原文:https://blog.csdn.net/u011068702/article/details/62881315
1、epoll介绍
epoll是在2.6内核中提出的,是之前的select和poll的增强版本。相对于select和poll来说,epoll更加灵活,没有描述符限制。epoll使用一个文件描述符管理多个描述符,将用户关系的文件描述符的事件存放到内核的一个事件表中,这样在用户空间和内核空间的copy只需一次。
2、epollAPI
epoll操作过程需要三个接口,分别如下:
#include <sys/epoll.h>
int epoll_create(int size);
int epoll_ctl(int epfd, int op, int fd, struct epoll_event *event);
int epoll_wait(int epfd, struct epoll_event * events, int maxevents, int timeout);
(1) int epoll_create(int size);
创建一个epoll的句柄,size用来告诉内核这个监听的数目一共有多大。这个参数不同于select()中的第一个参数,给出最大监听的fd+1的值。需要注意的是,当创建好epoll句柄后,它就是会占用一个fd值,在linux下如果查看/proc/进程id/fd/,是能够看到这个fd的,所以在使用完epoll后,必须调用close()关闭,否则可能导致fd被耗尽。
(2)int epoll_ctl(int epfd, int op, int fd, struct epoll_event *event);
epoll的事件注册函数,它不同与select()是在监听事件时告诉内核要监听什么类型的事件epoll的事件注册函数,而是在这里先注册要监听的事件类型。第一个参数是epoll_create()的返回值,第二个参数表示动作,用三个宏来表示:
EPOLL_CTL_ADD:注册新的fd到epfd中;
EPOLL_CTL_MOD:修改已经注册的fd的监听事件;
EPOLL_CTL_DEL:从epfd中删除一个fd;
注意这里EPOLL_CTL_ADD和EPOLL_CTL_MOD和EPOLL_CTL_DEL都是整形数据,他们的值分别是1,2, 3。读者可以自己测试。
第三个参数是需要监听的fd,第四个参数是告诉内核需要监听什么事,struct epoll_event结构如下:
1. struct epoll_event
结构体epoll_event被用于注册所感兴趣的事件和回传所发生待处理的事件,定义如下:
typedef union epoll_data {
void *ptr;
int fd;
__uint32_t u32;
__uint64_t u64;
} epoll_data_t;//保存触发事件的某个文件描述符相关的数据
struct epoll_event {
__uint32_t events; /* epoll event */
epoll_data_t data; /* User data variable */
};
events可以是以下几个宏的集合:
EPOLLIN :表示对应的文件描述符可以读(包括对端SOCKET正常关闭);
EPOLLOUT:表示对应的文件描述符可以写;
EPOLLPRI:表示对应的文件描述符有紧急的数据可读(这里应该表示有带外数据到来);
EPOLLERR:表示对应的文件描述符发生错误;
EPOLLHUP:表示对应的文件描述符被挂断;
EPOLLET: 将EPOLL设为边缘触发(Edge Triggered)模式,这是相对于水平触发(Level Triggered)来说的。
EPOLLONESHOT:只监听一次事件,当监听完这次事件之后,如果还需要继续监听这个socket的话,需要再次把这个socket加入到EPOLL队列里
(3) int epoll_wait(int epfd, struct epoll_event * events, int maxevents, int timeout);
等待事件的产生,类似于select()调用。参数events用来从内核得到事件的集合,maxevents告之内核这个events有多大,参数timeout是超时时间(毫秒,0会立即返回,-1将不确定,也有说法说是永久阻塞)。该函数返回需要处理的事件数目,如返回0表示已超时。
3、工作模式
epoll对文件描述符的操作有两种模式:LT(level trigger)和ET(edge trigger)。LT模式是默认模式,LT模式与ET模式的区别如下:
LT模式:当epoll_wait检测到描述符事件发生并将此事件通知应用程序,应用程序可以不立即处理该事件。下次调用epoll_wait时,会再次响应应用程序并通知此事件。
ET模式:当epoll_wait检测到描述符事件发生并将此事件通知应用程序,应用程序必须立即处理该事件。如果不处理,下次调用epoll_wait时,不会再次响应应用程序并通知此事件。
ET模式在很大程度上减少了epoll事件被重复触发的次数,因此效率要比LT模式高。epoll工作在ET模式的时候,必须使用非阻塞套接口,以避免由于一个文件句柄的阻塞读/阻塞写操作把处理多个文件描述符的任务饿死。
4、源码实现
服务端
#include <stdio.h>
#include <unistd.h>
#include <sys/socket.h>
#include <string.h>
#include <stdlib.h>
#include <sys/epoll.h>
#include <netinet/in.h>
#include <arpa/inet.h>
#include <sys/types.h>
#include <errno.h>
#define IPADDRESS "127.0.0.1"
#define PORT 8888
#define MAXSIZE 1024
#define LISTENQ 5
#define FDSIZE 1000
#define EPOLLEVENTS 100
//bind socket
