clm@fb.com>
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index f1a7744..c0c0dc8 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4203,6 +4203,49 @@
return ret;
}
+static int maybe_insert_hole(struct btrfs_root *root, struct inode *inode,
+ u64 offset, u64 len)
+{
+ struct btrfs_trans_handle *trans;
+ int ret;
+
+ /*
+ * Still need to make sure the inode looks like it's been updated so
+ * that any holes get logged if we fsync.
+ */
+ if (btrfs_fs_incompat(root->fs_info, NO_HOLES)) {
+ BTRFS_I(inode)->last_trans = root->fs_info->generation;
+ BTRFS_I(inode)->last_sub_trans = root->log_transid;
+ BTRFS_I(inode)->last_log_commit = root->last_log_commit;
+ return 0;
+ }
+
+ /*
+ * 1 - for the one we're dropping
+ * 1 - for the one we're adding
+ * 1 - for updating the inode.
+ */
+ trans = btrfs_start_transaction(root, 3);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
+
+ ret = btrfs_drop_extents(trans, root, inode, offset, offset + len, 1);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ btrfs_end_transaction(trans, root);
+ return ret;
+ }
+
+ ret = btrfs_insert_file_extent(trans, root, btrfs_ino(inode), offset,
+ 0, 0, len, 0, len, 0, 0, 0);
+ if (ret)
+ btrfs_abort_transaction(trans, root, ret);
+ else
+ btrfs_update_inode(trans, root, inode);
+ btrfs_end_transaction(trans, root);
+ return ret;
+}
+
/*
* This function puts in dummy file extents for the area we're creating a hole
* for. So if we are truncating this file to a larger size we need to insert
@@ -4211,7 +4254,6 @@
*/
int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
{
- struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
struct extent_map *em = NULL;
@@ -4266,31 +4308,10 @@
struct extent_map *hole_em;
hole_size = last_byte - cur_offset;
- trans = btrfs_start_transaction(root, 3);
- if (IS_ERR(trans)) {
- err = PTR_ERR(trans);
+ err = maybe_insert_hole(root, inode, cur_offset,
+ hole_size);
+ if (err)
break;
- }
-
- 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);
- break;
- }
-
- err = btrfs_insert_file_extent(trans, root,
- btrfs_ino(inode), cur_offset, 0,
- 0, hole_size, 0, hole_size,
- 0, 0, 0);
- if (err) {
- btrfs_abort_transaction(trans, root, err);
- btrfs_end_transaction(trans, root);
- break;
- }
-
btrfs_drop_extent_cache(inode, cur_offset,
cur_offset + hole_size - 1, 0);
hole_em = alloc_extent_map();
@@ -4309,7 +4330,7 @@
hole_em->ram_bytes = hole_size;
hole_em->bdev = root->fs_info->fs_devices->latest_bdev;
hole_em->compress_type = BTRFS_COMPRESS_NONE;
- hole_em->generation = trans->transid;
+ hole_em->generation = root->fs_info->generation;
while (1) {
write_lock(&em_tree->lock);
@@ -4322,17 +4343,14 @@
hole_size - 1, 0);
}
free_extent_map(hole_em);
-next:
- btrfs_update_inode(trans, root, inode);
- btrfs_end_transaction(trans, root);
}
+next:
free_extent_map(em);
em = NULL;
cur_offset = last_byte;
if (cur_offset >= block_end)
break;
}
-
free_extent_map(em);
unlock_extent_cached(io_tree, hole_start, block_end - 1, &cached_state,
GFP_NOFS);