On 2012-03-10, at 2:34 PM, Sami Liedes wrote:
> From 8d712d4e49cdc8a0b82f27b5b62a7691fafcacbf Mon Sep 17 00:00:00 2001
> From: Sami Liedes <sami.liedes@xxxxxx>
> Date: Sat, 10 Mar 2012 22:13:12 +0200
> Subject: [PATCH] libext2fs: Move a modulo operation out of a hot loop.
>
> Filesystem shrinking in particular is a heavy user of this loop in
> ext2fs_new_inode(). This change makes resize2fs use 24% less CPU time
> for shrinking a 100G filesystem.
>
> Signed-off-by: Sami Liedes <sami.liedes@xxxxxx>
>
> diff --git a/lib/ext2fs/alloc.c b/lib/ext2fs/alloc.c
> index 948a0ec..eb4e0f5 100644
> --- a/lib/ext2fs/alloc.c
> +++ b/lib/ext2fs/alloc.c
> @@ -109,6 +109,7 @@ errcode_t ext2fs_new_inode(ext2_filsys fs, ext2_ino_t dir,
> ext2_ino_t dir_group = 0;
> ext2_ino_t i;
> ext2_ino_t start_inode;
> + ext2_ino_t modulo;
Good to see someone working on optimizing this code. One comment on this patch -
"modulo" isn't a very good name for this variable. Sure it is a modulus, but it
doesn't really say _what_ it is. Better would be something like "inode_in_group".
> EXT2_CHECK_MAGIC(fs, EXT2_ET_MAGIC_EXT2FS_FILSYS);
>
> @@ -126,17 +127,21 @@ errcode_t ext2fs_new_inode(ext2_filsys fs, ext2_ino_t dir,
> if (start_inode > fs->super->s_inodes_count)
> return EXT2_ET_INODE_ALLOC_FAIL;
> i = start_inode;
> + modulo = (i - 1) % EXT2_INODES_PER_GROUP(fs->super);
>
> do {
> - if (((i - 1) % EXT2_INODES_PER_GROUP(fs->super)) == 0)
> + if (modulo == 0)
> check_inode_uninit(fs, map, (i - 1) /
> EXT2_INODES_PER_GROUP(fs->super));
>
> if (!ext2fs_fast_test_inode_bitmap2(map, i))
> break;
> - i++;
> - if (i > fs->super->s_inodes_count)
> + if (++modulo == EXT2_INODES_PER_GROUP(fs->super))
> + modulo = 0;
> + if (++i > fs->super->s_inodes_count) {
> i = EXT2_FIRST_INODE(fs->super);
> + modulo = (i - 1) % EXT2_INODES_PER_GROUP(fs->super);
> + }
> } while (i != start_inode);
>
> if (ext2fs_test_inode_bitmap2(map, i))
Cheers, Andreas
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
