Sebastian Schuberth <sschuberth@xxxxxxxxx> writes:
> As ntohl()/htonl() are function calls (that internally do shifts), I
> doubt they're faster than the shift macros, though I haven't measured
> it. However, I do not suggest to go for the macros on Windows/Intel,
> but to apply the following patch on top of your patch:
Your proposed commit log message makes it sound as if this change is
limited to Windows, but it is not protected with "#ifdef WIN32"; the
change should be applicable to non-windows but the message is misleading.
It should help any i386/amd64 platform whose ntohl()/htonl() is crappy, as
long as __builtin_bswap32() is supported by the compiler. And it should
not harm other platforms, nor i386/amd64 whose ntohl()/htonl() are sane.
As i386/amd64 part of block-sha1/sha1.c has gcc dependency already, I
think it would be safe to assume __builtin_bswap32() is available.
But I'd want an Ack/Nack from the original authors (Cc'ed).
It seems that your patch is linewrapped, so please be careful _if_ it
needs to be modified and resent (if this version gets trivially acked I
can fix it up when applying and in such a case there is no need to
resend).
> From: Sebastian Schuberth <sschuberth@xxxxxxxxx>
> Date: Tue, 18 Aug 2009 12:33:35 +0200
> Subject: [PATCH] block-sha1: On Intel, use bswap built-in in favor of
> ntohl()/htonl()
>
> On Windows/Intel, ntohl()/htonl() are function calls that do shifts to
> swap the
> byte order. Using the native bswap instruction boths gets rid of the
> shifts and
> the function call overhead to gain some performance.
>
> Signed-off-by: Sebastian Schuberth <sschuberth@xxxxxxxxx>
> ---
> block-sha1/sha1.c | 15 ++++++++++-----
> 1 files changed, 10 insertions(+), 5 deletions(-)
>
> diff --git a/block-sha1/sha1.c b/block-sha1/sha1.c
> index f2830c0..07f2937 100644
> --- a/block-sha1/sha1.c
> +++ b/block-sha1/sha1.c
> @@ -66,15 +66,20 @@
>
> /*
> * Performance might be improved if the CPU architecture is OK with
> - * unaligned 32-bit loads and a fast ntohl() is available.
> + * unaligned 32-bit loads and a fast ntohl() is available. On Intel,
> + * use the bswap built-in to get rid of the function call overhead.
> * Otherwise fall back to byte loads and shifts which is portable,
> * and is faster on architectures with memory alignment issues.
> */
>
> -#if defined(__i386__) || defined(__x86_64__) || \
> - defined(__ppc__) || defined(__ppc64__) || \
> - defined(__powerpc__) || defined(__powerpc64__) || \
> - defined(__s390__) || defined(__s390x__)
> +#if defined(__i386__) || defined(__x86_64__)
> +
> +#define get_be32(p) __builtin_bswap32(*(unsigned int *)(p))
> +#define put_be32(p, v) do { *(unsigned int *)(p) =
> __builtin_bswap32(v); } while (0)
> +
> +#elif defined(__ppc__) || defined(__ppc64__) || \
> + defined(__powerpc__) || defined(__powerpc64__) || \
> + defined(__s390__) || defined(__s390x__)
>
> #define get_be32(p) ntohl(*(unsigned int *)(p))
> #define put_be32(p, v) do { *(unsigned int *)(p) = htonl(v); } while (0)
> --
> 1.6.4.169.g64d5.dirty
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html