說(shuō)明:本文來(lái)自翻譯epoll man文檔。
1、ET,LT這兩種事件分發(fā)機(jī)制的不同。我們假定一個(gè)環(huán)境:
1. The file descriptor that represents the read side of a pipe ( RFD ) is added inside the epoll device.
2. Pipe writer writes 2Kb of data on the write side of the pipe.
3. A call to epoll_wait(2) is done that will return RFD as ready file descriptor.
4. The pipe reader reads 1Kb of data from RFD.
5. A call to epoll_wait(2) is done.
Edge Triggered 工作模式:
如果我們?cè)诘?步將RFD添加到epoll描述符的時(shí)候使用了EPOLLET標(biāo)志,那么在第5步調(diào)用epoll_wait(2)之后將有可能會(huì)掛起,因?yàn)槭S嗟臄?shù)據(jù)還存在于文件的輸入緩沖區(qū)內(nèi),而且數(shù)據(jù)發(fā)出端還在等待一個(gè)針對(duì)已經(jīng)發(fā)出數(shù)據(jù)的反饋信息。只有在監(jiān)視的文件句柄上發(fā)生了某個(gè)事件的時(shí)候,ET工作模式才會(huì)匯報(bào)事件(Edge Triggered event distribution delivers events)。因此在第5步的時(shí)候,調(diào)用者可能會(huì)放棄等待仍在存在于文件輸入緩沖區(qū)內(nèi)的剩余數(shù)據(jù)(the caller might end up waiting for some data that is already present inside the input buffer.)。在上面的例子中,會(huì)有一個(gè)事件產(chǎn)生在RFD句柄上,是因?yàn)樵诘?步執(zhí)行了一個(gè)寫(xiě)操作,然后,事件將會(huì)在第3步被銷毀(consumed)。因?yàn)榈?步的讀取操作沒(méi)有讀空文件輸入緩沖區(qū)內(nèi)的數(shù)據(jù),因此我們?cè)诘?步調(diào)用 epoll_wait(2)完成后,might lock indefinitely。epoll工作在ET模式的時(shí)候,必須使用非阻塞套接口,以避免由于一個(gè)文件句柄的阻塞讀/阻塞寫(xiě)操作把處理多個(gè)文件描述符的任務(wù)餓死。最好以下面的方式調(diào)用ET模式的epoll接口,在后面會(huì)介紹避免可能的缺陷。
i 基于非阻塞文件句柄
ii 只有當(dāng)read(2)或者write(2)返回EAGAIN時(shí)才需要掛起、等待(意為此時(shí),緩存區(qū)滿或無(wú)數(shù)據(jù))。
Level Triggered 工作模式
相反的,以LT方式調(diào)用epoll接口的時(shí)候,它就相當(dāng)于一個(gè)速度比較快的poll,and can be used wherever the latter is used since it shares the same semantics。因?yàn)榧词故褂肊T模式的epoll,在收到多個(gè)chunk的數(shù)據(jù)的時(shí)候仍然會(huì)產(chǎn)生多個(gè)事件(Since even with the Edge Triggered epoll multiple events can be generated up on receival of multiple chunks of data)。the caller has the option to specify the EPOLLONESHOT flag, to tell epoll to disable the associated file descriptor after the receival of an event with epoll_wait(2). When the EPOLLONESHOT flag is specified, it is caller responsibility to rearm(重新設(shè)置) the file descriptor using epoll_ctl(2) with EPOLL_CTL_MOD.
2、While the usage of epoll when employed(使用) like a Level Triggered interface does have the same semantics of poll(2), an Edge Triggered usage requires more clarification(澄清) to avoid stalls(拖延) in the application event loop.
In this example, listener is a non-blocking socket on which listen(2) has been called. The function do_use_fd() uses the new ready file descriptor until EAGAIN is returned by either read(2) or write(2). An event driven state machine(事件驅(qū)動(dòng)狀態(tài)機(jī)) application should, after having received EAGAIN, record its current state so that at the next call to do_use_fd() it will continue to read(2) or write(2) from where it stopped before.
示例代碼
View Code

struct epoll_event ev, *events;

for(;;)
{
nfds = epoll_wait(kdpfd, events, maxevents, -1);

for(n = 0; n < nfds; ++n)
{

if(events[n].data.fd == listener)
{
client = accept(listener, (struct sockaddr *) &local,
&addrlen);

if(client < 0)
{
perror("accept");
continue;
}
setnonblocking(client);
ev.events = EPOLLIN | EPOLLET;
ev.data.fd = client;

if (epoll_ctl(kdpfd, EPOLL_CTL_ADD, client, &ev) < 0)
{
fprintf(stderr, "epoll set insertion error: fd=%d0,
client);
return -1;
}
}
else
do_use_fd(events[n].data.fd);
}
}


When used as an Edge triggered interface, for performance reasons, it is possible to add the file descriptor inside the epoll interface ( EPOLL_CTL_ADD ) once by specifying ( EPOLLIN|EPOLLOUT ). This allows you to avoid continuously switching between EPOLLIN and EPOLLOUT calling epoll_ctl(2) with EPOLL_CTL_MOD.
水平觸發(fā)(LT, Level Triggered),默認(rèn)方式
支持阻塞/非阻塞socket。
內(nèi)核通知某fd就緒,如果不對(duì)fd操作,內(nèi)核會(huì)繼續(xù)通知
邊緣觸發(fā)(Edge-Triggered)
只支持非阻塞socket
內(nèi)核通知某fd就緒,如果不對(duì)fd操作,內(nèi)核不再繼續(xù)通知