linux下程序运行几天莫名其妙宕机了，不能还原现场，找到宕机原因就很无语了。

一个解决办法是使用core文件，但是对于大型服务器文件，动辄几百M的core文件是在有点伤不起，于是想到程序宕机时自动打印调用堆栈。简单实用。

废话不多说，直接上方案：

方案1：使用gdb指令列表文件启动程序并监控之

启动指令 gdb -x gdb_start.ini

以下是gdb_start.ini文件内容：

file MyApplication

set pagination off

shell rm ./logs/gdb_crash.log

set logging file ./logs/gdb_crash.log

set logging on

handle SIG32 nostop noprint

handle SIGPIPE nostop noprint

handle SIGSEGV stop

handle SIGFPE stop

handle SIGILL stop

handle SIGABRT stop

handle SIGSYS stop

r 6660

print "crash time:"

shell date >> ./logs/gdb_crash.log

print "crash frame:"

info f

print "crash locals:"

info locals

print "crash callstack:"

bt

set logging off

quit

这样程序宕机时将在gdb_crash.log中记录以下信息，太完美了：

Program received signal SIGFPE, Arithmetic exception.

0x0804874a in make_crash (nCrashType=1) at crash.c:50

50 z = 1/z;

$1 = "crash time:"

2014年 08月 13日 星期三 15:47:37 CST

$2 = "crash frame:"

Stack level 0, frame at 0xbf9e7900:

eip = 0x804874a in make_crash (crash.c:50); saved eip 0x8048911

called by frame at 0xbf9e7930

source language c.

Arglist at 0xbf9e78f8, args: nCrashType=1

Locals at 0xbf9e78f8, Previous frame's sp is 0xbf9e7900

Saved registers:

ebp at 0xbf9e78f8, edi at 0xbf9e78f4, eip at 0xbf9e78fc

$3 = "crash locals:"

z = 0

__PRETTY_FUNCTION__ = "make_crash"

$4 = "crash callstack:"

#0  0x0804874a in make_crash (nCrashType=1) at crash.c:50

#1  0x08048911 in test_crash () at crash.c:105

#2  0x08048930 in main () at crash.c:119

方案2：程序内设置宕机监控代码

#include [stdio.h]

#include [stdlib.h]

#include [signal.h]

#include [string.h]

#include[unistd.h]

void crash_dump(int signo)

{

char buf[1024];

char cmd[1024];

FILE *fh;

printf("crash detected: crash_dump(%d)", signo);

snprintf(buf, sizeof(buf), "/proc/%d/cmdline", getpid());

if(!(fh = fopen(buf, "r")))

exit(0);

if(!fgets(buf, sizeof(buf), fh))

exit(0);

fclose(fh);

if(buf[strlen(buf) - 1] == '/n')

buf[strlen(buf) - 1] = '/0';

snprintf(cmd, sizeof(cmd), "date > ./crash.log", buf, getpid());

printf(cmd);

system(cmd);

snprintf(cmd, sizeof(cmd), "gdb %s %d -ex=bt >> ./crash.log", buf, getpid());

printf(cmd);

system(cmd);

exit(-1);

}

int main(int argc, char* argv[])

{

//宕机输出调用堆栈

signal(SIGSEGV, &crash_dump);

signal(SIGFPE, &crash_dump);

signal(SIGINT, &crash_dump);

signal(SIGILL, &crash_dump);

signal(SIGABRT, &crash_dump);

//real code...

return 0;

}

参考链接：

[1]http://www.cppblog.com/sunicdavy/archive/2012/12/29/196809.html

[2]http://blog.csdn.net/cindy9902/article/details/6146816

[3]http://blog.csdn.net/ifengle/article/details/3849783

[4]http://sourceware.org/gdb/wiki/FAQ#How_do_I_disable_the_.22Type_.3Creturn.3E_to_continue.2C_or_q_.3Creturn.3E_to_quit.22_pagination_prompt_in_GDB.3F