From: Steven Whitehouse <swhiteho@redhat.com>
Date: Wed, 10 Jun 2009 15:23:33 +0100
Subject: [gfs2] fix truncate buffered/direct I/O issue
Message-id: 1244643813.3355.20.camel@localhost.localdomain
O-Subject: [RHEL 5.4] Fix truncate buffered/direct I/O issue (bz #504676) #2
Bugzilla: 504676
RH-Acked-by: Eric Sandeen <sandeen@redhat.com>

This (updated) patch fixes bz #504676

The fix is identical to the upstream patch modulo the context
and fixes a place where we were not marking a buffer dirty
which could result in it not being written if accessed via
O_DIRECT later.

diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c
index c628832..58a21a2 100644
--- a/fs/gfs2/bmap.c
+++ b/fs/gfs2/bmap.c
@@ -898,7 +898,7 @@ static int gfs2_block_truncate_page(struct address_space *mapping)
 	memset(kaddr + offset, 0, length);
 	flush_dcache_page(page);
 	kunmap_atomic(kaddr, KM_USER0);
-
+	mark_buffer_dirty(bh);
 unlock:
 	unlock_page(page);
 	page_cache_release(page);