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 04/17] Add string vectorized find and detection functions
Date: Fri, 2 Sep 2022 20:20:01 -0700 [thread overview]
Message-ID: <CAFUsyfJEoU2_Gbv0TkJqcWSaZ+SrQOVN8d7-jFACJENqkuuYKQ@mail.gmail.com> (raw)
In-Reply-To: <20220902203940.2385967-5-adhemerval.zanella@linaro.org>
On Fri, Sep 2, 2022 at 1:42 PM Adhemerval Zanella via Libc-alpha
<libc-alpha@sourceware.org> wrote:
>
> This patch adds generic string find and detection meant to be used in
> generic vectorized string implementation. The idea is to decompose the
> basic string operation so each architecture can reimplement if it
> provides any specialized hardware instruction.
>
> The 'string-fza.h' provides zero byte detection functions (find_zero_low,
> find_zero_all, find_eq_low, find_eq_all, find_zero_eq_low, find_zero_eq_all,
> find_zero_ne_low, and find_zero_ne_all). They are used on both functions
> provided by 'string-fzb.h' and 'string-fzi'.
>
> The 'string-fzb.h' provides boolean zero byte detection with the
> functions:
>
> - has_zero: determine if any byte within a word is zero.
> - has_eq: determine byte equality between two words.
> - has_zero_eq: determine if any byte within a word is zero along with
> byte equality between two words.
>
> The 'string-fzi.h' provides zero byte detection along with its positions:
>
> - index_first_zero: return index of first zero byte within a word.
> - index_first_eq: return index of first byte different between two words.
> - index_first_zero_eq: return index of first zero byte within a word or
> first byte different between two words.
> - index_first_zero_ne: return index of first zero byte within a word or
> first byte equal between two words.
> - index_last_zero: return index of last zero byte within a word.
> - index_last_eq: return index of last byte different between two words.
>
> Also, to avoid libcalls in the '__builtin_c{t,l}z{l}' calls (which may
> add performance degradation), inline implementation based on De Bruijn
> sequences are added (enabled by a configure check).
>
> Co-authored-by: Richard Henderson <rth@twiddle.net>
> ---
> config.h.in | 8 ++
> configure | 54 ++++++++
> configure.ac | 34 +++++
> sysdeps/generic/string-extbyte.h | 37 ++++++
> sysdeps/generic/string-fza.h | 106 ++++++++++++++++
> sysdeps/generic/string-fzb.h | 49 ++++++++
> sysdeps/generic/string-fzi.h | 208 +++++++++++++++++++++++++++++++
> 7 files changed, 496 insertions(+)
> create mode 100644 sysdeps/generic/string-extbyte.h
> create mode 100644 sysdeps/generic/string-fza.h
> create mode 100644 sysdeps/generic/string-fzb.h
> create mode 100644 sysdeps/generic/string-fzi.h
>
> diff --git a/config.h.in b/config.h.in
> index 43d32518ab..8f0540d85a 100644
> --- a/config.h.in
> +++ b/config.h.in
> @@ -202,6 +202,14 @@
> /* An integer used to scale the timeout of test programs. */
> #define TIMEOUTFACTOR 1
>
> +/* If compiler supports __builtin_ctz{l} without any external dependencies
> + (libgcc for instance). */
> +#define HAVE_BUILTIN_CTZ 0
> +
> +/* If compiler supports __builtin_clz{l} without any external dependencies
> + (libgcc for instance). */
> +#define HAVE_BUILTIN_CLZ 0
> +
> /*
> */
>
> diff --git a/configure b/configure
> index ff2c406b3b..99cdec1bef 100755
> --- a/configure
> +++ b/configure
> @@ -6726,6 +6726,60 @@ if test $libc_cv_builtin_trap = yes; then
>
> fi
>
> +{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for __builtin_ctz{l} with no external dependencies" >&5
> +$as_echo_n "checking for __builtin_ctz{l} with no external dependencies... " >&6; }
> +if ${libc_cv_builtin_ctz+:} false; then :
> + $as_echo_n "(cached) " >&6
> +else
> + libc_cv_builtin_ctz=yes
> +echo 'int foo (unsigned long x) { return __builtin_ctz (x); }' > conftest.c
> +if { ac_try='${CC-cc} $CFLAGS $CPPFLAGS -S conftest.c -o conftest.s 1>&5'
> + { { eval echo "\"\$as_me\":${as_lineno-$LINENO}: \"$ac_try\""; } >&5
> + (eval $ac_try) 2>&5
> + ac_status=$?