static int socket_bind(const char* ip, int port);
//io epoll
static void do_epoll(int listenfd);
//deal with events
static void handle_events(int epollfd, struct epoll_event *event, int num, int listenfd, char *buf);
//deal with accpet events
static void handle_accpet(int epolled, int listenfd);
//deal with read
static void do_read(int epollfd, int fd, char *buf);
//deal with write
static void do_write(int epollfd, int fd, char *buf);
//change event
static void change_event(int epolled, int fd, int state, int op);
int main()
{
printf("EPOLL_CTL_ADD is %d\n", EPOLL_CTL_ADD);
printf("EPOLL_CTL_DEL is %d\n", EPOLL_CTL_DEL);
printf("EPOLL_CTL_MOD is %d\n", EPOLL_CTL_MOD);
int listenfd;
listenfd = socket_bind(IPADDRESS, PORT);
listen(listenfd, LISTENQ);
do_epoll(listenfd); //while 结构
puts("epoll server end");
return 0;
}
//bind socket
static int socket_bind(const char* ip, int port)
{
int listenfd;
struct sockaddr_in serveraddr;
listenfd = socket(AF_INET, SOCK_STREAM, 0);
if (listenfd == -1)
{
perror("create socket failed\n");
exit(EXIT_FAILURE);
}
//please do not forget bzero serveraddr
bzero(&serveraddr, sizeof(serveraddr));
//init serveraddr
serveraddr.sin_family = AF_INET;
inet_pton(AF_INET, ip, &serveraddr.sin_addr);
//please do not forget this is htons(port). not is port
serveraddr.sin_port = htons(port);
if (bind(listenfd, (struct sockaddr*)&serveraddr, sizeof(serveraddr)) == -1)
{
perror("bind failed");
exit(EXIT_FAILURE);
}
return listenfd;
}
//io epoll
static void do_epoll(int listenfd)
{
int epollfd;
struct epoll_event events[EPOLLEVENTS];
int result;
char buf[MAXSIZE];
memset(buf, 0, MAXSIZE);
//函数原型:int epollfd = epoll_create(int size);
//函数解释:创建一个epoll的句柄,size用来告诉内核这个监听的数目一共有多大
epollfd = epoll_create(FDSIZE);
//add listen listenfd event
change_event(epollfd, listenfd, EPOLLIN, EPOLL_CTL_ADD);
while (1)
{
//函数原型: int epoll_wait(int epfd, struct epoll_event * events, int maxevents, int timeout);
//函数解释:等待事件的产生,类似于select()调用
//参数events用来从内核得到事件的集合,maxevents告之内核这个events有多大
//参数timeout是超时时间(毫秒,0会立即返回,-1将不确定,也有说法说是永久阻塞)
//该函数返回需要处理的事件数目,如返回0表示已超时。
result = epoll_wait(epollfd, events, EPOLLEVENTS, -1);
printf("need deal with events is %d\n", result);
handle_events(epollfd, events, result, listenfd, buf);
}
close(epollfd);
}
//deal with events
static void handle_events(int epollfd, struct epoll_event *events, int num, int listenfd, char *buf)
{
int i;
int fd;
//foreach
for (i = 0; i < num; ++i)
{
fd = events[i].data.fd;
//deal with events by fd
if ((fd == listenfd) && (events[i].events & EPOLLIN)) //表示在监听描述符上有数据 三次握手accept
{
puts("has go in handle_accpet");
handle_accpet(epollfd, listenfd);
}
else if (events[i].events & EPOLLIN)
