redis概述
redis作为一个高性能的kv数据库,基于c语言实现,数据主要基于内存存放,所以查询效率较高,并且支持较多的数据类型有字符串、字典、列表与集合等数据结构。另外redis叶支持持久化的日志,支持集群等特性,这些内容并不在本文的概述中,本文只是概述早期版本的redis的基本实现思路。
redis的架构概述
早期版本的redis,该版本的协议还是telnet协议,支持的数据类型只有string、dict与set,不过该版本支持db的定时load到文件中,定时保存数据时间间隔为1000毫秒,redis的选用的模型是单线程select事件驱动框架;
伪代码执行流程:
server初始化过程
1.监听端口
2.事件框架的初始化
3.添加事件监听驱动
while (1) {
select() // 监听是否有读写请求
1.如果有事件发生调用回调函数处理
2.如果时间时间到了调用时间注册的回调处理函数
}
以上就是整个业务的处理流程,现在的事件驱动模型可以选用select,epoll等不同操作系统提供的事件驱动的系统调用。
redis启动流程分析
main函数启动
启动函数位于redis.c中的main函数
int main(int argc, char **argv) {
initServerConfig(); // 读取配置信息
initServer(); // 初始化server
if (argc == 2) {
ResetServerSaveParams(); // 读取传入的配置信息并更新默认值
loadServerConfig(argv[1]);
redisLog(REDIS_NOTICE,"Configuration loaded");
} else if (argc > 2) { // 如果超过两两个输入值则报错
fprintf(stderr,"Usage: ./redis-server [/path/to/redis.conf]\n");
exit(1);
}
redisLog(REDIS_NOTICE,"Server started"); // 打印日志
if (loadDb("dump.rdb") == REDIS_OK) // 检查是否有加载dump内容
redisLog(REDIS_NOTICE,"DB loaded from disk");
if (aeCreateFileEvent(server.el, server.fd, AE_READABLE,
acceptHandler, NULL, NULL) == AE_ERR) oom("creating file event"); // 创建服务端监听事件的监听
redisLog(REDIS_NOTICE,"The server is now ready to accept connections");
aeMain(server.el); // 启动事件循环
aeDeleteEventLoop(server.el); // 移除监听事件
return 0;
}
main函数的主要工作就是初始化并加载配置文件,检查是否需要加载保存的rdb,创建监听对象,进入事件循环,这些步骤就是标准的事件驱动框架的执行流程。
initServer函数初始化
该函数主要工作就是初始化相关的内容监听,并初始化事件监听。
static void initServer() {
int j;
signal(SIGHUP, SIG_IGN); // 防止出现僵死进程
signal(SIGPIPE, SIG_IGN);
server.clients = listCreate(); // 创建一个列表用于保存clients 就是一个链表
server.objfreelist = listCreate();
createSharedObjects();
server.el = aeCreateEventLoop(); // 创建事件循环
server.dict = malloc(sizeof(dict*)*server.dbnum); // 生成一个字典保存不同数据库中存入的值
if (!server.dict || !server.clients || !server.el || !server.objfreelist)
oom("server initialization"); /* Fatal OOM */ // 如果三个中任意一个未生成成功则内存报错
server.fd = anetTcpServer(server.neterr, server.port, NULL); // 建立服务端监听请求
if (server.fd == -1) { // 如果创建监听失败则报错
redisLog(REDIS_WARNING, "Opening TCP port: %s", server.neterr);
exit(1);
}
for (j = 0; j < server.dbnum; j++) { // 根据配置的数据库db生成对应的保存的数据结构
server.dict[j] = dictCreate(&sdsDictType,NULL);
if (!server.dict[j])
oom("server initialization"); /* Fatal OOM */
}
server.cronloops = 0;
server.bgsaveinprogress = 0;
server.lastsave = time(NULL);
server.dirty = 0;
aeCreateTimeEvent(server.el, 1000, serverCron, NULL, NULL); // 创建事件监听
}
aeCreateEventLoop函数主要就是创建一个事件驱动loop,用于管理事件驱动里面包括了驱动事件与时间事件。
/* File event structure */
typedef struct aeFileEvent {
int fd; // 文件描述符
int mask; /* one of AE_(READABLE|WRITABLE|EXCEPTION) */ // 事件标志
aeFileProc *fileProc; // 事件回调函数
aeEventFinalizerProc *finalizerProc; // 事件处理完成后回调函数
void *clientData; // 数据
struct aeFileEvent *next; // 下一个事件
} aeFileEvent;
/* Time event structure */
typedef struct aeTimeEvent {
long long id; /* time event identifier. */ // 事件id
long when_sec; /* seconds */ // 时间秒
long when_ms; /* milliseconds */ // 时间毫秒
aeTimeProc *timeProc; // 时间回调函数
aeEventFinalizerProc *finalizerProc;
void *clientData;
struct aeTimeEvent *next; // 下一个回调事件
} aeTimeEvent;
/* State of an event based program */
typedef struct aeEventLoop {
long long timeEventNextId;
aeFileEvent *fileEventHead; // 事件列表
aeTimeEvent *timeEventHead; // 时间列表
int stop;
} aeEventLoop;
aeEventLoop *aeCreateEventLoop(void) {
aeEventLoop *eventLoop;
eventLoop = malloc(sizeof(*eventLoop)); // 申请内存
if (!eventLoop) return NULL;
eventLoop->fileEventHead = NULL; // 初始化链表
eventLoop->timeEventHead = NULL;
eventLoop->timeEventNextId = 0;
eventLoop->stop = 0;
return eventLoop;
}
继续分析anetTcpServer函数;
int anetTcpServer(char *err, int port, char *bindaddr)
{
int s, on = 1;
struct sockaddr_in sa;
if ((s = socket(AF_INET, SOCK_STREAM, 0)) == -1) { // 生成socket实例s
anetSetError(err, "socket: %s\n", strerror(errno));
return ANET_ERR;
}
if (setsockopt(s, SOL_SOCKET, SO_REUSEADDR, &on, sizeof(on)) == -1) { // 设置端口可复用
anetSetError(err, "setsockopt SO_REUSEADDR: %s\n", strerror(errno));
close(s);
return ANET_ERR;
}
sa.sin_family = AF_INET; // 设置协议
sa.sin_port = htons(port); // 设置端口
sa.sin_addr.s_addr = htonl(INADDR_ANY);
if (bindaddr) inet_aton(bindaddr, &sa.sin_addr);
if (bind(s, (struct sockaddr*)&sa, sizeof(sa)) == -1) { // 监听该套接字
anetSetError(err, "bind: %s\n", strerror(errno));
close(s);
return ANET_ERR;
}
if (listen(s, 5) == -1) { // 指定监听队列
anetSetError(err, "listen: %s\n", strerror(errno));
close(s);
return ANET_ERR;
}
return s;
}
主要就是根据端口ip监听队列并返回该套接字。创建完成之后就创建了定时回写到文件的内容备份时间事件;
aeCreateTimeEvent(server.el, 1000, serverCron, NULL, NULL);
long long aeCreateTimeEvent(aeEventLoop *eventLoop, long long milliseconds,
aeTimeProc *proc, void *clientData,
aeEventFinalizerProc *finalizerProc)
{
long long id = eventLoop->timeEventNextId++;
aeTimeEvent *te;
te = malloc(sizeof(*te));
if (te == NULL) return AE_ERR;
te->id = id;
aeAddMillisecondsToNow(milliseconds,&te->when_sec,&te->when_ms);
te->timeProc = proc;
te->finalizerProc = finalizerProc;
te->clientData = clientData;
te->next = eventLoop->timeEventHead;
eventLoop->timeEventHead = te;
return id;
}
该时间的回调函数是serverCron,时间是1000毫秒;
int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) {
int j, size, used, loops = server.cronloops++;
REDIS_NOTUSED(eventLoop);
REDIS_NOTUSED(id);
REDIS_NOTUSED(clientData);
/* If the percentage of used slots in the HT reaches REDIS_HT_MINFILL
* we resize the hash table to save memory */
for (j = 0; j < server.dbnum; j++) { // 重新计算字典的空间太小好调整内容以达到节省内存
