c++ - 多线程下对一个串多次调用strok后strlen报段错误

Question

while(1)
{
    char buf[800] ={0};
    ret=fgets(buf, sizeof(buf) -1, fp);
    if(!ret){ 
    ...lseek to head of file
    continue;
    }
    //问题很可能出在这里，改成strtok_r还是会同样出错
    char *url=strtok(buf,"\n");
    index++;
    thread=&stk_threads[index];
    stk_thread_push_task(thread, url);

    if(++index==max_thread)
    {
       index=0;
    }
}

先贴代码如上，问题如下
程序在对一个文件遍历一段时间（我确定还没遇到过尾部）之后，在push到选定thread的队列时【主程序会用strlen判断url处（来源指针）的字串长度，从而拷贝一份出来，防止各线程读到重复的url内容】在strlen时候报段错误。

我知道strtok的实现原理，但是不明白这个段错误在这个情形下发生的具体原因是什么呢？【注意，执行strtok和push到选定线程以及调用strlen拷贝源串的都是主线程自己】

Answer 1

url points to a global shared memory. Before your thread processes it, the pointers pushed into the queue have been destroyed by the next strtok call. If these pointers are accessed again, a segfault will occur.

Just use strtok_r instead.

Answer 2

My understanding is: char buf[800] is allocated on the stack space of the main thread, and after each strtok, url points to the address of buf. So for the child thread, there is here Cross-thread Stack Access.

https://software.intel.com/sites/products/documentation/hpc/inspectorxe/en-us/2011Update/lin/ug_docs/GUID-F0B0D34B-6691-4033-931F-4BBBD1A1F791.htm

strtok_r To no avail. Try to copy the content of the url in the heap each time, pass it to the child thread, and then delete it.
Of course, a better way is to allocate corresponding memory to the url in the thread data structure.