0xAA55 发表于 2019-2-3 11:50:49

【C语言】给微软MSVC使用的inttypes.h和stdatomic.h

MSVC一直以来对C语言标准支持不全,但有时候又具备一些奇怪的(对于当年而言的)超前特性(比如匿名联合体和结构体等)使其依然具有一定的实用性。即便如此,你要想让代码写得舒服,你还是得按照标准来写。

那么对于最基础的整数类型,我们是使用MSVC里面能用的stdint.h好呢,还是用C99标准新添加的inttypes.h好呢?虽说后者增加的东西或许并不都是你需要的,比如对printf格式代码的跨平台支持,以及完整除法运算所需的imaxdiv_t类型和imaxdiv函数等。

此外,当你尝到了GCC的全程序优化的甜头后,你再跑回去看MSVC,你会怎么想呢?

[*]C标准支持不全。局部变量甚至必须被定义在语句的前面。
[*]没有足够强大的全程序优化,虽说有/LTCG,但它看起来就和没有一样。
[*]对SIMD指令集的优化贼弱,强行给你擅自加入一堆movaps等,来解决由于它自身优化很差而带来的寄存器不够的问题。

以及

[*]完善的代码高亮和自动提示、自动补全功能。
(虽说对于C语言调用COM类的状况下而言,它的自动提示功能并不知道你写的是C还是C++,会在你试图通过虚表调用COM类的时候告诉你“没有lpVtbl成员”)
[*]令人上瘾的F12转到定义功能。这个功能不仅仅对于编程开发,对于编程的学习而言更具有关键性的使用价值。
[*]虽说不完整,但基本够用的一整套支持多线程和多机器的调试器工具。
[*]漂亮的界面
[*]一帮惹不起的信徒(咳咳)

由于Cygwin过于Tony带水,拖家带口,有悖于我们希望编译器足够好用的初衷,我选择MinGW。
然而我发现我在使用MinGW的时候我的代码里既可以使用Windows.h(需要设置-mwindows编译器命令行参数)又可以使用inttypes.h和stdatomic.h。我甚至还可以使用pthread.h,虽说其实根本不好用——它依赖unistd.h,但这样一来,我搞不好就要把Unix全家桶搬到Windows里面了,这十分南辕北辙。不如直接用Windows的API。

对于如此纠结的情况,我是使用stdint.h好呢,还是使用inttypes.h好呢,我该不该使用stdatomic.h呢?

我决定采用打补丁的方式,对于缺乏这些头文件的情况下,使用替补的去代替。我在我的工程目录里添加两个目录,一个是“备用头文件”,用于存放这些备用的头文件。另一个则是“额外包含头文件”,这个是给编译器查找头文件目录指定的,让编译器从这里面查找额外的头文件。任何人如果想要编译我这个工程,他应该先用自己的工具编译一下。如果编译不通过,他会在我的文档里看到说明,然后把缺失的头文件从“备用头文件”目录里拷贝到“额外包含头文件”目录里,然后再次编译即可。

那么问题来了。这两个头文件我是应该自己写呢,还是从现有的找呢?

人生宝贵,造重复轮子(还不一定比别人的圆)不可取。我决定使用别人写好的头文件。

其中,谷歌的开源库里有inttypes.h是给MSVC用的。
https://code.google.com/archive/p/msinttypes/downloads
此处我直接搬运一下:


我搬运的是我在写这篇帖子的时候看到的最新的版本。它是有可能更新的,所以请从原网址查看为妙。

它里面其实同时包含了stdint.h和inttypes.h。
目测是为了防止不同MSVC的stdint.h版本不同而带来的问题。

然后,我从FFMpeg的开源库里找到了给MSVC用的stdatomic.h
https://www.ffmpeg.org/doxygen/trunk/win32_2stdatomic_8h_source.html
https://github.com/FFmpeg/FFmpeg/blob/master/compat/atomics/win32/stdatomic.h
ffmpeg的更新是十分频繁的。建议自己上它的网站去看。
此处依然搬运一下:


也不长,我就直接复制到帖子里了。可以看出,stdatomic原有的内存顺序部分的特性被丢了个精光,全都变成了最严格的原子操作方式:memory_order_seq_cst。
我们其实并不应该给stdatomic的行为擅自加料,考虑到Windows运行在各种不同的平台上,还是以严格为好。典型例子,牙膏厂Atom系列CPU和NUMA架构多处理器Xeon对于原子操作的严格性要求就不一样,而且它们的实际行为也大不一样——Atom更多是对自己内部的缓存进行原子操作,而Xeon则更多是强制读写内存,如果你在NUMA架构上让Xeon不强制读写内存的话,两个处理器之间的内存同步性就会丢失。
stdatomic.h/*
* This file is part of FFmpeg.
*
* FFmpeg 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.
*
* FFmpeg 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 FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/

#ifndef COMPAT_ATOMICS_WIN32_STDATOMIC_H
#define COMPAT_ATOMICS_WIN32_STDATOMIC_H

#define WIN32_LEAN_AND_MEAN
#include <stddef.h>
#include <stdint.h>
#include <windows.h>

#define ATOMIC_FLAG_INIT 0

#define ATOMIC_VAR_INIT(value) (value)

#define atomic_init(obj, value) \
do {                            \
    *(obj) = (value);         \
} while(0)

#define kill_dependency(y) ((void)0)

#define atomic_thread_fence(order) \
    MemoryBarrier();

#define atomic_signal_fence(order) \
    ((void)0)

#define atomic_is_lock_free(obj) 0

typedef intptr_t atomic_flag;
typedef intptr_t atomic_bool;
typedef intptr_t atomic_char;
typedef intptr_t atomic_schar;
typedef intptr_t atomic_uchar;
typedef intptr_t atomic_short;
typedef intptr_t atomic_ushort;
typedef intptr_t atomic_int;
typedef intptr_t atomic_uint;
typedef intptr_t atomic_long;
typedef intptr_t atomic_ulong;
typedef intptr_t atomic_llong;
typedef intptr_t atomic_ullong;
typedef intptr_t atomic_wchar_t;
typedef intptr_t atomic_int_least8_t;
typedef intptr_t atomic_uint_least8_t;
typedef intptr_t atomic_int_least16_t;
typedef intptr_t atomic_uint_least16_t;
typedef intptr_t atomic_int_least32_t;
typedef intptr_t atomic_uint_least32_t;
typedef intptr_t atomic_int_least64_t;
typedef intptr_t atomic_uint_least64_t;
typedef intptr_t atomic_int_fast8_t;
typedef intptr_t atomic_uint_fast8_t;
typedef intptr_t atomic_int_fast16_t;
typedef intptr_t atomic_uint_fast16_t;
typedef intptr_t atomic_int_fast32_t;
typedef intptr_t atomic_uint_fast32_t;
typedef intptr_t atomic_int_fast64_t;
typedef intptr_t atomic_uint_fast64_t;
typedef intptr_t atomic_intptr_t;
typedef intptr_t atomic_uintptr_t;
typedef intptr_t atomic_size_t;
typedef intptr_t atomic_ptrdiff_t;
typedef intptr_t atomic_intmax_t;
typedef intptr_t atomic_uintmax_t;

#define atomic_store(object, desired)   \
do {                                    \
    *(object) = (desired);            \
    MemoryBarrier();                  \
} while (0)

#define atomic_store_explicit(object, desired, order) \
    atomic_store(object, desired)

#define atomic_load(object) \
    (MemoryBarrier(), *(object))

#define atomic_load_explicit(object, order) \
    atomic_load(object)

#define atomic_exchange(object, desired) \
    InterlockedExchangePointer(object, desired);

#define atomic_exchange_explicit(object, desired, order) \
    atomic_exchange(object, desired)

static inline int atomic_compare_exchange_strong(intptr_t *object, intptr_t *expected,
                                                 intptr_t desired)
{
    intptr_t old = *expected;
    *expected = (intptr_t)InterlockedCompareExchangePointer(
      (PVOID *)object, (PVOID)desired, (PVOID)old);
    return *expected == old;
}

#define atomic_compare_exchange_strong_explicit(object, expected, desired, success, failure) \
    atomic_compare_exchange_strong(object, expected, desired)

#define atomic_compare_exchange_weak(object, expected, desired) \
    atomic_compare_exchange_strong(object, expected, desired)

#define atomic_compare_exchange_weak_explicit(object, expected, desired, success, failure) \
    atomic_compare_exchange_weak(object, expected, desired)

#ifdef _WIN64
#define atomic_fetch_add(object, operand) \
    InterlockedExchangeAdd64(object, operand)

#define atomic_fetch_sub(object, operand) \
    InterlockedExchangeAdd64(object, -(operand))

#define atomic_fetch_or(object, operand) \
    InterlockedOr64(object, operand)

#define atomic_fetch_xor(object, operand) \
    InterlockedXor64(object, operand)

