[XFS] If you mount an XFS filesystem with no mount options at all, then

the "ikeep" option is set rather than "noikeep".

This regression was introduced in 970451.

With no mount options specified, xfs_parseargs() does the following:

int ikeep = 0;

args->flags |= XFSMNT_BARRIER;

args->flags2 |= XFSMNT2_COMPAT_IOSIZE;

if (!options)

goto done;

It only sets the above two options by default and before, it also used to
set XFSMNT_IDELETE by default.

If options are specified, then

if (!(args->flags & XFSMNT_DMAPI) && !ikeep)

args->flags |= XFSMNT_IDELETE;

is executed later on which is skipped by the "goto done;" above.

The solution is to invert the logic.

SGI-PV: 977771
SGI-Modid: xfs-linux-melb:xfs-kern:30590a

Signed-off-by: Niv Sardi <xaiki@sgi.com>
Signed-off-by: Barry Naujok <bnaujok@sgi.com>
Signed-off-by: Josef 'Jeff' Sipek <jeffpc@josefsipek.net>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>

1 files changed, 1 insertions, 1 deletions

diff --git a/fs/xfs/xfs_clnt.h b/fs/xfs/xfs_clnt.h
index d16c1b971074..d5d1e60ee224 100644
--- a/fs/xfs/xfs_clnt.h
+++ b/fs/xfs/xfs_clnt.h
@@ -86,7 +86,7 @@ struct xfs_mount_args {
 #define XFSMNT_NOUUID           0x01000000      /* Ignore fs uuid */
 #define XFSMNT_DMAPI            0x02000000      /* enable dmapi/xdsm */
 #define XFSMNT_BARRIER          0x04000000      /* use write barriers */
-#define XFSMNT_IDELETE          0x08000000      /* inode cluster delete */
+#define XFSMNT_IKEEP            0x08000000      /* inode cluster delete */
 #define XFSMNT_SWALLOC          0x10000000      /* turn on stripe width
                                                  * allocation */
 #define XFSMNT_DIRSYNC          0x40000000      /* sync creat,link,unlink,rename