gcc源代码分析，build_string ()函数分析

本文说明如何从源程序到一个tree节点的产生过程。

以"Hello,world!\n"字符串为例。

c-parse.tab.c文件中的yylex()函数是gcc编译器的词法分析部分。

这里的一段代码：

    else
      {
        *p = 0;
        yylval.ttype = build_string (p - token_buffer, token_buffer + 1);
        TREE_TYPE (yylval.ttype) = char_array_type_node;
      }

里面的build_string ()函数【tree.c文件中】会调用make_node ( )函数产生一个字符串常量tree节点。

当然只是分配空间和标识code为STRING_CST。

后面的  TREE_STRING_LENGTH (s) = len;

会设置节点的长度。

  TREE_STRING_POINTER (s) = obstack_copy0 (saveable_obstack, str, len);

这一句会设置好字符串的指针。

/* Return a newly constructed STRING_CST node whose value is
   the LEN characters at STR.
   The TREE_TYPE is not initialized.  */

tree
build_string (len, str)
     int len;
     char *str;
{
  register tree s = make_node (STRING_CST);
  TREE_STRING_LENGTH (s) = len;
  TREE_STRING_POINTER (s) = obstack_copy0 (saveable_obstack, str, len);
  return s;
}

 <string_cst 84014 literal "(null)"
 <string_cst 84014 literal "Hello, world!
"
红色的是 make_node (STRING_CST);之后的样子，

而紫色的是

  TREE_STRING_LENGTH (s) = len;
  TREE_STRING_POINTER (s) = obstack_copy0 (saveable_obstack, str, len);
之后的样子。

STRING  -> string
string  -> primary

primary:

    | string
        { $$ = combine_strings ($1); }


string:
      STRING
    | string STRING
        { $$ = chainon ($1, $2); }

根据上面的语法来看，规约到primary的时候会执行combine_strings ()函数。

combine_strings ()函数内部开始处加debug_tree ()函数变成了这个样子

 <string_cst 84014
    type <array_type 88a90
        type <integer_type 825bc char permanent QI
            size <integer_cst 82638 literal permanent 1
            align 8 size_unit 8 sep_unit 8 symtab 0
            sep <integer_cst 82608 literal permanent -128 precision 8 min <integer_cst 82608 -128>
            max <integer_cst 82620 literal permanent 127
            pointer_to_this <pointer_type 88a44> chain <integer_type 826a8* long int>
        permanent BLK
        size <integer_cst 88ae8 literal permanent 256
        align 8 size_unit 8 sep_unit 8 symtab 0 sep <integer_cst 82638 1>
        domain <integer_type 88480 unsigned char permanent unsigned QI size <integer_cst 82638 1>
            align 8 size_unit 8 sep_unit 8 symtab 0
            sep <integer_cst 884cc literal permanent 0 precision 8 min <integer_cst 884cc 0>
            max <integer_cst 884e4 literal permanent 255
            chain <real_type 88624* float>
        chain <pointer_type 88b4c>
    literal "Hello, world!
"

combine_strings ()函数内部结束处加debug_tree ()函数变成了这个样子
 <string_cst 84014
    type <array_type 94ef4
        type <integer_type 825bc char permanent QI
            size <integer_cst 82638 literal permanent 1
            align 8 size_unit 8 sep_unit 8 symtab 0
            sep <integer_cst 82608 literal permanent -128 precision 8 min <integer_cst 82608 -128>
            max <integer_cst 82620 literal permanent 127
            pointer_to_this <pointer_type 88a44> chain <integer_type 826a8* long int>
        BLK
        size <integer_cst 94f40 literal 15
        align 8 size_unit 8 sep_unit 8 symtab 0 sep <integer_cst 82638 1>
        domain <integer_type 94ea8 SI
            size <integer_cst 8254c literal permanent 4
            align 32 size_unit 8 sep_unit 32 symtab 0
            sep <integer_cst 84048 literal 0 precision 32 min <integer_cst 84048 0>
            max <integer_cst 84078 literal 14
    static literal "Hello, world!
"