#define atomic_fetch_and(object, operand) \
    InterlockedAnd64(object, operand)
#else
#define atomic_fetch_add(object, operand) \
    InterlockedExchangeAdd(object, operand)

#define atomic_fetch_sub(object, operand) \
    InterlockedExchangeAdd(object, -(operand))

#define atomic_fetch_or(object, operand) \
    InterlockedOr(object, operand)

#define atomic_fetch_xor(object, operand) \
    InterlockedXor(object, operand)

#define atomic_fetch_and(object, operand) \
    InterlockedAnd(object, operand)
#endif /* _WIN64 */

#define atomic_fetch_add_explicit(object, operand, order) \
    atomic_fetch_add(object, operand)

#define atomic_fetch_sub_explicit(object, operand, order) \
    atomic_fetch_sub(object, operand)

#define atomic_fetch_or_explicit(object, operand, order) \
    atomic_fetch_or(object, operand)

#define atomic_fetch_xor_explicit(object, operand, order) \
    atomic_fetch_xor(object, operand)

#define atomic_fetch_and_explicit(object, operand, order) \
    atomic_fetch_and(object, operand)

#define atomic_flag_test_and_set(object) \
    atomic_exchange(object, 1)

#define atomic_flag_test_and_set_explicit(object, order) \
    atomic_flag_test_and_set(object)

#define atomic_flag_clear(object) \
    atomic_store(object, 0)

#define atomic_flag_clear_explicit(object, order) \
    atomic_flag_clear(object)

#endif /* COMPAT_ATOMICS_WIN32_STDATOMIC_H */顺带我也把之前提到的inttypes.h也直接丢帖子里,便于Ctrl+C Ctrl+V。
inttypes.h// ISO C9xcompliant inttypes.h for Microsoft Visual Studio
// Based on ISO/IEC 9899:TC2 Committee draft (May 6, 2005) WG14/N1124
//
//Copyright (c) 2006 Alexander Chemeris
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are met:
//
//   1. Redistributions of source code must retain the above copyright notice,
//      this list of conditions and the following disclaimer.
//
//   2. Redistributions in binary form must reproduce the above copyright
//      notice, this list of conditions and the following disclaimer in the
//      documentation and/or other materials provided with the distribution.
//
//   3. The name of the author may be used to endorse or promote products
//      derived from this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
// WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
// MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
// EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
// OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
// WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
// OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
// ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
///////////////////////////////////////////////////////////////////////////////

#ifndef _MSC_VER // [
#error "Use this header only with Microsoft Visual C++ compilers!"
#endif // _MSC_VER ]

#ifndef _MSC_INTTYPES_H_ // [
#define _MSC_INTTYPES_H_

#if _MSC_VER > 1000
#pragma once
#endif

#include "stdint.h"

// 7.8 Format conversion of integer types

typedef struct {
   intmax_t quot;
   intmax_t rem;
} imaxdiv_t;

// 7.8.1 Macros for format specifiers

#if !defined(__cplusplus) || defined(__STDC_FORMAT_MACROS) // [   See footnote 185 at page 198

// The fprintf macros for signed integers are:
#define PRId8       "d"
#define PRIi8       "i"
#define PRIdLEAST8"d"
#define PRIiLEAST8"i"
#define PRIdFAST8   "d"
#define PRIiFAST8   "i"

#define PRId16       "hd"
#define PRIi16       "hi"
#define PRIdLEAST16"hd"
#define PRIiLEAST16"hi"
#define PRIdFAST16   "hd"
#define PRIiFAST16   "hi"

#define PRId32       "I32d"
#define PRIi32       "I32i"
#define PRIdLEAST32"I32d"
#define PRIiLEAST32"I32i"
#define PRIdFAST32   "I32d"
#define PRIiFAST32   "I32i"

#define PRId64       "I64d"
#define PRIi64       "I64i"
#define PRIdLEAST64"I64d"
#define PRIiLEAST64"I64i"
#define PRIdFAST64   "I64d"
#define PRIiFAST64   "I64i"

#define PRIdMAX   "I64d"
#define PRIiMAX   "I64i"

#define PRIdPTR   "Id"
#define PRIiPTR   "Ii"

// The fprintf macros for unsigned integers are:
#define PRIo8       "o"
#define PRIu8       "u"
#define PRIx8       "x"
#define PRIX8       "X"
#define PRIoLEAST8"o"
#define PRIuLEAST8"u"
#define PRIxLEAST8"x"
#define PRIXLEAST8"X"
#define PRIoFAST8   "o"
#define PRIuFAST8   "u"
#define PRIxFAST8   "x"
#define PRIXFAST8   "X"

#define PRIo16       "ho"
#define PRIu16       "hu"
#define PRIx16       "hx"
#define PRIX16       "hX"
#define PRIoLEAST16"ho"
#define PRIuLEAST16"hu"
#define PRIxLEAST16"hx"
#define PRIXLEAST16"hX"
#define PRIoFAST16   "ho"
#define PRIuFAST16   "hu"
#define PRIxFAST16   "hx"
#define PRIXFAST16   "hX"

#define PRIo32       "I32o"
#define PRIu32       "I32u"
#define PRIx32       "I32x"
#define PRIX32       "I32X"
#define PRIoLEAST32"I32o"
#define PRIuLEAST32"I32u"
#define PRIxLEAST32"I32x"
#define PRIXLEAST32"I32X"
#define PRIoFAST32   "I32o"
#define PRIuFAST32   "I32u"
#define PRIxFAST32   "I32x"
#define PRIXFAST32   "I32X"

#define PRIo64       "I64o"
#define PRIu64       "I64u"
#define PRIx64       "I64x"
#define PRIX64       "I64X"
#define PRIoLEAST64"I64o"
#define PRIuLEAST64"I64u"
#define PRIxLEAST64"I64x"
#define PRIXLEAST64"I64X"
#define PRIoFAST64   "I64o"
#define PRIuFAST64   "I64u"
#define PRIxFAST64   "I64x"
#define PRIXFAST64   "I64X"

#define PRIoMAX   "I64o"
#define PRIuMAX   "I64u"
#define PRIxMAX   "I64x"
#define PRIXMAX   "I64X"

#define PRIoPTR   "Io"
#define PRIuPTR   "Iu"
#define PRIxPTR   "Ix"
#define PRIXPTR   "IX"

// The fscanf macros for signed integers are:
#define SCNd8       "d"
#define SCNi8       "i"
#define SCNdLEAST8"d"
#define SCNiLEAST8"i"
#define SCNdFAST8   "d"
#define SCNiFAST8   "i"

#define SCNd16       "hd"
#define SCNi16       "hi"
#define SCNdLEAST16"hd"
#define SCNiLEAST16"hi"
#define SCNdFAST16   "hd"
#define SCNiFAST16   "hi"

#define SCNd32       "ld"
#define SCNi32       "li"
#define SCNdLEAST32"ld"
#define SCNiLEAST32"li"
#define SCNdFAST32   "ld"
#define SCNiFAST32   "li"

#define SCNd64       "I64d"
#define SCNi64       "I64i"
#define SCNdLEAST64"I64d"
#define SCNiLEAST64"I64i"
#define SCNdFAST64   "I64d"
#define SCNiFAST64   "I64i"

#define SCNdMAX   "I64d"
#define SCNiMAX   "I64i"

#ifdef _WIN64 // [
#define SCNdPTR   "I64d"
#define SCNiPTR   "I64i"
#else// _WIN64 ][
#define SCNdPTR   "ld"
#define SCNiPTR   "li"
#endif// _WIN64 ]

// The fscanf macros for unsigned integers are:
#define SCNo8       "o"
#define SCNu8       "u"
#define SCNx8       "x"
#define SCNX8       "X"
#define SCNoLEAST8"o"
#define SCNuLEAST8"u"
#define SCNxLEAST8"x"
#define SCNXLEAST8"X"
#define SCNoFAST8   "o"
#define SCNuFAST8   "u"
#define SCNxFAST8   "x"
#define SCNXFAST8   "X"

#define SCNo16       "ho"
#define SCNu16       "hu"
#define SCNx16       "hx"
#define SCNX16       "hX"
#define SCNoLEAST16"ho"
#define SCNuLEAST16"hu"
#define SCNxLEAST16"hx"
#define SCNXLEAST16"hX"
#define SCNoFAST16   "ho"
#define SCNuFAST16   "hu"
#define SCNxFAST16   "hx"
#define SCNXFAST16   "hX"

#define SCNo32       "lo"
#define SCNu32       "lu"
#define SCNx32       "lx"
#define SCNX32       "lX"
#define SCNoLEAST32"lo"
#define SCNuLEAST32"lu"
#define SCNxLEAST32"lx"
#define SCNXLEAST32"lX"
#define SCNoFAST32   "lo"
#define SCNuFAST32   "lu"
#define SCNxFAST32   "lx"
#define SCNXFAST32   "lX"

