gnu gdb

如果编译时没有-g参数，则不能实现基于C源码的符号调试功能，
看不到程序的函数名、变量名，显示的全是运行时的内存地址。

strip - Discard symbols from object files.
如果一个目标被strip了的话，在去dbg他，就不好了
一些gdb命令执行会抱怨没有符号表。
Reading symbols from /root/pgm...(no debugging symbols found)...done.
(gdb) b 20
No symbol table is loaded.  Use the "file" command.

即使带了-g，没有strip，源码文件是需要在位的，否则
Reading symbols from /root/td...done.
(gdb) l
15      td.c: No such file or directory.

条件满足时，我们来dbg它。l看一下代码

22      int main (int argc, char **argv) {
23          char buf[BS], clear[BS];
24          struct ip *iph = (struct ip *) buf;

然后b main设置断点 再run起来

(gdb) b main
Breakpoint 1 at 0x409ad7: file td.c, line 24.
(gdb) r
Starting program: /root/td

Breakpoint 1, main (argc=1, argv=0x7fffffffe5b8) at td.c:24
24          struct ip *iph = (struct ip *) buf;
Missing separate debuginfos, use: debuginfo-install glibc-2.17-317.el7.x86_64

//然后再打一个断点，想继续运行到断点，但是却退出了。因为中间执行了fork
(gdb) b 65
Breakpoint 2 at 0x409c07: file td.c, line 65.
(gdb) c
Continuing.
[Detaching after fork from child process 2615]
[Inferior 1 (process 2604) exited normally]

//没关系，可以再attach上去子进程2615
//也可以再开始时就设置跟踪子进程 set follow-fork-mode child

(gdb) set follow-fork-mode child
(gdb) b 65
Breakpoint 2 at 0x409c07: file td.c, line 65.
(gdb) c
Continuing.
[Attaching after process 3687 fork to child process 3687]
[New inferior 2 (process 3687)]
[Detaching after fork from parent process 3682]
[Inferior 1 (process 3682) detached]
[Switching to process 3687]

Breakpoint 2, main (argc=1, argv=0x7fffffffe5a8) at td.c:65
65              FD_ZERO (&rfds);

有些程序是要带参数启动的，要这样使用-args参数来说明，如果不用，gdb把后面也看做要调试的程序
gdb -args ./mypgm -h xx -p yy

//看变量值的时候，可以指定想要的格式，比如p/c opt
(gdb) p opt
$2 = 104
(gdb) p/x opt
$4 = 0x68
(gdb) p /c opt
$5 = 104 'h'

/*可以执行一个shell命令，如 shell date
都是单步调试，二者区别；s 是step into 的意思；n 是 step over

变量的类型是什么？*/
(gdb) whatis opt
type = int

//修改变量的值。使用set var更好一下，可避免变量名和gdb自己的参数重名导致的意外情况
(gdb) print opt='h'
$4 = 104
(gdb) set var opt='h'

/*当ip是一个指针char *时，怎么查看；直接查看是没有问题的，但是当加上一些格式后，就要注意客体
涉及展示长度的选项 show print elements*/
(gdb) p /c ip
$12 = 240 '\360'
(gdb) p /x ip
$13 = 0x7fffffffe7f0
(gdb) whatis ip
type = char *
(gdb) p ip
$14 = 0x7fffffffe7f0 "h1"
(gdb) p/c *ip
$16 = 104 'h'

//不想在这个函数里纠缠了，继续执行完函数
(gdb) finish
Run till exit from #0  isip (ip=0x7fffffffe7f0 "h1") at ip.c:40
0x0000000000407a6e in resolve (host=0x7fffffffe7f0 "h1") at ip.c:20
20          if (isip (host))
Value returned is $17 = 1

//查看结构体
(gdb) p udh
$1 = (struct udp *) 0x7fffffffa3a4
(gdb) p *udh
$2 = {src = 8693, dst = 11631, len = 0, sum = 42922}
(gdb) set print pretty on
(gdb) p *udh
$3 = {
  src = 8693,
  dst = 11631,
  len = 0,
  sum = 42922
}
(gdb) ptype udh
type = struct udp {
    u16 src;
    u16 dst;
    u16 len;
    u16 sum;
} *

//看某变量或类型的大小
(gdb) p sizeof(unsigned long)
$19 = 8

一些其他的
info b/th (break thread)
bt 函数调用栈，当前在哪里
help 层层递进的帮助
记得开头不记得后续，或不想打时用TAB键
info locals 看本地变量，全部的

其他的其他
输入h1，用sscanf解析输入（预期点分数字的IP形式），然后判断输入是否为IP（错误的变成了0.0.0.0）；
如果是IP，用inet_addr()来转换成binary data in network byte order（这个函数本就有缺陷，返回-1也是0xffffffff）
最终将h1错误的变成255.255.255.255

一个疑问，.h文件修改了，make没有检测到变化，因此没有进行重新编译

if (select (usock + 1, &rfds, NULL, NULL, NULL) < 1)

第1个参数 nfds is the highest-numbered file descriptor in any of the three sets, plus 1.

fd_set *readfds, fd_set *writefds, fd_set *exceptfds
Three independent sets of file descriptors are watched. 不要watch的就设置NULL

if (FD_ISSET (tsock, &rfds))

FD_ZERO()  clears  a  set.
FD_SET() and FD_CLR() respectively add and remove a given file descriptor from a set.
FD_ISSET() tests to see if a file descriptor is part of the set; this is useful after select() returns.