size = dictGetHashTableSize(server.dict[j]);
used = dictGetHashTableUsed(server.dict[j]);
if (!(loops % 5) && used > 0) {
redisLog(REDIS_DEBUG,"DB %d: %d keys in %d slots HT.",j,used,size);
// dictPrintStats(server.dict);
}
if (size && used && size > REDIS_HT_MINSLOTS &&
(used*100/size < REDIS_HT_MINFILL)) {
redisLog(REDIS_NOTICE,"The hash table %d is too sparse, resize it...",j); // 检查是否达到重新缩小字典大小的阈值达到就重新设置大小
dictResize(server.dict[j]);
redisLog(REDIS_NOTICE,"Hash table %d resized.",j);
}
}
/* Show information about connected clients */
if (!(loops % 5)) redisLog(REDIS_DEBUG,"%d clients connected",listLength(server.clients)); // 展示链接信息
/* Close connections of timedout clients */
if (!(loops % 10))
closeTimedoutClients(); // 关闭超时链接的客户端
/* Check if a background saving in progress terminated */
if (server.bgsaveinprogress) { // 检查是否在存储
int statloc;
if (wait4(-1,&statloc,WNOHANG,NULL)) {
int exitcode = WEXITSTATUS(statloc);
if (exitcode == 0) {
redisLog(REDIS_NOTICE,
"Background saving terminated with success");
server.dirty = 0;
server.lastsave = time(NULL);
} else {
redisLog(REDIS_WARNING,
"Background saving error");
}
server.bgsaveinprogress = 0;
}
} else {
/* If there is not a background saving in progress check if
* we have to save now */
time_t now = time(NULL);
for (j = 0; j < server.saveparamslen; j++) {
struct saveparam *sp = server.saveparams+j;
if (server.dirty >= sp->changes &&
now-server.lastsave > sp->seconds) {
redisLog(REDIS_NOTICE,"%d changes in %d seconds. Saving...",
sp->changes, sp->seconds); // 如果需要重新dump
saveDbBackground("dump.rdb"); // 将内存中保存的数据dump到dump.rdb中
break;
}
}
}
return 1000; // 返回时间,在时间驱动中会继续创建该时间事件
}
继续查看一下saveDbBackground的函数操作;
static int saveDbBackground(char *filename) {
pid_t childpid;
if (server.bgsaveinprogress) return REDIS_ERR;
if ((childpid = fork()) == 0) { // 生成一个子进程
/* Child */
close(server.fd);
if (saveDb(filename) == REDIS_OK) { // 在子进程中保存数据保存完成之后就退出
exit(0);
} else {
exit(1);
}
} else {
/* Parent */
redisLog(REDIS_NOTICE,"Background saving started by pid %d",childpid);
server.bgsaveinprogress = 1; // 父进程修改值后退出
return REDIS_OK;
}
return REDIS_OK; /* unreached */
}
由此可知,早期redis版本是通过子进程来进行数据备份的。
创建服务端监听事件
通过server初始化之后,现在就需要把server监听的sock注册到事件驱动中去;
aeCreateFileEvent(server.el, server.fd, AE_READABLE,
acceptHandler, NULL, NULL)
int aeCreateFileEvent(aeEventLoop *eventLoop, int fd, int mask,
aeFileProc *proc, void *clientData,
aeEventFinalizerProc *finalizerProc)
{
aeFileEvent *fe;
fe = malloc(sizeof(*fe)); // 申请aeFileEvent内存
if (fe == NULL) return AE_ERR;
fe->fd = fd; // 设置文件描述符
fe->mask = mask; // 设置监听描述符
fe->fileProc = proc; // 设置回调函数
fe->finalizerProc = finalizerProc;
fe->clientData = clientData;
fe->next = eventLoop->fileEventHead; // 添加到eventloop的列表中去
