程序员的自我修养: 静态链接的过程

1. 空间与地址分配

是一个相似段合并的过程。

step one，空间与地址分配

step two，符号解析与重定位

//a.c extern int shared; int main() { int a = 100; swap(&a, &shared); }

//b.c int shared = 1; void swap(int* a, int* b) { *a ^= *b ^= *a ^= *b; }

pinder@ubuntu:~/program$ gcc -c a.c b.c pinder@ubuntu:~/program$ objdump -h a.o a.o: file format elf32-i386 Sections: Idx Name Size VMA LMA File off Algn 0 .text 00000027 00000000 00000000 00000034 2**2 CONTENTS, ALLOC, LOAD, RELOC, READONLY, CODE 1 .data 00000000 00000000 00000000 0000005c 2**2 CONTENTS, ALLOC, LOAD, DATA 2 .bss 00000000 00000000 00000000 0000005c 2**2 ALLOC 3 .comment 00000024 00000000 00000000 0000005c 2**0 CONTENTS, READONLY 4 .note.GNU-stack 00000000 00000000 00000000 00000080 2**0 CONTENTS, READONLY pinder@ubuntu:~/program$ objdump -h b.o b.o: file format elf32-i386 Sections: Idx Name Size VMA LMA File off Algn 0 .text 0000003a 00000000 00000000 00000034 2**2 CONTENTS, ALLOC, LOAD, READONLY, CODE 1 .data 00000004 00000000 00000000 00000070 2**2 CONTENTS, ALLOC, LOAD, DATA 2 .bss 00000000 00000000 00000000 00000074 2**2 ALLOC 3 .comment 00000024 00000000 00000000 00000074 2**0 CONTENTS, READONLY 4 .note.GNU-stack 00000000 00000000 00000000 00000098 2**0 CONTENTS, READONLY pinder@ubuntu:~/program$ ld a.o b.o -e main -o ab pinder@ubuntu:~/program$ objdump -h ab ab: file format elf32-i386 Sections: Idx Name Size VMA LMA File off Algn 0 .text 00000062 08048094 08048094 00000094 2**2 CONTENTS, ALLOC, LOAD, READONLY, CODE 1 .data 00000004 080490f8 080490f8 000000f8 2**2 CONTENTS, ALLOC, LOAD, DATA 2 .comment 00000023 00000000 00000000 000000fc 2**0 CONTENTS, READONLY

链接之前，目标文件中的所有段的VMA都是0。

.text or .data的VMA(虚拟地址)为什么不为0？这涉及操作系统的进程虚拟地址空间的分配原则。

当虚拟空间地址分配完之后，就可以根据所分配的虚拟空间地址确定符号的地址。

2. 符号解析与重定位

pinder@ubuntu:~/program$ objdump -d a.o a.o: file format elf32-i386 Disassembly of section .text: 00000000 <main>: 0: 55 push %ebp 1: 89 e5 mov %esp,%ebp 3: 83 e4 f0 and $0xfffffff0,%esp 6: 83 ec 20 sub $0x20,%esp 9: c7 44 24 1c 64 00 00 movl $0x64,0x1c(%esp) 10: 00 11: c7 44 24 04 /*00 00 00 movl $0x0,0x4(%esp) 18: 00*/ 19: 8d 44 24 1c lea 0x1c(%esp),%eax 1d: 89 04 24 mov %eax,(%esp) 20: e8 /*fc ff ff ff call 21 <main+0x21>*/ 25: c9 leave 26: c3 ret pinder@ubuntu:~/program$ objdump -d ab ab: file format elf32-i386 Disassembly of section .text: 08048094 <main>: 8048094: 55 push %ebp 8048095: 89 e5 mov %esp,%ebp 8048097: 83 e4 f0 and $0xfffffff0,%esp 804809a: 83 ec 20 sub $0x20,%esp 804809d: c7 44 24 1c 64 00 00 movl $0x64,0x1c(%esp) 80480a4: 00 80480a5: c7 44 24 04 /*f8 90 04 movl $0x80490f8,0x4(%esp) 80480ac: 08*/ 80480ad: 8d 44 24 1c lea 0x1c(%esp),%eax 80480b1: 89 04 24 mov %eax,(%esp) 80480b4: e8 /*03 00 00 00 call 80480bc*/ <swap> 80480b9: c9 leave 80480ba: c3 ret 80480bb: 90 nop 080480bc <swap>: 80480bc: 55 push %ebp 80480bd: 89 e5 mov %esp,%ebp 80480bf: 53 push %ebx 80480c0: 8b 45 08 mov 0x8(%ebp),%eax 80480c3: 8b 10 mov (%eax),%edx 80480c5: 8b 45 0c mov 0xc(%ebp),%eax 80480c8: 8b 08 mov (%eax),%ecx 80480ca: 8b 45 08 mov 0x8(%ebp),%eax 80480cd: 8b 18 mov (%eax),%ebx 80480cf: 8b 45 0c mov 0xc(%ebp),%eax 80480d2: 8b 00 mov (%eax),%eax 80480d4: 31 c3 xor %eax,%ebx 80480d6: 8b 45 08 mov 0x8(%ebp),%eax 80480d9: 89 18 mov %ebx,(%eax) 80480db: 8b 45 08 mov 0x8(%ebp),%eax 80480de: 8b 00 mov (%eax),%eax 80480e0: 31 c1 xor %eax,%ecx 80480e2: 8b 45 0c mov 0xc(%ebp),%eax 80480e5: 89 08 mov %ecx,(%eax) 80480e7: 8b 45 0c mov 0xc(%ebp),%eax 80480ea: 8b 00 mov (%eax),%eax 80480ec: 31 c2 xor %eax,%edx 80480ee: 8b 45 08 mov 0x8(%ebp),%eax 80480f1: 89 10 mov %edx,(%eax) 80480f3: 5b pop %ebx 80480f4: 5d pop %ebp 80480f5: c3 ret 

重定位表

pinder@ubuntu:~/program$ objdump -r a.o a.o: file format elf32-i386 RELOCATION RECORDS FOR [.text]: OFFSET TYPE VALUE 00000015 R_386_32 shared 00000021 R_386_PC32 swap

重定位表的结构：

typedef struct { ELf32_Addr r_offset; //重定位入口的偏移 Elf 32_Word r_info; //重定位入口的类型和符号。低8位表示重定位入口的类型，高24位表示重定位入口的符号在符号表中的下标 }Elf32_Rel;

R_366_32        绝对寻址修正  S+A

R_386_PC32  相对寻址修正  S+A-P

C++相关问题

全局构造函数与析构函数

ELF文件定义了两种特殊的段.init, .fini分别存放初始化代码和进程终止代码指令.

最小的程序

不使用库函数

char* str = "Hello world/n"; void print() { asm("movl $13, %%edx /n/t" "movl %0, %%ecx /n/t" "movl $0, %%ebx /n/t" "movl $4, %%eax /n/t" "int $0x80 /n/t" ::"r"(str):"edx", "ecx", "ebx"); } void exit() { asm( "movl $42, %ebx /n/t" "movl $1, %eax /n/t" "int $0x80 /n/t"); } void nomain() { print(); exit(); }