{
puts("do read");
do_read(epollfd, fd, buf);
}
else if (events[i].events & EPOLLOUT)
{
puts("do write");
do_write(epollfd, fd, buf);
}
}
}
//deal with accpet events
static void handle_accpet(int epolled, int listenfd)
{
int clifd;
struct sockaddr_in cli_addr;
socklen_t length = sizeof(cli_addr);
clifd = accept(listenfd, (struct sockaddr*)&cli_addr, &length);
// if (clifd < 0) {
if (clifd == -1)
perror("accept failed\n");
// exit(EXIT_FAILURE);
// }
else
{
printf("accpet a new client:ip is %s and portsi %d\n", inet_ntoa(cli_addr.sin_addr), cli_addr.sin_port);
//add client event
change_event(epolled, clifd, EPOLLIN, EPOLL_CTL_ADD);
}
}
//deal with read
static void do_read(int epollfd, int fd, char *buf)
{
int nread;
nread = read(fd, buf, MAXSIZE);
if (nread == -1)
{
perror("read data failed\n");
close(fd);
//delelet event
change_event(epollfd, fd, EPOLLIN, EPOLL_CTL_DEL);
}
else if (nread == 0)
{
fprintf(stderr, "client close\n");
close(fd);
//delete event
change_event(epollfd, fd, EPOLLIN, EPOLL_CTL_DEL);
}
else
{
printf("read message is %s\n", buf);
//change event. from read to write
change_event(epollfd, fd, EPOLLOUT, EPOLL_CTL_MOD);
}
}
//deal with write
static void do_write(int epollfd, int fd, char *buf)
{
int nwrite;
nwrite = write(fd, buf, strlen(buf));
if (nwrite == -1)
{
perror("write falied\n");
close(fd);
change_event(epollfd, fd, EPOLLOUT, EPOLL_CTL_DEL); //不可写时从epollfd中删除监听的fd
}
else
{
change_event(epollfd, fd, EPOLLIN, EPOLL_CTL_MOD);
memset(buf, 0, MAXSIZE);
}
}
//change event
static void change_event(int epollfd, int fd, int state, int op)
{
struct epoll_event ev;
ev.events = state;
ev.data.fd = fd;
//函数原型:int epoll_ctl(int epfd, int op, int fd, struct epoll_event *event);
//函数解释:参数一epfd为epoll_create返回的创建fd
// 参数二op为需要执行的操作(EPOLL_CTL_ADD:注册新的fd到epfd中;EPOLL_CTL_MOD:修改已经注册的fd的监听事件;EPOLL_CTL_DEL:从epfd中删除一个fd;)
// 参数三fd为要监听的fd套接字
// 参数四event为要监听的状态 在ev.states=state,中赋值 包括EPOLLIN、EPOLLOUT
epoll_ctl(epollfd, op, fd, &ev);
}
客户端:
#include <netinet/in.h>
#include <time.h>
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/epoll.h>
#include <arpa/inet.h>
#include <sys/socket.h>
#include <sys/types.h>
#define MAXSIZE 1024
#define IPADDRESS "127.0.0.1"
#define SERV_PORT 8888
#define FDSIZE 1024
#define EPOLLEVENTS 20
static void handle_connection(int sockfd);
static void handle_events(int epollfd, struct epoll_event *events, int num, int sockfd, char *buf);
static void do_read(int epollfd, int fd, int sockfd, char *buf);
static void do_write(int epollfd, int fd, int sockfd, char *buf);
static void change_event(int epollfd, int fd, int state, int op);
int main()
{
int sockfd;
struct sockaddr_in server_addr;
sockfd = socket(AF_INET, SOCK_STREAM, 0);
if (sockfd < 0)
{
perror("client create socket failed\n");
exit(EXIT_SUCCESS);
}
bzero(&server_addr, sizeof(server_addr));
server_addr.sin_family = AF_INET;
server_addr.sin_port = htons(SERV_PORT);
inet_pton(AF_INET, IPADDRESS, &server_addr.sin_addr);
socklen_t length = sizeof(server_addr);
if (connect(sockfd, (struct sockaddr*)&server_addr, length) < 0)
{
perror("clint connect failed\n");
exit(EXIT_SUCCESS);
}
//deal with connection
handle_connection(sockfd);
//do not forget to close sockfd
close(sockfd);
return 0;
}
static void handle_connection(int sockfd)
{
int epollfd;
struct epoll_event events[EPOLLEVENTS];
char buf[MAXSIZE];
memset(buf, 0, sizeof(buf));
int result;
epollfd = epoll_create(FDSIZE);
change_event(epollfd, STDIN_FILENO,EPOLLIN, EPOLL_CTL_ADD); //STDIN_FILENO 接受键盘的输入
while (1)
{
result = epoll_wait(epollfd, events, EPOLLEVENTS, -1);
printf("clinet need deal events is %d\n", result);
handle_events(epollfd, events, result, sockfd, buf);
}
//do not forget to close epollfd
close(epollfd);
}
static void handle_events(int epollfd, struct epoll_event *events, int num, int sockfd, char *buf)
{
int fd;
int i;
for (i = 0; i < num; i++)
{
fd = events[i].data.fd;
if (events[i].events & EPOLLIN)
{
do_read(epollfd, fd, sockfd, buf); //可能读标准输入 也可能读服务端传来的socket输入
}
else if (events[i].events & EPOLLOUT)
{
do_write(epollfd, fd, sockfd, buf); //可能写标准输出 也可能向服务端写输出
}
}
}
static void do_read(int epollfd, int fd, int sockfd, char *buf)
{
int nread;
nread = read(fd, buf, MAXSIZE);
if (nread == -1)
{
perror("client read fail\n");
close(fd);
}
else if (nread == 0)
{
fprintf(stderr, "server close\n");
close(fd);
}
else
{
if (fd == STDIN_FILENO)
{
puts("do read fd == STDIN_FILENO");
change_event(epollfd, sockfd, EPOLLOUT, EPOLL_CTL_ADD); //read完成键盘的输入之后 将客户端的sockfd更改为监听可写状态 用于向服务端socket写数据
}
else
{
puts("do read fd != STDIN_FILENO");
change_event(epollfd, sockfd, EPOLLIN, EPOLL_CTL_DEL);
change_event(epollfd, STDOUT_FILENO, EPOLLOUT, EPOLL_CTL_ADD);//read完socket数据后将标准输出更改为可写 准备输出数据
}
}
}
static void do_write(int epollfd, int fd, int sockfd, char *buf)
{
int nwrite;
nwrite = write(fd, buf, strlen(buf));
if (nwrite == -1)
{
perror("client write failed\n");
close(fd);
}
else
{
if (fd == STDOUT_FILENO)
{
puts("do write fd == STDOUT_FILENO");
change_event(epollfd, fd, EPOLLOUT, EPOLL_CTL_DEL);
}
else
{
puts("do write fd != STDOUT_FILENO");
change_event(epollfd, fd, EPOLLIN, EPOLL_CTL_MOD);
}
}
memset(buf, 0, MAXSIZE);
}
static void change_event(int epollfd, int fd, int state, int op)
{
struct epoll_event ev;
ev.events = state;
ev.data.fd = fd;
epoll_ctl(epollfd, op, fd, &ev);
}