eventLoop->fileEventHead = fe;
return AE_OK;
}
从创建流程中可知,当客户端新连接进来之后调用了acceptHandler函数进行处理
static void acceptHandler(aeEventLoop *el, int fd, void *privdata, int mask) {
int cport, cfd;
char cip[128];
REDIS_NOTUSED(el);
REDIS_NOTUSED(mask);
REDIS_NOTUSED(privdata);
cfd = anetAccept(server.neterr, fd, cip, &cport); // 接受进来的请求
if (cfd == AE_ERR) {
redisLog(REDIS_DEBUG,"Accepting client connection: %s", server.neterr);
return;
}
redisLog(REDIS_DEBUG,"Accepted %s:%d", cip, cport);
if (createClient(cfd) == REDIS_ERR) { // 创建一个client来处理请求
redisLog(REDIS_WARNING,"Error allocating resoures for the client");
close(cfd); /* May be already closed, just ingore errors */
return;
}
}
int anetAccept(char *err, int serversock, char *ip, int *port)
{
int fd;
struct sockaddr_in sa;
unsigned int saLen;
while(1) {
saLen = sizeof(sa);
fd = accept(serversock, (struct sockaddr*)&sa, &saLen); // 接受新进来的连接
if (fd == -1) {
if (errno == EINTR)
continue;
else {
anetSetError(err, "accept: %s\n", strerror(errno)); // 如果出错则报错
return ANET_ERR;
}
}
break;
}
if (ip) strcpy(ip,inet_ntoa(sa.sin_addr)); // 将数据传入ip与port中
if (port) *port = ntohs(sa.sin_port);
return fd;
}
static int createClient(int fd) {
redisClient *c = malloc(sizeof(*c));
anetNonBlock(NULL,fd); // 设置连接为非阻塞
anetTcpNoDelay(NULL,fd);
if (!c) return REDIS_ERR;
selectDb(c,0); // 选择数据库号
c->fd = fd;
c->querybuf = sdsempty();
c->argc = 0;
c->bulklen = -1;
c->sentlen = 0;
c->lastinteraction = time(NULL);
if ((c->reply = listCreate()) == NULL) oom("listCreate");
listSetFreeMethod(c->reply,decrRefCount);
if (aeCreateFileEvent(server.el, c->fd, AE_READABLE,
readQueryFromClient, c, NULL) == AE_ERR) { // 创建读事件请求
freeClient(c); // 如果出错则释放client
return REDIS_ERR;
}
if (!listAddNodeTail(server.clients,c)) oom("listAddNodeTail");
return REDIS_OK;
}
主要在createClient中对新建立的连接设置非阻塞之后,创建读事件并注册进去。此时注册的读事件的回调函数为readQueryFromClient函数,
readQueryFromClient函数解析
static int processCommand(redisClient *c) {
struct redisCommand *cmd;
sdstolower(c->argv[0]);
/* The QUIT command is handled as a special case. Normal command
* procs are unable to close the client connection safely */
if (!strcmp(c->argv[0],"quit")) { // 如果是quit退出
freeClient(c);
return 0;
}
cmd = lookupCommand(c->argv[0]); // 寻找匹配的命令
if (!cmd) {
addReplySds(c,sdsnew("-ERR unknown command\r\n")); // 如果没有找到则返回内容并重置client
resetClient(c);
return 1;
} else if (cmd->arity != c->argc) { // 检查cmd是否与客户端传入的参数相等如果不想等则错误的输入参数
addReplySds(c,sdsnew("-ERR wrong number of arguments\r\n"));
resetClient(c);
return 1;
} else if (cmd->type == REDIS_CMD_BULK && c->bulklen == -1) {
int bulklen = atoi(c->argv[c->argc-1]);
sdsfree(c->argv[c->argc-1]);
if (bulklen < 0 || bulklen > 1024*1024*1024) {
c->argc--;
c->argv[c->argc] = NULL;
