C++学习---cstdio的源码学习分析01-类型定义

182 阅读5分钟

“我报名参加金石计划1期挑战——瓜分10万奖池,这是我的第4篇文章,点击查看活动详情

引言

cstdio文件是C++对stdio.h头文件的封装,Standard Input and Output Library,定义了一系列标准输入输出函数,包括文件操作(fopen/fclose等),格式化打印(printf/scanf)等。 通过该文件的学习,我们能够洞悉一些底层机制的本质,同时,也能够学习了解这些函数的实现方式,接下来,我们就一一学习这些函数的源码。

代码参考:因为这个头文件是相对比较特殊的,我们没有再采取阅读AOSP源码中bionic代码的方式,取而代之,我们参考glibc中的源码,这也是linux环境下,使用最为广泛的C标准库。

glibc源码下载:www.gnu.org/software/li…

git clone https://sourceware.org/git/glibc.git
cd glibc
git checkout master

后文的源码阅读也都基于glibc代码的实现方式。

头文件位置

在glibc代码中搜索stdio.h,可以看到如下文件:

   1      1  /home/sangyu/WorkSpace/Code/C++/glibc/include/bits/stdio.h <<<unknown>>>
   2      1  /home/sangyu/WorkSpace/Code/C++/glibc/include/stdio.h <<<unknown>>>
   3      1  /home/sangyu/WorkSpace/Code/C++/glibc/libio/bits/stdio.h <<<unknown>>>
   4      1  /home/sangyu/WorkSpace/Code/C++/glibc/libio/stdio.h <<<unknown>>>

通过查看头文件的包含关系:

glibc/include/bits/stdio.h include glibc/libio/bits/stdio.h实现

glibc/include/stdio.h include glibc/libio/stdio.h实现

实际上,实现函数定义以及相关接口的是glibc/libio/stdio.h

类型定义与宏定义

stdio.h头文件中定义了后续即将用到的类型与宏定义。

类型定义

FILE类型---文件流对象

/glibc/libio/bits/types/FILE.h,发现了FILE的实际定义_IO_FILE,继续进行追踪

  4 struct _IO_FILE;
  5                                                                         
  6 /* The opaque type of streams.  This is the definition used elsewhere.  */   
  7 typedef struct _IO_FILE FILE;

/glibc/libio/bits/types/struct_FILE.h中最终定义了_IO_FILE,其中的数据类型分为以下几类:

  • int _flags:用于控制文件流的访问权限(如读写执行,注意高位被_IO_MAGIC占用,剩余位用来表示flags)
  • streambuf控制的:用于标识当前读写位置,buffer大小等。
    • char *_IO_read_ptr:当前读取的指针
    • char *_IO_read_end:get区域的末尾
    • char *_IO_read_base:putback+get区域的开始
    • char *_IO_write_base:put区域的开始
    • char *_IO_write_ptr:当前写入的指针
    • char *_IO_write_end:put区域的末尾
    • char *_IO_buf_base:剩余数据的开始
    • char *_IO_buf_end:剩余数据的末尾
  • 备份和回退控制的:用于文件备份和回退控制的指针
    • char *_IO_save_base:非当前get区域的开始位置
    • char *_IO_backup_base:备份区域第一个合法字符
    • char *_IO_save_end:非当前get区域的末尾位置
  • struct _IO_marker *_markers:记录buffer中的位置(后面查看源码分析)
  • struct _IO_FILE *_chain:指向链接的下一个_IO_FILE对象
  • int _fileno:file的number
  • int _flags2:扩展flags
  • __off_t _old_offset:旧版的offset,最终定义为(long int,32位)
  • unsigned short _cur_column:当前列
  • signed char _vtable_offset:vtable表的偏移
  • char _shortbuf[1]:只存储一个字符的buffer
  • _IO_lock_t *_lock:访问IO锁 注意:这里针对旧版的IO_FILE做了兼容,下面是新版增加的内容
  • __off64_t _offset:64位offset(定义为__int64_t,64位)
  • 针对宽字符的处理buff
    • struct _IO_codecvt *_codecvt:字符转换
    • struct _IO_wide_data *_wide_data:宽数据
  • struct _IO_FILE *_freeres_list:剩余的_IO_FILE list
  • size_t __pad5:对齐信息
  • int _mode:mode信息
  • char _unused2[15 * sizeof (int) - 4 * sizeof (void *) - sizeof (size_t)]:预留位置保证追加部分是16个(加上int _mode)sizeof(int)对齐的。
 46 /* The tag name of this struct is _IO_FILE to preserve historic
 47    C++ mangled names for functions taking FILE* arguments.
 48    That name should not be used in new code.  */
 49 struct _IO_FILE
 50 {
 51   int _flags;       /* High-order word is _IO_MAGIC; rest is flags. */
 52 
 53   /* The following pointers correspond to the C++ streambuf protocol. */
 54   char *_IO_read_ptr;   /* Current read pointer */
 55   char *_IO_read_end;   /* End of get area. */
 56   char *_IO_read_base;  /* Start of putback+get area. */
 57   char *_IO_write_base; /* Start of put area. */
 58   char *_IO_write_ptr;  /* Current put pointer. */
 59   char *_IO_write_end;  /* End of put area. */
 60   char *_IO_buf_base;   /* Start of reserve area. */
 61   char *_IO_buf_end;    /* End of reserve area. */
 62 
 63   /* The following fields are used to support backing up and undo. */
 64   char *_IO_save_base; /* Pointer to start of non-current get area. */
 65   char *_IO_backup_base;  /* Pointer to first valid character of backup area */
 66   char *_IO_save_end; /* Pointer to end of non-current get area. */
 67 
 68   struct _IO_marker *_markers;
 69 
 70   struct _IO_FILE *_chain;
 71 
 72   int _fileno;
 73   int _flags2;
 74   __off_t _old_offset; /* This used to be _offset but it's too small.  */
 75 
 76   /* 1+column number of pbase(); 0 is unknown. */
 77   unsigned short _cur_column;
 78   signed char _vtable_offset;
 79   char _shortbuf[1];
 80 
 81   _IO_lock_t *_lock;
 82 #ifdef _IO_USE_OLD_IO_FILE
 83 };
 84 
 85 struct _IO_FILE_complete
 86 {
 87   struct _IO_FILE _file;
 88 #endif
 89   __off64_t _offset;
 90   /* Wide character stream stuff.  */
 91   struct _IO_codecvt *_codecvt;
 92   struct _IO_wide_data *_wide_data;
 93   struct _IO_FILE *_freeres_list;
 94   void *_freeres_buf;
 95   size_t __pad5;
 96   int _mode;
 97   /* Make sure we don't get into trouble again.  */
 98   char _unused2[15 * sizeof (int) - 4 * sizeof (void *) - sizeof (size_t)];
 99 };

