[Cluster-devel] [Patch 34/44] libgfs2: when mapping gfs1 dinode blocks, use dinode buffer

[Bob Peterson <rpeterso@redhat.com>]

From bc534f1d55b9b80b795356f7c5e3bc170bd75919 Mon Sep 17 00:00:00 2001
From: Bob Peterson <rpeterso@redhat.com>
Date: Wed, 10 Aug 2011 10:33:49 -0500
Subject: [PATCH 34/44] libgfs2: when mapping gfs1 dinode blocks, use dinode
 buffer

When function gfs1_block_map found a block for a stuffed file it
wasn't returning the dinode buffer, it was returning the block as a
new buffer.  The problem is, if changes are made to the block, they
will be overwritten by the dinode buffer when that is released.
The result is a fixed block that doesn't get fixed.  This patch
changes the code to return the block properly like gfs2_block_map.

rhbz#675723
---
 gfs2/libgfs2/gfs1.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/gfs2/libgfs2/gfs1.c b/gfs2/libgfs2/gfs1.c
index d09a8ee..5304b56 100644
--- a/gfs2/libgfs2/gfs1.c
+++ b/gfs2/libgfs2/gfs1.c
@@ -120,7 +120,10 @@ void gfs1_block_map(struct gfs2_inode *ip, uint64_t lblock, int *new,
 			mh.mh_format = GFS2_FORMAT_IN;
 			gfs2_meta_header_out(&mh, bh);
 		} else {
-			bh = bread(sdp, *dblock);
+			if (*dblock == ip->i_di.di_num.no_addr)
+				bh = ip->i_bh;
+			else
+				bh = bread(sdp, *dblock);
 		}
 	}
 
-- 
1.7.4.4