f2fs: sync dir->i_size with its block allocation
authorJaegeuk Kim <jaegeuk.kim@samsung.com>
Fri, 7 Jun 2013 13:08:23 +0000 (22:08 +0900)
committerJaegeuk Kim <jaegeuk.kim@samsung.com>
Tue, 11 Jun 2013 23:18:35 +0000 (08:18 +0900)
commit699489bbbea4fc3b9b735d69941cf4fca91ce1d5
tree4b3d474252dfc93f0f4b55145d6d7193d0cbcd13
parent2d4d9fb591fe83d9f0559afaa9736ebc8edad0aa
f2fs: sync dir->i_size with its block allocation

If new dentry block is allocated and its i_size is updated, we should update
its inode block together in order to sync i_size and its block allocation.
Otherwise, we can loose additional dentry block due to the unconsistent i_size.

Errorneous Scenario
-------------------

In the recovery routine,
 - recovery_dentry
 | - __f2fs_add_link
 | | - get_new_data_page
 | | | - i_size_write(new_i_size)
 | | | - mark_inode_dirty_sync(dir)
 | | - update_parent_metadata
 | | | - mark_inode_dirty(dir)
 |
 - write_checkpoint
   - sync_dirty_dir_inodes
     - filemap_flush(dentry_blocks)
       - f2fs_write_data_page
         - skip to write the last dentry block due to index < i_size

In the above flow, new_i_size is not updated to its inode block so that the
last dentry block will be lost accordingly.

Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
fs/f2fs/data.c
fs/f2fs/dir.c
fs/f2fs/f2fs.h