Skip to content

Latest commit

 

History

History
165 lines (114 loc) · 3.82 KB

volatile-and-setjmp.md

File metadata and controls

165 lines (114 loc) · 3.82 KB

14.3.1 Volatile 和 setjmp

编译器认为 setjmp 只是一个函数。然而 setjmp 并不真的是一个函数,其实际上是一个程序能够开始执行的“点”。一般情况下,在 setjmp 执行之前,一些局部变量被缓存在寄存器中(或未被分配内存)。当我们从 longjmp 调用返回到这个"点"时,这些变量不会被恢复。

关闭优化会改变上述行为。所以关闭优化可能会隐藏使用 setjmp 所导致的 bug。

想要正确完成程序的话,记住只有 volatile 类型的局部变量在 longjmp 之后还持有正确的值。这些变量并不像一般的变量一样会被恢复为旧的值,因为 jmp buf 并不保存栈上的变量,而是保存那些 longjmp 之前的变量值。

列表 14-15 展示了一个示例。

Listing 14-15.setjmp_volatile.c

#include <stdio.h>
#include <setjmp.h>

jmp_buf buf;

int main( int argc, char** argv ) {
    int var = 0;
    volatile int b = 0;
    setjmp( buf );
    if (b < 3) {
        b++;
        var ++;
        printf( "\n\n%d\n", var );
        longjmp( buf, 1 );
    }
    return 0; 
}

我们使用 gcc -O0(列表 14-16) 和 gcc -O2(列表 14-17)级别优化分别编译这个程序

Listing 14-16.volatile_setjmp_o0.asm

main:
push     rbp
mov      rbp,rsp
sub      rsp,0x20

; `argc` and `argv` are saved in stack to make `rdi` and `rsi` available
mov    DWORD PTR [rbp-0x14],edi
mov    QWORD PTR [rbp-0x20],rsi

; var = 0
mov    DWORD PTR [rbp-0x4],0x0

;b= 0
mov DWORD PTR [rbp-0x8],0x0

; 0x600a40 is the address of `buf` (a global variable of type `jmp_buf`)
mov    edi,0x600a40
call   400470 <_setjmp@plt>

; if (b < 3), the good branch is executed
; This is encoded by skipping several instructions to the `.endlabel` if b > 2
mov    eax,DWORD PTR [rbp-0x8]
cmp    eax,0x2
jg     .endlabel

; A fair increment
; b++
mov    eax,DWORD PTR [rbp-0x8]
add    eax,0x1
mov    DWORD PTR [rbp-0x8],eax

; var++
add    DWORD PTR [rbp-0x4],0x1

; `printf` call
mov    eax,DWORD PTR [rbp-0x4]
mov    esi,eax
mov    edi,0x400684
; There are no floating point arguments, thus rax = 0
mov    eax,0x0
call   400450 <printf@plt>

; calling `longjmp`
mov    esi,0x1
mov    edi,0x600a40
call   400490 <longjmp@plt>

.endlabel:
Mov    eax,0x0
Leave
ret

在开启优化的情况下,程序输出是:

1

2

3

Listing 14-17.volatile_setjmp_o2.asm

main:

; allocating memory in stack
sub    rsp,0x18

; a `setjmp` argument, the address of `buf`

mov    edi,0x600a40

;b= 0
mov DWORD PTR [rsp+0xc],0x0
; instructions are placed in the order different
; from C statements to make better use of pipeline and other inner ; CPU mechanisms.
call 400470 <_setjmp@plt>

; `b` is read and checked in a fair way
mov    eax,DWORD PTR [rsp+0xc]
cmp    eax,0x2
jle    .branch

; return 0
xor    eax,eax
add    rsp,0x18
ret

.branch:

mov    eax,DWORD PTR [rsp+0xc]

; the second argument of `printf` is var + 1
; It was not even read from memory nor allocated.
; The computations were performed in compile time
mov    esi,0x1

; The first argument of `printf`
mov    edi,0x400674

;b=b+ 1
add eax,0x1
mov DWORD PTR [rsp+0xc],eax

xor eax,eax
call   400450 <printf@plt>

; longjmp( buf, 1 )
mov    esi,0x1
mov    edi,0x600a40
call   400490 <longjmp@plt>

程序的输出将是

1

1

1

volatile 变量 b,行为和预期相符(否则,循环就永远结束不了了)。即使你在程序中对其进行 increment 操作,变量 var 始终还是等于 1。


■Question 265 如何用 setjmp 和 longjmp 实现类似 “try-catch” 式的结构呢?