Linux下利用条件变量实现读写锁

首先介绍下pthread_cond_t。 在Linux下称之为状态变量，与之相关的有下面几个API:
    int pthread_cond_init (pthread_cond_t *COND,pthread_condattr_t *cond_ATTR);
    int pthread_cond_signal (pthread_cond_t *COND);
    int pthread_cond_broadcast (pthread_cond_t *COND);
    int pthread_cond_wait (pthread_cond_t *COND, pthread_mutex_t *MUTEX);
    int pthread_cond_timedwait (pthread_cond_t *COND, pthread_mutex_t *MUTEX, const struct timespec *ABSTIME);
    int pthread_cond_destroy (pthread_cond_t *COND);
这里就讲下2个api，pthread_cond_signal和pthread_cond_wait，一般的用法如下：
{
    pthread_mutex_lock(lock)
...
    pthread_cond_wait(cond, lock);
...
    pthread_cond_mutex_unlock(lock);
}
pthread_cond_wait会解锁lock，然后在cond上等待,这两步是atomic operation. 当pthread_cond_wait返回时，会同时对lock上锁. 这里我的理解是，如果获取不到lock锁，即使cond已经被激活，pthread_cond_wait依然不会返回。
{
    pthread_mutex_lock(lock);
...
    pthread_cond_signal(cond);
...
    pthread_mutex_unlock(lock);
}
pthread_cond_signal调用之前一定会先拿到lock锁。pthread_cond_signal不会去管lock锁，只是将cond激活，接下去释放lock锁。这时候pthread_cond_wait就可以得到lock从而返回了。
言归正传，下面是利用条件变量实现的一个读写锁的例子：
typedef struct pthread_rwlock
{
int active_readers; /* -1 when writer lock locked, >0 when read lock locked */
int pending_readers;
int pending_writers;
pthread_mutex_t mutex;
pthread_cond_t ok_to_read;
pthread_cond_t ok_to_write;
} pthread_rwlock_t ;
int pthread_rwlock_init(pthread_rwlock_t * lock, pthread_rwlockattr_t *attr)
{
active_readers = 0;
pending_readers = 0;
pending_writers = 0;
pthread_mutex_init(&lock->mutex, NULL);
pthread_cond_init(&lock->ok_to_read, NULL);
pthread_cond_init(&lock->ok_to_write, NULL)