title: 从源码角度分析string内存分布
date: 2020-12-04 20:49:32
tags: C++ | string
categories: source code | C++
文章目录
前言
一直以来都认为：string str = “abc”，str本身是存在栈中的，内部有一个指针指向"abc"，"abc"存放在堆中。
今天逛某论坛时，看到一种说法：当字符串小于24字节时，存放在栈中，大于时存放在堆栈中。对此说法表示疑惑，接下来就通过实验和一步步的阅读源码来探析string中的内存到底是如何分配存储的。
string是basic_string的特化版本
string是basic_string传入字符类型char的特化版本：
typedef basic_string<char> string;
接下来看下basic_string的接口：
template < class charT,
class traits = char_traits<charT>,// basic_string::traits_type
class Alloc = allocator<charT> // basic_string::allocator_type
> class basic_string;
可以看到basic_string接收一个Alloc参数，即内存分配器参数，默认是allocator，也就是说内存分配任务交给了allocator，那么下一步就是一探allocator的究竟了。
阅读allocator源码
先给出GNU关于allocator的源码链接：allocator.h源码
阅读源码，在allocator.h文件中没有关于内存的任何操作，同时第92行可以看到class allocator继承了 __glibcxx_base_allocator<_Tp>，那么关于内存的操作必然是在class __glibcxx_base_allocator里面了。
__glibcxx_base_allocator<_Tp>是什么
在c++allocator.h文件第35行可以找到#define __glibcxx_base_allocator __gnu_cxx::new_allocator，也就是说
我们要找的__glibcxx_base_allocator就是__glibcxx_new_allocator。附上c++allocator.h的源码链接：
new_allocator.h——真正进行内存分配工作的地方
00001 // Allocator that wraps operator new -*- C++ -*-
00002
00003 // Copyright (C) 2001, 2002, 2003, 2004, 2005, 2009, 2010
00004 // Free Software Foundation, Inc.
00005 //
00006 // This file is part of the GNU ISO C++ Library. This library is free
00007 // software; you can redistribute it and/or modify it under the
00008 // terms of the GNU General Public License as published by the
00009 // Free Software Foundation; either version 3, or (at your option)
00010 // any later version.
00011
00012 // This library is distributed in the hope that it will be useful,
00013 // but WITHOUT ANY WARRANTY; without even the implied warranty of
00014 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
00015 // GNU General Public License for more details.
00016
00017 // Under Section 7 of GPL version 3, you are granted additional
00018 // permissions described in the GCC Runtime Library Exception, version
00019 // 3.1, as published by the Free Software Foundation.
00020
00021 // You should have received a copy of the GNU General Public License and
00022 // a copy of the GCC Runtime Library Exception along with this program;
00023 // see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
00024 // <http://www.gnu.org/licenses/>.
00025
00026 /** @file ext/new_allocator.h
00027 * This file is a GNU extension to the Standard C++ Library.
00028 */
00029
00030 #ifndef _NEW_ALLOCATOR_H
00031 #define _NEW_ALLOCATOR_H 1
00032
00033 #include <bits/c++config.h>
00034 #include <new>
00035 #include <bits/functexcept.h>
00036 #include <bits/move.h>
00037
00038 namespace __gnu_cxx _GLIBCXX_VISIBILITY(default)
00039 {
00040 _GLIBCXX_BEGIN_NAMESPACE_VERSION
00041
00042 using std::size_t;
00043 using std::ptrdiff_t;
00044
00045 /**
00046 * @brief An allocator that uses global new, as per [20.4].
00047 * @ingroup allocators
00048 *
00049 * This is precisely the allocator defined in the C++ Standard.
