本篇文章内容基于go1.14.2分析
golang的chan是一个内置类型,作为csp编程的核心数据结构,其底层数据结构是一个叫hchan的struct:
type hchan struct {
qcount uint // 队列中的元素数量
dataqsiz uint // (环形)队列的大小
buf unsafe.Pointer // 队列的指针
elemsize uint16 // 元素大小
closed uint32 // 是否已close
elemtype *_type // 元素类型
sendx uint // 环形队列中,send的位置
recvx uint // 环形队列中 recv的位置
recvq waitq // 读取等待队列
sendq waitq // 发送等待队列
lock mutex // 互斥锁
}
如图所示,chan最核心的部分由一个环形队列和2个waitq组成,环形队列用于存放数据(带缓冲的情况下),waitq用于实现阻塞和恢复goroutine。
chan的相关操作
对chan的操作有:make、读、写、close,当然还有select,这里只讨论前面四个操作。
创建 chan
当在代码中使用make创建chan时,编译器会根据情况自动替换成makechan64 或者makechan,makechan64 其实还是调用了makechan函数。
func makechan(t *chantype, size int) *hchan {
elem := t.elem
// 确保元素类型的size < 2^16,
if elem.size >= 1<<16 {
throw("makechan: invalid channel element type")
}
// 检查内存对齐
if hchanSize%maxAlign != 0 || elem.align > maxAlign {
throw("makechan: bad alignment")
}
// 计算缓冲区所需分配内存大小
mem, overflow := math.MulUintptr(elem.size, uintptr(size))
if overflow || mem > maxAlloc-hchanSize || size < 0 {
panic(plainError("makechan: size out of range"))
}
var c *hchan
switch {
case mem == 0:
// 即不带缓冲区的情况,只需要调用mallocgc分配
c = (*hchan)(mallocgc(hchanSize, nil, true))
// 理解为空地址
c.buf = c.raceaddr()
case elem.ptrdata == 0:
// 元素类型不包含指针的情况
c = (*hchan)(mallocgc(hchanSize+mem, nil, true))
c.buf = add(unsafe.Pointer(c), hchanSize)
default:
// 默认情况下:包含指针
c = new(hchan)
c.buf = mallocgc(mem, elem, true)
}
c.elemsize = uint16(elem.size)
c.elemtype = elem
c.dataqsiz = uint(size)
if debugChan {
print("makechan: chan=", c, "; elemsize=", elem.size, "; dataqsiz=", size, "\n")
}
return c
}
chan 写操作
当对chan进行写入“ch <- interface{}” 时,会被编译器替换成chansend1函数的调用,最终还是调用了chansend函数:
//elem 是待写入元素的地址
func chansend1(c *hchan, elem unsafe.Pointer) {
chansend(c, elem, true, getcallerpc())
}
先看看chansend的函数签名,只需关注ep和block这个两个参数即可,ep是要写入数据的地址,block表示是否阻塞式的调用
func chansend(c *hchan, ep unsafe.Pointer, block bool, callerpc uintptr) bool
chansend有以下几种处理流程:
-
当对一个nil chan进行写操作时,如果是非阻塞调用,直接返回;否则将当前协程挂起
// chansend 对一个 nil chan发送数据时,如果是非阻塞则直接返回,否则将当前协程挂起 if c == nil { if !block { return false } gopark(nil, nil, waitReasonChanSendNilChan, traceEvGoStop, 2) throw("unreachable") }
-
非阻塞模式且chan未close,没有缓冲区且没有等待接收或者缓冲区满的情况下,直接return false。
// 1. 非阻塞模式且chan未close // 2. 没有缓冲区且没有等待接收 或者 缓冲区满的情况下 // 满足以上条件直接return false if !block && c.closed == 0 && ((c.dataqsiz == 0 && c.recvq.first == nil) || (c.dataqsiz > 0 && c.qcount == c.dataqsiz)) { return false }
-
c.recvq中有等待读的接收者,将其出队,将数据直接copy给接收者,并唤醒接收者。
// 有等待的接收的goroutine // 出队,传递数据 if sg := c.recvq.dequeue(); sg != nil { // Found a waiting receiver. We pass the value we want to send // directly to the receiver, bypassing the channel buffer (if any). send(c, sg, ep, func() { unlock(&c.lock) }, 3) return true }
recvq是一个双向链表,每个sudog会关联上一个reader(被阻塞的g)
当sudog出队后,会调用send方法,通过sendDirect 实现数据在两个地址之间拷贝,最后调用goready唤醒reader(被阻塞的g)
func send(c *hchan, sg *sudog, ep unsafe.Pointer, unlockf func(), skip int) { // ... 剔除无关代码 if sg.elem != nil { // 直接将数据拷贝到变量ep所在的地址 sendDirect(c.elemtype, sg, ep) sg.elem = nil } gp := sg.g unlockf() gp.param = unsafe.Pointer(sg) if sg.releasetime != 0 { sg.releasetime = cputicks() } //将reader的goroutine唤起 goready(gp, skip+1) }
-
缓冲区未满的情况下,数据放入环形缓冲区即可。
// 缓冲区未满 // 将数据放到缓冲区 if c.qcount < c.dataqsiz { // Space is available in the channel buffer. Enqueue the element to send. // 存放位置 qp := chanbuf(c, c.sendx) if raceenabled { raceacquire(qp) racerelease(qp) } typedmemmove(c.elemtype, qp, ep) // 指针自增 c.sendx++ if c.sendx == c.dataqsiz { c.sendx = 0 } c.qcount++ unlock(&c.lock) return true }
-
缓冲区已满,阻塞模式下关联一个sudog数据结构并进入c.sendq队列,挂起当前协程。
