1、初始化
初始化是通过InitServerLast中完成的。如果配置的线程数为1，则不创建创建，限制线程数为128。循环创建io线程列表 (主要用于存放需要写或者读的client)，初始化io线程pending标识(0表io线程已经处理完毕)，线程互斥量，线程。在创建io线程前，锁住io_thread_mutex[i]，避免io线程运行。
void InitServerLast() {
bioInit();
initThreadedIO();
set_jemalloc_bg_thread(server.jemalloc_bg_thread);
server.initial_memory_usage = zmalloc_used_memory();
}
void initThreadedIO(void) {
server.io_threads_active = 0; /* We start with threads not active. */
/* Don't spawn any thread if the user selected a single thread:
* we'll handle I/O directly from the main thread. */
if (server.io_threads_num == 1) return;
if (server.io_threads_num > IO_THREADS_MAX_NUM) {
serverLog(LL_WARNING,"Fatal: too many I/O threads configured. "
"The maximum number is %d.", IO_THREADS_MAX_NUM);
exit(1);
}
/* Spawn and initialize the I/O threads. */
for (int i = 0; i < server.io_threads_num; i++) {
/* Things we do for all the threads including the main thread. */
io_threads_list[i] = listCreate();
if (i == 0) continue; /* Thread 0 is the main thread. */
/* Things we do only for the additional threads. */
pthread_t tid;
pthread_mutex_init(&io_threads_mutex[i],NULL);
io_threads_pending[i] = 0;
pthread_mutex_lock(&io_threads_mutex[i]); /* Thread will be stopped. */
if (pthread_create(&tid,NULL,IOThreadMain,(void*)(long)i) != 0) {
serverLog(LL_WARNING,"Fatal: Can't initialize IO thread.");
exit(1);
}
io_threads[i] = tid;
}
}
io线程工作时即线程对应的io_threads_pending不等于0，在线程对应的io_threads_list中遍历获取client，根据操作类型io_thread_op(读或者写)来写入或者读取，操作完后，清空对应的io_threads_list并且把io_theads_pending设置为0，等待在下一次的事件循环中处理。
void *IOThreadMain(void *myid) {
/* The ID is the thread number (from 0 to server.iothreads_num-1), and is
* used by the thread to just manipulate a single sub-array of clients. */
long id = (unsigned long)myid;
char thdname[16];
snprintf(thdname, sizeof(thdname), "io_thd_%ld", id);
redis_set_thread_title(thdname);
redisSetCpuAffinity(server.server_cpulist);
makeThreadKillable();
while(1) {
/* Wait for start */
for (int j = 0; j < 1000000; j++) {
if (io_threads_pending[id] != 0) break;
}
/* Give the main thread a chance to stop this thread. */
if (io_threads_pending[id] == 0) {
pthread_mutex_lock(&io_threads_mutex[id]);
pthread_mutex_unlock(&io_threads_mutex[id]);
continue;
}
serverAssert(io_threads_pending[id] != 0);
if (tio_debug) printf("[%ld] %d to handle\n", id, (int)listLength(io_threads_list[id]));
/* Process: note that the main thread will never touch our list
* before we drop the pending count to 0. */
listIter li;
listNode *ln;
listRewind(io_threads_list[id],&li);
while((ln = listNext(&li))) {
client *c = listNodeValue(ln);
if (io_threads_op == IO_THREADS_OP_WRITE) {
writeToClient(c,0);
} else if (io_threads_op == IO_THREADS_OP_READ) {
readQueryFromClient(c->conn);
} else {
serverPanic("io_threads_op value is unknown");
}
}
listEmpty(io_threads_list[id]);
io_threads_pending[id] = 0;
if (tio_debug) printf("[%ld] Done\n", id);
}
}
2、运行
在aeMain事件循环中会开启io线程
void aeMain(aeEventLoop *eventLoop) {
eventLoop->stop = 0;
while (!eventLoop->stop) {
aeProcessEvents(eventLoop, AE_ALL_EVENTS|
AE_CALL_BEFORE_SLEEP|
AE_CALL_AFTER_SLEEP);
}
}
int aeProcessEvents(aeEventLoop *eventLoop, int flags)
{
int processed = 0, numevents;
/* Nothing to do? return ASAP */
if (!(flags & AE_TIME_EVENTS) && !(flags & AE_FILE_EVENTS)) return 0;
/* 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 (eventLoop->maxfd != -1 ||
((flags & AE_TIME_EVENTS) && !(flags & AE_DONT_WAIT))) {
int j;
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;
aeGetTime(&now_sec, &now_ms);
tvp = &tv;
/* How many milliseconds we need to wait for the next
* time event to fire? */
long long ms =
(shortest->when_sec - now_sec)*1000 +
shortest->when_ms - now_ms;
if (ms > 0) {
tvp->tv_sec = ms/1000;
tvp->tv_usec = (ms % 1000)*1000;
} else {
tvp->tv_sec = 0;
tvp->tv_usec = 0;
}
} else {
/* If we have to check for events but need to return
* ASAP because of AE_DONT_WAIT we need to set 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 */
}
}
if (eventLoop->flags & AE_DONT_WAIT) {
tv.tv_sec = tv.tv_usec = 0;
tvp = &tv;
}
if (eventLoop->beforesleep != NULL && flags & AE_CALL_BEFORE_SLEEP)
eventLoop->beforesleep(eventLoop);
/* Call the multiplexing API, will return only on timeout or when
* some event fires. */
numevents = aeApiPoll(eventLoop, tvp);
/* After sleep callback. */
if (eventLoop->aftersleep != NULL && flags & AE_CALL_AFTER_SLEEP)
eventLoop->aftersleep(eventLoop);
for (j = 0; j < numevents; j++) {
aeFileEvent *fe = &eventLoop->events[eventLoop->fired[j].fd];
int mask = eventLoop->fired[j].mask;
int fd = eventLoop->fired[j].fd;
int fired = 0; /* Number of events fired for current fd. */
/* Normally we execute the readable event first, and the writable
* event later. This is useful as sometimes we may be able
* to serve the reply of a query immediately after processing the
* query.
*
* However if AE_BARRIER is set in the mask, our application is
* asking us to do the reverse: never fire the writable event
* after the readable. In such a case, we invert the calls.
* This is useful when, for instance, we want to do things
* in the beforeSleep() hook, like fsyncing a file to disk,
* before replying to a client. */
int invert = fe->mask & AE_BARRIER;
/* Note the "fe->mask & mask & ..." code: maybe an already
* processed event removed an element that fired and we still
* didn't processed, so we check if the event is still valid.
*
* Fire the readable event if the call sequence is not
* inverted. */
if (!invert && fe->mask & mask & AE_READABLE) {
fe->rfileProc(eventLoop,fd,fe->clientData,mask);
fired++;
fe = &eventLoop->events[fd]; /* Refresh in case of resize. */
}
/* Fire the writable event. */
if (fe->mask & mask & AE_WRITABLE) {
if (!fired || fe->wfileProc != fe->rfileProc) {
fe->wfileProc(eventLoop,fd,fe->clientData,mask);
fired++;
}
}
/* If we have to invert the call, fire the readable event now
* after the writable one. */
if (invert) {
fe = &eventLoop->events[fd]; /* Refresh in case of resize. */
if ((fe->mask & mask & AE_READABLE) &&
(!fired || fe->wfileProc != fe->rfileProc))
{
fe->rfileProc(eventLoop,fd,fe->clientData,mask);
fired++;
}
}
processed++;
}
}
/* Check time events */
if (flags & AE_TIME_EVENTS)
processed += processTimeEvents(eventLoop);
return processed; /* return the number of processed file/time events */
}
在aeApiPoll之前会调用eventLoop->beforeSleep，beforeSleep在服务器初始化时通过initServer设置的
aeSetBeforeSleepProc(server.el,beforeSleep);
aeSetAfterSleepProc(server.el,afterSleep);
beforeSlepp处理代码为
void beforeSleep(struct aeEventLoop *eventLoop) {
UNUSED(eventLoop);
size_t zmalloc_used = zmalloc_used_memory();
if (zmalloc_used > server.stat_peak_memory)
server.stat_peak_memory = zmalloc_used;
/* Just call a subset of vital functions in case we are re-entering
* the event loop from processEventsWhileBlocked(). Note that in this
* case we keep track of the number of events we are processing, since
* processEventsWhileBlocked() wants to stop ASAP if there are no longer
* events to handle. */
if (ProcessingEventsWhileBlocked) {
uint64_t processed = 0;
processed += handleClientsWithPendingReadsUsingThreads();
processed += tlsProcessPendingData();
processed += handleClientsWithPendingWrites();
processed += freeClientsInAsyncFreeQueue();
server.events_processed_while_blocked += processed;
return;
}
/* Handle precise timeouts of blocked clients. */
handleBlockedClientsTimeout();
/* We should handle pending reads clients ASAP after event loop. */
handleClientsWithPendingReadsUsingThreads();
/* Handle TLS pending data. (must be done before flushAppendOnlyFile) */
tlsProcessPendingData();
/* If tls still has pending unread data don't sleep at all. */
aeSetDontWait(server.el, tlsHasPendingData());
/* Call the Redis Cluster before sleep function. Note that this function
* may change the state of Redis Cluster (from ok to fail or vice versa),
* so it's a good idea to call it before serving the unblocked clients
* later in this function. */
if (server.cluster_enabled) clusterBeforeSleep();
/* Run a fast expire cycle (the called function will return
* ASAP if a fast cycle is not needed). */
if (server.active_expire_enabled && server.masterhost == NULL)
activeExpireCycle(ACTIVE_EXPIRE_CYCLE_FAST);
/* Unblock all the clients blocked for synchronous replication
* in WAIT. */
if (listLength(server.clients_waiting_acks))
processClientsWaitingReplicas();
/* Check if there are clients unblocked by modules that implement
* blocking commands. */
if (moduleCount()) moduleHandleBlockedClients();
/* Try to process pending commands for clients that were just unblocked. */
if (listLength(server.unblocked_clients))
processUnblockedClients();
/* Send all the slaves an ACK request if at least one client blocked
* during the previous event loop iteration. Note that we do this after
* processUnblockedClients(), so if there are multiple pipelined WAITs
* and the just unblocked WAIT gets blocked again, we don't have to wait
* a server cron cycle in absence of other event loop events. See #6623. */
if (server.get_ack_from_slaves) {
robj *argv[3];
argv[0] = createStringObject("REPLCONF",8);
argv[1] = createStringObject("GETACK",6);
argv[2] = createStringObject("*",1); /* Not used argument. */
replicationFeedSlaves(server.slaves, server.slaveseldb, argv, 3);
decrRefCount(argv[0]);
decrRefCount(argv[1]);
decrRefCount(argv[2]);
server.get_ack_from_slaves = 0;
}
/* Send the invalidation messages to clients participating to the
* client side caching protocol in broadcasting (BCAST) mode. */
trackingBroadcastInvalidationMessages();
/* Write the AOF buffer on disk */
flushAppendOnlyFile(0);
/* Handle writes with pending output buffers. */
handleClientsWithPendingWritesUsingThreads();
/* Close clients that need to be closed asynchronous */
freeClientsInAsyncFreeQueue();
/* Try to process blocked clients every once in while. Example: A module
* calls RM_SignalKeyAsReady from within a timer callback (So we don't
* visit processCommand() at all). */
handleClientsBlockedOnKeys();
/* Before we are going to sleep, let the threads access the dataset by
* releasing the GIL. Redis main thread will not touch anything at this
* time. */
if (moduleCount()) moduleReleaseGIL();
/* Do NOT add anything below moduleReleaseGIL !!! */
}
当有写操作时，会将client添加到server.clients_pending_write中，同时将数据添加到client的缓冲区中。
int prepareClientToWrite(client *c) {
if (c->flags & (CLIENT_LUA|CLIENT_MODULE)) return C_OK;
if (c->flags & CLIENT_CLOSE_ASAP) return C_ERR;
if (c->flags & (CLIENT_REPLY_OFF|CLIENT_REPLY_SKIP)) return C_ERR;