> + $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
> + test $ac_status = 0; }; }; then
> + if grep '__ctz[s,d]i2' conftest.s > /dev/null; then
> + libc_cv_builtin_ctz=no
> + fi
> +fi
> +rm -f conftest.c conftest.s
> +
> +fi
> +{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $libc_cv_builtin_ctz" >&5
> +$as_echo "$libc_cv_builtin_ctz" >&6; }
> +if test x$libc_cv_builtin_ctz = xyes; then
> + $as_echo "#define HAVE_BUILTIN_CTZ 1" >>confdefs.h
> +
> +fi
> +
> +{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for __builtin_clz{l} with no external dependencies" >&5
> +$as_echo_n "checking for __builtin_clz{l} with no external dependencies... " >&6; }
> +if ${libc_cv_builtin_clz+:} false; then :
> + $as_echo_n "(cached) " >&6
> +else
> + libc_cv_builtin_clz=yes
> +echo 'int foo (unsigned long x) { return __builtin_clz (x); }' > conftest.c
> +if { ac_try='${CC-cc} $CFLAGS $CPPFLAGS -S conftest.c -o conftest.s 1>&5'
> + { { eval echo "\"\$as_me\":${as_lineno-$LINENO}: \"$ac_try\""; } >&5
> + (eval $ac_try) 2>&5
> + ac_status=$?
> + $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
> + test $ac_status = 0; }; }; then
> + if grep '__clz[s,d]i2' conftest.s > /dev/null; then
> + libc_cv_builtin_clz=no
> + fi
> +fi
> +rm -f conftest.c conftest.s
> +
> +fi
> +{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $libc_cv_builtin_clz" >&5
> +$as_echo "$libc_cv_builtin_clz" >&6; }
> +if test x$libc_cv_builtin_clz = xyes; then
> + $as_echo "#define HAVE_BUILTIN_CLZ 1" >>confdefs.h
> +
> +fi
> +
> ac_ext=cpp
> ac_cpp='$CXXCPP $CPPFLAGS'
> ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
> diff --git a/configure.ac b/configure.ac
> index eb5bc6a131..66e02a0566 100644
> --- a/configure.ac
> +++ b/configure.ac
> @@ -1618,6 +1618,40 @@ if test $libc_cv_builtin_trap = yes; then
> AC_DEFINE([HAVE_BUILTIN_TRAP])
> fi
>
> +AC_CACHE_CHECK(for __builtin_ctz{l} with no external dependencies,
> + libc_cv_builtin_ctz, [dnl
> +libc_cv_builtin_ctz=yes
> +echo 'int foo (unsigned long x) { return __builtin_ctz (x); }' > conftest.c
> +if AC_TRY_COMMAND(${CC-cc} $CFLAGS $CPPFLAGS -S conftest.c -o conftest.s 1>&AS_MESSAGE_LOG_FD); then
> +changequote(,)dnl
> + if grep '__ctz[s,d]i2' conftest.s > /dev/null; then
> + libc_cv_builtin_ctz=no
> + fi
> +changequote([,])dnl
> +fi
> +rm -f conftest.c conftest.s
> +])
> +if test x$libc_cv_builtin_ctz = xyes; then
> + AC_DEFINE(HAVE_BUILTIN_CTZ)
> +fi
> +
> +AC_CACHE_CHECK(for __builtin_clz{l} with no external dependencies,
> + libc_cv_builtin_clz, [dnl
> +libc_cv_builtin_clz=yes
> +echo 'int foo (unsigned long x) { return __builtin_clz (x); }' > conftest.c
> +if AC_TRY_COMMAND(${CC-cc} $CFLAGS $CPPFLAGS -S conftest.c -o conftest.s 1>&AS_MESSAGE_LOG_FD); then
> +changequote(,)dnl
> + if grep '__clz[s,d]i2' conftest.s > /dev/null; then
> + libc_cv_builtin_clz=no
> + fi
> +changequote([,])dnl
> +fi
> +rm -f conftest.c conftest.s
> +])
> +if test x$libc_cv_builtin_clz = xyes; then
> + AC_DEFINE(HAVE_BUILTIN_CLZ)
> +fi
> +
> dnl C++ feature tests.
