Mask root object ID into f_fsid in btrfs_statfs()
authorDavid Woodhouse <David.Woodhouse@intel.com>
Mon, 18 Aug 2008 12:10:20 +0000 (13:10 +0100)
committerChris Mason <chris.mason@oracle.com>
Thu, 25 Sep 2008 15:04:06 +0000 (11:04 -0400)
Date: Mon, 18 Aug 2008 13:10:20 +0100
This means that subvolumes get a different fsid, and NFS exporting them
works properly.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
fs/btrfs/super.c

index 6446ab73f35aeaff25047f060a9662c5f5c975a4..55f4d00fda3aeb7b1db287d8b15257114343bb54 100644 (file)
@@ -503,6 +503,10 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
           on a big-endian or little-endian host */
        buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
        buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
+       /* Mask in the root object ID too, to disambiguate subvols */
+       buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
+       buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
+
        return 0;
 }