GFS2: Use MAX_LFS_FILESIZE for meta inode size
authorSteven Whitehouse <swhiteho@redhat.com>
Fri, 8 Jan 2010 13:44:49 +0000 (13:44 +0000)
committerSteven Whitehouse <swhiteho@redhat.com>
Mon, 11 Jan 2010 08:57:55 +0000 (08:57 +0000)
Using ~0ULL was cauing sign issues in filemap_fdatawrite_range, so
use MAX_LFS_FILESIZE instead.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
fs/gfs2/meta_io.c

index cb8d7a93d5ec257204c93aac033c95146fdd1a07..6f68a5f18eb8ccdb0c2b85744def9f9476c77af0 100644 (file)
@@ -121,7 +121,7 @@ struct inode *gfs2_aspace_get(struct gfs2_sbd *sdp)
        if (aspace) {
                mapping_set_gfp_mask(aspace->i_mapping, GFP_NOFS);
                aspace->i_mapping->a_ops = &aspace_aops;
-               aspace->i_size = ~0ULL;
+               aspace->i_size = MAX_LFS_FILESIZE;
                ip = GFS2_I(aspace);
                clear_bit(GIF_USER, &ip->i_flags);
                insert_inode_hash(aspace);