> AC_LANG_PUSH([C++])
>
> diff --git a/sysdeps/generic/string-extbyte.h b/sysdeps/generic/string-extbyte.h
> new file mode 100644
> index 0000000000..c8fecd259f
> --- /dev/null
> +++ b/sysdeps/generic/string-extbyte.h
> @@ -0,0 +1,37 @@
> +/* Extract by from memory word. Generic C version.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <http://www.gnu.org/licenses/>. */
> +
> +#ifndef _STRING_EXTBYTE_H
> +#define _STRING_EXTBYTE_H 1
> +
> +#include <limits.h>
> +#include <endian.h>
> +#include <string-optype.h>
> +
> +/* Extract the byte at index IDX from word X, with index 0 being the
> + least significant byte. */
> +static inline unsigned char
> +extractbyte (op_t x, unsigned int idx)
> +{
> + if (__BYTE_ORDER == __LITTLE_ENDIAN)
> + return x >> (idx * CHAR_BIT);
> + else
> + return x >> (sizeof (x) - 1 - idx) * CHAR_BIT;
> +}
> +
> +#endif /* _STRING_EXTBYTE_H */
> diff --git a/sysdeps/generic/string-fza.h b/sysdeps/generic/string-fza.h
> new file mode 100644
> index 0000000000..8470437824
> --- /dev/null
> +++ b/sysdeps/generic/string-fza.h
> @@ -0,0 +1,106 @@
> +/* Basic zero byte detection. Generic C version.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <http://www.gnu.org/licenses/>. */
> +
> +#ifndef _STRING_FZA_H
> +#define _STRING_FZA_H 1
> +
> +#include <limits.h>
> +#include <string-optype.h>
> +#include <string-maskoff.h>
> +
> +/* This function returns non-zero if any byte in X is zero.
> + More specifically, at least one bit set within the least significant
> + byte that was zero; other bytes within the word are indeterminate. */
> +static inline op_t
> +find_zero_low (op_t x)
> +{
> + /* This expression comes from
> + https://graphics.stanford.edu/~seander/bithacks.html#ZeroInWord
> + Subtracting 1 sets 0x80 in a byte that was 0; anding ~x clears
> + 0x80 in a byte that was >= 128; anding 0x80 isolates that test bit. */
> + op_t lsb = repeat_bytes (0x01);
> + op_t msb = repeat_bytes (0x80);
> + return (x - lsb) & ~x & msb;
> +}
> +
> +/* This function returns at least one bit set within every byte of X that
> + is zero. The result is exact in that, unlike find_zero_low, all bytes
> + are determinate. This is usually used for finding the index of the
> + most significant byte that was zero. */
> +static inline op_t
> +find_zero_all (op_t x)
> +{
> + /* For each byte, find not-zero by
> + (0) And 0x7f so that we cannot carry between bytes,
> + (1) Add 0x7f so that non-zero carries into 0x80,
> + (2) Or in the original byte (which might have had 0x80 set).
> + Then invert and mask such that 0x80 is set iff that byte was zero. */
> + op_t m = ((op_t)-1 / 0xff) * 0x7f;
> + return ~(((x & m) + m) | x | m);
> +}
> +
> +/* With similar caveats, identify bytes that are equal between X1 and X2. */
> +static inline op_t
> +find_eq_low (op_t x1, op_t x2)
> +{
> + return find_zero_low (x1 ^ x2);
> +}
> +
> +static inline op_t
> +find_eq_all (op_t x1, op_t x2)
> +{
> + return find_zero_all (x1 ^ x2);
> +}
> +
> +/* With similar caveats, identify zero bytes in X1 and bytes that are
> + equal between in X1 and X2. */
> +static inline op_t
> +find_zero_eq_low (op_t x1, op_t x2)
> +{
> + return find_zero_low (x1) | find_zero_low (x1 ^ x2);
> +}
> +
> +static inline op_t
> +find_zero_eq_all (op_t x1, op_t x2)
> +{
> + return find_zero_all (x1) | find_zero_all (x1 ^ x2);
> +}
> +
> +/* With similar caveats, identify zero bytes in X1 and bytes that are
> + not equal between in X1 and X2. */
> +static inline op_t
> +find_zero_ne_low (op_t x1, op_t x2)
> +{
> + op_t m = ((op_t)-1 / 0xff) * 0x7f;
Can this use repeat_bytes?
> + op_t eq = x1 ^ x2;
> + op_t nz1 = (x1 + m) | x1; /* msb set if byte not zero. */
> + op_t ne2 = (eq + m) | eq; /* msb set if byte not equal. */
> + return (ne2 | ~nz1) & ~m; /* msb set if x1 zero or x2 not equal. */
> +}
> +
> +static inline op_t
> +find_zero_ne_all (op_t x1, op_t x2)
> +{
> + op_t m = ((op_t)-1 / 0xff) * 0x7f;
Likewise. Elsewhere too.
> + op_t eq = x1 ^ x2;
> + op_t nz1 = ((x1 & m) + m) | x1;
> + op_t ne2 = ((eq & m) + m) | eq;
> + return (ne2 | ~nz1) & ~m;
> +}
> +
> +#endif /* _STRING_FZA_H */
> diff --git a/sysdeps/generic/string-fzb.h b/sysdeps/generic/string-fzb.h
> new file mode 100644
> index 0000000000..f1c0ae0922
> --- /dev/null
> +++ b/sysdeps/generic/string-fzb.h
> @@ -0,0 +1,49 @@
> +/* Zero byte detection, boolean. Generic C version.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <http://www.gnu.org/licenses/>. */
> +
> +#ifndef _STRING_FZB_H
> +#define _STRING_FZB_H 1
> +
> +#include <endian.h>
> +#include <string-fza.h>
> +
> +/* Determine if any byte within X is zero. This is a pure boolean test. */
> +
> +static inline _Bool
> +has_zero (op_t x)
> +{
> + return find_zero_low (x) != 0;
> +}
> +
> +/* Likewise, but for byte equality between X1 and X2. */
> +
> +static inline _Bool
> +has_eq (op_t x1, op_t x2)
> +{
> + return find_eq_low (x1, x2) != 0;
> +}
> +
> +/* Likewise, but for zeros in X1 and equal bytes between X1 and X2. */
> +
> +static inline _Bool
> +has_zero_eq (op_t x1, op_t x2)
> +{
> + return find_zero_eq_low (x1, x2);
> +}
> +
> +#endif /* _STRING_FZB_H */
> diff --git a/sysdeps/generic/string-fzi.h b/sysdeps/generic/string-fzi.h
> new file mode 100644
> index 0000000000..8c6b6dc3d8
> --- /dev/null
> +++ b/sysdeps/generic/string-fzi.h
> @@ -0,0 +1,208 @@
> +/* Zero byte detection; indexes. Generic C version.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <http://www.gnu.org/licenses/>. */
> +
> +#ifndef _STRING_FZI_H
> +#define _STRING_FZI_H 1
> +
> +#include <limits.h>
> +#include <endian.h>
> +#include <string-fza.h>
> +
> +/* An improved bitscan routine, multiplying the De Bruijn sequence with a
> + 0-1 mask separated by the least significant one bit of a scanned integer
> + or bitboard [1].
> +
> + [1] https://chessprogramming.wikispaces.com/Kim+Walisch */
> +static inline unsigned int
> +index_access (const op_t i)
> +{
> + static const char index[] =
> + {
> +# if __WORDSIZE == 64
> + 0, 47, 1, 56, 48, 27, 2, 60,
> + 57, 49, 41, 37, 28, 16, 3, 61,
> + 54, 58, 35, 52, 50, 42, 21, 44,
> + 38, 32, 29, 23, 17, 11, 4, 62,
> + 46, 55, 26, 59, 40, 36, 15, 53,
> + 34, 51, 20, 43, 31, 22, 10, 45,
> + 25, 39, 14, 33, 19, 30, 9, 24,
> + 13, 18, 8, 12, 7, 6, 5, 63
> +# else
> + 0, 9, 1, 10, 13, 21, 2, 29,
> + 11, 14, 16, 18, 22, 25, 3, 30,
> + 8, 12, 20, 28, 15, 17, 24, 7,
> + 19, 27, 23, 6, 26, 5, 4, 31
> +# endif
> + };
> + return index[i];
> +}
> +
> +/* For architectures which only provides __builtin_clz{l} (HAVE_BUILTIN_CLZ)
> + and/or __builtin_ctz{l} (HAVE_BUILTIN_CTZ) which uses external libcalls
> + (for intance __c{l,t}z{s,d}i2 from libgcc) the following wrapper provides
> + inline implementation for both count leading zeros and count trailing
> + zeros using branchless computation. As for builtin, if x is 0 the
> + result is undefined.*/
> +static inline unsigned int
> +__ctz (op_t x)
> +{
> +#if !HAVE_BUILTIN_CTZ
> + op_t i;
> +# if __WORDSIZE == 64
> + i = (x ^ (x - 1)) * 0x03F79D71B4CB0A89ull >> 58;
> +# else
> + i = (x ^ (x - 1)) * 0x07C4ACDDU >> 27;
> +# endif
> + return index_access (i);
> +#else
> + if (sizeof (op_t) == sizeof (long int))
> + return __builtin_ctzl (x);
> + else
> + return __builtin_ctzll (x);
> +#endif
> +};
> +
> +static inline unsigned int
> +__clz (op_t x)
> +{
> +#if !HAVE_BUILTIN_CLZ
> + unsigned r;
> + op_t i;
> +
> + x |= x >> 1;
> + x |= x >> 2;
> + x |= x >> 4;
> + x |= x >> 8;
> + x |= x >> 16;
> +# if __WORDSIZE == 64
> + x |= x >> 32;