00050 * - all allocation calls operator new
00051 * - all deallocation calls operator delete
00052 */
00053 template<typename _Tp>
00054 class new_allocator
00055 {
00056 public:
00057 typedef size_t size_type;
00058 typedef ptrdiff_t difference_type;
00059 typedef _Tp* pointer;
00060 typedef const _Tp* const_pointer;
00061 typedef _Tp& reference;
00062 typedef const _Tp& const_reference;
00063 typedef _Tp value_type;
00064
00065 template<typename _Tp1>
00066 struct rebind
00067 { typedef new_allocator<_Tp1> other; };
00068
00069 new_allocator() throw() { }
00070
00071 new_allocator(const new_allocator&) throw() { }
00072
00073 template<typename _Tp1>
00074 new_allocator(const new_allocator<_Tp1>&) throw() { }
00075
00076 ~new_allocator() throw() { }
00077
00078 pointer
00079 address(reference __x) const { return std::__addressof(__x); }
00080
00081 const_pointer
00082 address(const_reference __x) const { return std::__addressof(__x); }
00083
00084 // NB: __n is permitted to be 0. The C++ standard says nothing
00085 // about what the return value is when __n == 0.
00086 pointer
00087 allocate(size_type __n, const void* = 0)
00088 {
00089 if (__n > this->max_size())
00090 std::__throw_bad_alloc();
00091
00092 return static_cast<_Tp*>(::operator new(__n * sizeof(_Tp)));
00093 }
00094
00095 // __p is not permitted to be a null pointer.
00096 void
00097 deallocate(pointer __p, size_type)
00098 { ::operator delete(__p); }
00099
00100 size_type
00101 max_size() const throw()
00102 { return size_t(-1) / sizeof(_Tp); }
00103
00104 // _GLIBCXX_RESOLVE_LIB_DEFECTS
00105 // 402. wrong new expression in [some_] allocator::construct
00106 void
00107 construct(pointer __p, const _Tp& __val)
00108 { ::new((void *)__p) _Tp(__val); }
00109
00110 #ifdef __GXX_EXPERIMENTAL_CXX0X__
00111 template<typename... _Args>
00112 void
00113 construct(pointer __p, _Args&&... __args)
00114 { ::new((void *)__p) _Tp(std::forward<_Args>(__args)...); }
00115 #endif
00116
00117 void
00118 destroy(pointer __p) { __p->~_Tp(); }
00119 };
00120
00121 template<typename _Tp>
00122 inline bool
00123 operator==(const new_allocator<_Tp>&, const new_allocator<_Tp>&)
00124 { return true; }
00125
00126 template<typename _Tp>
00127 inline bool
00128 operator!=(const new_allocator<_Tp>&, const new_allocator<_Tp>&)
00129 { return false; }
00130
00131 _GLIBCXX_END_NAMESPACE_VERSION
00132 } // namespace
00133
00134 #endif
阅读以上源码，可以发现所有关于内存的操作都是通过global new来实现的，在关于new_allocator文件的官方参阅手册上也有这样一段话：
Detailed Description
template class __gnu_cxx::new_allocator< _Tp >
An allocator that uses global new, as per [20.4].
This is precisely the allocator defined in the C++ Standard.
- all allocation calls operator new
- all deallocation calls operator delete.
Definition at line 54 of file new_allocator.h.
结论
至此，通过阅读GNU相关allocator的源码，我们可以小结一下：
-
string是basic_string的特化版本。
-
basic_string接收一个Alloc参数进行内存分配，默认是allocator。
-
allocator继承了basic_allocator<_Tp>，basic_allocator<_Tp>就是new_allocator<_Tp>
-
new_allocator<_Tp>是一个使用global new的内存分配器，内部所有关于内存分配和内存释放的操作都是通过operator new 和operator delete来进行的，而operator new是向堆请求内存的。
因此：对于string str = “abc”，"abc"是存放在堆中的。
写在最后
本文通过一步步的阅读源码，从string逐渐解析到最终为之分配内存的内存分配器new_allocator，确定string中存储的字符串是位于堆中的。在上述分析中，如有错误还请不吝赐教。