小编Ced*_*lja的帖子

http://en.cppreference.com/w/cpp/atomic/memory_order和其他C++ 11在线参考,将memory_order_acquire和memory_order_release定义为:

• 获取操作:在此加载之前,不能对当前线程中的读取进行重新排序.
• 释放操作:在此存储之后,当前线程中的写入不能重新排序.

这似乎允许在获取操作之前执行获取后写入,这看起来很奇怪(通常的获取/释放操作语义限制所有内存操作的移动).

相同的在线资源(http://en.cppreference.com/w/cpp/atomic/atomic_flag)表明可以使用C++原子和上面提到的宽松内存排序规则构建自旋锁互斥:

lock mutex: while (lock.test_and_set(std::memory_order_acquire))

unlock mutex: lock.clear(std::memory_order_release);               

有了这个锁定/解锁的定义,如果确实以这种方式定义了memory_order_acquire/release,那么下面的简单代码就不会被破坏(即,不禁止对获取后写入进行重新排序):

Thread1:
  (0) lock
    (1) x = 1;
    (2) if (x != 1) PANIC
  (3) unlock

Thread2:
  (4) lock
    (5) x = 0;
  (6) unlock

以下执行是否可行:(0)锁定,(1)x = 1,(5)x = 0,(2)PANIC？我错过了什么？

我试图使用LD_PRELOAD机制插入对pthread_cond_broadcast的调用.我插入的pthread_cond_broadcast函数只调用原始的pthread_cond_broadcast.但是,对于一个非常简单的pthread代码,其中pthread_cond_wait和pthread_cond_broadcast都被调用,我最终在glibc(对于glibc 2.11.1)或程序​​挂起(对于glibc 2.15)中有一个段错误.关于这方面的任何线索都在发生？

插入代码(编译为共享库):

#define _GNU_SOURCE
#include <stdio.h>
#include <stdlib.h>
#include <pthread.h>
#include <dlfcn.h>

static int (*orig_pthread_cond_broadcast)(pthread_cond_t *cond) = NULL;

__attribute__((constructor))
static void start() {
    orig_pthread_cond_broadcast =
        (int (*)()) dlsym(RTLD_NEXT, "pthread_cond_broadcast");
    if (orig_pthread_cond_broadcast == NULL) {
        printf("pthread_cond_broadcast not found!!!\n");
        exit(1);
    }
}

__attribute__((__visibility__("default")))
int pthread_cond_broadcast(pthread_cond_t *cond) {
    return orig_pthread_cond_broadcast(cond);
}

简单的pthread程序:

#include <stdio.h>
#include <pthread.h>
#include <unistd.h>

pthread_mutex_t cond_mutex;
pthread_cond_t cond_var;
int condition;

void *thread0_work(void *arg) {
    pthread_mutex_lock(&cond_mutex);
    printf("Signal\n");
    condition = 1;
    pthread_cond_broadcast(&cond_var);
    pthread_mutex_unlock(&cond_mutex);
    return NULL;
}

void *thread1_work(void *arg) …