From: Noah Goldstein <goldstein.w.n@gmail.com>
To: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Cc: GNU C Library <libc-alpha@sourceware.org>,
Richard Henderson <rth@twiddle.net>,
Joseph Myers <joseph@codesourcery.com>,
caiyinyu <caiyinyu@loongson.cn>
Subject: Re: [PATCH 10/17] string: Improve generic memchr
Date: Fri, 2 Sep 2022 20:47:50 -0700 [thread overview]
Message-ID: <CAFUsyfJnneN8wJE4Rbi7qyrAPBXaoU5P536j_2hHJxk8UPys3g@mail.gmail.com> (raw)
In-Reply-To: <20220902203940.2385967-11-adhemerval.zanella@linaro.org>
On Fri, Sep 2, 2022 at 1:45 PM Adhemerval Zanella via Libc-alpha
<libc-alpha@sourceware.org> wrote:
>
> New algorithm have the following key differences:
>
> - Reads first word unaligned and use string-maskoff function to
> remove unwanted data. This strategy follow arch-specific
> optimization used on aarch64 and powerpc.
>
> - Use string-fz{b,i} and string-opthr functions.
>
> Checked on x86_64-linux-gnu, i686-linux-gnu, powerpc-linux-gnu,
> and powerpc64-linux-gnu by removing the arch-specific assembly
> implementation and disabling multi-arch (it covers both LE and BE
> for 64 and 32 bits).
>
> Co-authored-by: Richard Henderson <rth@twiddle.net>
> ---
> string/memchr.c | 168 +++++-------------
> .../powerpc32/power4/multiarch/memchr-ppc32.c | 14 +-
> .../powerpc64/multiarch/memchr-ppc64.c | 9 +-
> 3 files changed, 48 insertions(+), 143 deletions(-)
>
> diff --git a/string/memchr.c b/string/memchr.c
> index 422bcd0cd6..8fe0ac48ab 100644
> --- a/string/memchr.c
> +++ b/string/memchr.c
> @@ -1,10 +1,6 @@
> -/* Copyright (C) 1991-2022 Free Software Foundation, Inc.
> +/* Scan memory for a character. Generic version
> + Copyright (C) 1991-2022 Free Software Foundation, Inc.
> This file is part of the GNU C Library.
> - Based on strlen implementation by Torbjorn Granlund (tege@sics.se),
> - with help from Dan Sahlin (dan@sics.se) and
> - commentary by Jim Blandy (jimb@ai.mit.edu);
> - adaptation to memchr suggested by Dick Karpinski (dick@cca.ucsf.edu),
> - and implemented by Roland McGrath (roland@ai.mit.edu).
>
> The GNU C Library is free software; you can redistribute it and/or
> modify it under the terms of the GNU Lesser General Public
> @@ -20,143 +16,65 @@
> License along with the GNU C Library; if not, see
> <https://www.gnu.org/licenses/>. */
>
> -#ifndef _LIBC
> -# include <config.h>
> -#endif
> -
> +#include <intprops.h>
> +#include <string-fza.h>
> +#include <string-fzb.h>
> +#include <string-fzi.h>
> +#include <string-maskoff.h>
> +#include <string-opthr.h>
> #include <string.h>
>
> -#include <stddef.h>
> +#undef memchr
>
> -#include <limits.h>
> -
> -#undef __memchr
> -#ifdef _LIBC
> -# undef memchr
> +#ifdef MEMCHR
> +# define __memchr MEMCHR
> #endif
>
> -#ifndef weak_alias
> -# define __memchr memchr
> -#endif
> -
> -#ifndef MEMCHR
> -# define MEMCHR __memchr
> -#endif
> +static inline const char *
> +sadd (uintptr_t x, uintptr_t y)
> +{
> + uintptr_t ret = INT_ADD_OVERFLOW (x, y) ? (uintptr_t)-1 : x + y;
> + return (const char *)ret;
> +}
>
> /* Search no more than N bytes of S for C. */
> void *
> -MEMCHR (void const *s, int c_in, size_t n)
> +__memchr (void const *s, int c_in, size_t n)
> {
> - /* On 32-bit hardware, choosing longword to be a 32-bit unsigned
> - long instead of a 64-bit uintmax_t tends to give better
> - performance. On 64-bit hardware, unsigned long is generally 64
> - bits already. Change this typedef to experiment with
> - performance. */
> - typedef unsigned long int longword;
> + if (__glibc_unlikely (n == 0))
> + return NULL;
>
> - const unsigned char *char_ptr;
> - const longword *longword_ptr;
> - longword repeated_one;
> - longword repeated_c;
> - unsigned char c;
> + uintptr_t s_int = (uintptr_t) s;
>
> - c = (unsigned char) c_in;
> + /* Set up a word, each of whose bytes is C. */
> + op_t repeated_c = repeat_bytes (c_in);
> + op_t before_mask = create_mask (s_int);
>
> - /* Handle the first few bytes by reading one byte at a time.
