ext4: Assure that metadata blocks are written during fsync in no journal mode
authorTheodore Ts'o <tytso@mit.edu>
Sat, 12 Sep 2009 17:41:55 +0000 (13:41 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Sat, 12 Sep 2009 17:41:55 +0000 (13:41 -0400)
commitfe188c0e084bdf3038dc0ac963c21d764f53f7da
treebb712f58f4c0356058825da75da57d0af450e01b
parentc7acb4c16646943180bd221c167a077e0a084f9c
ext4: Assure that metadata blocks are written during fsync in no journal mode

When there is no journal present, we must attach buffer heads
associated with extent tree and indirect blocks to the inode's
mapping->private_list via mark_buffer_dirty_inode() so that
ext4_sync_file() --- which is called to service fsync() and
fdatasync() system calls --- can write out the inode's metadata blocks
by calling sync_mapping_buffers().

Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
fs/ext4/ext4_jbd2.c
fs/ext4/fsync.c