_IO_marker定义为包含下一个_IO_marker指针,指向当前的FILE* _sbuf,以及标识的位置_pos。

104 /* A streammarker remembers a position in a buffer. */
105 struct _IO_marker {
106   struct _IO_marker *_next;
107   FILE *_sbuf;
108   /* If _pos >= 0
109  it points to _buf->Gbase()+_pos. FIXME comment */
110   /* if _pos < 0, it points to _buf->eBptr()+_pos. FIXME comment */
111   int _pos;
112 };

_IO_lock_t定义包含int lock; int cnt; void *owner;lock用来指定锁,cnt记录访问数量,owner执向对应的IO。

typedef struct { int lock; int cnt; void *owner; } _IO_lock_t;

fpos_t类型---file特定位置的类型

根据具体的offset位数进行定义

 82 /* The type of the second argument to `fgetpos' and `fsetpos'.  */
 83 #ifndef __USE_FILE_OFFSET64
 84 typedef __fpos_t fpos_t;                                                                                                                             
 85 #else
 86 typedef __fpos64_t fpos_t;
 87 #endif
 88 #ifdef __USE_LARGEFILE64
 89 typedef __fpos64_t fpos64_t;
 90 #endif

对应查看glibc/libio/bits/types/__fpos_t.h,区别__fpos64_t中__pos被定义为__off64_t __mbstate_t是转换状态信息,包括一个__count信息和__value信息(union结构体进行uint32位到4个字符char的转换)

  7 /* The tag name of this struct is _G_fpos_t to preserve historic
  8    C++ mangled names for functions taking fpos_t arguments.
  9    That name should not be used in new code.  */
 10 typedef struct _G_fpos_t
 11 {
 12   __off_t __pos; //32位偏移,long int
 13   __mbstate_t __state;
 14 } __fpos_t;
 
 10 typedef struct _G_fpos64_t
 11 {
 12   __off64_t __pos;//64位偏移,int_64t
 13   __mbstate_t __state;
 14 } __fpos64_t;
 
 //glibc/wcsmbs/bits/types/__mbstate_t.h
  8 #ifndef __WINT_TYPE__
  9 # define __WINT_TYPE__ unsigned int
 10 #endif
 11 
 12 /* Conversion state information.  */
 13 typedef struct
 14 { 
 15   int __count;
 16   union
 17   {
 18     __WINT_TYPE__ __wch;
 19     char __wchb[4];
 20   } __value;        /* Value so far.  */
 21 } __mbstate_t;

size_t类型---unsigned int类型

这个就不需过多解释了,#define size_t unsigned int