Re: ext4: call blkdev_issue_flush on fsync

[Theodore Tso <tytso@mit.edu>]

On Thu, Feb 12, 2009 at 07:40:45PM +0900, Fernando Luis Vázquez Cao wrote:
> @@ -76,25 +77,34 @@ int ext4_sync_file(struct file *file, st
>  	 */
>  	if (ext4_should_journal_data(inode)) {
>  		ret = ext4_force_commit(inode->i_sb);
> +		if (!(journal->j_flags & JBD2_BARRIER))
> +			goto no_journal_barrier;
>  		goto out;
>  	}

All of these goto statements makes it one gigantic pile of spaghetti.
The code will be much more understable if you do:

		if (!(journal->j_flags & JBD2_BARRIER))
			block_flush_device(inode->i_sb);
		return ret;	

>  
> -	if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))
> -		goto out;
> +	if (datasync && !(i_state & I_DIRTY_DATASYNC))
> +		goto flush_blkdev;
>  

Maybe instead:
	if (datasync && !(i_state & I_DIRTY_DATASYNC)) {
		if (i_state & I_DIRTY_PAGES)
			block_flush_device(inode->i_sb);
		return ret;
	}


> -		if (journal && (journal->j_flags & JBD2_BARRIER))
> -			blkdev_issue_flush(inode->i_sb->s_bdev, NULL);
> +		if (journal && !(journal->j_flags & JBD2_BARRIER))
> +			goto no_journal_barrier;
> +		goto out;

Maybe instead:
		if (journal && !(journal->j_flags & JBD2_BARRIER))
			block_flush_device(inode->i_sb);
		return ret;
  	}

I'm not a fanatic about eliminating all goto's, but "goto out" which
could be replaced by "return ret;" is just silly.

      	 	     	     	      - Ted