#define SCNo64       "I64o"
#define SCNu64       "I64u"
#define SCNx64       "I64x"
#define SCNX64       "I64X"
#define SCNoLEAST64"I64o"
#define SCNuLEAST64"I64u"
#define SCNxLEAST64"I64x"
#define SCNXLEAST64"I64X"
#define SCNoFAST64   "I64o"
#define SCNuFAST64   "I64u"
#define SCNxFAST64   "I64x"
#define SCNXFAST64   "I64X"

#define SCNoMAX   "I64o"
#define SCNuMAX   "I64u"
#define SCNxMAX   "I64x"
#define SCNXMAX   "I64X"

#ifdef _WIN64 // [
#define SCNoPTR   "I64o"
#define SCNuPTR   "I64u"
#define SCNxPTR   "I64x"
#define SCNXPTR   "I64X"
#else// _WIN64 ][
#define SCNoPTR   "lo"
#define SCNuPTR   "lu"
#define SCNxPTR   "lx"
#define SCNXPTR   "lX"
#endif// _WIN64 ]

#endif // __STDC_FORMAT_MACROS ]

// 7.8.2 Functions for greatest-width integer types

// 7.8.2.1 The imaxabs function
#define imaxabs _abs64

// 7.8.2.2 The imaxdiv function

// This is modified version of div() function from Microsoft's div.c found
// in %MSVC.NET%\crt\src\div.c
#ifdef STATIC_IMAXDIV // [
static
#else // STATIC_IMAXDIV ][
_inline
#endif // STATIC_IMAXDIV ]
imaxdiv_t __cdecl imaxdiv(intmax_t numer, intmax_t denom)
{
   imaxdiv_t result;

   result.quot = numer / denom;
   result.rem = numer % denom;

   if (numer < 0 && result.rem > 0) {
      // did division wrong; must fix up
      ++result.quot;
      result.rem -= denom;
   }

   return result;
}

// 7.8.2.3 The strtoimax and strtoumax functions
#define strtoimax _strtoi64
#define strtoumax _strtoui64

// 7.8.2.4 The wcstoimax and wcstoumax functions
#define wcstoimax _wcstoi64
#define wcstoumax _wcstoui64


#endif // _MSC_INTTYPES_H_ ]还有它所依赖的stdint.h
stdint.h// ISO C9xcompliant stdint.h for Microsoft Visual Studio
// Based on ISO/IEC 9899:TC2 Committee draft (May 6, 2005) WG14/N1124
//
//Copyright (c) 2006-2008 Alexander Chemeris
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are met:
//
//   1. Redistributions of source code must retain the above copyright notice,
//      this list of conditions and the following disclaimer.
//
//   2. Redistributions in binary form must reproduce the above copyright
//      notice, this list of conditions and the following disclaimer in the
//      documentation and/or other materials provided with the distribution.
//
//   3. The name of the author may be used to endorse or promote products
//      derived from this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
// WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
// MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
// EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
// OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
// WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
// OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
// ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
///////////////////////////////////////////////////////////////////////////////

#ifndef _MSC_VER // [
#error "Use this header only with Microsoft Visual C++ compilers!"
#endif // _MSC_VER ]

#ifndef _MSC_STDINT_H_ // [
#define _MSC_STDINT_H_

#if _MSC_VER > 1000
#pragma once
#endif

#include <limits.h>

// For Visual Studio 6 in C++ mode and for many Visual Studio versions when
// compiling for ARM we should wrap <wchar.h> include with 'extern "C++" {}'
// or compiler give many errors like this:
//   error C2733: second C linkage of overloaded function 'wmemchr' not allowed
#ifdef __cplusplus
extern "C" {
#endif
#include <wchar.h>
#ifdef __cplusplus
}
#endif

// Define _W64 macros to mark types changing their size, like intptr_t.
#ifndef _W64
#if !defined(__midl) && (defined(_X86_) || defined(_M_IX86)) && _MSC_VER >= 1300
#   define _W64 __w64
#else
#   define _W64
#endif
#endif


// 7.18.1 Integer types

// 7.18.1.1 Exact-width integer types

// Visual Studio 6 and Embedded Visual C++ 4 doesn't
// realize that, e.g. char has the same size as __int8
// so we give up on __intX for them.
#if (_MSC_VER < 1300)
   typedef signed char       int8_t;
   typedef signed short      int16_t;
   typedef signed int      int32_t;
   typedef unsigned char   uint8_t;
   typedef unsigned short    uint16_t;
   typedef unsigned int      uint32_t;
#else
   typedef signed __int8   int8_t;
   typedef signed __int16    int16_t;
   typedef signed __int32    int32_t;
   typedef unsigned __int8   uint8_t;
   typedef unsigned __int16uint16_t;
   typedef unsigned __int32uint32_t;
#endif
typedef signed __int64       int64_t;
typedef unsigned __int64   uint64_t;


// 7.18.1.2 Minimum-width integer types
typedef int8_t    int_least8_t;
typedef int16_t   int_least16_t;
typedef int32_t   int_least32_t;
typedef int64_t   int_least64_t;
typedef uint8_t   uint_least8_t;
typedef uint16_tuint_least16_t;
typedef uint32_tuint_least32_t;
typedef uint64_tuint_least64_t;

// 7.18.1.3 Fastest minimum-width integer types
typedef int8_t    int_fast8_t;
typedef int16_t   int_fast16_t;
typedef int32_t   int_fast32_t;
typedef int64_t   int_fast64_t;
typedef uint8_t   uint_fast8_t;
typedef uint16_tuint_fast16_t;
typedef uint32_tuint_fast32_t;
typedef uint64_tuint_fast64_t;

// 7.18.1.4 Integer types capable of holding object pointers
#ifdef _WIN64 // [
   typedef signed __int64    intptr_t;
   typedef unsigned __int64uintptr_t;
#else // _WIN64 ][
   typedef _W64 signed int   intptr_t;
   typedef _W64 unsigned int uintptr_t;
#endif // _WIN64 ]

// 7.18.1.5 Greatest-width integer types
typedef int64_t   intmax_t;
typedef uint64_tuintmax_t;


// 7.18.2 Limits of specified-width integer types

#if !defined(__cplusplus) || defined(__STDC_LIMIT_MACROS) // [   See footnote 220 at page 257 and footnote 221 at page 259

// 7.18.2.1 Limits of exact-width integer types
#define INT8_MIN   ((int8_t)_I8_MIN)
#define INT8_MAX   _I8_MAX
#define INT16_MIN    ((int16_t)_I16_MIN)
#define INT16_MAX    _I16_MAX
#define INT32_MIN    ((int32_t)_I32_MIN)
#define INT32_MAX    _I32_MAX
#define INT64_MIN    ((int64_t)_I64_MIN)
#define INT64_MAX    _I64_MAX
#define UINT8_MAX    _UI8_MAX
#define UINT16_MAX   _UI16_MAX
#define UINT32_MAX   _UI32_MAX
#define UINT64_MAX   _UI64_MAX

// 7.18.2.2 Limits of minimum-width integer types
#define INT_LEAST8_MIN    INT8_MIN
#define INT_LEAST8_MAX    INT8_MAX
#define INT_LEAST16_MIN   INT16_MIN
#define INT_LEAST16_MAX   INT16_MAX
#define INT_LEAST32_MIN   INT32_MIN
#define INT_LEAST32_MAX   INT32_MAX
#define INT_LEAST64_MIN   INT64_MIN
#define INT_LEAST64_MAX   INT64_MAX
#define UINT_LEAST8_MAX   UINT8_MAX
#define UINT_LEAST16_MAXUINT16_MAX
#define UINT_LEAST32_MAXUINT32_MAX
#define UINT_LEAST64_MAXUINT64_MAX

// 7.18.2.3 Limits of fastest minimum-width integer types
#define INT_FAST8_MIN    INT8_MIN
#define INT_FAST8_MAX    INT8_MAX
#define INT_FAST16_MIN   INT16_MIN
#define INT_FAST16_MAX   INT16_MAX
#define INT_FAST32_MIN   INT32_MIN
#define INT_FAST32_MAX   INT32_MAX
#define INT_FAST64_MIN   INT64_MIN
#define INT_FAST64_MAX   INT64_MAX
#define UINT_FAST8_MAX   UINT8_MAX
#define UINT_FAST16_MAXUINT16_MAX
#define UINT_FAST32_MAXUINT32_MAX
#define UINT_FAST64_MAXUINT64_MAX

// 7.18.2.4 Limits of integer types capable of holding object pointers
#ifdef _WIN64 // [
#define INTPTR_MIN   INT64_MIN
#define INTPTR_MAX   INT64_MAX
#define UINTPTR_MAXUINT64_MAX
#else // _WIN64 ][
#define INTPTR_MIN   INT32_MIN
#define INTPTR_MAX   INT32_MAX
#define UINTPTR_MAXUINT32_MAX
#endif // _WIN64 ]

