Re: [PATCH 3/5] direct-io: honor dio->boundary a little more strictly

[Josef Bacik <josef@redhat.com>]

On Fri, May 07, 2010 at 01:41:04PM -0400, Josef Bacik wrote:
> Because BTRFS needs to be able to lookup checksums when we submit the bio's, we
> need to be able to look up the logical offset in the inode we're submitting the
> bio for.  The way we do this is in our get_blocks function is return the map_bh
> with b_blocknr of the logical offset in the file, and then in the submit path
> turn that into an actual block number on the device.  This causes problems with
> the DIO stuff since it will try and merge requests that look like they are
> contiguous, even though they are not actually contiguous on disk.  So BTRFS sets
> buffer_boundary on the map_bh.  Unfortunately if there is not a bio already
> setup in the DIO stuff, dio->boundary gets cleared and then the next time a
> request is made they will get merged.  So instead of clearing dio->boundary in
> dio_new_bio, save the boundary value before doing anything, that way if
> dio->boundary gets cleared, we still submit the IO.  Thanks,
> 
> Signed-off-by: Josef Bacik <josef@redhat.com>
> ---
>  fs/direct-io.c |    5 ++++-
>  1 files changed, 4 insertions(+), 1 deletions(-)
> 
> diff --git a/fs/direct-io.c b/fs/direct-io.c
> index 2dbf2e9..98f6f42 100644
> --- a/fs/direct-io.c
> +++ b/fs/direct-io.c
> @@ -615,6 +615,7 @@ static int dio_bio_add_page(struct dio *dio)
>   */
>  static int dio_send_cur_page(struct dio *dio)
>  {
> +	int boundary = dio->boundary;
>  	int ret = 0;
>  
>  	if (dio->bio) {
> @@ -627,7 +628,7 @@ static int dio_send_cur_page(struct dio *dio)
>  		 * Submit now if the underlying fs is about to perform a
>  		 * metadata read
>  		 */
> -		if (dio->boundary)
> +		if (boundary)
>  			dio_bio_submit(dio);
>  	}
>  
> @@ -644,6 +645,8 @@ static int dio_send_cur_page(struct dio *dio)
>  			ret = dio_bio_add_page(dio);
>  			BUG_ON(ret != 0);
>  		}
> +	} else if (boundary) {
> +		dio_bio_submit(dio);
>  	}
>  out:
>  	return ret;

Self-NACK on this one.  Seems to have an unwanted side-effect of forcing every
page to be submitted individually.  I'm going to fix this a different way.
Thanks,

Josef

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>