From: Eric Dumazet <dada1@cosmosbay.com>
To: Stephen Hemminger <shemminger@linux-foundation.org>
Cc: Andi Kleen <ak@suse.de>,
discuss@x86-64.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] x86-64: memset optimization
Date: Sat, 18 Aug 2007 09:17:58 +0200 [thread overview]
Message-ID: <46C69D26.3000200@cosmosbay.com> (raw)
In-Reply-To: <20070817163446.3e63f208@freepuppy.rosehill.hemminger.net>
Stephen Hemminger a écrit :
> Optimize uses of memset with small constant offsets.
> This will generate smaller code, and avoid the slow rep/string instructions.
> Code copied from i386 with a little cleanup.
>
You obviously didnt test it, did you ?
How can you be sure this is going to speedup things then ?
> Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
>
> --- a/include/asm-x86_64/string.h 2007-08-17 15:14:32.000000000 -0700
> +++ b/include/asm-x86_64/string.h 2007-08-17 15:36:30.000000000 -0700
> @@ -42,9 +42,51 @@ extern void *__memcpy(void *to, const vo
> __ret = __builtin_memcpy((dst),(src),__len); \
> __ret; })
> #endif
> -
> #define __HAVE_ARCH_MEMSET
> -void *memset(void *s, int c, size_t n);
> +void *__memset(void *s, int c, size_t n);
> +
> +/* Optimize for cases of trivial memset's
> + * Compiler should optimize away all but the case used.
> + */
> +static __always_inline void *
> +__constant_c_and_count_memset(void *s, int c, size_t count)
> +{
> + unsigned long pattern = 0x01010101UL * (unsigned char) c;
Main difference between x86_64 and i386 is sizeof(long) being 8 instead of 4
Why not let gcc do its job about memset() ?
On x86_64 at least, modern gcc are smart enough.
> +
> + switch (count) {
> + case 0:
> + return s;
> + case 1:
> + *(unsigned char *)s = pattern;
> + return s;
> + case 2:
> + *(unsigned short *)s = pattern;
> + return s;
> + case 3:
> + *(unsigned short *)s = pattern;
> + *(2+(unsigned char *)s) = pattern;
> + return s;
> + case 4:
> + *(unsigned long *)s = pattern;
> + return s;
> + case 6:
> + *(unsigned long *)s = pattern;
> + *(2+(unsigned short *)s) = pattern;
> + return s;
> + case 8:
> + *(unsigned long *)s = pattern;
> + *(1+(unsigned long *)s) = pattern;
> + return s;
> + default:
> + return __memset(s, c, count);
> + }
> +}
> +#define memset(s, c, count) \
> + (__builtin_constant_p(c) \
> + ? __constant_c_and_count_memset((s),(c),(count)) \
> + : __memset((s),(c),(count)))
> +
> +
>
> #define __HAVE_ARCH_MEMMOVE
> void * memmove(void * dest,const void *src,size_t count);
> --- a/arch/x86_64/kernel/x8664_ksyms.c 2007-08-17 15:14:32.000000000 -0700
> +++ b/arch/x86_64/kernel/x8664_ksyms.c 2007-08-17 15:44:58.000000000 -0700
> @@ -48,10 +48,12 @@ EXPORT_SYMBOL(__read_lock_failed);
> #undef memmove
>
> extern void * memset(void *,int,__kernel_size_t);
> +extern void * __memset(void *,int,__kernel_size_t);
> extern void * memcpy(void *,const void *,__kernel_size_t);
> extern void * __memcpy(void *,const void *,__kernel_size_t);
>
> EXPORT_SYMBOL(memset);
> +EXPORT_SYMBOL(__memset);
> EXPORT_SYMBOL(memcpy);
> EXPORT_SYMBOL(__memcpy);
>
next prev parent reply other threads:[~2007-08-18 7:18 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-08-17 23:34 [PATCH] x86-64: memset optimization Stephen Hemminger
2007-08-18 7:17 ` Eric Dumazet [this message]
2007-08-18 9:46 ` Andi Kleen
2007-08-18 14:56 ` Stephen Hemminger
2007-08-18 18:55 ` Andi Kleen
2007-08-19 5:04 ` Stephen Hemminger
2007-08-19 18:24 ` [discuss] " Andi Kleen
2007-08-20 15:52 ` Stephen Hemminger
2007-08-20 15:51 ` Arjan van de Ven
2007-08-20 17:03 ` Roland Dreier
2007-08-20 18:16 ` Andi Kleen
2007-08-20 18:56 ` Jan Hubicka
2007-08-21 10:16 ` Denys Vlasenko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=46C69D26.3000200@cosmosbay.com \
--to=dada1@cosmosbay.com \
--cc=ak@suse.de \
--cc=discuss@x86-64.org \
--cc=linux-kernel@vger.kernel.org \
--cc=shemminger@linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox