diff --git a/libcxx/docs/ReleaseNotes/21.rst b/libcxx/docs/ReleaseNotes/21.rst index 91123ffa3e34b..6e66b1177fbfb 100644 --- a/libcxx/docs/ReleaseNotes/21.rst +++ b/libcxx/docs/ReleaseNotes/21.rst @@ -72,6 +72,8 @@ Improvements and New Features - The ``num_put::do_put`` integral overloads have been optimized, resulting in a performance improvement of up to 2.4x. +- The ``num_get::do_get`` integral overloads have been optimized, resulting in a performance improvement of up to 2.8x. + - The ``std::stable_sort`` algorithm uses radix sort for floating-point types now, which can improve the performance up to 10x, depending on type of sorted elements and the initial state of the sorted array. diff --git a/libcxx/include/CMakeLists.txt b/libcxx/include/CMakeLists.txt index c6b87a34a43e9..264602934dbc2 100644 --- a/libcxx/include/CMakeLists.txt +++ b/libcxx/include/CMakeLists.txt @@ -516,7 +516,6 @@ set(files __locale_dir/locale_base_api.h __locale_dir/locale_base_api/bsd_locale_fallbacks.h __locale_dir/locale_base_api/ibm.h - __locale_dir/locale_base_api/musl.h __locale_dir/locale_base_api/openbsd.h __locale_dir/messages.h __locale_dir/money.h diff --git a/libcxx/include/__algorithm/simd_utils.h b/libcxx/include/__algorithm/simd_utils.h index 07fef20f6166d..ac96f024adb41 100644 --- a/libcxx/include/__algorithm/simd_utils.h +++ b/libcxx/include/__algorithm/simd_utils.h @@ -114,11 +114,37 @@ template }(make_index_sequence<__simd_vector_size_v<_VecT>>{}); } +// Load the first _Np elements, zero the rest +_LIBCPP_DIAGNOSTIC_PUSH +_LIBCPP_CLANG_DIAGNOSTIC_IGNORED("-Wpsabi") +template +[[__nodiscard__]] _LIBCPP_ALWAYS_INLINE _LIBCPP_HIDE_FROM_ABI _VecT __partial_load(_Iter __iter) noexcept { + return [=]( + index_sequence<_LoadIndices...>, index_sequence<_ZeroIndices...>) _LIBCPP_ALWAYS_INLINE noexcept { + return _VecT{__iter[_LoadIndices]..., ((void)_ZeroIndices, 0)...}; + }(make_index_sequence<_Np>{}, make_index_sequence<__simd_vector_size_v<_VecT> - _Np>{}); +} + +// Create a vector where every elements is __val +template +[[__nodiscard__]] _LIBCPP_ALWAYS_INLINE _LIBCPP_HIDE_FROM_ABI _VecT +__broadcast(__simd_vector_underlying_type_t<_VecT> __val) { + return [&](index_sequence<_Indices...>) { + return _VecT{((void)_Indices, __val)...}; + }(make_index_sequence<__simd_vector_size_v<_VecT>>()); +} +_LIBCPP_DIAGNOSTIC_POP + template [[__nodiscard__]] _LIBCPP_HIDE_FROM_ABI bool __all_of(__simd_vector<_Tp, _Np> __vec) noexcept { return __builtin_reduce_and(__builtin_convertvector(__vec, __simd_vector)); } +template +[[__nodiscard__]] _LIBCPP_HIDE_FROM_ABI bool __none_of(__simd_vector<_Tp, _Np> __vec) noexcept { + return !__builtin_reduce_or(__builtin_convertvector(__vec, __simd_vector)); +} + template [[__nodiscard__]] _LIBCPP_HIDE_FROM_ABI size_t __find_first_set(__simd_vector<_Tp, _Np> __vec) noexcept { using __mask_vec = __simd_vector; diff --git a/libcxx/include/__locale_dir/locale_base_api.h b/libcxx/include/__locale_dir/locale_base_api.h index 9f3ce02a3af20..823cbb667b68c 100644 --- a/libcxx/include/__locale_dir/locale_base_api.h +++ b/libcxx/include/__locale_dir/locale_base_api.h @@ -57,8 +57,6 @@ // float __strtof(const char*, char**, __locale_t); // double __strtod(const char*, char**, __locale_t); // long double __strtold(const char*, char**, __locale_t); -// long long __strtoll(const char*, char**, __locale_t); -// unsigned long long __strtoull(const char*, char**, __locale_t); // } // // Character manipulation functions @@ -106,7 +104,6 @@ // // int __snprintf(char*, size_t, __locale_t, const char*, ...); // required by the headers // int __asprintf(char**, __locale_t, const char*, ...); // required by the headers -// int __sscanf(const char*, __locale_t, const char*, ...); // required by the headers // } #if _LIBCPP_HAS_LOCALIZATION @@ -131,8 +128,6 @@ # include <__locale_dir/locale_base_api/ibm.h> # elif defined(__OpenBSD__) # include <__locale_dir/locale_base_api/openbsd.h> -# elif defined(__wasi__) || _LIBCPP_HAS_MUSL_LIBC -# include <__locale_dir/locale_base_api/musl.h> # endif # include <__locale_dir/locale_base_api/bsd_locale_fallbacks.h> @@ -192,15 +187,6 @@ inline _LIBCPP_HIDE_FROM_ABI long double __strtold(const char* __nptr, char** __ return strtold_l(__nptr, __endptr, __loc); } -inline _LIBCPP_HIDE_FROM_ABI long long __strtoll(const char* __nptr, char** __endptr, int __base, __locale_t __loc) { - return strtoll_l(__nptr, __endptr, __base, __loc); -} - -inline _LIBCPP_HIDE_FROM_ABI unsigned long long -__strtoull(const char* __nptr, char** __endptr, int __base, __locale_t __loc) { - return strtoull_l(__nptr, __endptr, __base, __loc); -} - // // Character manipulation functions // @@ -302,11 +288,6 @@ _LIBCPP_HIDE_FROM_ABI _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT(__printf__, 3, 4) int __ char** __s, __locale_t __loc, const char* __format, _Args&&... __args) { return std::__libcpp_asprintf_l(__s, __loc, __format, std::forward<_Args>(__args)...); } -template -_LIBCPP_HIDE_FROM_ABI _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT(__scanf__, 3, 4) int __sscanf( - const char* __s, __locale_t __loc, const char* __format, _Args&&... __args) { - return std::__libcpp_sscanf_l(__s, __loc, __format, std::forward<_Args>(__args)...); -} _LIBCPP_DIAGNOSTIC_POP # undef _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT diff --git a/libcxx/include/__locale_dir/locale_base_api/bsd_locale_fallbacks.h b/libcxx/include/__locale_dir/locale_base_api/bsd_locale_fallbacks.h index b62a1b737e97f..8cdbe0cd15051 100644 --- a/libcxx/include/__locale_dir/locale_base_api/bsd_locale_fallbacks.h +++ b/libcxx/include/__locale_dir/locale_base_api/bsd_locale_fallbacks.h @@ -125,16 +125,6 @@ inline _LIBCPP_ATTRIBUTE_FORMAT(__printf__, 3, 4) int __libcpp_asprintf_l( return __res; } -inline _LIBCPP_ATTRIBUTE_FORMAT(__scanf__, 3, 4) int __libcpp_sscanf_l( - const char* __s, locale_t __l, const char* __format, ...) { - va_list __va; - va_start(__va, __format); - __locale_guard __current(__l); - int __res = vsscanf(__s, __format, __va); - va_end(__va); - return __res; -} - _LIBCPP_END_NAMESPACE_STD #endif // _LIBCPP___LOCALE_DIR_LOCALE_BASE_API_BSD_LOCALE_FALLBACKS_H diff --git a/libcxx/include/__locale_dir/locale_base_api/ibm.h b/libcxx/include/__locale_dir/locale_base_api/ibm.h index 1d1d15df9f799..47a83eac7df36 100644 --- a/libcxx/include/__locale_dir/locale_base_api/ibm.h +++ b/libcxx/include/__locale_dir/locale_base_api/ibm.h @@ -53,11 +53,6 @@ struct __setAndRestore { // The following are not POSIX routines. These are quick-and-dirty hacks // to make things pretend to work -inline _LIBCPP_HIDE_FROM_ABI long long strtoll_l(const char* __nptr, char** __endptr, int __base, locale_t locale) { - __setAndRestore __newloc(locale); - return ::strtoll(__nptr, __endptr, __base); -} - inline _LIBCPP_HIDE_FROM_ABI double strtod_l(const char* __nptr, char** __endptr, locale_t locale) { __setAndRestore __newloc(locale); return ::strtod(__nptr, __endptr); @@ -73,12 +68,6 @@ inline _LIBCPP_HIDE_FROM_ABI long double strtold_l(const char* __nptr, char** __ return ::strtold(__nptr, __endptr); } -inline _LIBCPP_HIDE_FROM_ABI unsigned long long -strtoull_l(const char* __nptr, char** __endptr, int __base, locale_t locale) { - __setAndRestore __newloc(locale); - return ::strtoull(__nptr, __endptr, __base); -} - inline _LIBCPP_HIDE_FROM_ABI _LIBCPP_ATTRIBUTE_FORMAT(__printf__, 2, 0) int vasprintf(char** strp, const char* fmt, va_list ap) { const size_t buff_size = 256; diff --git a/libcxx/include/__locale_dir/locale_base_api/musl.h b/libcxx/include/__locale_dir/locale_base_api/musl.h deleted file mode 100644 index 1653214cdba1e..0000000000000 --- a/libcxx/include/__locale_dir/locale_base_api/musl.h +++ /dev/null @@ -1,31 +0,0 @@ -// -*- C++ -*- -//===-----------------------------------------------------------------------===// -// -// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. -// See https://llvm.org/LICENSE.txt for license information. -// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception -// -//===----------------------------------------------------------------------===// -// This adds support for the extended locale functions that are currently -// missing from the Musl C library. -// -// This only works when the specified locale is "C" or "POSIX", but that's -// about as good as we can do without implementing full xlocale support -// in Musl. -//===----------------------------------------------------------------------===// - -#ifndef _LIBCPP___LOCALE_DIR_LOCALE_BASE_API_MUSL_H -#define _LIBCPP___LOCALE_DIR_LOCALE_BASE_API_MUSL_H - -#include -#include - -inline _LIBCPP_HIDE_FROM_ABI long long strtoll_l(const char* __nptr, char** __endptr, int __base, locale_t) { - return ::strtoll(__nptr, __endptr, __base); -} - -inline _LIBCPP_HIDE_FROM_ABI unsigned long long strtoull_l(const char* __nptr, char** __endptr, int __base, locale_t) { - return ::strtoull(__nptr, __endptr, __base); -} - -#endif // _LIBCPP___LOCALE_DIR_LOCALE_BASE_API_MUSL_H diff --git a/libcxx/include/__locale_dir/num.h b/libcxx/include/__locale_dir/num.h index 7ca8ffe348959..6ae2f64175966 100644 --- a/libcxx/include/__locale_dir/num.h +++ b/libcxx/include/__locale_dir/num.h @@ -11,6 +11,7 @@ #include <__algorithm/find.h> #include <__algorithm/reverse.h> +#include <__algorithm/simd_utils.h> #include <__charconv/to_chars_integral.h> #include <__charconv/traits.h> #include <__config> @@ -46,9 +47,9 @@ struct _LIBCPP_EXPORTED_FROM_ABI __num_get_base { static int __get_base(ios_base&); static const char __src[33]; // "0123456789abcdefABCDEFxX+-pPiInN" // count of leading characters in __src used for parsing integers ("012..X+-") - static const size_t __int_chr_cnt = 26; + static inline const size_t __int_chr_cnt = 26; // count of leading characters in __src used for parsing floating-point values ("012..-pP") - static const size_t __fp_chr_cnt = 28; + static inline const size_t __fp_chr_cnt = 28; }; template @@ -71,7 +72,8 @@ struct __num_get : protected __num_get_base { [[__deprecated__("This exists only for ABI compatibility")]] static string __stage2_int_prep(ios_base& __iob, _CharT* __atoms, _CharT& __thousands_sep); - static int __stage2_int_loop( + + [[__deprecated__("This exists only for ABI compatibility")]] static int __stage2_int_loop( _CharT __ct, int __base, char* __a, @@ -83,11 +85,22 @@ struct __num_get : protected __num_get_base { unsigned*& __g_end, _CharT* __atoms); - _LIBCPP_HIDE_FROM_ABI static string __stage2_int_prep(ios_base& __iob, _CharT& __thousands_sep) { - locale __loc = __iob.getloc(); - const numpunct<_CharT>& __np = use_facet >(__loc); - __thousands_sep = __np.thousands_sep(); - return __np.grouping(); + _LIBCPP_HIDE_FROM_ABI static ptrdiff_t __atoms_offset(const _CharT* __atoms, _CharT __val) { +# if _LIBCPP_HAS_ALGORITHM_VECTOR_UTILS + if constexpr (is_same<_CharT, char>::value) { + _LIBCPP_DIAGNOSTIC_PUSH + _LIBCPP_CLANG_DIAGNOSTIC_IGNORED("-Wpsabi") // This should be removed once https://llvm.org/PR128361 is fixed + using __vec = __simd_vector; + __vec __chars = std::__broadcast<__vec>(__val); + __vec __cmp = std::__partial_load<__vec, __int_chr_cnt>(__atoms); + auto __res = __chars == __cmp; + if (std::__none_of(__res)) + return __int_chr_cnt; + return std::min(__int_chr_cnt, std::__find_first_set(__res)); + _LIBCPP_DIAGNOSTIC_POP + } +# endif + return std::find(__atoms, __atoms + __int_chr_cnt, __val) - __atoms; } _LIBCPP_HIDE_FROM_ABI const _CharT* __do_widen(ios_base& __iob, _CharT* __atoms) const { @@ -120,54 +133,6 @@ string __num_get<_CharT>::__stage2_float_prep( return __np.grouping(); } -template -int __num_get<_CharT>::__stage2_int_loop( - _CharT __ct, - int __base, - char* __a, - char*& __a_end, - unsigned& __dc, - _CharT __thousands_sep, - const string& __grouping, - unsigned* __g, - unsigned*& __g_end, - _CharT* __atoms) { - if (__a_end == __a && (__ct == __atoms[24] || __ct == __atoms[25])) { - *__a_end++ = __ct == __atoms[24] ? '+' : '-'; - __dc = 0; - return 0; - } - if (__grouping.size() != 0 && __ct == __thousands_sep) { - if (__g_end - __g < __num_get_buf_sz) { - *__g_end++ = __dc; - __dc = 0; - } - return 0; - } - ptrdiff_t __f = std::find(__atoms, __atoms + __int_chr_cnt, __ct) - __atoms; - if (__f >= 24) - return -1; - switch (__base) { - case 8: - case 10: - if (__f >= __base) - return -1; - break; - case 16: - if (__f < 22) - break; - if (__a_end != __a && __a_end - __a <= 2 && __a_end[-1] == '0') { - __dc = 0; - *__a_end++ = __src[__f]; - return 0; - } - return -1; - } - *__a_end++ = __src[__f]; - ++__dc; - return 0; -} - template int __num_get<_CharT>::__stage2_float_loop( _CharT __ct, @@ -229,9 +194,9 @@ int __num_get<_CharT>::__stage2_float_loop( } extern template struct _LIBCPP_EXTERN_TEMPLATE_TYPE_VIS __num_get; -# if _LIBCPP_HAS_WIDE_CHARACTERS +# if _LIBCPP_HAS_WIDE_CHARACTERS extern template struct _LIBCPP_EXTERN_TEMPLATE_TYPE_VIS __num_get; -# endif +# endif template _LIBCPP_HIDE_FROM_ABI _Tp __do_strtod(const char* __a, char** __p2); @@ -272,65 +237,6 @@ _LIBCPP_HIDE_FROM_ABI _Tp __num_get_float(const char* __a, const char* __a_end, return 0; } -template -_LIBCPP_HIDE_FROM_ABI _Tp -__num_get_signed_integral(const char* __a, const char* __a_end, ios_base::iostate& __err, int __base) { - if (__a != __a_end) { - __libcpp_remove_reference_t __save_errno = errno; - errno = 0; - char* __p2; - long long __ll = __locale::__strtoll(__a, &__p2, __base, _LIBCPP_GET_C_LOCALE); - __libcpp_remove_reference_t __current_errno = errno; - if (__current_errno == 0) - errno = __save_errno; - if (__p2 != __a_end) { - __err = ios_base::failbit; - return 0; - } else if (__current_errno == ERANGE || __ll < numeric_limits<_Tp>::min() || numeric_limits<_Tp>::max() < __ll) { - __err = ios_base::failbit; - if (__ll > 0) - return numeric_limits<_Tp>::max(); - else - return numeric_limits<_Tp>::min(); - } - return static_cast<_Tp>(__ll); - } - __err = ios_base::failbit; - return 0; -} - -template -_LIBCPP_HIDE_FROM_ABI _Tp -__num_get_unsigned_integral(const char* __a, const char* __a_end, ios_base::iostate& __err, int __base) { - if (__a != __a_end) { - const bool __negate = *__a == '-'; - if (__negate && ++__a == __a_end) { - __err = ios_base::failbit; - return 0; - } - __libcpp_remove_reference_t __save_errno = errno; - errno = 0; - char* __p2; - unsigned long long __ll = __locale::__strtoull(__a, &__p2, __base, _LIBCPP_GET_C_LOCALE); - __libcpp_remove_reference_t __current_errno = errno; - if (__current_errno == 0) - errno = __save_errno; - if (__p2 != __a_end) { - __err = ios_base::failbit; - return 0; - } else if (__current_errno == ERANGE || numeric_limits<_Tp>::max() < __ll) { - __err = ios_base::failbit; - return numeric_limits<_Tp>::max(); - } - _Tp __res = static_cast<_Tp>(__ll); - if (__negate) - __res = -__res; - return __res; - } - __err = ios_base::failbit; - return 0; -} - template > class num_get : public locale::facet, private __num_get<_CharT> { public: @@ -468,137 +374,188 @@ class num_get : public locale::facet, private __num_get<_CharT> { return __b; } - template - _LIBCPP_HIDE_FROM_ABI iter_type - __do_get_signed(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, _Signed& __v) const { + template + iter_type __do_get_integral( + iter_type __first, iter_type __last, ios_base& __iob, ios_base::iostate& __err, _MaybeSigned& __v) const { + using _Unsigned = __make_unsigned_t<_MaybeSigned>; + // Stage 1 int __base = this->__get_base(__iob); - // Stage 2 - char_type __thousands_sep; - const int __atoms_size = __num_get_base::__int_chr_cnt; - char_type __atoms1[__atoms_size]; - const char_type* __atoms = this->__do_widen(__iob, __atoms1); - string __grouping = this->__stage2_int_prep(__iob, __thousands_sep); - string __buf; - __buf.resize(__buf.capacity()); - char* __a = &__buf[0]; - char* __a_end = __a; + + // Stages 2 & 3 + // These are combined into a single step where we parse the characters and calculate the value in one go instead of + // storing the relevant characters first (in an allocated buffer) and parse the characters after we extracted them. + // This makes the whole process significantly faster, since we avoid potential allocations and copies. + + const auto& __numpunct = use_facet >(__iob.getloc()); + char_type __thousands_sep = __numpunct.thousands_sep(); + string __grouping = __numpunct.grouping(); + + char_type __atoms_buffer[__num_get_base::__int_chr_cnt]; + const char_type* __atoms = this->__do_widen(__iob, __atoms_buffer); unsigned __g[__num_get_base::__num_get_buf_sz]; unsigned* __g_end = __g; unsigned __dc = 0; - for (; __b != __e; ++__b) { - if (__a_end == __a + __buf.size()) { - size_t __tmp = __buf.size(); - __buf.resize(2 * __buf.size()); - __buf.resize(__buf.capacity()); - __a = &__buf[0]; - __a_end = __a + __tmp; + + if (__first == __last) { + __err |= ios_base::eofbit; + return __first; + } + + while (!__grouping.empty() && *__first == __thousands_sep) { + ++__first; + if (__g_end - __g < this->__num_get_buf_sz) { + *__g_end++ = __dc; + __dc = 0; } - if (this->__stage2_int_loop( - *__b, - __base, - __a, - __a_end, - __dc, - __thousands_sep, - __grouping, - __g, - __g_end, - const_cast(__atoms))) - break; } - if (__grouping.size() != 0 && __g_end - __g < __num_get_base::__num_get_buf_sz) - *__g_end++ = __dc; - // Stage 3 - __v = std::__num_get_signed_integral<_Signed>(__a, __a_end, __err, __base); - // Digit grouping checked - __check_grouping(__grouping, __g, __g_end, __err); - // EOF checked - if (__b == __e) + + bool __negate = false; + // __c == '+' || __c == '-' + if (auto __c = *__first; __c == __atoms[24] || __c == __atoms[25]) { + __negate = __c == __atoms[25]; + ++__first; + } + + if (__first == __last) { __err |= ios_base::eofbit; - return __b; - } + return __first; + } - template - _LIBCPP_HIDE_FROM_ABI iter_type - __do_get_unsigned(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, _Unsigned& __v) const { - // Stage 1 - int __base = this->__get_base(__iob); - // Stage 2 - char_type __thousands_sep; - const int __atoms_size = __num_get_base::__int_chr_cnt; - char_type __atoms1[__atoms_size]; - const char_type* __atoms = this->__do_widen(__iob, __atoms1); - string __grouping = this->__stage2_int_prep(__iob, __thousands_sep); - string __buf; - __buf.resize(__buf.capacity()); - char* __a = &__buf[0]; - char* __a_end = __a; - unsigned __g[__num_get_base::__num_get_buf_sz]; - unsigned* __g_end = __g; - unsigned __dc = 0; - for (; __b != __e; ++__b) { - if (__a_end == __a + __buf.size()) { - size_t __tmp = __buf.size(); - __buf.resize(2 * __buf.size()); - __buf.resize(__buf.capacity()); - __a = &__buf[0]; - __a_end = __a + __tmp; + bool __parsed_num = false; + + // If we don't have a pre-set base, figure it out + if (__base == 0) { + auto __c = *__first; + // __c == '0' + if (__c == __atoms[0]) { + ++__first; + if (__first == __last) { + __err |= ios_base::eofbit; + return __first; + } + // __c2 == 'x' || __c2 == 'X' + if (auto __c2 = *__first; __c2 == __atoms[22] || __c2 == __atoms[23]) { + __base = 16; + ++__first; + } else { + __base = 8; + } + } else { + __base = 10; } - if (this->__stage2_int_loop( - *__b, - __base, - __a, - __a_end, - __dc, - __thousands_sep, - __grouping, - __g, - __g_end, - const_cast(__atoms))) + } else if (__base == 16) { + // Try to swallow '0x' + + // *__first == '0' + if (*__first == __atoms[0]) { + ++__first; + if (__first == __last) { + __err |= ios_base::eofbit; + return __first; + } + // __c == 'x' || __c == 'X' + if (auto __c = *__first; __c == __atoms[22] || __c == __atoms[23]) + ++__first; + else + __parsed_num = true; // We only swallowed '0', so we've started to parse a number + } + } + + // Calculate the actual number + _Unsigned __val = 0; + bool __overflowed = false; + for (; __first != __last; ++__first) { + auto __c = *__first; + if (!__grouping.empty() && __c == __thousands_sep) { + if (__g_end - __g < this->__num_get_buf_sz) { + *__g_end++ = __dc; + __dc = 0; + } + continue; + } + auto __offset = this->__atoms_offset(__atoms, __c); + if (__offset >= 22) break; + + if (__base == 16 && __offset >= 16) + __offset -= 6; + if (__offset >= __base) + break; + __overflowed |= __builtin_mul_overflow(__val, __base, std::addressof(__val)) || + __builtin_add_overflow(__val, __offset, std::addressof(__val)); + __parsed_num = true; + ++__dc; } + + if (!__parsed_num) { + __err |= ios_base::failbit; + __v = 0; + } else if (__overflowed) { + __err |= ios_base::failbit; + __v = is_signed<_MaybeSigned>::value && __negate + ? numeric_limits<_MaybeSigned>::min() + : numeric_limits<_MaybeSigned>::max(); + } else if (!__negate) { + if (__val > static_cast<_Unsigned>(numeric_limits<_MaybeSigned>::max())) { + __err |= ios_base::failbit; + __v = numeric_limits<_MaybeSigned>::max(); + } else { + __v = __val; + } + } else if (is_signed<_MaybeSigned>::value) { + if (__val > static_cast<_Unsigned>(numeric_limits<_MaybeSigned>::max()) + 1) { + __err |= ios_base::failbit; + __v = numeric_limits<_MaybeSigned>::min(); + } else if (__val == static_cast<_Unsigned>(numeric_limits<_MaybeSigned>::max()) + 1) { + __v = numeric_limits<_MaybeSigned>::min(); + } else { + __v = -__val; + } + } else { + __v = -__val; + } + if (__grouping.size() != 0 && __g_end - __g < __num_get_base::__num_get_buf_sz) *__g_end++ = __dc; - // Stage 3 - __v = std::__num_get_unsigned_integral<_Unsigned>(__a, __a_end, __err, __base); + // Digit grouping checked __check_grouping(__grouping, __g, __g_end, __err); // EOF checked - if (__b == __e) + if (__first == __last) __err |= ios_base::eofbit; - return __b; + return __first; } virtual iter_type do_get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, bool& __v) const; virtual iter_type do_get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, long& __v) const { - return this->__do_get_signed(__b, __e, __iob, __err, __v); + return this->__do_get_integral(__b, __e, __iob, __err, __v); } virtual iter_type do_get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, long long& __v) const { - return this->__do_get_signed(__b, __e, __iob, __err, __v); + return this->__do_get_integral(__b, __e, __iob, __err, __v); } virtual iter_type do_get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, unsigned short& __v) const { - return this->__do_get_unsigned(__b, __e, __iob, __err, __v); + return this->__do_get_integral(__b, __e, __iob, __err, __v); } virtual iter_type do_get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, unsigned int& __v) const { - return this->__do_get_unsigned(__b, __e, __iob, __err, __v); + return this->__do_get_integral(__b, __e, __iob, __err, __v); } virtual iter_type do_get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, unsigned long& __v) const { - return this->__do_get_unsigned(__b, __e, __iob, __err, __v); + return this->__do_get_integral(__b, __e, __iob, __err, __v); } virtual iter_type do_get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, unsigned long long& __v) const { - return this->__do_get_unsigned(__b, __e, __iob, __err, __v); + return this->__do_get_integral(__b, __e, __iob, __err, __v); } virtual iter_type do_get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, float& __v) const { @@ -652,46 +609,19 @@ _InputIterator num_get<_CharT, _InputIterator>::do_get( template _InputIterator num_get<_CharT, _InputIterator>::do_get( iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, void*& __v) const { - // Stage 1 - int __base = 16; - // Stage 2 - char_type __atoms[__num_get_base::__int_chr_cnt]; - char_type __thousands_sep = char_type(); - string __grouping; - std::use_facet >(__iob.getloc()) - .widen(__num_get_base::__src, __num_get_base::__src + __num_get_base::__int_chr_cnt, __atoms); - string __buf; - __buf.resize(__buf.capacity()); - char* __a = &__buf[0]; - char* __a_end = __a; - unsigned __g[__num_get_base::__num_get_buf_sz]; - unsigned* __g_end = __g; - unsigned __dc = 0; - for (; __b != __e; ++__b) { - if (__a_end == __a + __buf.size()) { - size_t __tmp = __buf.size(); - __buf.resize(2 * __buf.size()); - __buf.resize(__buf.capacity()); - __a = &__buf[0]; - __a_end = __a + __tmp; - } - if (this->__stage2_int_loop(*__b, __base, __a, __a_end, __dc, __thousands_sep, __grouping, __g, __g_end, __atoms)) - break; - } - // Stage 3 - __buf.resize(__a_end - __a); - if (__locale::__sscanf(__buf.c_str(), _LIBCPP_GET_C_LOCALE, "%p", &__v) != 1) - __err = ios_base::failbit; - // EOF checked - if (__b == __e) - __err |= ios_base::eofbit; - return __b; + auto __flags = __iob.flags(); + __iob.flags((__flags & ~ios_base::basefield & ~ios_base::uppercase) | ios_base::hex); + uintptr_t __ptr; + auto __res = __do_get_integral(__b, __e, __iob, __err, __ptr); + __iob.flags(__flags); + __v = reinterpret_cast(__ptr); + return __res; } extern template class _LIBCPP_EXTERN_TEMPLATE_TYPE_VIS num_get; -# if _LIBCPP_HAS_WIDE_CHARACTERS +# if _LIBCPP_HAS_WIDE_CHARACTERS extern template class _LIBCPP_EXTERN_TEMPLATE_TYPE_VIS num_get; -# endif +# endif struct _LIBCPP_EXPORTED_FROM_ABI __num_put_base { protected: diff --git a/libcxx/include/__locale_dir/support/bsd_like.h b/libcxx/include/__locale_dir/support/bsd_like.h index ac402924709e5..5e4593e87f951 100644 --- a/libcxx/include/__locale_dir/support/bsd_like.h +++ b/libcxx/include/__locale_dir/support/bsd_like.h @@ -77,15 +77,6 @@ inline _LIBCPP_HIDE_FROM_ABI long double __strtold(const char* __nptr, char** __ return ::strtold_l(__nptr, __endptr, __loc); } -inline _LIBCPP_HIDE_FROM_ABI long long __strtoll(const char* __nptr, char** __endptr, int __base, __locale_t __loc) { - return ::strtoll_l(__nptr, __endptr, __base, __loc); -} - -inline _LIBCPP_HIDE_FROM_ABI unsigned long long -__strtoull(const char* __nptr, char** __endptr, int __base, __locale_t __loc) { - return ::strtoull_l(__nptr, __endptr, __base, __loc); -} - // // Character manipulation functions // @@ -213,12 +204,6 @@ _LIBCPP_HIDE_FROM_ABI _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT(__printf__, 3, 4) int __ char** __s, __locale_t __loc, const char* __format, _Args&&... __args) { return ::asprintf_l(__s, __loc, __format, std::forward<_Args>(__args)...); // non-standard } - -template -_LIBCPP_HIDE_FROM_ABI _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT(__scanf__, 3, 4) int __sscanf( - const char* __s, __locale_t __loc, const char* __format, _Args&&... __args) { - return ::sscanf_l(__s, __loc, __format, std::forward<_Args>(__args)...); -} _LIBCPP_DIAGNOSTIC_POP #undef _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT diff --git a/libcxx/include/__locale_dir/support/fuchsia.h b/libcxx/include/__locale_dir/support/fuchsia.h index 4b9e63facb19e..528bfeb0cb6e1 100644 --- a/libcxx/include/__locale_dir/support/fuchsia.h +++ b/libcxx/include/__locale_dir/support/fuchsia.h @@ -141,13 +141,6 @@ _LIBCPP_HIDE_FROM_ABI _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT(__printf__, 3, 4) int __ __locale_guard __current(__loc); return ::asprintf(__s, __format, std::forward<_Args>(__args)...); // non-standard } -template -_LIBCPP_HIDE_FROM_ABI _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT(__scanf__, 3, 4) int __sscanf( - const char* __s, __locale_t __loc, const char* __format, _Args&&... __args) { - __locale_guard __current(__loc); - return std::sscanf(__s, __format, std::forward<_Args>(__args)...); -} - _LIBCPP_DIAGNOSTIC_POP #undef _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT diff --git a/libcxx/include/__locale_dir/support/linux.h b/libcxx/include/__locale_dir/support/linux.h index 23bcf44c31dbf..f0fece2e451d4 100644 --- a/libcxx/include/__locale_dir/support/linux.h +++ b/libcxx/include/__locale_dir/support/linux.h @@ -94,25 +94,6 @@ inline _LIBCPP_HIDE_FROM_ABI long double __strtold(const char* __nptr, char** __ return ::strtold_l(__nptr, __endptr, __loc); } -inline _LIBCPP_HIDE_FROM_ABI long long __strtoll(const char* __nptr, char** __endptr, int __base, __locale_t __loc) { -#if !_LIBCPP_HAS_MUSL_LIBC - return ::strtoll_l(__nptr, __endptr, __base, __loc); -#else - (void)__loc; - return ::strtoll(__nptr, __endptr, __base); -#endif -} - -inline _LIBCPP_HIDE_FROM_ABI unsigned long long -__strtoull(const char* __nptr, char** __endptr, int __base, __locale_t __loc) { -#if !_LIBCPP_HAS_MUSL_LIBC - return ::strtoull_l(__nptr, __endptr, __base, __loc); -#else - (void)__loc; - return ::strtoull(__nptr, __endptr, __base); -#endif -} - // // Character manipulation functions // @@ -261,20 +242,6 @@ inline _LIBCPP_ATTRIBUTE_FORMAT(__printf__, 3, 4) int __asprintf( va_end(__va); return __res; } - -#ifndef _LIBCPP_COMPILER_GCC // GCC complains that this can't be always_inline due to C-style varargs -_LIBCPP_HIDE_FROM_ABI -#endif -inline _LIBCPP_ATTRIBUTE_FORMAT(__scanf__, 3, 4) int __sscanf( - const char* __s, __locale_t __loc, const char* __format, ...) { - va_list __va; - va_start(__va, __format); - __locale_guard __current(__loc); - int __res = std::vsscanf(__s, __format, __va); - va_end(__va); - return __res; -} - } // namespace __locale _LIBCPP_END_NAMESPACE_STD diff --git a/libcxx/include/__locale_dir/support/no_locale/strtonum.h b/libcxx/include/__locale_dir/support/no_locale/strtonum.h index 0e7a32993e736..59544e10e4a4c 100644 --- a/libcxx/include/__locale_dir/support/no_locale/strtonum.h +++ b/libcxx/include/__locale_dir/support/no_locale/strtonum.h @@ -34,15 +34,6 @@ inline _LIBCPP_HIDE_FROM_ABI long double __strtold(const char* __nptr, char** __ return std::strtold(__nptr, __endptr); } -inline _LIBCPP_HIDE_FROM_ABI long long __strtoll(const char* __nptr, char** __endptr, int __base, __locale_t) { - return std::strtoll(__nptr, __endptr, __base); -} - -inline _LIBCPP_HIDE_FROM_ABI unsigned long long -__strtoull(const char* __nptr, char** __endptr, int __base, __locale_t) { - return std::strtoull(__nptr, __endptr, __base); -} - } // namespace __locale _LIBCPP_END_NAMESPACE_STD diff --git a/libcxx/include/__locale_dir/support/windows.h b/libcxx/include/__locale_dir/support/windows.h index 0df8709f118d0..81e78b09b1446 100644 --- a/libcxx/include/__locale_dir/support/windows.h +++ b/libcxx/include/__locale_dir/support/windows.h @@ -186,14 +186,6 @@ inline _LIBCPP_HIDE_FROM_ABI double __strtod(const char* __nptr, char** __endptr return ::_strtod_l(__nptr, __endptr, __loc); } -inline _LIBCPP_HIDE_FROM_ABI long long __strtoll(const char* __nptr, char** __endptr, int __base, __locale_t __loc) { - return ::_strtoi64_l(__nptr, __endptr, __base, __loc); -} -inline _LIBCPP_HIDE_FROM_ABI unsigned long long -__strtoull(const char* __nptr, char** __endptr, int __base, __locale_t __loc) { - return ::_strtoui64_l(__nptr, __endptr, __base, __loc); -} - // // Character manipulation functions // @@ -280,23 +272,6 @@ _LIBCPP_EXPORTED_FROM_ABI _LIBCPP_ATTRIBUTE_FORMAT(__printf__, 4, 5) int __snpri _LIBCPP_EXPORTED_FROM_ABI _LIBCPP_ATTRIBUTE_FORMAT(__printf__, 3, 4) int __asprintf(char** __ret, __locale_t __loc, const char* __format, ...); -_LIBCPP_DIAGNOSTIC_PUSH -_LIBCPP_CLANG_DIAGNOSTIC_IGNORED("-Wgcc-compat") -_LIBCPP_GCC_DIAGNOSTIC_IGNORED("-Wformat-nonliteral") // GCC doesn't support [[gnu::format]] on variadic templates -#ifdef _LIBCPP_COMPILER_CLANG_BASED -# define _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT(...) _LIBCPP_ATTRIBUTE_FORMAT(__VA_ARGS__) -#else -# define _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT(...) /* nothing */ -#endif - -template -_LIBCPP_HIDE_FROM_ABI _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT(__scanf__, 3, 4) int __sscanf( - const char* __dest, __locale_t __loc, const char* __format, _Args&&... __args) { - return ::_sscanf_l(__dest, __format, __loc, std::forward<_Args>(__args)...); -} -_LIBCPP_DIAGNOSTIC_POP -#undef _LIBCPP_VARIADIC_ATTRIBUTE_FORMAT - #if defined(_LIBCPP_BUILDING_LIBRARY) struct __locale_guard { _LIBCPP_HIDE_FROM_ABI __locale_guard(__locale_t __l) : __status(_configthreadlocale(_ENABLE_PER_THREAD_LOCALE)) { diff --git a/libcxx/include/__support/xlocale/__strtonum_fallback.h b/libcxx/include/__support/xlocale/__strtonum_fallback.h index 5275aead35af9..90bd59d36c248 100644 --- a/libcxx/include/__support/xlocale/__strtonum_fallback.h +++ b/libcxx/include/__support/xlocale/__strtonum_fallback.h @@ -34,12 +34,4 @@ inline _LIBCPP_HIDE_FROM_ABI long double strtold_l(const char* __nptr, char** __ return ::strtold(__nptr, __endptr); } -inline _LIBCPP_HIDE_FROM_ABI long long strtoll_l(const char* __nptr, char** __endptr, int __base, locale_t) { - return ::strtoll(__nptr, __endptr, __base); -} - -inline _LIBCPP_HIDE_FROM_ABI unsigned long long strtoull_l(const char* __nptr, char** __endptr, int __base, locale_t) { - return ::strtoull(__nptr, __endptr, __base); -} - #endif // _LIBCPP___SUPPORT_XLOCALE_STRTONUM_FALLBACK_H diff --git a/libcxx/include/module.modulemap.in b/libcxx/include/module.modulemap.in index c431c0cb407f3..f577913b201d0 100644 --- a/libcxx/include/module.modulemap.in +++ b/libcxx/include/module.modulemap.in @@ -1594,7 +1594,6 @@ module std [system] { module locale_base_api { textual header "__locale_dir/locale_base_api/bsd_locale_fallbacks.h" textual header "__locale_dir/locale_base_api/ibm.h" - textual header "__locale_dir/locale_base_api/musl.h" textual header "__locale_dir/locale_base_api/openbsd.h" } export * diff --git a/libcxx/src/locale.cpp b/libcxx/src/locale.cpp index da735865c322c..0b78fe08a89de 100644 --- a/libcxx/src/locale.cpp +++ b/libcxx/src/locale.cpp @@ -5571,6 +5571,54 @@ string __num_get<_CharT>::__stage2_int_prep(ios_base& __iob, _CharT* __atoms, _C return __np.grouping(); } +template +int __num_get<_CharT>::__stage2_int_loop( + _CharT __ct, + int __base, + char* __a, + char*& __a_end, + unsigned& __dc, + _CharT __thousands_sep, + const string& __grouping, + unsigned* __g, + unsigned*& __g_end, + _CharT* __atoms) { + if (__a_end == __a && (__ct == __atoms[24] || __ct == __atoms[25])) { + *__a_end++ = __ct == __atoms[24] ? '+' : '-'; + __dc = 0; + return 0; + } + if (__grouping.size() != 0 && __ct == __thousands_sep) { + if (__g_end - __g < __num_get_buf_sz) { + *__g_end++ = __dc; + __dc = 0; + } + return 0; + } + ptrdiff_t __f = __atoms_offset(__atoms, __ct); + if (__f >= 24) + return -1; + switch (__base) { + case 8: + case 10: + if (__f >= __base) + return -1; + break; + case 16: + if (__f < 22) + break; + if (__a_end != __a && __a_end - __a <= 2 && __a_end[-1] == '0') { + __dc = 0; + *__a_end++ = __src[__f]; + return 0; + } + return -1; + } + *__a_end++ = __src[__f]; + ++__dc; + return 0; +} + template class _LIBCPP_CLASS_TEMPLATE_INSTANTIATION_VIS collate; _LIBCPP_IF_WIDE_CHARACTERS(template class _LIBCPP_CLASS_TEMPLATE_INSTANTIATION_VIS collate;) diff --git a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_long.pass.cpp b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_long.pass.cpp index 03b74ebf53936..6bf38bd8c72a1 100644 --- a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_long.pass.cpp +++ b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_long.pass.cpp @@ -102,6 +102,18 @@ int main(int, char**) assert(err == ios.goodbit); assert(v == 291); } + { + const char str[] = "a123"; + std::dec(ios); + std::ios_base::iostate err = ios.goodbit; + cpp17_input_iterator iter = + f.get(cpp17_input_iterator(str), + cpp17_input_iterator(str+sizeof(str)), + ios, err, v); + assert(base(iter) == str); + assert(err == ios.failbit); + assert(v == 0); + } { const char str[] = "0x123"; std::hex(ios); diff --git a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_int.pass.cpp b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_int.pass.cpp index bee1be08baafc..f9cef08e247d0 100644 --- a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_int.pass.cpp +++ b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_int.pass.cpp @@ -68,6 +68,17 @@ int main(int, char**) assert(err == ios.goodbit); assert(v == 1); } + { + const char str[] = "-1"; + std::ios_base::iostate err = ios.goodbit; + cpp17_input_iterator iter = + f.get(cpp17_input_iterator(str), + cpp17_input_iterator(str+sizeof(str)), + ios, err, v); + assert(base(iter) == str+sizeof(str)-1); + assert(err == ios.goodbit); + assert(v == std::numeric_limits::max()); + } std::hex(ios); { const char str[] = "0xFFFFFFFF"; diff --git a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_long.pass.cpp b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_long.pass.cpp index b087bdcd94017..fed6fc0246d82 100644 --- a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_long.pass.cpp +++ b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_long.pass.cpp @@ -68,6 +68,17 @@ int main(int, char**) assert(err == ios.goodbit); assert(v == 1); } + { + const char str[] = "-1"; + std::ios_base::iostate err = ios.goodbit; + cpp17_input_iterator iter = + f.get(cpp17_input_iterator(str), + cpp17_input_iterator(str+sizeof(str)), + ios, err, v); + assert(base(iter) == str+sizeof(str)-1); + assert(err == ios.goodbit); + assert(v == std::numeric_limits::max()); + } std::hex(ios); { const char str[] = "0xFFFFFFFF"; diff --git a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_long_long.pass.cpp b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_long_long.pass.cpp index 6769aebe424e3..0bdb6c1c38606 100644 --- a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_long_long.pass.cpp +++ b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_long_long.pass.cpp @@ -68,6 +68,17 @@ int main(int, char**) assert(err == ios.goodbit); assert(v == 1); } + { + const char str[] = "-1"; + std::ios_base::iostate err = ios.goodbit; + cpp17_input_iterator iter = + f.get(cpp17_input_iterator(str), + cpp17_input_iterator(str+sizeof(str)), + ios, err, v); + assert(base(iter) == str+sizeof(str)-1); + assert(err == ios.goodbit); + assert(v == std::numeric_limits::max()); + } std::hex(ios); { const char str[] = "0xFFFFFFFFFFFFFFFF"; diff --git a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_short.pass.cpp b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_short.pass.cpp index bec9a7ff9e3bb..decfbe943461f 100644 --- a/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_short.pass.cpp +++ b/libcxx/test/std/localization/locale.categories/category.numeric/locale.num.get/facet.num.get.members/get_unsigned_short.pass.cpp @@ -68,6 +68,17 @@ int main(int, char**) assert(err == ios.goodbit); assert(v == 1); } + { + const char str[] = "-1"; + std::ios_base::iostate err = ios.goodbit; + cpp17_input_iterator iter = + f.get(cpp17_input_iterator(str), + cpp17_input_iterator(str+sizeof(str)), + ios, err, v); + assert(base(iter) == str+sizeof(str)-1); + assert(err == ios.goodbit); + assert(v == std::numeric_limits::max()); + } std::hex(ios); { const char str[] = "0xFFFF";