ocfs2: Force use of GFP_NOFS in ocfs2_write()
authorMark Fasheh <mark.fasheh@oracle.com>
Wed, 2 May 2007 00:44:20 +0000 (17:44 -0700)
committerMark Fasheh <mark.fasheh@oracle.com>
Wed, 2 May 2007 22:08:34 +0000 (15:08 -0700)
We can otherwise recurse into the file system.

Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
fs/ocfs2/aops.c

index 36b3d2aaabbdc4e630e984f3272a4130f254f058..8e7cafb5fc6c638cded86d0a04c369f62daa5b25 100644 (file)
@@ -1087,7 +1087,7 @@ static ssize_t ocfs2_write(struct file *file, u32 phys, handle_t *handle,
        for(i = 0; i < numpages; i++) {
                index = start + i;
 
-               cpages[i] = grab_cache_page(mapping, index);
+               cpages[i] = find_or_create_page(mapping, index, GFP_NOFS);
                if (!cpages[i]) {
                        ret = -ENOMEM;
                        mlog_errno(ret);