if ((c->flags & CLIENT_MASTER) &&
!(c->flags & CLIENT_MASTER_FORCE_REPLY)) return C_ERR;
if (!c->conn) return C_ERR;
if (!clientHasPendingReplies(c) && !(c->flags & CLIENT_PENDING_READ))
clientInstallWriteHandler(c);
return C_OK;
}
void clientInstallWriteHandler(client *c) {
if (!(c->flags & CLIENT_PENDING_WRITE) &&
(c->replstate == REPL_STATE_NONE ||
(c->replstate == SLAVE_STATE_ONLINE && !c->repl_put_online_on_ack)))
{
c->flags |= CLIENT_PENDING_WRITE;
listAddNodeHead(server.clients_pending_write,c);
}
}
int _addReplyToBuffer(client *c, const char *s, size_t len) {
size_t available = sizeof(c->buf)-c->bufpos;
if (c->flags & CLIENT_CLOSE_AFTER_REPLY) return C_OK;
if (listLength(c->reply) > 0) return C_ERR;
if (len > available) return C_ERR;
memcpy(c->buf+c->bufpos,s,len);
c->bufpos+=len;
return C_OK;
}
beforeSleep在处理是否有等待写的客户端时handleClientsWithPendingWritesUsingThreads，如果io线程没有激活，此时会开启io线程。将需要写的客户端分配到线程对应的io_threads_list中。同时设置io_threads_op为IO_THREADS_OP_WRITE，将各个线程的io_threads_pending设置为非0来开启线程写操作。主线程也会处理一部分客户端的写操作。等待线程操作处理完成。最后遍历client，如果仍然有需要写操作，设置connection的WriteHandler
int handleClientsWithPendingWritesUsingThreads(void) {
int processed = listLength(server.clients_pending_write);
if (processed == 0) return 0; /* Return ASAP if there are no clients. */
/* If I/O threads are disabled or we have few clients to serve, don't
* use I/O threads, but thejboring synchronous code. */
if (server.io_threads_num == 1 || stopThreadedIOIfNeeded()) {
return handleClientsWithPendingWrites();
}
/* Start threads if needed. */
if (!server.io_threads_active) startThreadedIO();
if (tio_debug) printf("%d TOTAL WRITE pending clients\n", processed);
/* Distribute the clients across N different lists. */
listIter li;
listNode *ln;
listRewind(server.clients_pending_write,&li);
int item_id = 0;
while((ln = listNext(&li))) {
client *c = listNodeValue(ln);
c->flags &= ~CLIENT_PENDING_WRITE;
/* Remove clients from the list of pending writes since
* they are going to be closed ASAP. */
if (c->flags & CLIENT_CLOSE_ASAP) {
listDelNode(server.clients_pending_write, ln);
continue;
}
int target_id = item_id % server.io_threads_num;
listAddNodeTail(io_threads_list[target_id],c);
item_id++;
}
/* Give the start condition to the waiting threads, by setting the
* start condition atomic var. */
io_threads_op = IO_THREADS_OP_WRITE;
for (int j = 1; j < server.io_threads_num; j++) {
int count = listLength(io_threads_list[j]);
io_threads_pending[j] = count;
}
/* Also use the main thread to process a slice of clients. */
listRewind(io_threads_list[0],&li);
while((ln = listNext(&li))) {
client *c = listNodeValue(ln);
writeToClient(c,0);
}
listEmpty(io_threads_list[0]);
/* Wait for all the other threads to end their work. */
while(1) {
unsigned long pending = 0;
for (int j = 1; j < server.io_threads_num; j++)
pending += io_threads_pending[j];
if (pending == 0) break;
}
if (tio_debug) printf("I/O WRITE All threads finshed\n");
/* Run the list of clients again to install the write handler where
* needed. */
listRewind(server.clients_pending_write,&li);
while((ln = listNext(&li))) {
client *c = listNodeValue(ln);
/* Install the write handler if there are pending writes in some
* of the clients. */
if (clientHasPendingReplies(c) &&
connSetWriteHandler(c->conn, sendReplyToClient) == AE_ERR)
{
freeClientAsync(c);
}
}
listEmpty(server.clients_pending_write);
/* Update processed count on server */
server.stat_io_writes_processed += processed;
return processed;
}
|