> - Do this until CHAR_PTR is aligned on a longword boundary. */
> - for (char_ptr = (const unsigned char *) s;
> - n > 0 && (size_t) char_ptr % sizeof (longword) != 0;
> - --n, ++char_ptr)
> - if (*char_ptr == c)
> - return (void *) char_ptr;
> + /* Compute the address of the last byte taking in consideration possible
> + overflow. */
> + const char *lbyte = sadd (s_int, n - 1);
>
> - longword_ptr = (const longword *) char_ptr;
> + /* Compute the address of the word containing the last byte. */
> + const op_t *lword = word_containing (lbyte);
>
> - /* All these elucidatory comments refer to 4-byte longwords,
> - but the theory applies equally well to any size longwords. */
> + /* Read the first word, but munge it so that bytes before the array
> + will not match goal. */
> + const op_t * word_ptr = word_containing (s);
> + op_t word = (*word_ptr | before_mask) ^ (repeated_c & before_mask);
Why do you xor with repeated_c & before_mask here?
Doesn't the has_eq(word, repeated_c) do that?
>
> - /* Compute auxiliary longword values:
> - repeated_one is a value which has a 1 in every byte.
> - repeated_c has c in every byte. */
> - repeated_one = 0x01010101;
> - repeated_c = c | (c << 8);
> - repeated_c |= repeated_c << 16;
> - if (0xffffffffU < (longword) -1)
> + while (has_eq (word, repeated_c) == 0)
> {
> - repeated_one |= repeated_one << 31 << 1;
> - repeated_c |= repeated_c << 31 << 1;
> - if (8 < sizeof (longword))
> - {
> - size_t i;
> -
> - for (i = 64; i < sizeof (longword) * 8; i *= 2)
> - {
> - repeated_one |= repeated_one << i;
> - repeated_c |= repeated_c << i;
> - }
> - }
> + if (word_ptr == lword)
> + return NULL;
> + word = *++word_ptr;
> }
>
> - /* Instead of the traditional loop which tests each byte, we will test a
> - longword at a time. The tricky part is testing if *any of the four*
> - bytes in the longword in question are equal to c. We first use an xor
> - with repeated_c. This reduces the task to testing whether *any of the
> - four* bytes in longword1 is zero.
> -
> - We compute tmp =
> - ((longword1 - repeated_one) & ~longword1) & (repeated_one << 7).
> - That is, we perform the following operations:
> - 1. Subtract repeated_one.
> - 2. & ~longword1.
> - 3. & a mask consisting of 0x80 in every byte.
> - Consider what happens in each byte:
> - - If a byte of longword1 is zero, step 1 and 2 transform it into 0xff,
> - and step 3 transforms it into 0x80. A carry can also be propagated
> - to more significant bytes.
> - - If a byte of longword1 is nonzero, let its lowest 1 bit be at
> - position k (0 <= k <= 7); so the lowest k bits are 0. After step 1,
> - the byte ends in a single bit of value 0 and k bits of value 1.
> - After step 2, the result is just k bits of value 1: 2^k - 1. After
> - step 3, the result is 0. And no carry is produced.
> - So, if longword1 has only non-zero bytes, tmp is zero.
> - Whereas if longword1 has a zero byte, call j the position of the least
> - significant zero byte. Then the result has a zero at positions 0, ...,
> - j-1 and a 0x80 at position j. We cannot predict the result at the more
> - significant bytes (positions j+1..3), but it does not matter since we
> - already have a non-zero bit at position 8*j+7.
