lk/linux-kernel-test
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Btrfs: Split the extent_map code into two parts

Browse Source 
There is now extent_map for mapping offsets in the file to disk and
extent_io for state tracking, IO submission and extent_bufers.

The new extent_map code shifts from [start,end] pairs to [start,len], and
pushes the locking out into the caller.  This allows a few performance
optimizations and is easier to use.

A number of extent_map usage bugs were fixed, mostly with failing
to remove extent_map entries when changing the file.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
This commit is contained in:
Chris Mason
2008-01-24 16:13:08 -05:00
parent 5f56406aab
commit d1310b2e0c
14 changed files with 3614 additions and 3415 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
fs/btrfs/transaction.c
View File

@ -70,7 +70,7 @@ static int join_transaction(struct btrfs_root *root)
INIT_LIST_HEAD(&cur_trans->pending_snapshots);
list_add_tail(&cur_trans->list, &root->fs_info->trans_list);
btrfs_ordered_inode_tree_init(&cur_trans->ordered_inode_tree);
extent_map_tree_init(&cur_trans->dirty_pages,
extent_io_tree_init(&cur_trans->dirty_pages,
root->fs_info->btree_inode->i_mapping,
GFP_NOFS);
} else {
@ -153,7 +153,7 @@ int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
int ret;
int err;
int werr = 0;
struct extent_map_tree *dirty_pages;
struct extent_io_tree *dirty_pages;
struct page *page;
struct inode *btree_inode = root->fs_info->btree_inode;
u64 start;
@ -610,7 +610,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
struct btrfs_transaction *cur_trans;
struct btrfs_transaction *prev_trans = NULL;
struct list_head dirty_fs_roots;
struct extent_map_tree *pinned_copy;
struct extent_io_tree *pinned_copy;
DEFINE_WAIT(wait);
int ret;
@ -639,7 +639,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
if (!pinned_copy)
return -ENOMEM;
extent_map_tree_init(pinned_copy,
extent_io_tree_init(pinned_copy,
root->fs_info->btree_inode->i_mapping, GFP_NOFS);
trans->transaction->in_commit = 1;