Linux内核调试方法总结之调试宏
本文介绍的内核调试宏属于静态调试方法,通过调试宏主动触发oops从而打印出函数调用栈信息。
1) BUG_ON 查看bug处堆栈内容,主动制造oops
Linux中BUG_ON,WARN_ON用于调试,比如
#define BUG_ON(condition) do { /
if (unlikely((condition)!=0)) /
BUG(); /
} while(0)
如果觉得该condition下是一个BUG,可以添加此调试信息,查看对应堆栈内容
具体的BUG_ON最终调用__bug
__bug
{
*(int*)0=0;
}
从而地址非法访问,例如在你的驱动中调用BUG_ON(1),dmesg会出现下面类似信息:
[ 19.360000] Unable to handle kernel NULL pointer dereference at virtual address 00000000
[ 19.360000] pgd = c0004000
[ 19.360000] [00000000] *pgd=00000000
函数的调用流程为
fault.c
__do_kernel_fault--------[ 19.360000] Unable to handle kernel NULL pointer dereference at virtual address 00000000
traps.c
die--->__die--->__show_regs / dump_mem
2) WARN_ON 调用dump_stack打印堆栈信息,不会oops
#define WARN_ON(condition) do{/
if(unlikely((condition) != 0)) {/
printk(“Badness in %s at %s:%d\n”, __FUNCTION__, __FILE__, __LINE__);/
dump_stack();/
}
}while(0)
3) BUILD_BUG_ON 编译调试宏
#define BUILD_BUG_ON(condition) ((void)sizeof(char[1 - 2*!!(condition)]))
BUILD_BUG_ON宏中的condition如果为真就会报错。假设条件为真,则BUILD_BUG_ON变化为:(void) sizeof (char[-1]),这种语法是在难理解,因为char[-1]本来就是非法的。结果导致编译错误。
例子:
#define condition 0
static int __init main_init(void)
{
printk("in %s function\n", __func__);
BUILD_BUG_ON(condition);/*if the macro "confition" is not zero, the program can not be compiled to success. so the BUILD_BUG_ON used for the build of program*/
WARN_ON(!condition);/*if the macro "confition" is not zero, there will dump the satck information for this program. so the WARN_ON used for debugof program*/
BUG_ON(condition); /*if the macro "confition" is not zero, The kernel will occur an Oops errro "Unable to handle kernel NULL pointer dereference at virtual address 00000000". so the BUG_ON used for ending the bug program*/
return 0;
}