记录一次epoll事件未触发的经历,由于是tcp通信,使用了io多路复用epoll,客户端只发送数据,不处理服务端响应,服务端处理对应的事件,并且会响应消息给客户端,遇到了EPOLLOUT事件未触发的情况,进行排查。

经过查阅相关资料得知,socket通信时是全双工的,都有send、recv缓冲区,当recv缓冲区满了,就会响应ACK给对方,将发送窗口置为0,这个时候EPOLLOUT事件就无法触发,EPOLLOUT触发的前提条件是发送缓冲区可写,即对应的内核缓冲区为可写状态,这个时候不满足条件,EPOLLOUT事件就触发不了,响应消息就发送不过去。
附相关代码
客户端代码
#include<stdio.h>
#include<stdlib.h>
#include<unistd.h>
#include<arpa/inet.h>
#include<sys/socket.h>
#include<netinet/in.h>
#include<string.h>
#include<errno.h>
#define MAXLINE 1024
void cli_send_msg(FILE *fp, int sockfd)
{
char sendline[MAXLINE], recvline[MAXLINE];
//while (fgets(sendline, MAXLINE, fp) != NULL) {
while(1) {
memset(sendline, 97, MAXLINE);
int n = write(sockfd, sendline, strlen(sendline));
if (n < 0) {
fprintf(stderr, "write err errno(%d) %s\n", errno, strerror(errno));
}
memset(sendline, 0, sizeof(sendline));
usleep(10);
//if (read(sockfd, recvline, MAXLINE) == -1) {
// fprintf(stderr, "read err errno(%d) %s\n", errno, strerror(errno));
// close(sockfd);
//}
//memset(recvline, 0, MAXLINE);
//Fputs(recvline, stdout);
}
}
int main(int argc, const char **argv)
{
int sockfd;
struct sockaddr_in servaddr;
if (argc != 3) {
fprintf(stderr, "usage: tcpcli <ip> <port>\n");
return -1;
}
sockfd = socket(AF_INET, SOCK_STREAM, 0);
bzero(&servaddr, sizeof(servaddr));
servaddr.sin_family = AF_INET;
servaddr.sin_port = htons(atoi(argv[2]));
int ret = inet_pton(AF_INET, argv[1], &servaddr.sin_addr);
if (ret <= 0) {
fprintf(stderr, "inet_pton err\n");
}
int len = 4;
int rcvbuf_size = 3145728;
setsockopt(sockfd, SOL_SOCKET, SO_RCVBUF, &rcvbuf_size, len);
rcvbuf_size = 0;
getsockopt(sockfd, SOL_SOCKET, SO_RCVBUF, &rcvbuf_size, &len);
printf("recive buff size is %d\n", rcvbuf_size);
ret = connect(sockfd, (struct sockaddr *)&servaddr, sizeof(servaddr));
if (ret < 0) {
fprintf(stderr, "connect err\n");
}
cli_send_msg(stdin, sockfd);
return 0;
}
服务端代码
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#include<errno.h>
#include<netinet/in.h>
#include<arpa/inet.h>
#include<sys/socket.h>
#include<sys/epoll.h>
#include<unistd.h>
#include<sys/types.h>
#define IPADDRESS "0.0.0.0"
#define PORT 7001
#define MAXSIZE 1024
#define LISTENQ 5
#define FDSIZE 1000
#define EPOLLEVENTS 1024
int rcv_c = 0;
int wt_c = 0;
int socket_bind(const char *ip, int port);
void do_epoll(int listenfd);
void handle_events(int epollfd, struct epoll_event *event, int num,int listenfd,char *buf);
void handle_accept(int epollfd, int listenfd);
void do_read(int epollfd, int connfd, char *buf);
void do_write(int epollfd, int connfd, char *buf);
void add_event(int epollfd, int connfd, int state);
void delete_event(int epollfd, int connfd, int state);
void modify_event(int epollfd, int connfd, int state);
int main(void)
{
int listenfd;
listenfd = socket_bind(IPADDRESS, PORT);
listen(listenfd, LISTENQ);
do_epoll(listenfd);
return 0;
}
int socket_bind(const char *ip, int port)
{
int fd;
struct sockaddr_in servaddr;
if ((fd = socket(AF_INET, SOCK_STREAM, 0)) == -1)
{
perror("socket create error: ");
exit(1);
}
servaddr.sin_family = AF_INET;
inet_pton(AF_INET, ip ,&servaddr.sin_addr);
servaddr.sin_port = htons(port);
int reuse = 1;
setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &reuse, sizeof(reuse));
if(bind(fd, (struct sockaddr*) &servaddr, sizeof(servaddr)) == -1)
{
perror("bind error: ");
exit(1);
}
return fd;
}
void do_epoll(int listenfd)
{
int epollfd;
struct epoll_event events[EPOLLEVENTS];
char buf[MAXSIZE];
int ready = 0;
memset(buf, 0, MAXSIZE);
epollfd = epoll_create(FDSIZE);
//添加服务监听事件
add_event(epollfd,listenfd,EPOLLIN);
//轮询处理事件
for( ; ; )
{
printf("waiting...\n");
ready = epoll_wait(epollfd, events, EPOLLEVENTS, -1);
if(ready == 0)
{
fprintf(stderr, "epoll_wait timeout\n");
continue;
}
printf("wait_fd %d\n", ready);
handle_events(epollfd, events, ready, listenfd, buf);
}
}
void add_event(int epollfd, int fd, int state)
{
struct epoll_event ev;
ev.events = state;
ev.data.fd = fd;
epoll_ctl(epollfd, EPOLL_CTL_ADD, fd, &ev);
}
void delete_event(int epollfd, int fd, int state)
{
struct epoll_event ev;
ev.events = state;
ev.data.fd = fd;
epoll_ctl(epollfd, EPOLL_CTL_DEL, fd, &ev);
}
void modify_event(int epollfd, int fd, int state)
{
struct epoll_event ev;
ev.events = state;
ev.data.fd = fd;
epoll_ctl(epollfd, EPOLL_CTL_MOD, fd, &ev);
}
void handle_events(int epollfd, struct epoll_event *events, int num,int listenfd,char *buf)
{
int fd = -1;
for(int i = 0; i < num; i++ )
{
fd = events[i].data.fd;
if (fd == listenfd && events[i].events & EPOLLIN )
{
handle_accept(epollfd, listenfd);
}else if (events[i].events & EPOLLIN )
{
//处理读事件fd->buf
do_read(epollfd, fd, buf);
}else if ( events[i].events & EPOLLOUT )
{
//处理写事件buf->fd
do_write(epollfd, fd, buf);
} else if( events[i].events & EPOLLERR) {
close(fd);
fprintf(stderr, "close connect, error causeed!\n");
}
}
}
void handle_accept(int epollfd, int listenfd)
{
int connfd;
struct sockaddr_in cliaddr;
socklen_t socklen;
socklen = sizeof(cliaddr);
if ( (connfd = accept(listenfd,(struct sockaddr*)&cliaddr, &socklen)) == -1 )
{
perror("accept error: ");
return;
}
fprintf(stdout, "accept a client : %s:%d\n", inet_ntoa(cliaddr.sin_addr), cliaddr.sin_port);
//添加对客户端监听事件
add_event(epollfd, connfd, EPOLLIN|EPOLLERR);
}
void do_read(int epollfd, int fd, char *buf)
{
int n;
n = recv(fd, buf, MAXSIZE, 0);
if ( n == -1 )
{
char msg[64] = {0};
int n = 0;
if (errno == ECONNRESET) {
n = sprintf(msg, "a client( %d ) close connection. ", fd);
if (n < 0) {
fprintf(stderr, "sprintf err\n");
}
}
sprintf(msg+n, "[ERROR][%d]", errno);
perror(msg);
close(fd);
delete_event(epollfd, fd, EPOLLIN);
}else if ( n == 0 )
{
fprintf(stdout, "a client( %d ) close connection. [OK]\n", fd);
close(fd);
delete_event(epollfd, fd, EPOLLIN);
}else
{
rcv_c++;
printf("receive[%d]: %s\n", rcv_c, buf);
modify_event(epollfd, fd, EPOLLOUT);
printf("add write\n");
}
}
void do_write(int epollfd, int fd, char *buf)
{
int nwrite;
nwrite = write(fd, buf, strlen(buf));
if (nwrite == -1)
{
perror("write error:");
close(fd);
delete_event(epollfd, fd, EPOLLOUT);
} else {
wt_c++;
printf("write count %d\n", wt_c);
modify_event(epollfd, fd, EPOLLIN);
}
memset(buf, 0, MAXSIZE);
}

本文记录了一次在使用epoll进行TCP通信时遇到的EPOLLOUT事件未触发的问题。当服务端的recv缓冲区满,导致发送窗口关闭,EPOLLOUT事件无法触发,进而影响服务端响应消息到客户端。解决的关键在于理解全双工通信中send和recv缓冲区的工作原理。

被折叠的 条评论
为什么被折叠?