> + i = x * 0x03F79D71B4CB0A89ull >> 58;
> +# else
> + i = x * 0x07C4ACDDU >> 27;
> +# endif
> + r = index_access (i);
> + return r ^ (sizeof (op_t) * CHAR_BIT - 1);
> +#else
> + if (sizeof (op_t) == sizeof (long int))
> + return __builtin_clzl (x);
> + else
> + return __builtin_clzll (x);
> +#endif
> +}
> +
> +/* A subroutine for the index_zero functions. Given a test word C, return
> + the (memory order) index of the first byte (in memory order) that is
> + non-zero. */
> +static inline unsigned int
> +index_first_ (op_t c)
> +{
> + _Static_assert (sizeof (op_t) == sizeof (long int)
> + || sizeof (op_t) == sizeof (long long int),
> + "Unhandled word size");
> +
> + unsigned r;
> + if (__BYTE_ORDER == __LITTLE_ENDIAN)
> + r = __ctz (c);
> + else
> + r = __clz (c);
> + return r / CHAR_BIT;
> +}
> +
> +/* Similarly, but return the (memory order) index of the last byte that is
> + non-zero. */
> +static inline unsigned int
> +index_last_ (op_t c)
> +{
> + _Static_assert (sizeof (op_t) == sizeof (long int)
> + || sizeof (op_t) == sizeof (long long int),
> + "Unhandled word size");
> +
> + unsigned r;
> + if (__BYTE_ORDER == __LITTLE_ENDIAN)
> + r = __clz (c);
> + else
> + r = __ctz (c);
> + return sizeof (op_t) - 1 - (r / CHAR_BIT);
> +}
> +
> +/* Given a word X that is known to contain a zero byte, return the
> + index of the first such within the word in memory order. */
> +
> +static inline unsigned int
> +index_first_zero (op_t x)
> +{
> + if (__BYTE_ORDER == __LITTLE_ENDIAN)
> + x = find_zero_low (x);
> + else
> + x = find_zero_all (x);
> + return index_first_ (x);
> +}
> +
> +/* Similarly, but perform the search for byte equality between X1 and X2. */
> +static inline unsigned int
> +index_first_eq (op_t x1, op_t x2)
> +{
> + if (__BYTE_ORDER == __LITTLE_ENDIAN)
> + x1 = find_eq_low (x1, x2);
> + else
> + x1 = find_eq_all (x1, x2);
> + return index_first_ (x1);
> +}
> +
> +/* Similarly, but perform the search for zero within X1 or equality between
> + X1 and X2. */
> +static inline unsigned int
> +index_first_zero_eq (op_t x1, op_t x2)
> +{
> + if (__BYTE_ORDER == __LITTLE_ENDIAN)
> + x1 = find_zero_eq_low (x1, x2);
> + else
> + x1 = find_zero_eq_all (x1, x2);
> + return index_first_ (x1);
> +}
> +
> +/* Similarly, but perform the search for zero within X1 or
> + inequality between X1 and X2. */
> +static inline unsigned int
> +index_first_zero_ne (op_t x1, op_t x2)
> +{
> + if (__BYTE_ORDER == __LITTLE_ENDIAN)
> + x1 = find_zero_ne_low (x1, x2);
> + else
> + x1 = find_zero_ne_all (x1, x2);
> + return index_first_ (x1);
> +}
> +
> +/* Similarly, but search for the last zero within X. */
> +static inline unsigned int
> +index_last_zero (op_t x)
> +{
> + if (__BYTE_ORDER == __LITTLE_ENDIAN)
> + x = find_zero_all (x);
> + else
> + x = find_zero_low (x);
> + return index_last_ (x);
> +}
> +
> +static inline unsigned int
> +index_last_eq (op_t x1, op_t x2)
> +{
> + return index_last_zero (x1 ^ x2);
> +}
> +
> +#endif /* STRING_FZI_H */
> --
> 2.34.1
>
next prev parent reply other threads:[~2022-09-03 3:20 UTC|newest]
Thread overview: 33+ 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 [this message]
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
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
2022-09-03 13:13 [PATCH 04/17] Add string vectorized find and detection functions Wilco Dijkstra
2022-09-19 13:59 ` Adhemerval Zanella Netto
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=CAFUsyfJEoU2_Gbv0TkJqcWSaZ+SrQOVN8d7-jFACJENqkuuYKQ@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).