[PATCH v2 6/8] xfs: use xfs_icluster_size_fsb in xfs_ialloc_inode_init

[PATCH v2 6/8] xfs: use xfs_icluster_size_fsb in xfs_ialloc_inode_init

[Jeff Liu]

From: Jie Liu <jeff.liu@oracle.com>

Use xfs_icluster_size_fsb() in xfs_ialloc_inode_init(), rename variable
ninodes to inodes_per_cluster, the later is more meaningful.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
---
 fs/xfs/xfs_ialloc.c | 23 ++++++++---------------
 1 file changed, 8 insertions(+), 15 deletions(-)

diff --git a/fs/xfs/xfs_ialloc.c b/fs/xfs/xfs_ialloc.c
index 5acba64..492ac88 100644
--- a/fs/xfs/xfs_ialloc.c
+++ b/fs/xfs/xfs_ialloc.c
@@ -170,27 +170,20 @@ xfs_ialloc_inode_init(
 {
 	struct xfs_buf		*fbuf;
 	struct xfs_dinode	*free;
-	int			blks_per_cluster, nbufs, ninodes;
+	int			nbufs, blks_per_cluster, inodes_per_cluster;
 	int			version;
 	int			i, j;
 	xfs_daddr_t		d;
 	xfs_ino_t		ino = 0;
 
 	/*
-	 * Loop over the new block(s), filling in the inodes.
-	 * For small block sizes, manipulate the inodes in buffers
-	 * which are multiples of the blocks size.
+	 * Loop over the new block(s), filling in the inodes.  For small block
+	 * sizes, manipulate the inodes in buffers  which are multiples of the
+	 * blocks size.
 	 */
-	if (mp->m_sb.sb_blocksize >= mp->m_inode_cluster_size) {
-		blks_per_cluster = 1;
-		nbufs = length;
-		ninodes = mp->m_sb.sb_inopblock;
-	} else {
-		blks_per_cluster = mp->m_inode_cluster_size /
-				   mp->m_sb.sb_blocksize;
-		nbufs = length / blks_per_cluster;
-		ninodes = blks_per_cluster * mp->m_sb.sb_inopblock;
-	}
+	blks_per_cluster = xfs_icluster_size_fsb(mp);
+	inodes_per_cluster = blks_per_cluster << mp->m_sb.sb_inopblog;
+	nbufs = length / blks_per_cluster;
 
 	/*
 	 * Figure out what version number to use in the inodes we create.  If
@@ -246,7 +239,7 @@ xfs_ialloc_inode_init(
 		/* Initialize the inode buffers and log them appropriately. */
 		fbuf->b_ops = &xfs_inode_buf_ops;
 		xfs_buf_zero(fbuf, 0, BBTOB(fbuf->b_length));
-		for (i = 0; i < ninodes; i++) {
+		for (i = 0; i < inodes_per_cluster; i++) {
 			int	ioffset = i << mp->m_sb.sb_inodelog;
 			uint	isize = xfs_dinode_size(version);
 
-- 
1.8.3.2

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH v2 6/8] xfs: use xfs_icluster_size_fsb in xfs_ialloc_inode_init

[Christoph Hellwig]

> +	 * Loop over the new block(s), filling in the inodes.  For small block
> +	 * sizes, manipulate the inodes in buffers  which are multiples of the
> +	 * blocks size.
>  	 */
> +	blks_per_cluster = xfs_icluster_size_fsb(mp);
> +	inodes_per_cluster = blks_per_cluster << mp->m_sb.sb_inopblog;
> +	nbufs = length / blks_per_cluster;

I don't think there's a point to have the nbufs variable as it's used
just once as the loop termination condition.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH v2 6/8] xfs: use xfs_icluster_size_fsb in xfs_ialloc_inode_init

[Jeff Liu]

On 12/13 2013 19:14 PM, Christoph Hellwig wrote:
>> +	 * Loop over the new block(s), filling in the inodes.  For small block
>> +	 * sizes, manipulate the inodes in buffers  which are multiples of the
>> +	 * blocks size.
>>  	 */
>> +	blks_per_cluster = xfs_icluster_size_fsb(mp);
>> +	inodes_per_cluster = blks_per_cluster << mp->m_sb.sb_inopblog;
>> +	nbufs = length / blks_per_cluster;
> 
> I don't think there's a point to have the nbufs variable as it's used
> just once as the loop termination condition.
Thanks for the review, and that sounds make sense to me given that compiler
would mostly likely optimize this situation to calculate nbufs only once in
the loop, but I also consider that might affect a bit readabilities in either
case...

Thanks,
-Jeff

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs