2009-07-14 11:59:57

by Xiaotian Feng

Subject: [RFC PATCH] block: slightly optimize blkdev_ioctl BLKRA{SET,GET}

Slightly optimize blkdev_ioctl by using shift instead of multiply
and divide.

Signed-off-by: Xiaotian Feng <[email protected]>
---
block/ioctl.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/block/ioctl.c b/block/ioctl.c
index 500e4c7..f8c7362 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -305,7 +305,7 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
bdi = blk_get_backing_dev_info(bdev);
if (bdi == NULL)
return -ENOTTY;
- return put_long(arg, (bdi->ra_pages * PAGE_CACHE_SIZE) / 512);
+ return put_long(arg, bdi->ra_pages << (PAGE_CACHE_SHIFT - 9));
case BLKROGET:
return put_int(arg, bdev_read_only(bdev) != 0);
case BLKBSZGET: /* get the logical block size (cf. BLKSSZGET) */
@@ -321,7 +321,7 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
bdi = blk_get_backing_dev_info(bdev);
if (bdi == NULL)
return -ENOTTY;
- bdi->ra_pages = (arg * 512) / PAGE_CACHE_SIZE;
+ bdi->ra_pages = arg >> (PAGE_CACHE_SHIFT - 9);
return 0;
case BLKBSZSET:
/* set the logical block size */
--
1.6.2.5

2009-07-14 13:48:30

by Cong Wang

Subject: Re: [RFC PATCH] block: slightly optimize blkdev_ioctl BLKRA{SET,GET}

On Tue, Jul 14, 2009 at 7:59 PM, Xiaotian Feng<[email protected]> wrote:
> Slightly optimize blkdev_ioctl by using shift instead of multiply
> and divide.

Not necessary, I believe gcc can do this by itself.

Have you checked its asm code?

>
> Signed-off-by: Xiaotian Feng <[email protected]>
> ---
>  block/ioctl.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/block/ioctl.c b/block/ioctl.c
> index 500e4c7..f8c7362 100644
> --- a/block/ioctl.c
> +++ b/block/ioctl.c
> @@ -305,7 +305,7 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
>                bdi = blk_get_backing_dev_info(bdev);
>                if (bdi == NULL)
>                        return -ENOTTY;
> -               return put_long(arg, (bdi->ra_pages * PAGE_CACHE_SIZE) / 512);
> +               return put_long(arg, bdi->ra_pages << (PAGE_CACHE_SHIFT - 9));
>        case BLKROGET:
>                return put_int(arg, bdev_read_only(bdev) != 0);
>        case BLKBSZGET: /* get the logical block size (cf. BLKSSZGET) */
> @@ -321,7 +321,7 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
>                bdi = blk_get_backing_dev_info(bdev);
>                if (bdi == NULL)
>                        return -ENOTTY;
> -               bdi->ra_pages = (arg * 512) / PAGE_CACHE_SIZE;
> +               bdi->ra_pages = arg >> (PAGE_CACHE_SHIFT - 9);
>                return 0;
>        case BLKBSZSET:
>                /* set the logical block size */
> --
> 1.6.2.5
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

2009-07-14 14:09:19

by Johannes Weiner

Subject: Re: [RFC PATCH] block: slightly optimize blkdev_ioctl BLKRA{SET,GET}

On Tue, Jul 14, 2009 at 07:59:54PM +0800, Xiaotian Feng wrote:
> Slightly optimize blkdev_ioctl by using shift instead of multiply
> and divide.

Does it make an actual difference in the object code for you? I don't
have a single multiplication in that function after compilation.

Hannes