[RFC PATCH] xfs: fix calculation of freed inode cluster blocks

[RFC PATCH] xfs: fix calculation of freed inode cluster blocks

[Ben Myers]

rec.ir_startino is an agino rather than an ino.  Use the correct macro
when dealing with it in xfs_difree.

---

This is very lightly tested so it's RFC.  Looks like using "ino to
agbno" is actually harmless when you pass in an agino, so there
shouldn't have been any trouble caused by this one.

-Ben

 fs/xfs/xfs_ialloc.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: b/fs/xfs/xfs_ialloc.c
===================================================================
--- a/fs/xfs/xfs_ialloc.c	2013-12-03 12:07:24.485185962 -0600
+++ b/fs/xfs/xfs_ialloc.c	2013-12-03 12:07:40.994563706 -0600
@@ -1229,7 +1229,7 @@ xfs_difree(
 		}
 
 		xfs_bmap_add_free(XFS_AGB_TO_FSB(mp,
-				agno, XFS_INO_TO_AGBNO(mp,rec.ir_startino)),
+				agno, XFS_AGINO_TO_AGBNO(mp, rec.ir_startino)),
 				XFS_IALLOC_BLOCKS(mp), flist, mp);
 	} else {
 		*delete = 0;

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [RFC PATCH] xfs: fix calculation of freed inode cluster blocks

[Dave Chinner]

On Tue, Dec 03, 2013 at 03:00:54PM -0600, Ben Myers wrote:
> rec.ir_startino is an agino rather than an ino.  Use the correct macro
> when dealing with it in xfs_difree.
> 
> ---
> 
> This is very lightly tested so it's RFC.  Looks like using "ino to
> agbno" is actually harmless when you pass in an agino, so there
> shouldn't have been any trouble caused by this one.

Right, the XFS_INO_TO_AGBNO() macro will behave correctly when
passed an agino - it's the same as an inode from AG 0. Changing the
macro to the correct one is effectively a no-op, so just send a
patch and get it reviewed and integrated...

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs