From: "H.J. Lu" <hjl.tools@gmail.com>
To: Noah Goldstein <goldstein.w.n@gmail.com>
Cc: GNU C Library <libc-alpha@sourceware.org>,
"Carlos O'Donell" <carlos@systemhalted.org>
Subject: Re: [PATCH v1] x86: Move wcschr SSE2 implementation to multiarch/wcschr-sse2.S
Date: Tue, 12 Jul 2022 13:55:54 -0700 [thread overview]
Message-ID: <CAMe9rOpF8f6Uir_2qnx51WE-gx7DKo1-7tejOEFikHjX4Qz8zg@mail.gmail.com> (raw)
In-Reply-To: <20220712192910.351121-7-goldstein.w.n@gmail.com>
On Tue, Jul 12, 2022 at 12:29 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>
> This commit doesn't affect libc.so.6, its just housekeeping to prepare
> for adding explicit ISA level support.
>
> Tested build on x86_64 and x86_32 with/without multiarch.
> ---
> sysdeps/x86_64/multiarch/wcschr-sse2.S | 145 +++++++++++++++++++++++--
> sysdeps/x86_64/wcschr.S | 135 +----------------------
> 2 files changed, 138 insertions(+), 142 deletions(-)
>
> diff --git a/sysdeps/x86_64/multiarch/wcschr-sse2.S b/sysdeps/x86_64/multiarch/wcschr-sse2.S
> index 218ea609b9..c872926ba9 100644
> --- a/sysdeps/x86_64/multiarch/wcschr-sse2.S
> +++ b/sysdeps/x86_64/multiarch/wcschr-sse2.S
> @@ -17,14 +17,141 @@
> <https://www.gnu.org/licenses/>. */
>
> #if IS_IN (libc)
> -# define __wcschr __wcschr_sse2
> -
> -# undef weak_alias
> -# define weak_alias(__wcschr, wcschr)
> -# undef libc_hidden_def
> -# define libc_hidden_def(__wcschr)
> -# undef libc_hidden_weak
> -# define libc_hidden_weak(wcschr)
> +# ifndef WCSCHR
> +# define WCSCHR __wcschr_sse2
> +# endif
> #endif
>
> -#include "../wcschr.S"
> +#include <sysdep.h>
> +
> + .text
> +ENTRY (WCSCHR)
> +
> + movd %rsi, %xmm1
> + pxor %xmm2, %xmm2
> + mov %rdi, %rcx
> + punpckldq %xmm1, %xmm1
> + punpckldq %xmm1, %xmm1
> +
> + and $63, %rcx
> + cmp $48, %rcx
> + ja L(cross_cache)
> +
> + movdqu (%rdi), %xmm0
> + pcmpeqd %xmm0, %xmm2
> + add $16, %rdi
> + pcmpeqd %xmm1, %xmm0
> + pmovmskb %xmm2, %rdx
> + pmovmskb %xmm0, %rax
> + or %rax, %rdx
> + jnz L(matches)
> +
> + and $-16, %rdi
> +
> + movdqa (%rdi), %xmm0
> + pcmpeqd %xmm0, %xmm2
> + add $16, %rdi
> + pcmpeqd %xmm1, %xmm0
> + pmovmskb %xmm2, %rdx
> + pmovmskb %xmm0, %rax
> + or %rax, %rdx
> + jnz L(matches)
> +
> + jmp L(loop)
> +
> +L(cross_cache):
> + and $15, %rcx
> + and $-16, %rdi
> + movdqa (%rdi), %xmm0
> + pcmpeqd %xmm0, %xmm2
> + pcmpeqd %xmm1, %xmm0
> + pmovmskb %xmm2, %rdx
> + pmovmskb %xmm0, %rax
> +
> + sar %cl, %rdx
> + sar %cl, %rax
> + test %rax, %rax
> + je L(unaligned_no_match)
> +
> + bsf %rax, %rax
> + test %rdx, %rdx
> + je L(unaligned_match)
> + bsf %rdx, %rdx
> + cmp %rdx, %rax
> + ja L(return_null)
> +
> +L(unaligned_match):
> + add %rdi, %rax
> + add %rcx, %rax
> + ret
> +
> + .p2align 4
> +L(unaligned_no_match):
> + test %rdx, %rdx
> + jne L(return_null)
> + pxor %xmm2, %xmm2
> +
> + add $16, %rdi
> +
> + .p2align 4
> +/* Loop start on aligned string. */
> +L(loop):
> + movdqa (%rdi), %xmm0
> + pcmpeqd %xmm0, %xmm2
> + add $16, %rdi
> + pcmpeqd %xmm1, %xmm0
> + pmovmskb %xmm2, %rdx
> + pmovmskb %xmm0, %rax
> + or %rax, %rdx
> + jnz L(matches)
> +
> + movdqa (%rdi), %xmm0
> + pcmpeqd %xmm0, %xmm2
> + add $16, %rdi
> + pcmpeqd %xmm1, %xmm0
> + pmovmskb %xmm2, %rdx
> + pmovmskb %xmm0, %rax
> + or %rax, %rdx
> + jnz L(matches)
> +
> + movdqa (%rdi), %xmm0
> + pcmpeqd %xmm0, %xmm2
> + add $16, %rdi
> + pcmpeqd %xmm1, %xmm0
> + pmovmskb %xmm2, %rdx
> + pmovmskb %xmm0, %rax
> + or %rax, %rdx
> + jnz L(matches)
> +
> + movdqa (%rdi), %xmm0
> + pcmpeqd %xmm0, %xmm2
> + add $16, %rdi
> + pcmpeqd %xmm1, %xmm0
> + pmovmskb %xmm2, %rdx
> + pmovmskb %xmm0, %rax
> + or %rax, %rdx
> + jnz L(matches)
> + jmp L(loop)
> +
> + .p2align 4
> +L(matches):
> + pmovmskb %xmm2, %rdx
> + test %rax, %rax
> + jz L(return_null)
> + bsf %rax, %rax
> + test %rdx, %rdx
> + je L(match)
> + bsf %rdx, %rcx
> + cmp %rcx, %rax
> + ja L(return_null)
> +L(match):
> + sub $16, %rdi
> + add %rdi, %rax
> + ret
> +
> + .p2align 4
> +L(return_null):
> + xor %rax, %rax
> + ret
> +
> +END (WCSCHR)
> diff --git a/sysdeps/x86_64/wcschr.S b/sysdeps/x86_64/wcschr.S
> index 2131220382..80b12c4286 100644
> --- a/sysdeps/x86_64/wcschr.S
> +++ b/sysdeps/x86_64/wcschr.S
> @@ -16,140 +16,9 @@
> License along with the GNU C Library; if not, see
> <https://www.gnu.org/licenses/>. */
>
> -#include <sysdep.h>
> -
> - .text
> -ENTRY (__wcschr)
> -
> - movd %rsi, %xmm1
> - pxor %xmm2, %xmm2
> - mov %rdi, %rcx
> - punpckldq %xmm1, %xmm1
> - punpckldq %xmm1, %xmm1
> -
> - and $63, %rcx
> - cmp $48, %rcx
> - ja L(cross_cache)
> -
> - movdqu (%rdi), %xmm0
> - pcmpeqd %xmm0, %xmm2
> - add $16, %rdi
> - pcmpeqd %xmm1, %xmm0
> - pmovmskb %xmm2, %rdx
> - pmovmskb %xmm0, %rax
> - or %rax, %rdx
> - jnz L(matches)
> -
> - and $-16, %rdi
> -
> - movdqa (%rdi), %xmm0
> - pcmpeqd %xmm0, %xmm2
> - add $16, %rdi
> - pcmpeqd %xmm1, %xmm0
> - pmovmskb %xmm2, %rdx
> - pmovmskb %xmm0, %rax
> - or %rax, %rdx
> - jnz L(matches)
> -
> - jmp L(loop)
> -
> -L(cross_cache):
> - and $15, %rcx
> - and $-16, %rdi
> - movdqa (%rdi), %xmm0
> - pcmpeqd %xmm0, %xmm2
> - pcmpeqd %xmm1, %xmm0
> - pmovmskb %xmm2, %rdx
> - pmovmskb %xmm0, %rax
> -
> - sar %cl, %rdx
> - sar %cl, %rax
> - test %rax, %rax
> - je L(unaligned_no_match)
> -
> - bsf %rax, %rax
> - test %rdx, %rdx
> - je L(unaligned_match)
> - bsf %rdx, %rdx
> - cmp %rdx, %rax
> - ja L(return_null)
> -
> -L(unaligned_match):
> - add %rdi, %rax
> - add %rcx, %rax
> - ret
> -
> - .p2align 4
> -L(unaligned_no_match):
> - test %rdx, %rdx
> - jne L(return_null)
> - pxor %xmm2, %xmm2
> -
> - add $16, %rdi
> -
> - .p2align 4
> -/* Loop start on aligned string. */
> -L(loop):
> - movdqa (%rdi), %xmm0
> - pcmpeqd %xmm0, %xmm2
> - add $16, %rdi
> - pcmpeqd %xmm1, %xmm0
> - pmovmskb %xmm2, %rdx
> - pmovmskb %xmm0, %rax
> - or %rax, %rdx
> - jnz L(matches)
> -
> - movdqa (%rdi), %xmm0
> - pcmpeqd %xmm0, %xmm2
> - add $16, %rdi
