linux send 失败_Epoll学习服务器的实现-Linux内核原始Epoll结构

1.Begins~

有的人学习linux编程很久，只知道网络编程是socket，bind， listen。。。，然而这些都是网络通信软件最基本的接口。在某网络公司待了y，也了解到公司的基础就是网络转发，然而网络转发实现并非我们平时所见的简单的send，recv。公司设备的转发都是建立在稳定并且高效的内部业务的基础上的，例如一个可靠性的服务：BFD（链路双向快速检测），进程内业务就是建立在内核与用户态，后台处理与前台配置交互的过程上实现。在做项目的日子里，渐渐的对公司BFD模块有一定了解，特别是涉及到框架的地方，仔细看发现，一些技巧其实就是我们平日所用的技能外加一些优化。对于单进程用户态来说，处理前台进程交互与内核上送消息，如果没有一个高效的机制，是很难处理上以千计甚至万计的消息的。在参看公司代码的初始化阶段，发现一共用了至少3 个epoll回调机制，一个是处理前台发送的配置及查询，一个是处理与其他应用模块的交互，另一个是向其他应用通报状态变化的机制。公司的epoll_event 机制不知道是不是修改了内核，结构大致是这样：

typedef struct epoll_event { uint32_t events;      // Epoll events  epoll_data_t data;      // User data variable ep_callback  callback;    /* 这跟linux内核不一样 ？ */}EPOLL_PACK; /*在添加epoll事件的时候就多了一个回调函数注册，这个十分方便了，回调函数可以用一个全局变量搞定，一个框架搭好，随后添加就大大的简单多了*/

实际内核提供给我们使用的用户态结构是这样：


typedef union epoll_data {  void *ptr;  int fd;  uint32_t u32;  uint64_t u64;  } epoll_data_t;struct epoll_event { uint32_t events;      // Epoll events  epoll_data_t data;      // User data variable  };

不过我们也可以将epoll_data_t这个联合体封装在一个结构，只需要用指针*ptr指向封装的带fd及回调函数就可以。打算自己学习一些必要的技能，暂且就记录一下吧

2.Epoll API


extern int epoll_create(int  size);extern int epoll_ctl(int epfd, int op, int fd, struct epoll_event *event);extern 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。自从linux2.6.8之后，size参数是被忽略的。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。
②int epoll_ctl(int epfd, int op, int fd, struct epoll_event *event);
epoll的事件注册函数，它不同于select()是在监听事件时告诉内核要监听什么类型的事件，而是在这里先
注册要监听的事件类型。
参数：
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；
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：

enum EPOLL_EVENTS{

EPOLLIN = 0x001,

#define EPOLLIN EPOLLIN

EPOLLPRI = 0x002,

#define EPOLLPRI EPOLLPRI

EPOLLOUT = 0x004,

#define EPOLLOUT EPOLLOUT

EPOLLRDNORM = 0x040,

#define EPOLLRDNORM EPOLLRDNORM

EPOLLRDBAND = 0x080,

#define EPOLLRDBAND EPOLLRDBAND

EPOLLWRNORM = 0x100,

#define EPOLLWRNORM EPOLLWRNORM

EPOLLWRBAND = 0x200,

#define EPOLLWRBAND EPOLLWRBAND

EPOLLMSG = 0x400,

#define EPOLLMSG EPOLLMSG

EPOLLERR = 0x008,

#define EPOLLERR EPOLLERR

EPOLLHUP = 0x010,

#define EPOLLHUP EPOLLHUP

EPOLLRDHUP = 0x2000,

#define EPOLLRDHUP EPOLLRDHUP

EPOLLWAKEUP = 1u << 29,

#define EPOLLWAKEUP EPOLLWAKEUP

EPOLLONESHOT = 1u << 30,

#define EPOLLONESHOT EPOLLONESHOT

EPOLLET = 1u << 31

#define EPOLLET EPOLLET

};

(3)epoll_wait();

参数events是分配好的epoll_event结构体数组；
epoll将会把发生的事件复制到events数组中（events不可以是空指针，内核只负责把数据复制到）
参数timeout 是超时时间（毫秒，0表示立即返回，-1表示永久阻塞）；
如果函数调用成功，返回对应IO上已经准备好的文件描述符数目；如果返回0，表示已超时；如果返回值小于0，表示函数失败。

