如何为每个具有该类型的结构变量观察结构的成员？

Question

我正在腐蚀结构成员,其中struct用于构建一个大树.我正在尝试设置一个观察点,但"变量"有各种名称(我见过的所有例子都使用显式全局变量,所以不要显示我正在寻找的例子).

(gdb) watch *(OP*)->op.sibling
A syntax error in expression, near `->op.sibling'.
(gdb) watch *(struct op*)->op.sibling
A syntax error in expression, near `->op.sibling'.
(gdb) watch (struct op*)->op.sibling
A syntax error in expression, near `->op.sibling'.

最接近我接受的语法不起作用

(gdb) watch (struct op*)o->op.sibling
No symbol "o" in current context.

是否存在一个表达形式,我错过了与变量名称无关的表达形式,并注意到它是一种特殊的结构？

这是否能够检测到结构的memset覆盖？(即使不是,它会帮助我排除一些原因)

是否有超出基本用法的gdb引用？(对于某些"基本"的定义)

Answer 1

正如ninjalj所说,你必须有一个地址值得关注; 这就是观察点的重点.请考虑以下示例:

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

typedef struct node {
  long value;
  struct node *next;
} node;

node *build_node(int x, node *next)
{
  node *n = malloc(sizeof(*n));
  n->value = x;
  n->next = next;
  return n;
}

int main()
{
  node *n = build_node(1, NULL);
  n = build_node(2, n);
  n = build_node(3, n);

  memset(&n->next->value, 0xFF, sizeof(long) + 3);  // corrupt the list

  return 0;
}

在这里,我生成一个整数列表,然后腐蚀它.让我们看看GDB中的内容:

gdb -q ./a.out
Reading symbols for shared libraries .. done
(gdb) b main
Breakpoint 1 at 0x100000e0c: file foo.c, line 20.
(gdb) r
Starting program: /tmp/a.out 
Reading symbols for shared libraries +. done

Breakpoint 1, main () at foo.c:20
20    node *n = build_node(1, NULL);

(gdb) b 26
Breakpoint 2 at 0x100000e8d: file foo.c, line 26.
(gdb) c
Continuing.

Breakpoint 2, main () at foo.c:26
26    return 0;
(gdb) p *n
$1 = {
  value = 3, 
  next = 0x100100090
}
(gdb) p *n.next
$2 = {
  value = -1, 
  next = 0x100ffffff
}

在这里,我们清楚地看到,n->next已彻底腐败.假设我们不知道发生了什么,并希望通过GDB观察点找到答案.

首先,我们需要建立已损坏的地址:

(gdb) print &n.next.value
$3 = (long int *) 0x100100090
(gdb) watch *$3
Hardware watchpoint 3: *$3

在这里,我只需在地址0x100100090上设置一个8字节的观察点.

(gdb) run
The program being debugged has been started already.
Start it from the beginning? (y or n) y
Starting program: /private/tmp/a.out 
warning: Could not set watchpoint 3
warning: Could not set watchpoint 3

Breakpoint 1, main () at foo.c:20
20    node *n = build_node(1, NULL);

我使用的是旧版本的GDB,它不了解如何在程序重启期间正确禁用和重新启用硬件观察点.如果您使用的是更新版本,则可能看不到上述警告.当我在断点1处停下时,我可以简单地重新启用观察点:

(gdb) enable 3
(gdb) c
Continuing.
Hardware watchpoint 3: *$3

Old value = 0
New value = 2
build_node (x=2, next=0x100100080) at foo.c:14
14    n->next = next;

好的,我们已达到预期的点和预期值.下一次修改会破坏它,我们想知道发生了什么.

(gdb) c
Continuing.
Hardware watchpoint 3: *$3

Old value = 2
New value = 255
0x00007fff82fae450 in memset ()
(gdb) bt
#0  0x00007fff82fae450 in memset ()
#1  0x0000000100000ebe in __inline_memset_chk (__dest=0x100100090, __val=255, __len=11) at _string.h:80
#2  0x0000000100000e8d in main () at foo.c:24

Voilà,您现在知道意外修改的来源.