> - pcmpeqd %xmm1, %xmm0
> - pmovmskb %xmm2, %rdx
> - pmovmskb %xmm0, %rax
> - or %rax, %rdx
> - jnz L(matches)
> -
> - movdqa (%rdi), %xmm0
> - pcmpeqd %xmm0, %xmm2
> - add $16, %rdi
> - pcmpeqd %xmm1, %xmm0
> - pmovmskb %xmm2, %rdx
> - pmovmskb %xmm0, %rax
> - or %rax, %rdx
> - jnz L(matches)
> -
> - movdqa (%rdi), %xmm0
> - pcmpeqd %xmm0, %xmm2
> - add $16, %rdi
> - pcmpeqd %xmm1, %xmm0
> - pmovmskb %xmm2, %rdx
> - pmovmskb %xmm0, %rax
> - or %rax, %rdx
> - jnz L(matches)
> - jmp L(loop)
> -
> - .p2align 4
> -L(matches):
> - pmovmskb %xmm2, %rdx
> - test %rax, %rax
> - jz L(return_null)
> - bsf %rax, %rax
> - test %rdx, %rdx
> - je L(match)
> - bsf %rdx, %rcx
> - cmp %rcx, %rax
> - ja L(return_null)
> -L(match):
> - sub $16, %rdi
> - add %rdi, %rax
> - ret
> -
> - .p2align 4
> -L(return_null):
> - xor %rax, %rax
> - ret
> -
> -END (__wcschr)
>
> +#define WCSCHR __wcschr
> +#include "multiarch/wcschr-sse2.S"
> libc_hidden_def(__wcschr)
> weak_alias (__wcschr, wcschr)
> libc_hidden_weak (wcschr)
> --
> 2.34.1
>
LGTM.
Thanks.
--
H.J.
next prev parent reply other threads:[~2022-07-12 20:56 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-07-12 19:29 [PATCH v1] x86: Move strlen SSE2 implementation to multiarch/strlen-sse2.S Noah Goldstein
2022-07-12 19:29 ` [PATCH v1] x86: Move strcpy SSE2 implementation to multiarch/strcpy-sse2.S Noah Goldstein
2022-07-12 23:23 ` H.J. Lu
2022-07-12 19:29 ` [PATCH v1] x86: Move memrchr SSE2 implementation to multiarch/memrchr-sse2.S Noah Goldstein
2022-07-12 22:58 ` H.J. Lu
2022-07-12 19:29 ` [PATCH v1] x86: Move strrchr SSE2 implementation to multiarch/strrchr-sse2.S Noah Goldstein
2022-07-12 22:28 ` H.J. Lu
2022-07-12 19:29 ` [PATCH v1] x86: Move strchr SSE2 implementation to multiarch/strchr-sse2.S Noah Goldstein
2022-07-12 21:27 ` H.J. Lu
2022-07-12 19:29 ` [PATCH v1] x86: Move strcat SSE2 implementation to multiarch/strcat-sse2.S Noah Goldstein
2022-07-12 21:16 ` H.J. Lu
2022-07-12 19:29 ` [PATCH v1] x86: Move wcschr SSE2 implementation to multiarch/wcschr-sse2.S Noah Goldstein
2022-07-12 20:55 ` H.J. Lu [this message]
2022-07-12 19:29 ` [PATCH v1] x86: Move wcslen SSE2 implementation to multiarch/wcslen-sse2.S Noah Goldstein
2022-07-12 20:26 ` H.J. Lu
2022-07-12 19:29 ` [PATCH v1] x86: Remove unneeded rtld-wmemcmp Noah Goldstein
2022-07-12 19:44 ` H.J. Lu
2022-07-12 19:29 ` [PATCH v1] x86: Add missing rtm tests for strcmp family Noah Goldstein
2022-07-12 19:59 ` H.J. Lu
2022-07-12 23:29 ` [PATCH v1] x86: Move strlen SSE2 implementation to multiarch/strlen-sse2.S H.J. Lu
2022-07-13 4:06 ` Noah Goldstein
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=CAMe9rOpF8f6Uir_2qnx51WE-gx7DKo1-7tejOEFikHjX4Qz8zg@mail.gmail.com \
--to=hjl.tools@gmail.com \
--cc=carlos@systemhalted.org \
--cc=goldstein.w.n@gmail.com \
--cc=libc-alpha@sourceware.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;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).