> -
> - So, the test whether any byte in longword1 is zero is equivalent to
> - testing whether tmp is nonzero. */
> -
> - while (n >= sizeof (longword))
> - {
> - longword longword1 = *longword_ptr ^ repeated_c;
> -
> - if ((((longword1 - repeated_one) & ~longword1)
> - & (repeated_one << 7)) != 0)
> - break;
> - longword_ptr++;
> - n -= sizeof (longword);
> - }
> -
> - char_ptr = (const unsigned char *) longword_ptr;
> -
> - /* At this point, we know that either n < sizeof (longword), or one of the
> - sizeof (longword) bytes starting at char_ptr is == c. On little-endian
> - machines, we could determine the first such byte without any further
> - memory accesses, just by looking at the tmp result from the last loop
> - iteration. But this does not work on big-endian machines. Choose code
> - that works in both cases. */
> -
> - for (; n > 0; --n, ++char_ptr)
> - {
> - if (*char_ptr == c)
> - return (void *) char_ptr;
> - }
> -
> - return NULL;
> + /* We found a match, but it might be in a byte past the end
> + of the array. */
> + char *ret = (char *) word_ptr + index_first_eq (word, repeated_c);
> + return (ret <= lbyte) ? ret : NULL;
> }
> -#ifdef weak_alias
> +#ifndef MEMCHR
> weak_alias (__memchr, memchr)
> -#endif
> libc_hidden_builtin_def (memchr)
> +#endif
> diff --git a/sysdeps/powerpc/powerpc32/power4/multiarch/memchr-ppc32.c b/sysdeps/powerpc/powerpc32/power4/multiarch/memchr-ppc32.c
> index fc69df54b3..02877d3c98 100644
> --- a/sysdeps/powerpc/powerpc32/power4/multiarch/memchr-ppc32.c
> +++ b/sysdeps/powerpc/powerpc32/power4/multiarch/memchr-ppc32.c
> @@ -18,17 +18,11 @@
>
> #include <string.h>
>
> -#define MEMCHR __memchr_ppc
> +extern __typeof (memchr) __memchr_ppc attribute_hidden;
>
> -#undef weak_alias
> -#define weak_alias(a, b)
> +#define MEMCHR __memchr_ppc
> +#include <string/memchr.c>
>
> #ifdef SHARED
> -# undef libc_hidden_builtin_def
> -# define libc_hidden_builtin_def(name) \
> - __hidden_ver1(__memchr_ppc, __GI_memchr, __memchr_ppc);
> +__hidden_ver1(__memchr_ppc, __GI_memchr, __memchr_ppc);
> #endif
> -
> -extern __typeof (memchr) __memchr_ppc attribute_hidden;
> -
> -#include <string/memchr.c>
> diff --git a/sysdeps/powerpc/powerpc64/multiarch/memchr-ppc64.c b/sysdeps/powerpc/powerpc64/multiarch/memchr-ppc64.c
> index 3c966f4403..15beca787b 100644
> --- a/sysdeps/powerpc/powerpc64/multiarch/memchr-ppc64.c
> +++ b/sysdeps/powerpc/powerpc64/multiarch/memchr-ppc64.c
> @@ -18,14 +18,7 @@
>
> #include <string.h>
>
> -#define MEMCHR __memchr_ppc
> -
> -#undef weak_alias
> -#define weak_alias(a, b)
> -
> -# undef libc_hidden_builtin_def
> -# define libc_hidden_builtin_def(name)
> -
> extern __typeof (memchr) __memchr_ppc attribute_hidden;
>
> +#define MEMCHR __memchr_ppc
> #include <string/memchr.c>
> --
> 2.34.1
>
next prev parent reply other threads:[~2022-09-03 3:48 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-09-02 20:39 [PATCH 00/17] Improve generic string routines Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 01/17] Parameterize op_t from memcopy.h Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 02/17] Parameterize OP_T_THRES " Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 03/17] Add string-maskoff.h generic header Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 04/17] Add string vectorized find and detection functions Adhemerval Zanella
2022-09-03 3:20 ` Noah Goldstein
2022-09-19 14:00 ` Adhemerval Zanella Netto
2022-09-02 20:39 ` [PATCH 05/17] string: Improve generic strlen Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 06/17] string: Improve generic strnlen Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 07/17] string: Improve generic strchr Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 08/17] string: Improve generic strchrnul Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 09/17] string: Improve generic strcmp Adhemerval Zanella
2022-09-03 3:31 ` Noah Goldstein
2022-09-19 14:04 ` Adhemerval Zanella Netto
2022-09-03 8:54 ` Richard Henderson
2022-09-02 20:39 ` [PATCH 10/17] string: Improve generic memchr Adhemerval Zanella
2022-09-03 3:47 ` Noah Goldstein [this message]
2022-09-19 19:17 ` Adhemerval Zanella Netto
2022-09-19 21:59 ` Noah Goldstein
2022-09-22 17:51 ` Adhemerval Zanella Netto
2022-09-02 20:39 ` [PATCH 11/17] string: Improve generic memrchr Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 12/17] hppa: Add memcopy.h Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 13/17] hppa: Add string-fzb.h and string-fzi.h Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 14/17] alpha: " Adhemerval Zanella
2022-09-02 20:39 ` [PATCH 15/17] arm: Add string-fza.h Adhemerval Zanella
2022-09-05 15:40 ` Richard Earnshaw
2022-09-05 15:50 ` Richard Earnshaw
2022-09-02 20:39 ` [PATCH 16/17] powerpc: " Adhemerval Zanella
2022-09-06 14:48 ` Paul E Murphy
2022-09-19 19:55 ` Adhemerval Zanella Netto
2022-09-02 20:39 ` [PATCH 17/17] sh: Add string-fzb.h Adhemerval Zanella
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=CAFUsyfJnneN8wJE4Rbi7qyrAPBXaoU5P536j_2hHJxk8UPys3g@mail.gmail.com \
--to=goldstein.w.n@gmail.com \
--cc=adhemerval.zanella@linaro.org \
--cc=caiyinyu@loongson.cn \
--cc=joseph@codesourcery.com \
--cc=libc-alpha@sourceware.org \
--cc=rth@twiddle.net \
/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).