addReplySds(c,sdsnew("-ERR invalid bulk write count\r\n"));
resetClient(c);
return 1;
}
c->argv[c->argc-1] = NULL;
c->argc--;
c->bulklen = bulklen+2; /* add two bytes for CR+LF */
/* It is possible that the bulk read is already in the
* buffer. Check this condition and handle it accordingly */
if ((signed)sdslen(c->querybuf) >= c->bulklen) {
c->argv[c->argc] = sdsnewlen(c->querybuf,c->bulklen-2);
c->argc++;
c->querybuf = sdsrange(c->querybuf,c->bulklen,-1);
} else {
return 1;
}
}
/* Exec the command */
cmd->proc(c); // 执行命令
resetClient(c); // 重置客户端
return 1;
}
static void readQueryFromClient(aeEventLoop *el, int fd, void *privdata, int mask) {
redisClient *c = (redisClient*) privdata;
char buf[REDIS_QUERYBUF_LEN];
int nread;
REDIS_NOTUSED(el);
REDIS_NOTUSED(mask);
nread = read(fd, buf, REDIS_QUERYBUF_LEN); // 从fd中读数据
if (nread == -1) {
if (errno == EAGAIN) { // 如果出错就进行处理
nread = 0;
} else {
redisLog(REDIS_DEBUG, "Reading from client: %s",strerror(errno));
freeClient(c);
return;
}
} else if (nread == 0) { // 如果为0则客户端关闭
redisLog(REDIS_DEBUG, "Client closed connection");
freeClient(c);
return;
}
if (nread) {
c->querybuf = sdscatlen(c->querybuf, buf, nread); // 保存读入的数据
c->lastinteraction = time(NULL);
} else {
return;
}
again:
if (c->bulklen == -1) { // 是否是批量读模式
/* Read the first line of the query */
char *p = strchr(c->querybuf,'\n'); // 解析第一行命令
size_t querylen;
if (p) {
sds query, *argv;
int argc, j;
query = c->querybuf;
c->querybuf = sdsempty();
querylen = 1+(p-(query));
if (sdslen(query) > querylen) {
/* leave data after the first line of the query in the buffer */
c->querybuf = sdscatlen(c->querybuf,query+querylen,sdslen(query)-querylen);
}
*p = '\0'; /* remove "\n" */
if (*(p-1) == '\r') *(p-1) = '\0'; /* and "\r" if any */
sdsupdatelen(query);
/* Now we can split the query in arguments */
if (sdslen(query) == 0) {
/* Ignore empty query */
sdsfree(query);
return;
}
argv = sdssplitlen(query,sdslen(query)," ",1,&argc); // 解析命令
sdsfree(query);
if (argv == NULL) oom("Splitting query in token");
for (j = 0; j < argc && j < REDIS_MAX_ARGS; j++) {
if (sdslen(argv[j])) {
c->argv[c->argc] = argv[j];
c->argc++;
} else {
sdsfree(argv[j]);
}
}
free(argv);
/* Execute the command. If the client is still valid
* after processCommand() return and there is something
* on the query buffer try to process the next command. */
if (processCommand(c) && sdslen(c->querybuf)) goto again; // 执行命令如果还有内容则继续解析命令
return;
} else if (sdslen(c->querybuf) >= 1024) {
redisLog(REDIS_DEBUG, "Client protocol error");
freeClient(c);
return;
}
} else {
/* Bulk read handling. Note that if we are at this point
the client already sent a command terminated with a newline,
we are reading the bulk data that is actually the last
argument of the command. */
