[GFS2] Fix bmap to map extents properly
This fix means that bmap will map extents of the length requested
by the VFS rather than guessing at it, or just mapping one block
at a time. The other callers of gfs2_block_map are audited to ensure
they send the correct max extent lengths (i.e. set bh->b_size correctly).
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
diff --git a/fs/gfs2/quota.c b/fs/gfs2/quota.c
index c69b94a..a3deae7 100644
--- a/fs/gfs2/quota.c
+++ b/fs/gfs2/quota.c
@@ -251,7 +251,7 @@
unsigned int block, offset;
struct buffer_head *bh;
int error;
- struct buffer_head bh_map;
+ struct buffer_head bh_map = { .b_state = 0, .b_blocknr = 0 };
mutex_lock(&sdp->sd_quota_mutex);
@@ -263,7 +263,8 @@
block = qd->qd_slot / sdp->sd_qc_per_block;
offset = qd->qd_slot % sdp->sd_qc_per_block;;
- error = gfs2_block_map(&ip->i_inode, block, 0, &bh_map, 1);
+ bh_map.b_size = 1 << ip->i_inode.i_blkbits;
+ error = gfs2_block_map(&ip->i_inode, block, 0, &bh_map);
if (error)
goto fail;
error = gfs2_meta_read(ip->i_gl, bh_map.b_blocknr, DIO_WAIT, &bh);