GFS2: Set GFP_NOFS when allocating page on write
authorSteven Whitehouse <swhiteho@redhat.com>
Tue, 6 Jan 2009 10:08:33 +0000 (10:08 +0000)
committerSteven Whitehouse <swhiteho@redhat.com>
Wed, 7 Jan 2009 08:57:04 +0000 (08:57 +0000)
We need to ensure that we always set GFP_NOFS in this one
particular case when allocating pages for write.

Reported-by: Fabio M. Di Nitto <fdinitto@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
fs/gfs2/ops_address.c

index 6e4ea36c6605af6300bbe4854497ca7f4dacd8fc..4ddab67867eb273553b6afbf0a42399f56c51735 100644 (file)
@@ -675,6 +675,7 @@ static int gfs2_write_begin(struct file *file, struct address_space *mapping,
                goto out_trans_fail;
 
        error = -ENOMEM;
+       flags |= AOP_FLAG_NOFS;
        page = grab_cache_page_write_begin(mapping, index, flags);
        *pagep = page;
        if (unlikely(!page))