Question

我遇到以下代码问题：

#define NTHREADS 100

int main(int argc, const char * argv[]) {
    pthread_t threads[NTHREADS];

    // ...

    // Create threads
    for (size_t i = 0; i < NTHREADS; ++i) {
        pthread_create(threads + i, NULL, thread, NULL);
    }

    debugln("All %u threads are ready", NTHREADS);

    // ...

    for (size_t i = 0; i < NTHREADS; ++i) {
        pthread_join(threads[i], NULL);
    }

    debugln("All threads terminated");

    return 0;
}

debugln的定义如下：

#define debugln(_fmt, _args...) fprintf(stderr, _fmt "\n", ##_args)

问题：

执行NTHREADS设置为100的程序时，valgrind会多次输出此错误：

==2780== Invalid write of size 4
==2780==    at 0x50976A2: vfprintf (vfprintf.c:246)
==2780==    by 0x509CE53: buffered_vfprintf (vfprintf.c:2313)
==2780==    by 0x509788D: vfprintf (vfprintf.c:1316)
==2780==    by 0x50A24E6: fprintf (fprintf.c:33)
==2780==    by 0x400ACA: main (in /tmp/cbuild_1413987921)
==2780==  Address 0x7feffd158 is on thread 1's stack

但是当我将debugln更改为

时

#define debugln(_fmt, _args...) printf(_fmt "\n", ##_args)

然后错误就消失了。

我已经尝试flockfile＆amp; funlockfile并增加stacksize。两者都没有修复错误。 BTW valgrind --version提供valgrind-3.7.0

修改

更改

时错误消失

#define NTHREADS 100

到

const unsigned NTHREADS = 100;

有人可以解释一下吗？

fprintf生成valgrind错误

0 个答案: