btrfs-progs: Don't abuse READA_* for extent tree search

Commit Message

Qu Wenruo May 1, 2020, 5:29 a.m. UTC

For extent tree search, we are only search two things: either
EXTENT_ITEM/METADATA_ITEM (inlined) or SHARED_BLOCK_REF/SHARED_DATA_REF
(keyed).

Except certain situation like cache_block_group(), we never read tree
blocks in a forward or backward sequence.

So remove those reada abuse.

Signed-off-by: Qu Wenruo <[e-mail]>
---
 extent-tree.c | 8 --------
 1 file changed, 8 deletions(-)

Patch

diff --git a/extent-tree.c b/extent-tree.c
index 4af8f4ba8a47..732f90207b25 100644
--- a/extent-tree.c
+++ b/extent-tree.c
@@ -1247,8 +1247,6 @@  int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->reada = READA_BACK;
-
 	ret = insert_inline_extent_backref(trans, root->fs_info->extent_root,
 					   path, bytenr, num_bytes, parent,
 					   root_objectid, owner, offset, 1);
@@ -1268,8 +1266,6 @@  int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_release_path(path);
 
-	path->reada = READA_BACK;
-
 	/* now insert the actual backref */
 	ret = insert_extent_backref(trans, root->fs_info->extent_root,
 				    path, bytenr, parent, root_objectid,
@@ -1303,7 +1299,6 @@  int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-	path->reada = READA_BACK;
 
 	key.objectid = bytenr;
 	key.offset = offset;
@@ -1383,7 +1378,6 @@  int btrfs_set_block_flags(struct btrfs_trans_handle *trans, u64 bytenr,
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-	path->reada = READA_BACK;
 
 	key.objectid = bytenr;
 	if (skinny_metadata) {
@@ -1928,8 +1922,6 @@  static int __free_extent(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->reada = READA_BACK;
-
 	is_data = owner_objectid >= BTRFS_FIRST_FREE_OBJECTID;
 	if (is_data)
 		skinny_metadata = 0;