mailing list of musl libc
 help / color / mirror / Atom feed
From: Rich Felker <dalias@libc.org>
To: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>, musl@lists.openwall.com
Subject: Re: [musl] [PATCH] x86/memset: avoid performing final store twice
Date: Sat, 10 Oct 2020 20:25:14 -0400
Message-ID: <20201011002514.GF17637@brightrain.aerifal.cx> (raw)
In-Reply-To: <20201003223209.10307-1-vda.linux@googlemail.com>

On Sun, Oct 04, 2020 at 12:32:09AM +0200, Denys Vlasenko wrote:
> From: Denys Vlasenko <dvlasenk@redhat.com>
> 
> For not very short NBYTES case:
> 
> To handle the tail alignment, the code performs a potentially
> misaligned word store to fill the final 8 bytes of the buffer.
> This is done even if the buffer's end is aligned.
> 
> Eventually code fills the rest of the buffer, which is a multiple
> of 8 bytes now, with NBYTES / 8 aligned word stores.
> 
> However, this means that if NBYTES *was* divisible by 8,
> we store last word too, again.
> 
> This patch decrements byte count before dividing it by 8,
> making one less store in "NBYTES is divisible by 8" case,
> and not changing anything in all other cases.
> 
> CC: Rich Felker <dalias@libc.org>
> CC: musl@lists.openwall.com
> Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
> ---
>  src/string/i386/memset.s   | 7 ++++---
>  src/string/x86_64/memset.s | 2 +-
>  2 files changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/src/string/i386/memset.s b/src/string/i386/memset.s
> index d00422c4..b1c5c2f8 100644
> --- a/src/string/i386/memset.s
> +++ b/src/string/i386/memset.s
> @@ -47,7 +47,7 @@ memset:
>  	mov %edx,(-1-2-4-8-8)(%eax,%ecx)
>  	mov %edx,(-1-2-4-8-4)(%eax,%ecx)
>  
> -1:	ret 	
> +1:	ret
>  
>  2:	movzbl 8(%esp),%eax
>  	mov %edi,12(%esp)
> @@ -57,13 +57,14 @@ memset:
>  	mov %eax,-4(%edi,%ecx)
>  	jnz 2f
>  
> -1:	shr $2, %ecx
> +1:	dec %ecx
> +	shr $2, %ecx
>  	rep
>  	stosl
>  	mov 4(%esp),%eax
>  	mov 12(%esp),%edi
>  	ret
> -	
> +
>  2:	xor %edx,%edx
>  	sub %edi,%edx
>  	and $15,%edx
> diff --git a/src/string/x86_64/memset.s b/src/string/x86_64/memset.s
> index 2d3f5e52..85bb686c 100644
> --- a/src/string/x86_64/memset.s
> +++ b/src/string/x86_64/memset.s
> @@ -53,7 +53,7 @@ memset:
>  2:	test $15,%edi
>  	mov %rdi,%r8
>  	mov %rax,-8(%rdi,%rdx)
> -	mov %rdx,%rcx
> +	lea -1(%rdx),%rcx
>  	jnz 2f
>  
>  1:	shr $3,%rcx
> -- 
> 2.25.0

Does this have measurably better performance on a system you've tested
it on? Indeed it nominally stores less, but it also performs an extra
arithmetic operation. I doubt it makes anything worse but I'd like to
avoid making random changes to such functions unless there's a good
reason to believe it actually helps.

Rich

  reply	other threads:[~2020-10-11  0:25 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-03 22:32 Denys Vlasenko
2020-10-11  0:25 ` Rich Felker [this message]
2020-10-12 12:18   ` 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=20201011002514.GF17637@brightrain.aerifal.cx \
    --to=dalias@libc.org \
    --cc=dvlasenk@redhat.com \
    --cc=musl@lists.openwall.com \
    --cc=vda.linux@googlemail.com \
    /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

mailing list of musl libc

This inbox may be cloned and mirrored by anyone:

	git clone --mirror http://inbox.vuxu.org/musl

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V1 musl musl/ http://inbox.vuxu.org/musl \
		musl@inbox.vuxu.org
	public-inbox-index musl

Example config snippet for mirrors.
Newsgroup available over NNTP:
	nntp://inbox.vuxu.org/vuxu.archive.musl


code repositories for the project(s) associated with this inbox:

	https://git.vuxu.org/mirror/musl/

AGPL code for this site: git clone https://public-inbox.org/public-inbox.git