int qbl = sdslen(c->querybuf);
if (c->bulklen <= qbl) {
/* Copy everything but the final CRLF as final argument */
c->argv[c->argc] = sdsnewlen(c->querybuf,c->bulklen-2);
c->argc++;
c->querybuf = sdsrange(c->querybuf,c->bulklen,-1);
processCommand(c); // 执行命令
return;
}
}
}
主要就是从客户端读取数据并解析出对应的命令,然后根据解析的命令查找不同的command去执行。对应的command查找过程如下;
static struct redisCommand cmdTable[] = {
{"get",getCommand,2,REDIS_CMD_INLINE},
{"set",setCommand,3,REDIS_CMD_BULK},
{"setnx",setnxCommand,3,REDIS_CMD_BULK},
{"del",delCommand,2,REDIS_CMD_INLINE},
{"exists",existsCommand,2,REDIS_CMD_INLINE},
{"incr",incrCommand,2,REDIS_CMD_INLINE},
{"decr",decrCommand,2,REDIS_CMD_INLINE},
{"rpush",rpushCommand,3,REDIS_CMD_BULK},
{"lpush",lpushCommand,3,REDIS_CMD_BULK},
{"rpop",rpopCommand,2,REDIS_CMD_INLINE},
{"lpop",lpopCommand,2,REDIS_CMD_INLINE},
{"llen",llenCommand,2,REDIS_CMD_INLINE},
{"lindex",lindexCommand,3,REDIS_CMD_INLINE},
{"lrange",lrangeCommand,4,REDIS_CMD_INLINE},
{"ltrim",ltrimCommand,4,REDIS_CMD_INLINE},
{"randomkey",randomkeyCommand,1,REDIS_CMD_INLINE},
{"select",selectCommand,2,REDIS_CMD_INLINE},
{"move",moveCommand,3,REDIS_CMD_INLINE},
{"rename",renameCommand,3,REDIS_CMD_INLINE},
{"renamenx",renamenxCommand,3,REDIS_CMD_INLINE},
{"keys",keysCommand,2,REDIS_CMD_INLINE},
{"dbsize",dbsizeCommand,1,REDIS_CMD_INLINE},
{"ping",pingCommand,1,REDIS_CMD_INLINE},
{"echo",echoCommand,2,REDIS_CMD_BULK},
{"save",saveCommand,1,REDIS_CMD_INLINE},
{"bgsave",bgsaveCommand,1,REDIS_CMD_INLINE},
{"shutdown",shutdownCommand,1,REDIS_CMD_INLINE},
{"lastsave",lastsaveCommand,1,REDIS_CMD_INLINE},
/* lpop, rpop, lindex, llen */
/* dirty, lastsave, info */
{"",NULL,0,0}
};
static struct redisCommand *lookupCommand(char *name) {
int j = 0;
while(cmdTable[j].name != NULL) { // 遍历列表直到找到或者达到表尾
if (!strcmp(name,cmdTable[j].name)) return &cmdTable[j]; // 通过字符串对比name是否相同
j++;
}
return NULL;
}
command执行过程
举例分析,假如是setCommand函数执行,
static void setGenericCommand(redisClient *c, int nx) {
int retval;
robj *o;
o = createObject(REDIS_STRING,c->argv[2]);
c->argv[2] = NULL;
retval = dictAdd(c->dict,c->argv[1],o); // 向数据库添加数据
if (retval == DICT_ERR) {
if (!nx)
dictReplace(c->dict,c->argv[1],o); // 检查是否是替换
else
decrRefCount(o);
} else {
/* Now the key is in the hash entry, don't free it */
c->argv[1] = NULL;
}
server.dirty++;
addReply(c,shared.ok); // 将ok返回客户端
}
static void setCommand(redisClient *c) {
return setGenericCommand(c,0);
}
具体的字典操作大家可以详细查看,就是哈希表的操作,接着就继续分析addReply。
addRepl返回数据到客户端
static void addReply(redisClient *c, robj *obj) {
if (listLength(c->reply) == 0 &&
aeCreateFileEvent(server.el, c->fd, AE_WRITABLE,
sendReplyToClient, c, NULL) == AE_ERR) return; // 创建写事件到驱动中
if (!listAddNodeTail(c->reply,obj)) oom("listAddNodeTail");
incrRefCount(obj);
