Re: [PATCH 4/4] parisc: checksum: Optimize from32to16

From: Guenter Roeck
Date: Thu Feb 22 2024 - 11:05:33 EST


On Wed, Feb 21, 2024 at 06:37:14PM -0800, Charlie Jenkins wrote:
> Replace the shifting and masking of x with a rotation. This generates
> better assembly.
>
> Signed-off-by: Charlie Jenkins <charlie@xxxxxxxxxxxx>

Tested-by: Guenter Roeck <linux@xxxxxxxxxxxx>

> ---
> arch/parisc/lib/checksum.c | 7 ++-----
> 1 file changed, 2 insertions(+), 5 deletions(-)
>
> diff --git a/arch/parisc/lib/checksum.c b/arch/parisc/lib/checksum.c
> index eaa660491e24..1ae8cc730d13 100644
> --- a/arch/parisc/lib/checksum.c
> +++ b/arch/parisc/lib/checksum.c
> @@ -27,11 +27,8 @@
>
> static inline unsigned short from32to16(unsigned int x)
> {
> - /* 32 bits --> 16 bits + carry */
> - x = (x & 0xffff) + (x >> 16);
> - /* 16 bits + carry --> 16 bits including carry */
> - x = (x & 0xffff) + (x >> 16);
> - return (unsigned short)x;
> + x += ror32(x, 16);
> + return (unsigned short)(x >> 16);
> }
>
> unsigned int do_csum(const unsigned char *buff, int len)
>
> --
> 2.34.1
>