xfs: fix xfs_get_extsz_hint for a zero extent size hint

We can easily set the extsize flag without setting an extent size
hint, or one that evaluates to zero.  Historically the di_extsize
field was only used when it was non-zero, but the commit

	"Cleanup inode extent size hint extraction"

broke this.  Restore the old behaviour, thus fixing xfsqa 090 with
a debug kernel.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>

1 files changed, 5 insertions, 13 deletions

diff --git a/fs/xfs/xfs_rw.c b/fs/xfs/xfs_rw.c
index 56861d5daae..ccd3adf640e 100644
--- a/fs/xfs/xfs_rw.c
+++ b/fs/xfs/xfs_rw.c
@@ -173,17 +173,9 @@ xfs_extlen_t
 xfs_get_extsz_hint(
         struct xfs_inode        *ip)
 {
-        xfs_extlen_t            extsz;
-
-        if (unlikely(XFS_IS_REALTIME_INODE(ip))) {
-                extsz = (ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE)
-                                ? ip->i_d.di_extsize
-                                : ip->i_mount->m_sb.sb_rextsize;
-                ASSERT(extsz);
-        } else {
-                extsz = (ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE)
-                                ? ip->i_d.di_extsize : 0;
-        }
-
-        return extsz;
+        if ((ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE) && ip->i_d.di_extsize)
+                return ip->i_d.di_extsize;
+        if (XFS_IS_REALTIME_INODE(ip))
+                return ip->i_mount->m_sb.sb_rextsize;
+        return 0;
 }