}
通过addReply创建写事件到事件驱动中,触发的回调函数是sendReplyToClient;
static void sendReplyToClient(aeEventLoop *el, int fd, void *privdata, int mask) {
redisClient *c = privdata;
int nwritten = 0, totwritten = 0, objlen;
robj *o;
REDIS_NOTUSED(el);
REDIS_NOTUSED(mask);
while(listLength(c->reply)) {
o = listNodeValue(listFirst(c->reply));
objlen = sdslen(o->ptr);
if (objlen == 0) {
listDelNode(c->reply,listFirst(c->reply));
continue;
}
nwritten = write(fd, o->ptr+c->sentlen, objlen - c->sentlen); // 将数据写入fd中
if (nwritten <= 0) break;
c->sentlen += nwritten; // 添加已经发送的数据长度
totwritten += nwritten;
/* If we fully sent the object on head go to the next one */
if (c->sentlen == objlen) {
listDelNode(c->reply,listFirst(c->reply));
c->sentlen = 0;
}
}
if (nwritten == -1) { // 如果写入报错
if (errno == EAGAIN) {
nwritten = 0;
} else {
redisLog(REDIS_DEBUG,
"Error writing to client: %s", strerror(errno));
freeClient(c); // 释放并报错
return;
}
}
if (totwritten > 0) c->lastinteraction = time(NULL);
if (listLength(c->reply) == 0) { // 如果数据写入完成
c->sentlen = 0;
aeDeleteFileEvent(server.el,c->fd,AE_WRITABLE); // 删除该写事件
}
}
通过该函数调用就可以将数据返回给客户端。
aeMain事件驱动启动函数
该函数就是启动整个事件循环,来推送上节分析的所有流程。
void aeMain(aeEventLoop *eventLoop)
{
eventLoop->stop = 0;
while (!eventLoop->stop) // 只要没有停止
aeProcessEvents(eventLoop, AE_ALL_EVENTS); // 处理该函数
}
所有的处理逻辑都在aeProcessEvents函数中处理;
aeProcessEvents事件处理过程
int aeProcessEvents(aeEventLoop *eventLoop, int flags)
{
int maxfd = 0, numfd = 0, processed = 0;
fd_set rfds, wfds, efds;
aeFileEvent *fe = eventLoop->fileEventHead; // 获取事件头部
aeTimeEvent *te;
long long maxId;
AE_NOTUSED(flags);
/* Nothing to do? return ASAP */
if (!(flags & AE_TIME_EVENTS) && !(flags & AE_FILE_EVENTS)) return 0;
FD_ZERO(&rfds); // 读事件置空
FD_ZERO(&wfds); // 写事件置空
FD_ZERO(&efds);
/* Check file events */
if (flags & AE_FILE_EVENTS) { // 检查事件列表
while (fe != NULL) { // 遍历该链表
if (fe->mask & AE_READABLE) FD_SET(fe->fd, &rfds); // 添加读事件监听fd
if (fe->mask & AE_WRITABLE) FD_SET(fe->fd, &wfds); // 添加写事件监听fd
if (fe->mask & AE_EXCEPTION) FD_SET(fe->fd, &efds);
if (maxfd < fe->fd) maxfd = fe->fd;
numfd++;
fe = fe->next; // 获取下一个事件
}
}
/* Note that we want call select() even if there are no
* file events to process as long as we want to process time
* events, in order to sleep until the next time event is ready
* to fire. */
if (numfd || ((flags & AE_TIME_EVENTS) && !(flags & AE_DONT_WAIT))) {
int retval;
aeTimeEvent *shortest = NULL;
struct timeval tv, *tvp;
if (flags & AE_TIME_EVENTS && !(flags & AE_DONT_WAIT))
shortest = aeSearchNearestTimer(eventLoop);
if (shortest) {
long now_sec, now_ms;
/* Calculate the time missing for the nearest
* timer to fire. */
aeGetTime(&now_sec, &now_ms);
tvp = &tv;
tvp->tv_sec = shortest->when_sec - now_sec;