Linuxc/c++服务器开发高阶视频学习资料+主页群获取

3.epoll mode

epoll的使用也很方便，基本都会用到以下模型，取自linux帮助手册：

/*Example for suggested usageWhile  the  usage  of  epoll when employed as a level-triggered interface does have the same semantics as poll(2), the edge-triggered usage requires moreclarification to avoid stalls in the application event loop.  In this example, listener is a nonblocking socket on which listen(2) has been called.   Thefunction  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 applica‐tion 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)  fromwhere it stopped before.*/#define MAX_EVENTS 10struct epoll_event ev, events[MAX_EVENTS];int listen_sock, conn_sock, nfds, epollfd;/* Code to set up listening socket, 'listen_sock',(socket(), bind(), listen()) omitted */static Epoll_Case(){epollfd = epoll_create1(0);if (epollfd == -1) {perror("epoll_create1");exit(EXIT_FAILURE);}ev.events = EPOLLIN;ev.data.fd = listen_sock;if (epoll_ctl(epollfd, EPOLL_CTL_ADD, listen_sock, &ev) == -1) {perror("epoll_ctl: listen_sock");exit(EXIT_FAILURE);}for (;;) {nfds = epoll_wait(epollfd, events, MAX_EVENTS, -1);if (nfds == -1) {perror("epoll_wait");exit(EXIT_FAILURE);}for (n = 0; n < nfds; ++n) {if (events[n].data.fd == listen_sock) {conn_sock = accept(listen_sock,(struct sockaddr *) &local, &addrlen);if (conn_sock == -1) {perror("accept");exit(EXIT_FAILURE);}setnonblocking(conn_sock);ev.events = EPOLLIN | EPOLLET;ev.data.fd = conn_sock;if (epoll_ctl(epollfd, EPOLL_CTL_ADD, conn_sock,&ev) == -1) {perror("epoll_ctl: conn_sock");exit(EXIT_FAILURE);}} 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)  withEPOLL_CTL_MOD.*/

下面给出一个epoll 服务器的简单例子,参考了网上不少方式，自己写了一个，算不上原创，把BIT_TEST部分注释打开，可以忽略basetype.h,可以编译通过。创建socket的地方可以自己优化省略，我实现的时候觉得这种方式比较容易看清楚socket的获取方式。


#include <netdb.h>#include <stdio.h>#include <stdlib.h>#include <sys/types.h>#include <sys/socket.h>#include <arpa/inet.h>#include <string.h>#include <netinet/in.h>#include <sys/epoll.h>#include <poll.h>#include <assert.h>#include <unistd.h>#include <fcntl.h>#include "basetype.h"static int g_Main_Epoll_Fd = -1;#define MAX_EPOLL_EVENT_COUNT 64#define MAX_LISTEN_NUM  10#define MAX_BUFF_LEN 512/*#define BIT_TEST(a, b)          ((a) & (b))#define BIT_RESET(a, b)         ((a) ~= (b))#define BIT_SET(a, b)           ((a) |= (b))#define BIT_MATCH(a,b)          ((a)&(b) = (b))#define BIT_COMPARE(a, b, c)    ((a)&(b) == (c))*/static void print_socket_info(struct addrinfo *ai){char ipstr[INET6_ADDRSTRLEN];uint16_t port;void *addr  = NULL;char *ipver = NULL;struct sockaddr_in *ipv4 = NULL;struct sockaddr_in6 *ipv6 = NULL;assert(NULL != ai);if (ai->ai_family == AF_INET) { // IPv4ipv4 = (struct sockaddr_in *)ai->ai_addr;addr = &(ipv4->sin_addr);ipver = "IPv4";port =ntohs(((struct sockaddr_in*)ai->ai_addr)->sin_port);} else { // IPv6ipv6 = (struct sockaddr_in6 *)ai->ai_addr;addr = &(ipv6->sin6_addr);ipver = "IPv6";port =ntohs(((struct sockaddr_in6*)ai->ai_addr)->sin6_port);}// convert the IP to a string and print it:(void)inet_ntop(ai->ai_family, addr, ipstr, sizeof ipstr);printf(" server initing... AF: %s IP: %s,PORT: %un", ipver, ipstr,port);return ;}static int SetSocketNoblocking(const int sockfd){int flags = 0;int iret  = 0;flags = fcntl (sockfd, F_GETFL, 0);if (flags == -1){perror ("fcntl");return -1;}flags |= O_NONBLOCK;iret = fcntl (sockfd, F_SETFL, flags);if (iret == -1){perror ("fcntl");return -1;}return 0;}static int Epoll_Control(const int sockfd, const int oper, const int events){struct epoll_event ep_event;int iret = 0;memset(&ep_event, 0, sizeof(ep_event));ep_event.events = events;ep_event.data.fd = sockfd;iret = epoll_ctl(g_Main_Epoll_Fd, oper, sockfd, &ep_event);if (0 > iret){perror("epoll_ctl error.");printf("epoll_ctl error opt:%dn", oper);return -1;}return 0;}static int CreateServerScoket(const char *host, const char *port){struct addrinfo hints;struct addrinfo*ailist= NULL; struct addrinfo *ai= NULL;int iret;int sockfd = 0;int sockoptval = 1;memset(&hints, 0, sizeof(hints));hints.ai_family = AF_UNSPEC; // AF_INET 或 AF_INET6 hints.ai_socktype = SOCK_STREAM;hints.ai_flags = AI_PASSIVE;     /* All interfaces */iret = getaddrinfo(host, port, &hints, &ailist);if (0 != iret){fprintf(stderr, "getaddrinfo: %sn", gai_strerror(iret));return -1;}for(ai = ailist;ai != NULL; ai = ai->ai_next) {sockfd = socket(ai->ai_family, ai->ai_socktype, ai->ai_protocol);if (0 > sockfd){continue;}/* set reuse addr */iret = setsockopt(sockfd, SOL_SOCKET, SO_REUSEADDR, &sockoptval, sizeof(sockoptval));if (0 > iret){perror("setsockopt errorn");freeaddrinfo(ailist);return -1;}iret = bind(sockfd, ai->ai_addr, ai->ai_addrlen);/*sizeof(struct sockaddr)*/if (0 == iret){print_socket_info(ai);break;}else{close(sockfd);}}if (NULL == ai){fprintf(stderr, "get server socket error.n");return -1;}freeaddrinfo(ailist); /* set no-block mode */iret = SetSocketNoblocking(sockfd);if (0 != iret){return -1;}/* listen socket */iret = listen(sockfd, MAX_LISTEN_NUM);if (0  > iret){perror("listen socket error.n");return -1;}printf("server socket init success. listenning...n");return sockfd;}static void HandleAccept(const int listen_scok, int Event){struct sockaddr_in addr;socklen_t  len = sizeof(addr);int client_sock  = -1;if (BIT_TEST(EPOLLIN, Event)){client_sock = accept(listen_scok, (struct sockaddr *)&addr, &len);if(client_sock < 0 ){perror("accept infd error");return;}(void)Epoll_Control(client_sock, EPOLL_CTL_ADD, EPOLLIN);}return ;}static void ProcRecvMsg(int sockfd){ssize_t recvlen = 0;char recvBuf[MAX_BUFF_LEN];//read ev  ready;recvlen = recv(sockfd, recvBuf, sizeof(recvBuf), 0);if(recvlen < 0){perror("recv error.");return ;}else if(recvlen == 0){printf("client quitn");   (void)Epoll_Control(sockfd, EPOLL_CTL_DEL, 0);close(sockfd);}else{(void)Epoll_Control(sockfd, EPOLL_CTL_MOD, EPOLLOUT);printf("recv msg fprom client:msg#:%sn",recvBuf);}return ;}static void SendMsg2Client(const int sockfd){ssize_t sendlen = 0;char sendBuf[MAX_BUFF_LEN];int event = 0;sprintf(sendBuf, "HTTP/1.0 200 OKrnrn<html><h1>Hello Epoll! [client_fd:%d]</h1></html>", sockfd);  int sendsize = send(sockfd, sendBuf, strlen(sendBuf)+1, 0);  if(sendsize <= 0)  {  perror("send error.");(void)Epoll_Control(sockfd, EPOLL_CTL_DEL, 0);close(sockfd);}  else {  printf("Server reply msg ok! data: %sn", sendBuf);  event = EPOLLIN | EPOLLERR | EPOLLHUP;(void)Epoll_Control(sockfd, EPOLL_CTL_MOD, event);}  return ;}int main(int argc, char *argv[]){int listenfd =0;int epfd = 0;int iret = 0;int event = 0;struct epoll_event ep_events[MAX_EPOLL_EVENT_COUNT];int ev_num = 0;int in_sockfd = 0;int loop = 0;if (3 > argc){printf("Usage: %s [ip_addr] [port]n",argv[0]);return -1;}listenfd = CreateServerScoket(argv[1], argv[2]);if (0 > listenfd){return -1;}epfd = epoll_create(1);if (-1 == epfd){perror("create epoll error.n");return -1;}else{g_Main_Epoll_Fd = epfd;}event = EPOLLIN;iret = Epoll_Control(listenfd, EPOLL_CTL_ADD, event);if (0 != iret){return -1;}/* proc epoll event regined */printf("server init success, epoll wait now.n");for(;;){memset(ep_events, 0, sizeof(ep_events));ev_num = epoll_wait(g_Main_Epoll_Fd, ep_events, MAX_EPOLL_EVENT_COUNT, -1);switch (ev_num){case 0:{printf("epoll_wait timeout.n");break;}case -1:{perror("epoll_wait error.n");break;}default:{for (loop = 0; loop < ev_num; loop++){in_sockfd = ep_events[loop].data.fd;event = ep_events[loop].events;if (in_sockfd == listenfd){HandleAccept(in_sockfd, event);}else{if (BIT_TEST(EPOLLIN, event)){/* have data to read */ProcRecvMsg(in_sockfd);}else if (BIT_TEST(EPOLLOUT, event)){SendMsg2Client(in_sockfd);}}}break;}}}return 0;}

结果如下，客户端任意都可以：注意，运行时需要手动输入服务器主机名及端口号。

原文链接：Epoll学习服务器的简单实现-Linux内核原始Epoll结构_li-CSDN博客