glibc/stdlib/stdbit.h

849 lines
26 KiB
C
Raw Permalink Normal View History

Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
/* ISO C23 Standard: 7.18 - Bit and byte utilities <stdbit.h>.
Copyright (C) 2024 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
<https://www.gnu.org/licenses/>. */
#ifndef _STDBIT_H
#define _STDBIT_H 1
#include <features.h>
#include <bits/endian.h>
#include <bits/stdint-intn.h>
#include <bits/stdint-uintn.h>
#include <bits/stdint-least.h>
/* In C23, <stdbool.h> defines only an implementation-namespace macro,
so is OK to include here. Before C23, including <stdbool.h> allows
the header to use bool rather than _Bool unconditionally, and so to
compile as C++ (although the type-generic macros are not a good
form of type-generic interface for C++). */
#include <stdbool.h>
#define __need_size_t
#include <stddef.h>
#define __STDC_VERSION_STDBIT_H__ 202311L
#define __STDC_ENDIAN_LITTLE__ __LITTLE_ENDIAN
#define __STDC_ENDIAN_BIG__ __BIG_ENDIAN
#define __STDC_ENDIAN_NATIVE__ __BYTE_ORDER
__BEGIN_DECLS
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
/* Use __pacify_uint16 (N) instead of (uint16_t) (N) when the cast is helpful
stdlib: Fix stdbit.h with -Wconversion for clang With clang 14 and also with main the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h:701:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:707:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:751:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:757:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ tst-stdbit-Wconversion.c:45:31: error: implicit conversion loses integer precision: 'unsigned short' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (us); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:47:31: error: implicit conversion loses integer precision: 'unsigned long' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ul); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ [...] It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Although not strickly required for older gcc, using the same __pacify macro simpify the required code. Checked on x86_64-linux-gnu and i686-linux-gnu.
2024-01-04 17:59:28 +00:00
only to pacify older GCC (e.g., GCC 10 -Wconversion) or non-GCC (e.g
clang -Wimplicit-int-conversion). */
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
#if __GNUC_PREREQ (11, 0)
# define __pacify_uint8(n) (n)
# define __pacify_uint16(n) (n)
#else
# define __pacify_uint8(n) ((uint8_t) (n))
# define __pacify_uint16(n) ((uint16_t) (n))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
/* Count leading zeros. */
extern unsigned int stdc_leading_zeros_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_leading_zeros_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_leading_zeros_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_leading_zeros_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_leading_zeros_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_leading_zeros)
# define stdc_leading_zeros(x) (__builtin_stdc_leading_zeros (x))
#else
# define stdc_leading_zeros(x) \
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
(stdc_leading_zeros_ull (x) \
- (unsigned int) (8 * (sizeof (0ULL) - sizeof (x))))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_clzll)
static __always_inline unsigned int
__clz64_inline (uint64_t __x)
{
return __x == 0 ? 64U : (unsigned int) __builtin_clzll (__x);
}
static __always_inline unsigned int
__clz32_inline (uint32_t __x)
{
return __x == 0 ? 32U : (unsigned int) __builtin_clz (__x);
}
static __always_inline unsigned int
__clz16_inline (uint16_t __x)
{
return __clz32_inline (__x) - 16;
}
static __always_inline unsigned int
__clz8_inline (uint8_t __x)
{
return __clz32_inline (__x) - 24;
}
# define stdc_leading_zeros_uc(x) (__clz8_inline (x))
# define stdc_leading_zeros_us(x) (__clz16_inline (x))
# define stdc_leading_zeros_ui(x) (__clz32_inline (x))
# if __WORDSIZE == 64
# define stdc_leading_zeros_ul(x) (__clz64_inline (x))
# else
# define stdc_leading_zeros_ul(x) (__clz32_inline (x))
# endif
# define stdc_leading_zeros_ull(x) (__clz64_inline (x))
#endif
/* Count leading ones. */
extern unsigned int stdc_leading_ones_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_leading_ones_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_leading_ones_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_leading_ones_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_leading_ones_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_leading_ones)
# define stdc_leading_ones(x) (__builtin_stdc_leading_ones (x))
#else
# define stdc_leading_ones(x) \
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
(stdc_leading_ones_ull ((unsigned long long int) (x) \
<< 8 * (sizeof (0ULL) - sizeof (x))))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_clzll)
static __always_inline unsigned int
__clo64_inline (uint64_t __x)
{
return __clz64_inline (~__x);
}
static __always_inline unsigned int
__clo32_inline (uint32_t __x)
{
return __clz32_inline (~__x);
}
static __always_inline unsigned int
__clo16_inline (uint16_t __x)
{
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
return __clz16_inline (__pacify_uint16 (~__x));
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
}
static __always_inline unsigned int
__clo8_inline (uint8_t __x)
{
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
return __clz8_inline (__pacify_uint8 (~__x));
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
}
# define stdc_leading_ones_uc(x) (__clo8_inline (x))
# define stdc_leading_ones_us(x) (__clo16_inline (x))
# define stdc_leading_ones_ui(x) (__clo32_inline (x))
# if __WORDSIZE == 64
# define stdc_leading_ones_ul(x) (__clo64_inline (x))
# else
# define stdc_leading_ones_ul(x) (__clo32_inline (x))
# endif
# define stdc_leading_ones_ull(x) (__clo64_inline (x))
#endif
/* Count trailing zeros. */
extern unsigned int stdc_trailing_zeros_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_trailing_zeros_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_trailing_zeros_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_trailing_zeros_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_trailing_zeros_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_trailing_zeros)
# define stdc_trailing_zeros(x) (__builtin_stdc_trailing_zeros (x))
#else
# define stdc_trailing_zeros(x) \
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
(sizeof (x) == 8 ? stdc_trailing_zeros_ull (x) \
: sizeof (x) == 4 ? stdc_trailing_zeros_ui (x) \
stdlib: Fix stdbit.h with -Wconversion for clang With clang 14 and also with main the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h:701:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:707:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:751:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:757:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ tst-stdbit-Wconversion.c:45:31: error: implicit conversion loses integer precision: 'unsigned short' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (us); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:47:31: error: implicit conversion loses integer precision: 'unsigned long' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ul); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ [...] It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Although not strickly required for older gcc, using the same __pacify macro simpify the required code. Checked on x86_64-linux-gnu and i686-linux-gnu.
2024-01-04 17:59:28 +00:00
: sizeof (x) == 2 ? stdc_trailing_zeros_us (__pacify_uint16 (x)) \
: stdc_trailing_zeros_uc (__pacify_uint8 (x)))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_ctzll)
static __always_inline unsigned int
__ctz64_inline (uint64_t __x)
{
return __x == 0 ? 64U : (unsigned int) __builtin_ctzll (__x);
}
static __always_inline unsigned int
__ctz32_inline (uint32_t __x)
{
return __x == 0 ? 32U : (unsigned int) __builtin_ctz (__x);
}
static __always_inline unsigned int
__ctz16_inline (uint16_t __x)
{
return __x == 0 ? 16U : (unsigned int) __builtin_ctz (__x);
}
static __always_inline unsigned int
__ctz8_inline (uint8_t __x)
{
return __x == 0 ? 8U : (unsigned int) __builtin_ctz (__x);
}
# define stdc_trailing_zeros_uc(x) (__ctz8_inline (x))
# define stdc_trailing_zeros_us(x) (__ctz16_inline (x))
# define stdc_trailing_zeros_ui(x) (__ctz32_inline (x))
# if __WORDSIZE == 64
# define stdc_trailing_zeros_ul(x) (__ctz64_inline (x))
# else
# define stdc_trailing_zeros_ul(x) (__ctz32_inline (x))
# endif
# define stdc_trailing_zeros_ull(x) (__ctz64_inline (x))
#endif
/* Count trailing ones. */
extern unsigned int stdc_trailing_ones_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_trailing_ones_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_trailing_ones_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_trailing_ones_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_trailing_ones_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_trailing_ones)
# define stdc_trailing_ones(x) (__builtin_stdc_trailing_ones (x))
#else
# define stdc_trailing_ones(x) (stdc_trailing_ones_ull (x))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_ctzll)
static __always_inline unsigned int
__cto64_inline (uint64_t __x)
{
return __ctz64_inline (~__x);
}
static __always_inline unsigned int
__cto32_inline (uint32_t __x)
{
return __ctz32_inline (~__x);
}
static __always_inline unsigned int
__cto16_inline (uint16_t __x)
{
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
return __ctz16_inline (__pacify_uint16 (~__x));
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
}
static __always_inline unsigned int
__cto8_inline (uint8_t __x)
{
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
return __ctz8_inline (__pacify_uint8 (~__x));
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
}
# define stdc_trailing_ones_uc(x) (__cto8_inline (x))
# define stdc_trailing_ones_us(x) (__cto16_inline (x))
# define stdc_trailing_ones_ui(x) (__cto32_inline (x))
# if __WORDSIZE == 64
# define stdc_trailing_ones_ul(x) (__cto64_inline (x))
# else
# define stdc_trailing_ones_ul(x) (__cto32_inline (x))
# endif
# define stdc_trailing_ones_ull(x) (__cto64_inline (x))
#endif
/* First leading zero. */
extern unsigned int stdc_first_leading_zero_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_leading_zero_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_leading_zero_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_leading_zero_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_first_leading_zero_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_first_leading_zero)
# define stdc_first_leading_zero(x) (__builtin_stdc_first_leading_zero (x))
#else
# define stdc_first_leading_zero(x) \
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
(sizeof (x) == 8 ? stdc_first_leading_zero_ull (x) \
: sizeof (x) == 4 ? stdc_first_leading_zero_ui (x) \
stdlib: Fix stdbit.h with -Wconversion for clang With clang 14 and also with main the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h:701:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:707:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:751:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:757:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ tst-stdbit-Wconversion.c:45:31: error: implicit conversion loses integer precision: 'unsigned short' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (us); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:47:31: error: implicit conversion loses integer precision: 'unsigned long' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ul); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ [...] It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Although not strickly required for older gcc, using the same __pacify macro simpify the required code. Checked on x86_64-linux-gnu and i686-linux-gnu.
2024-01-04 17:59:28 +00:00
: sizeof (x) == 2 ? stdc_first_leading_zero_us (__pacify_uint16 (x)) \
: stdc_first_leading_zero_uc (__pacify_uint8 (x)))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_clzll)
static __always_inline unsigned int
__flz64_inline (uint64_t __x)
{
return __x == (uint64_t) -1 ? 0 : 1 + __clo64_inline (__x);
}
static __always_inline unsigned int
__flz32_inline (uint32_t __x)
{
return __x == (uint32_t) -1 ? 0 : 1 + __clo32_inline (__x);
}
static __always_inline unsigned int
__flz16_inline (uint16_t __x)
{
return __x == (uint16_t) -1 ? 0 : 1 + __clo16_inline (__x);
}
static __always_inline unsigned int
__flz8_inline (uint8_t __x)
{
return __x == (uint8_t) -1 ? 0 : 1 + __clo8_inline (__x);
}
# define stdc_first_leading_zero_uc(x) (__flz8_inline (x))
# define stdc_first_leading_zero_us(x) (__flz16_inline (x))
# define stdc_first_leading_zero_ui(x) (__flz32_inline (x))
# if __WORDSIZE == 64
# define stdc_first_leading_zero_ul(x) (__flz64_inline (x))
# else
# define stdc_first_leading_zero_ul(x) (__flz32_inline (x))
# endif
# define stdc_first_leading_zero_ull(x) (__flz64_inline (x))
#endif
/* First leading one. */
extern unsigned int stdc_first_leading_one_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_leading_one_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_leading_one_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_leading_one_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_first_leading_one_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_first_leading_one)
# define stdc_first_leading_one(x) (__builtin_stdc_first_leading_one (x))
#else
# define stdc_first_leading_one(x) \
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
(sizeof (x) == 8 ? stdc_first_leading_one_ull (x) \
: sizeof (x) == 4 ? stdc_first_leading_one_ui (x) \
stdlib: Fix stdbit.h with -Wconversion for clang With clang 14 and also with main the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h:701:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:707:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:751:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:757:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ tst-stdbit-Wconversion.c:45:31: error: implicit conversion loses integer precision: 'unsigned short' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (us); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:47:31: error: implicit conversion loses integer precision: 'unsigned long' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ul); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ [...] It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Although not strickly required for older gcc, using the same __pacify macro simpify the required code. Checked on x86_64-linux-gnu and i686-linux-gnu.
2024-01-04 17:59:28 +00:00
: sizeof (x) == 2 ? stdc_first_leading_one_us (__pacify_uint16 (x)) \
: stdc_first_leading_one_uc (__pacify_uint8 (x)))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_clzll)
static __always_inline unsigned int
__flo64_inline (uint64_t __x)
{
return __x == 0 ? 0 : 1 + __clz64_inline (__x);
}
static __always_inline unsigned int
__flo32_inline (uint32_t __x)
{
return __x == 0 ? 0 : 1 + __clz32_inline (__x);
}
static __always_inline unsigned int
__flo16_inline (uint16_t __x)
{
return __x == 0 ? 0 : 1 + __clz16_inline (__x);
}
static __always_inline unsigned int
__flo8_inline (uint8_t __x)
{
return __x == 0 ? 0 : 1 + __clz8_inline (__x);
}
# define stdc_first_leading_one_uc(x) (__flo8_inline (x))
# define stdc_first_leading_one_us(x) (__flo16_inline (x))
# define stdc_first_leading_one_ui(x) (__flo32_inline (x))
# if __WORDSIZE == 64
# define stdc_first_leading_one_ul(x) (__flo64_inline (x))
# else
# define stdc_first_leading_one_ul(x) (__flo32_inline (x))
# endif
# define stdc_first_leading_one_ull(x) (__flo64_inline (x))
#endif
/* First trailing zero. */
extern unsigned int stdc_first_trailing_zero_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_trailing_zero_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_trailing_zero_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_trailing_zero_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_first_trailing_zero_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_first_trailing_zero)
# define stdc_first_trailing_zero(x) (__builtin_stdc_first_trailing_zero (x))
#else
# define stdc_first_trailing_zero(x) \
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
(sizeof (x) == 8 ? stdc_first_trailing_zero_ull (x) \
: sizeof (x) == 4 ? stdc_first_trailing_zero_ui (x) \
stdlib: Fix stdbit.h with -Wconversion for clang With clang 14 and also with main the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h:701:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:707:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:751:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:757:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ tst-stdbit-Wconversion.c:45:31: error: implicit conversion loses integer precision: 'unsigned short' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (us); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:47:31: error: implicit conversion loses integer precision: 'unsigned long' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ul); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ [...] It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Although not strickly required for older gcc, using the same __pacify macro simpify the required code. Checked on x86_64-linux-gnu and i686-linux-gnu.
2024-01-04 17:59:28 +00:00
: sizeof (x) == 2 ? stdc_first_trailing_zero_us (__pacify_uint16 (x)) \
: stdc_first_trailing_zero_uc (__pacify_uint8 (x)))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_ctzll)
static __always_inline unsigned int
__ftz64_inline (uint64_t __x)
{
return __x == (uint64_t) -1 ? 0 : 1 + __cto64_inline (__x);
}
static __always_inline unsigned int
__ftz32_inline (uint32_t __x)
{
return __x == (uint32_t) -1 ? 0 : 1 + __cto32_inline (__x);
}
static __always_inline unsigned int
__ftz16_inline (uint16_t __x)
{
return __x == (uint16_t) -1 ? 0 : 1 + __cto16_inline (__x);
}
static __always_inline unsigned int
__ftz8_inline (uint8_t __x)
{
return __x == (uint8_t) -1 ? 0 : 1 + __cto8_inline (__x);
}
# define stdc_first_trailing_zero_uc(x) (__ftz8_inline (x))
# define stdc_first_trailing_zero_us(x) (__ftz16_inline (x))
# define stdc_first_trailing_zero_ui(x) (__ftz32_inline (x))
# if __WORDSIZE == 64
# define stdc_first_trailing_zero_ul(x) (__ftz64_inline (x))
# else
# define stdc_first_trailing_zero_ul(x) (__ftz32_inline (x))
# endif
# define stdc_first_trailing_zero_ull(x) (__ftz64_inline (x))
#endif
/* First trailing one. */
extern unsigned int stdc_first_trailing_one_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_trailing_one_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_trailing_one_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_first_trailing_one_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_first_trailing_one_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_first_trailing_one)
# define stdc_first_trailing_one(x) (__builtin_stdc_first_trailing_one (x))
#else
# define stdc_first_trailing_one(x) \
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
(sizeof (x) == 8 ? stdc_first_trailing_one_ull (x) \
: sizeof (x) == 4 ? stdc_first_trailing_one_ui (x) \
stdlib: Fix stdbit.h with -Wconversion for clang With clang 14 and also with main the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h:701:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:707:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:751:40: error: implicit conversion loses integer precision: 'int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h:757:39: error: implicit conversion loses integer precision: 'int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ~~~~~~ ~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ tst-stdbit-Wconversion.c:45:31: error: implicit conversion loses integer precision: 'unsigned short' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (us); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:46:31: error: implicit conversion loses integer precision: 'unsigned int' to 'uint8_t' (aka 'unsigned char') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ui); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:164:30: note: expanded from macro 'stdc_trailing_zeros' : stdc_trailing_zeros_uc (x)) ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:191:52: note: expanded from macro 'stdc_trailing_zeros_uc' # define stdc_trailing_zeros_uc(x) (__ctz8_inline (x)) ~~~~~~~~~~~~~ ^ tst-stdbit-Wconversion.c:47:31: error: implicit conversion loses integer precision: 'unsigned long' to 'uint16_t' (aka 'unsigned short') [-Werror,-Wimplicit-int-conversion] (void) stdc_trailing_zeros (ul); ~~~~~~~~~~~~~~~~~~~~~^~~ ../stdlib/stdbit.h:163:48: note: expanded from macro 'stdc_trailing_zeros' : sizeof (x) == 2 ? stdc_trailing_zeros_us (x) \ ~~~~~~~~~~~~~~~~~~~~~~~~^~ ../stdlib/stdbit.h:192:53: note: expanded from macro 'stdc_trailing_zeros_us' # define stdc_trailing_zeros_us(x) (__ctz16_inline (x)) ~~~~~~~~~~~~~~ ^ [...] It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Although not strickly required for older gcc, using the same __pacify macro simpify the required code. Checked on x86_64-linux-gnu and i686-linux-gnu.
2024-01-04 17:59:28 +00:00
: sizeof (x) == 2 ? stdc_first_trailing_one_us (__pacify_uint16 (x)) \
: stdc_first_trailing_one_uc (__pacify_uint8 (x)))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_ctzll)
static __always_inline unsigned int
__fto64_inline (uint64_t __x)
{
return __x == 0 ? 0 : 1 + __ctz64_inline (__x);
}
static __always_inline unsigned int
__fto32_inline (uint32_t __x)
{
return __x == 0 ? 0 : 1 + __ctz32_inline (__x);
}
static __always_inline unsigned int
__fto16_inline (uint16_t __x)
{
return __x == 0 ? 0 : 1 + __ctz16_inline (__x);
}
static __always_inline unsigned int
__fto8_inline (uint8_t __x)
{
return __x == 0 ? 0 : 1 + __ctz8_inline (__x);
}
# define stdc_first_trailing_one_uc(x) (__fto8_inline (x))
# define stdc_first_trailing_one_us(x) (__fto16_inline (x))
# define stdc_first_trailing_one_ui(x) (__fto32_inline (x))
# if __WORDSIZE == 64
# define stdc_first_trailing_one_ul(x) (__fto64_inline (x))
# else
# define stdc_first_trailing_one_ul(x) (__fto32_inline (x))
# endif
# define stdc_first_trailing_one_ull(x) (__fto64_inline (x))
#endif
/* Count zeros. */
extern unsigned int stdc_count_zeros_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_count_zeros_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_count_zeros_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_count_zeros_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_count_zeros_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_count_zeros)
# define stdc_count_zeros(x) (__builtin_stdc_count_zeros (x))
#else
# define stdc_count_zeros(x) \
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
(stdc_count_zeros_ull (x) \
- (unsigned int) (8 * (sizeof (0ULL) - sizeof (x))))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_popcountll)
static __always_inline unsigned int
__cz64_inline (uint64_t __x)
{
return 64U - (unsigned int) __builtin_popcountll (__x);
}
static __always_inline unsigned int
__cz32_inline (uint32_t __x)
{
return 32U - (unsigned int) __builtin_popcount (__x);
}
static __always_inline unsigned int
__cz16_inline (uint16_t __x)
{
return 16U - (unsigned int) __builtin_popcount (__x);
}
static __always_inline unsigned int
__cz8_inline (uint8_t __x)
{
return 8U - (unsigned int) __builtin_popcount (__x);
}
# define stdc_count_zeros_uc(x) (__cz8_inline (x))
# define stdc_count_zeros_us(x) (__cz16_inline (x))
# define stdc_count_zeros_ui(x) (__cz32_inline (x))
# if __WORDSIZE == 64
# define stdc_count_zeros_ul(x) (__cz64_inline (x))
# else
# define stdc_count_zeros_ul(x) (__cz32_inline (x))
# endif
# define stdc_count_zeros_ull(x) (__cz64_inline (x))
#endif
/* Count ones. */
extern unsigned int stdc_count_ones_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_count_ones_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_count_ones_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_count_ones_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_count_ones_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_count_ones)
# define stdc_count_ones(x) (__builtin_stdc_count_ones (x))
#else
# define stdc_count_ones(x) (stdc_count_ones_ull (x))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_popcountll)
static __always_inline unsigned int
__co64_inline (uint64_t __x)
{
return (unsigned int) __builtin_popcountll (__x);
}
static __always_inline unsigned int
__co32_inline (uint32_t __x)
{
return (unsigned int) __builtin_popcount (__x);
}
static __always_inline unsigned int
__co16_inline (uint16_t __x)
{
return (unsigned int) __builtin_popcount (__x);
}
static __always_inline unsigned int
__co8_inline (uint8_t __x)
{
return (unsigned int) __builtin_popcount (__x);
}
# define stdc_count_ones_uc(x) (__co8_inline (x))
# define stdc_count_ones_us(x) (__co16_inline (x))
# define stdc_count_ones_ui(x) (__co32_inline (x))
# if __WORDSIZE == 64
# define stdc_count_ones_ul(x) (__co64_inline (x))
# else
# define stdc_count_ones_ul(x) (__co32_inline (x))
# endif
# define stdc_count_ones_ull(x) (__co64_inline (x))
#endif
/* Single-bit check. */
extern bool stdc_has_single_bit_uc (unsigned char __x)
__THROW __attribute_const__;
extern bool stdc_has_single_bit_us (unsigned short __x)
__THROW __attribute_const__;
extern bool stdc_has_single_bit_ui (unsigned int __x)
__THROW __attribute_const__;
extern bool stdc_has_single_bit_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern bool stdc_has_single_bit_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_has_single_bit)
# define stdc_has_single_bit(x) (__builtin_stdc_has_single_bit (x))
#else
# define stdc_has_single_bit(x) \
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
((bool) (sizeof (x) <= sizeof (unsigned int) \
? stdc_has_single_bit_ui (x) \
: stdc_has_single_bit_ull (x)))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
static __always_inline bool
__hsb64_inline (uint64_t __x)
{
return (__x ^ (__x - 1)) > __x - 1;
}
static __always_inline bool
__hsb32_inline (uint32_t __x)
{
return (__x ^ (__x - 1)) > __x - 1;
}
static __always_inline bool
__hsb16_inline (uint16_t __x)
{
return (__x ^ (__x - 1)) > __x - 1;
}
static __always_inline bool
__hsb8_inline (uint8_t __x)
{
return (__x ^ (__x - 1)) > __x - 1;
}
#define stdc_has_single_bit_uc(x) (__hsb8_inline (x))
#define stdc_has_single_bit_us(x) (__hsb16_inline (x))
#define stdc_has_single_bit_ui(x) (__hsb32_inline (x))
#if __WORDSIZE == 64
# define stdc_has_single_bit_ul(x) (__hsb64_inline (x))
#else
# define stdc_has_single_bit_ul(x) (__hsb32_inline (x))
#endif
#define stdc_has_single_bit_ull(x) (__hsb64_inline (x))
/* Bit width. */
extern unsigned int stdc_bit_width_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned int stdc_bit_width_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_bit_width_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned int stdc_bit_width_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned int stdc_bit_width_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_bit_width)
# define stdc_bit_width(x) (__builtin_stdc_bit_width (x))
#else
# define stdc_bit_width(x) (stdc_bit_width_ull (x))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_clzll)
static __always_inline unsigned int
__bw64_inline (uint64_t __x)
{
return 64 - __clz64_inline (__x);
}
static __always_inline unsigned int
__bw32_inline (uint32_t __x)
{
return 32 - __clz32_inline (__x);
}
static __always_inline unsigned int
__bw16_inline (uint16_t __x)
{
return 16 - __clz16_inline (__x);
}
static __always_inline unsigned int
__bw8_inline (uint8_t __x)
{
return 8 - __clz8_inline (__x);
}
# define stdc_bit_width_uc(x) (__bw8_inline (x))
# define stdc_bit_width_us(x) (__bw16_inline (x))
# define stdc_bit_width_ui(x) (__bw32_inline (x))
# if __WORDSIZE == 64
# define stdc_bit_width_ul(x) (__bw64_inline (x))
# else
# define stdc_bit_width_ul(x) (__bw32_inline (x))
# endif
# define stdc_bit_width_ull(x) (__bw64_inline (x))
#endif
/* Bit floor. */
extern unsigned char stdc_bit_floor_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned short stdc_bit_floor_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_bit_floor_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned long int stdc_bit_floor_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned long long int stdc_bit_floor_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_bit_floor)
# define stdc_bit_floor(x) (__builtin_stdc_bit_floor (x))
#else
# define stdc_bit_floor(x) ((__typeof (x)) stdc_bit_floor_ull (x))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_clzll)
static __always_inline uint64_t
__bf64_inline (uint64_t __x)
{
return __x == 0 ? 0 : ((uint64_t) 1) << (__bw64_inline (__x) - 1);
}
static __always_inline uint32_t
__bf32_inline (uint32_t __x)
{
return __x == 0 ? 0 : ((uint32_t) 1) << (__bw32_inline (__x) - 1);
}
static __always_inline uint16_t
__bf16_inline (uint16_t __x)
{
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
return __pacify_uint16 (__x == 0
? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1));
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
}
static __always_inline uint8_t
__bf8_inline (uint8_t __x)
{
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
return __pacify_uint8 (__x == 0
? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1));
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
}
# define stdc_bit_floor_uc(x) ((unsigned char) __bf8_inline (x))
# define stdc_bit_floor_us(x) ((unsigned short) __bf16_inline (x))
# define stdc_bit_floor_ui(x) ((unsigned int) __bf32_inline (x))
# if __WORDSIZE == 64
# define stdc_bit_floor_ul(x) ((unsigned long int) __bf64_inline (x))
# else
# define stdc_bit_floor_ul(x) ((unsigned long int) __bf32_inline (x))
# endif
# define stdc_bit_floor_ull(x) ((unsigned long long int) __bf64_inline (x))
#endif
/* Bit ceiling. */
extern unsigned char stdc_bit_ceil_uc (unsigned char __x)
__THROW __attribute_const__;
extern unsigned short stdc_bit_ceil_us (unsigned short __x)
__THROW __attribute_const__;
extern unsigned int stdc_bit_ceil_ui (unsigned int __x)
__THROW __attribute_const__;
extern unsigned long int stdc_bit_ceil_ul (unsigned long int __x)
__THROW __attribute_const__;
__extension__
extern unsigned long long int stdc_bit_ceil_ull (unsigned long long int __x)
__THROW __attribute_const__;
#if __glibc_has_builtin (__builtin_stdc_bit_ceil)
# define stdc_bit_ceil(x) (__builtin_stdc_bit_ceil (x))
#else
# define stdc_bit_ceil(x) ((__typeof (x)) stdc_bit_ceil_ull (x))
#endif
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
#if __GNUC_PREREQ (3, 4) || __glibc_has_builtin (__builtin_clzll)
static __always_inline uint64_t
__bc64_inline (uint64_t __x)
{
return __x <= 1 ? 1 : ((uint64_t) 2) << (__bw64_inline (__x - 1) - 1);
}
static __always_inline uint32_t
__bc32_inline (uint32_t __x)
{
return __x <= 1 ? 1 : ((uint32_t) 2) << (__bw32_inline (__x - 1) - 1);
}
static __always_inline uint16_t
__bc16_inline (uint16_t __x)
{
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
return __pacify_uint16 (__x <= 1
? 1
: ((uint16_t) 2)
<< (__bw16_inline ((uint16_t) (__x - 1)) - 1));
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
}
static __always_inline uint8_t
__bc8_inline (uint8_t __x)
{
stdlib: Fix stdbit.h with -Wconversion for older gcc With gcc 6.5.0, 7.5.0, 8.5.0, and 9.5.0 the tst-stdbit-Wconversion issues the warnings: ../stdlib/stdbit.h: In function ‘__clo16_inline’: ../stdlib/stdbit.h:128:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __clz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__clo8_inline’: ../stdlib/stdbit.h:134:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __clz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto16_inline’: ../stdlib/stdbit.h:232:26: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz16_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__cto8_inline’: ../stdlib/stdbit.h:238:25: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __ctz8_inline (~__x); ^ ../stdlib/stdbit.h: In function ‘__bf16_inline’: ../stdlib/stdbit.h:701:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint16_t) 1) << (__bw16_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bf8_inline’: ../stdlib/stdbit.h:707:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x == 0 ? 0 : ((uint8_t) 1) << (__bw8_inline (__x) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc16_inline’: ../stdlib/stdbit.h:751:59: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:751:23: error: conversion to ‘uint16_t {aka short unsigned int}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint16_t) 2) << (__bw16_inline (__x - 1) - 1); ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ../stdlib/stdbit.h: In function ‘__bc8_inline’: ../stdlib/stdbit.h:757:57: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); ^~~ ../stdlib/stdbit.h:757:23: error: conversion to ‘uint8_t {aka unsigned char}’ from ‘int’ may alter its value [-Werror=conversion] return __x <= 1 ? 1 : ((uint8_t) 2) << (__bw8_inline (__x - 1) - 1); It seems to boiler down to __builtin_clz not having a variant for 8 or 16 bits. Fix it by explicit casting to the expected types. Checked on x86_64-linux-gnu and i686-linux-gnu with gcc 9.5.0.
2024-01-04 17:44:42 +00:00
return __pacify_uint8 (__x <= 1
? 1
: ((uint8_t) 2)
<< (__bw8_inline ((uint8_t) (__x - 1)) - 1));
Implement C23 <stdbit.h> C23 adds a header <stdbit.h> with various functions and type-generic macros for bit-manipulation of unsigned integers (plus macro defines related to endianness). Implement this header for glibc. The functions have both inline definitions in the header (referenced by macros defined in the header) and copies with external linkage in the library (which are implemented in terms of those macros to avoid duplication). They are documented in the glibc manual. Tests, as well as verifying results for various inputs (of both the macros and the out-of-line functions), verify the types of those results (which showed up a bug in an earlier version with the type-generic macro stdc_has_single_bit wrongly returning a promoted type), that the macros can be used at top level in a source file (so don't use ({})), that they evaluate their arguments exactly once, and that the macros for the type-specific functions have the expected implicit conversions to the relevant argument type. Jakub previously referred to -Wconversion warnings in type-generic macros, so I've included a test with -Wconversion (but the only warnings I saw and fixed from that test were actually in inline functions in the <stdbit.h> header - not anything coming from use of the type-generic macros themselves). This implementation of the type-generic macros does not handle unsigned __int128, or unsigned _BitInt types with a width other than that of a standard integer type (and C23 doesn't require the header to handle such types either). Support for those types, using the new type-generic built-in functions Jakub's added for GCC 14, can reasonably be added in a followup (along of course with associated tests). This implementation doesn't do anything special to handle C++, or have any tests of functionality in C++ beyond the existing tests that all headers can be compiled in C++ code; it's not clear exactly what form this header should take in C++, but probably not one using macros. DIS ballot comment AT-107 asks for the word "count" to be added to the names of the stdc_leading_zeros, stdc_leading_ones, stdc_trailing_zeros and stdc_trailing_ones functions and macros. I don't think it's likely to be accepted (accepting any technical comments would mean having an FDIS ballot), but if it is accepted at the WG14 meeting (22-26 January in Strasbourg, starting with DIS ballot comment handling) then there would still be time to update glibc for the renaming before the 2.39 release. The new functions and header are placed in the stdlib/ directory in glibc, rather than creating a new toplevel stdbit/ or putting them in string/ alongside ffs. Tested for x86_64 and x86.
2024-01-03 12:07:14 +00:00
}
# define stdc_bit_ceil_uc(x) ((unsigned char) __bc8_inline (x))
# define stdc_bit_ceil_us(x) ((unsigned short) __bc16_inline (x))
# define stdc_bit_ceil_ui(x) ((unsigned int) __bc32_inline (x))
# if __WORDSIZE == 64
# define stdc_bit_ceil_ul(x) ((unsigned long int) __bc64_inline (x))
# else
# define stdc_bit_ceil_ul(x) ((unsigned long int) __bc32_inline (x))
# endif
# define stdc_bit_ceil_ull(x) ((unsigned long long int) __bc64_inline (x))
#endif
__END_DECLS
#endif /* _STDBIT_H */