if (shortest->when_ms < now_ms) {
tvp->tv_usec = ((shortest->when_ms+1000) - now_ms)*1000;
tvp->tv_sec --;
} else {
tvp->tv_usec = (shortest->when_ms - now_ms)*1000;
}
} else {
/* If we have to check for events but need to return
* ASAP because of AE_DONT_WAIT we need to se the timeout
* to zero */
if (flags & AE_DONT_WAIT) {
tv.tv_sec = tv.tv_usec = 0;
tvp = &tv;
} else {
/* Otherwise we can block */
tvp = NULL; /* wait forever */
}
}
retval = select(maxfd+1, &rfds, &wfds, &efds, tvp); // 调用select检查是否有监听的事件发生
if (retval > 0) {
fe = eventLoop->fileEventHead; // 获取事件头部
while(fe != NULL) { // 依次遍历该事件列表
int fd = (int) fe->fd;
if ((fe->mask & AE_READABLE && FD_ISSET(fd, &rfds)) ||
(fe->mask & AE_WRITABLE && FD_ISSET(fd, &wfds)) ||
(fe->mask & AE_EXCEPTION && FD_ISSET(fd, &efds)))
{
int mask = 0;
if (fe->mask & AE_READABLE && FD_ISSET(fd, &rfds)) // 计算当前事件的mask值
mask |= AE_READABLE;
if (fe->mask & AE_WRITABLE && FD_ISSET(fd, &wfds))
mask |= AE_WRITABLE;
if (fe->mask & AE_EXCEPTION && FD_ISSET(fd, &efds))
mask |= AE_EXCEPTION;
fe->fileProc(eventLoop, fe->fd, fe->clientData, mask); // 调用事件驱动的回调函数
processed++;
/* After an event is processed our file event list
* may no longer be the same, so what we do
* is to clear the bit for this file descriptor and
* restart again from the head. */
fe = eventLoop->fileEventHead; // 获取下一个
FD_CLR(fd, &rfds); // 将该fd清理掉
FD_CLR(fd, &wfds);
FD_CLR(fd, &efds);
} else {
fe = fe->next; // 如果没有事件发生则下一个
}
}
}
}
/* Check time events */
if (flags & AE_TIME_EVENTS) { // 处理时间事件
te = eventLoop->timeEventHead;
maxId = eventLoop->timeEventNextId-1;
while(te) {
long now_sec, now_ms;
long long id;
if (te->id > maxId) {
te = te->next;
continue;
}
aeGetTime(&now_sec, &now_ms); // 获取当前时间
if (now_sec > te->when_sec ||
(now_sec == te->when_sec && now_ms >= te->when_ms)) // 比较当前时间 如果获取的时间事件的时间已经过期
{
int retval;
id = te->id;
retval = te->timeProc(eventLoop, id, te->clientData); // 调用时间事件的回调函数
/* After an event is processed our time event list may
* no longer be the same, so we restart from head.
* Still we make sure to don't process events registered
* by event handlers itself in order to don't loop forever.
* To do so we saved the max ID we want to handle. */
if (retval != AE_NOMORE) {
aeAddMillisecondsToNow(retval,&te->when_sec,&te->when_ms); // 如果时间事件返回的是时间, 则修改该时间事件到返回的时间以继续在指定时间到了之后再次调用
} else {
aeDeleteTimeEvent(eventLoop, id); // 删除该时间事件
}
te = eventLoop->timeEventHead; // 获取下一个事件
} else {
te = te->next; // 获取下一个事件
}
}
}
return processed; /* return the number of processed file/time events */
}
所有的驱动事件与时间驱动都在该函数中通过select的系统调用来完成,至此redis的主要的事件驱动逻辑执行完成。
总结
通过早期版本的redis的代码查看,可知redis是一个单线程的事件驱动框架,只有在讲数据保存到本地文件中时,才使用进程来保存数据。本文只是简单的分析了执行的主要过程,大家如果对redis的其他特性感兴趣,可自行查看代码。由于本人才疏学浅,如有错误请批评指正。