| Message ID | 1309205864-13124-3-git-send-email-vgoyal@redhat.com |
|---|---|
| State | New, archived |
| Headers | show |
diff --git a/fs/ext4/fsync.c b/fs/ext4/fsync.c index ce66d2f..8b16a90 100644 --- a/fs/ext4/fsync.c +++ b/fs/ext4/fsync.c @@ -216,7 +216,10 @@ int ext4_sync_file(struct file *file, int datasync) !jbd2_trans_will_send_data_barrier(journal, commit_tid)) needs_barrier = true; jbd2_log_start_commit(journal, commit_tid); + blk_set_depends_on_task(journal->j_dev->bd_disk->queue, + journal->j_task); ret = jbd2_log_wait_commit(journal, commit_tid); + blk_reset_depends_on_task(journal->j_dev->bd_disk->queue); if (needs_barrier) blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL, NULL); out:
Set/reset fsync dependency of fsync on journalling thread. This allows CFQ to dispatch IO from journalling thread in fsync's time slice. Otherwise, lots of cfq queue idling takes place if fsync is running in a separate cgroup and journalling thread is running in root group. Signed-off-by: Vivek Goyal <vgoyal@redhat.com> --- fs/ext4/fsync.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)