poll--wait for some event on a file descriptor

poll同select，用于监控file descriptor事件，推荐用poll的升级版epool来实现功能，但在简单应用中使用poll更方便。

#include <poll.h>
int poll(struct pollfd *fds, nfds_t nfds, int timeout);
struct pollfd {
　　int fd;
　　short events;
　　short revents;
};

fds是文件描述符集的数组，nfds指定数组元素个数。

pollfd中fd是打开文件的文件描述符；events是输入参数，指定监控事件；revents是输出参数，被内核填充，只是实际发生的事件。

timeout指定超时毫秒数（milliseconds）。

成功返回正数，表示非0revents的fd个数，0表示超时，错误返回-1并设置errno。

On success, a positive number is returned; this is the number of structures which have  nonzero  revents  fields  
(in other  words,  those descriptors with events or errors reported).  
A value of 0 indicates that the call timed out and no file descriptors were ready.  
On error, -1 is returned, and errno is set appropriately.

应用举例

监控一个网络socket是否发生错误（刚链接时）。

int usock_wait_ready(int fd, int msecs) {
    struct pollfd fds[1];
    int res;

    fds[0].fd = fd; 
    fds[0].events = POLLOUT;

    res = poll(fds, 1, msecs);
    if (res < 0) {
        return errno;
    } else if (res == 0) {
        return -ETIMEDOUT;
    } else {
        int err = 0;
        socklen_t optlen = sizeof(err);

        res = getsockopt(fd, SOL_SOCKET, SO_ERROR, &err, &optlen);
        if (res)
            return errno;
        if (err)
            return err;
    }

    return 0;
}