// 7.18.2.5 Limits of greatest-width integer types
#define INTMAX_MIN   INT64_MIN
#define INTMAX_MAX   INT64_MAX
#define UINTMAX_MAXUINT64_MAX

// 7.18.3 Limits of other integer types

#ifdef _WIN64 // [
#define PTRDIFF_MIN_I64_MIN
#define PTRDIFF_MAX_I64_MAX
#else// _WIN64 ][
#define PTRDIFF_MIN_I32_MIN
#define PTRDIFF_MAX_I32_MAX
#endif// _WIN64 ]

#define SIG_ATOMIC_MININT_MIN
#define SIG_ATOMIC_MAXINT_MAX

#ifndef SIZE_MAX // [
#ifdef _WIN64 // [
#   define SIZE_MAX_UI64_MAX
#else // _WIN64 ][
#   define SIZE_MAX_UI32_MAX
#endif // _WIN64 ]
#endif // SIZE_MAX ]

// WCHAR_MIN and WCHAR_MAX are also defined in <wchar.h>
#ifndef WCHAR_MIN // [
#define WCHAR_MIN0
#endif// WCHAR_MIN ]
#ifndef WCHAR_MAX // [
#define WCHAR_MAX_UI16_MAX
#endif// WCHAR_MAX ]

#define WINT_MIN0
#define WINT_MAX_UI16_MAX

#endif // __STDC_LIMIT_MACROS ]


// 7.18.4 Limits of other integer types

#if !defined(__cplusplus) || defined(__STDC_CONSTANT_MACROS) // [   See footnote 224 at page 260

// 7.18.4.1 Macros for minimum-width integer constants

#define INT8_C(val)val##i8
#define INT16_C(val) val##i16
#define INT32_C(val) val##i32
#define INT64_C(val) val##i64

#define UINT8_C(val)val##ui8
#define UINT16_C(val) val##ui16
#define UINT32_C(val) val##ui32
#define UINT64_C(val) val##ui64

// 7.18.4.2 Macros for greatest-width integer constants
#define INTMAX_C   INT64_C
#define UINTMAX_CUINT64_C

#endif // __STDC_CONSTANT_MACROS ]


#endif // _MSC_STDINT_H_ ]以及它的更新日志:------------------------------------------------------------------------
r26 | 2009-10-02 13:36:47 +0400 | 2 lines

Change <stdint.h> to "stdint.h" to let compiler search for it in local directory.

------------------------------------------------------------------------
r25 | 2009-09-17 23:46:49 +0400 | 2 lines

Fix incorrect int8_t behaviour if compiled with /J flag.

------------------------------------------------------------------------
r24 | 2009-05-13 14:53:48 +0400 | 2 lines

Forgot about #ifdef __cplusplus guard around 'extern "C"', so inclusion to C files has been broken.

------------------------------------------------------------------------
r23 | 2009-05-12 01:27:45 +0400 | 3 lines

Always wrap <wchar钹?with external "C" {}.
It turns out that not only Visual Studio 6 requires this, but also newer versions when compiling for ARM.

------------------------------------------------------------------------
r22 | 2009-05-11 22:22:15 +0400 | 3 lines

Visual Studio 6 and Embedded Visual C++ 4 doesn't realize that, e.g. char has the same size as __int8 so we give up on __intX for them.
his should close Issue 3 in issue tracker.

------------------------------------------------------------------------
r21 | 2008-07-17 09:47:22 +0400 | 4 lines

Get rid of these compiler warnings when compiling for 32-bit:
warning C4311: 'type cast' : pointer truncation from 'void *' to 'uintptr_t'
warning C4312: 'type cast' : conversion from 'uintptr_t' to 'const void *' of greater size

------------------------------------------------------------------------
r20 | 2007-10-09 16:54:27 +0400 | 2 lines

Better C99 conformance: macros for format specifiers should only be included in C++ implementations if __STDC_FORMAT_MACROS is defined before <inttypes.h> is included.

------------------------------------------------------------------------
r19 | 2007-07-04 02:14:40 +0400 | 3 lines

Explicitly cast to appropriate type INT8_MIN, INT16_MIN, INT32_MIN and INT64_MIN constants.
Due to their unusual definition in Visual Studio headers (-_Ix_MAX-1) they are propagated to int and thus do not have expected type, causing VS6 strict compiler to claim about type inconsistency.

------------------------------------------------------------------------
r18 | 2007-06-26 16:53:23 +0400 | 2 lines

Better handling of (U)INTx_C macros - now they generate constants of exact width.

------------------------------------------------------------------------
r17 | 2007-03-29 20:16:14 +0400 | 2 lines

Fix typo: Miscrosoft -> Microsoft.

------------------------------------------------------------------------
r16 | 2007-02-24 17:32:58 +0300 | 4 lines

Remove <BaseTsd.h> include, as it is not present in Visual Studio 2005 Epxress Edition and required only for INT_PTR and UINT_PTR types.

'intptr_t' and 'uintptr_t' types now defined explicitly with #ifdef _WIN64.

------------------------------------------------------------------------
r15 | 2007-02-11 20:53:05 +0300 | 2 lines

More correct fix for compilation under VS6.

------------------------------------------------------------------------
r14 | 2007-02-11 20:04:32 +0300 | 2 lines

Bugfix: fix compiling under VS6, when stdint.h enclosed in 'extern "C" {}'.

------------------------------------------------------------------------
r13 | 2006-12-13 16:53:11 +0300 | 2 lines

Make _inline modifier for imaxdiv default option. Use STATIC_IMAXDIV to make it static.

------------------------------------------------------------------------
r12 | 2006-12-13 16:42:24 +0300 | 2 lines

Error message changed: VC6 supported from now.

------------------------------------------------------------------------
r11 | 2006-12-13 16:39:33 +0300 | 2 lines

All (U)INT* types changed to (unsigned) __int*. This should make stdint.h compatible with VC6.

------------------------------------------------------------------------
r10 | 2006-12-13 16:20:57 +0300 | 3 lines

Added INLINE_IMAXDIV define switch.
If INLINE_IMAXDIV is defined imaxdiv() have static modifier. If not - it is _inline.

------------------------------------------------------------------------
r9 | 2006-12-13 15:53:52 +0300 | 2 lines

Error message for non-MSC compiler changed.

------------------------------------------------------------------------
r8 | 2006-12-13 12:47:48 +0300 | 2 lines

Added #ifndef for SIZE_MAX (it is defined in limits.h on MSVSC 8).

------------------------------------------------------------------------
r7 | 2006-12-13 01:08:02 +0300 | 2 lines

License chaged to BSD-derivative.

------------------------------------------------------------------------
r6 | 2006-12-13 00:53:20 +0300 | 2 lines

Added <wchar.h> include to avoid warnings when it is included after stdint.h.

------------------------------------------------------------------------
r5 | 2006-12-12 00:58:05 +0300 | 2 lines

BUGFIX: Definitions of INTPTR_MIN, INTPTR_MAX and UINTPTR_MAX for WIN32 and WIN64 was mixed up.

------------------------------------------------------------------------
r4 | 2006-12-12 00:51:55 +0300 | 2 lines

Rise #error if _MSC_VER is not defined. I.e. compiler other then Microsoft Visual C++ is used.

------------------------------------------------------------------------
r3 | 2006-12-11 22:54:14 +0300 | 2 lines

Added <limits.h> include to stdint.h.

------------------------------------------------------------------------
r2 | 2006-12-11 21:39:27 +0300 | 2 lines

Initial check in.

------------------------------------------------------------------------
r1 | 2006-12-11 21:30:23 +0300 | 1 line

Initial directory structure.
------------------------------------------------------------------------

watermelon 发表于 2019-2-3 12:07:40

顶站长:D,小弟我记得pells C对C11标准支持的比较好,至于msvc它不管C99或者C11,我认为他自己就有一套msvc的C语言标准。
顺便提一下,补头文件对我来说是个恐怖的事情,因为大多数头文件包含别的头文件,而别的头文件又包含别的头文件,这个我经常处理不好。:Q

Ayala 发表于 2019-2-9 00:53:50

本帖最后由 Ayala 于 2019-2-9 01:02 编辑

新版本的vs不自己包含么
stdatomic.h和atomic区别很大么
还有
cinttypes和inttypes.h

0xAA55 发表于 2019-2-9 23:09:45

Ayala 发表于 2019-2-9 00:53
新版本的vs不自己包含么
stdatomic.h和atomic区别很大么
还有


cinttypes是C++的,inttypes.h是C的。cinttypes是C++从C那里搬运过来的。

stdatomic.h是C的,atomic是C++的,但atomic不是C++从C的stdatomic.h搬运过来的。搬运的版本是cstdatomic。
页: [1]
查看完整版本: 【C语言】给微软MSVC使用的inttypes.h和stdatomic.h