// 阻塞的情况 gp := getg() //拿到当前g mysg := acquireSudog() // 获取一个sudog mysg.releasetime = 0 if t0 != 0 { mysg.releasetime = -1 mysg.elem = ep //关联ep,即待写入的数据地址 mysg.waitlink = nil mysg.g = gp mysg.isSelect = false mysg.c = c gp.waiting = mysg gp.param = nil c.sendq.enqueue(mysg) // 入队 // Signal to anyone trying to shrink our stack that we're about // to park on a channel. The window between when this G's status // changes and when we set gp.activeStackChans is not safe for // stack shrinking. atomic.Store8(&gp.parkingOnChan, 1) // 将g休眠,让出cpu // gopark后,需等待reader来唤醒它 gopark(chanparkcommit, unsafe.Pointer(&c.lock), waitReasonChanSend, traceEvGoBlockSend, 2) // 唤醒过后 // Ensure the value being sent is kept alive until the // receiver copies it out. The sudog has a pointer to the // stack object, but sudogs aren't considered as roots of the // stack tracer. // 保持数据不被回收 KeepAlive(ep) // someone woke us up. if mysg != gp.waiting { throw("G waiting list is corrupted") } gp.waiting = nil gp.activeStackChans = false if gp.param == nil { if c.closed == 0 { throw("chansend: spurious wakeup") } panic(plainError("send on closed channel")) } gp.param = nil if mysg.releasetime > 0 { blockevent(mysg.releasetime-t0, 2) } mysg.c = nil releaseSudog(mysg) return true
chan 读操作
当对chan进行读操作时,编译器会替换成 chanrecv1或者chanrecv2函数,最终会调用chanrecv函数处理读取
// v := <- ch
func chanrecv1(c *hchan, elem unsafe.Pointer) {
chanrecv(c, elem, true)
}
// v, ok := <- ch
func chanrecv2(c *hchan, elem unsafe.Pointer) (received bool) {
_, received = chanrecv(c, elem, true)
return
}
和chansend一样,chanrecv也是支持非阻塞式的调用
func chanrecv(c *hchan, ep unsafe.Pointer, block bool) (selected, received bool)
chanrecv有以下几种处理流程:
-
读nil chan,如果是非阻塞,直接返回;如果是阻塞式,将当前协程挂起。
// 读阻塞 if c == nil { if !block { return } gopark(nil, nil, waitReasonChanReceiveNilChan, traceEvGoStop, 2) throw("unreachable") }
-
非阻塞模式下,没有缓冲区且没有等待写的writer或者缓冲区没数据,直接返回。
if !block && (c.dataqsiz == 0 && c.sendq.first == nil || c.dataqsiz > 0 && atomic.Loaduint(&c.qcount) == 0) && atomic.Load(&c.closed) == 0 { return }
-
chan已经被close,并且队列中没有数据时,会将存放值的变量清零,然后返回。
// c已经被close 并且 没有数据 // 清除ep指针 if c.closed != 0 && c.qcount == 0 { if raceenabled { raceacquire(c.raceaddr()) } unlock(&c.lock) if ep != nil { typedmemclr(c.elemtype, ep) } return true, false }
-
sendq中有等待的writer,writer出队,并调用recv函数
// 从sendq中取出sender if sg := c.sendq.dequeue(); sg != nil { // Found a waiting sender. If buffer is size 0, receive value // directly from sender. Otherwise, receive from head of queue // and add sender's value to the tail of the queue (both map to // the same buffer slot because the queue is full). // 从sender中读取数据 recv(c, sg, ep, func() { unlock(&c.lock) }, 3) return true, true }
recv在这分两种处理:如果ch不带缓冲区的话,直接将writer的sg.elem数据拷贝到ep;如果带缓冲区的话,此时缓冲区肯定满了,那么就从缓冲区队列头部取出数据拷贝至ep,然后将writer的sg.elem数据拷贝到缓冲区中,最后唤醒writer(g)
func recv(c *hchan, sg *sudog, ep unsafe.Pointer, unlockf func(), skip int) { // 不带缓冲区的情况 // 直接copy from sender if c.dataqsiz == 0 { if raceenabled { racesync(c, sg) } if ep != nil { // copy data from sender recvDirect(c.elemtype, sg, ep) } } else { // Queue is full. Take the item at the // head of the queue. Make the sender enqueue // its item at the tail of the queue. Since the // queue is full, those are both the same slot. // 队列已满 // 队列元素出队 qp := chanbuf(c, c.recvx) if raceenabled { raceacquire(qp) racerelease(qp) raceacquireg(sg.g, qp) racereleaseg(sg.g, qp) } // copy data from queue to receiver // 数据拷贝给ep if ep != nil { typedmemmove(c.elemtype, ep, qp) } // copy data from sender to queue // 将sender的数据拷贝到这个槽中 typedmemmove(c.elemtype, qp, sg.elem) c.recvx++ if c.recvx == c.dataqsiz { c.recvx = 0 } c.sendx = c.recvx // c.sendx = (c.sendx+1) % c.dataqsiz } // 置空 sg.elem = nil gp := sg.g unlockf() gp.param = unsafe.Pointer(sg) if sg.releasetime != 0 { sg.releasetime = cputicks() } // 唤醒sender协程 goready(gp, skip+1) }
-
直接从缓冲队列中读数。
// 带缓冲区 if c.qcount > 0 { // Receive directly from queue // 直接buf中取 qp := chanbuf(c, c.recvx) if raceenabled { raceacquire(qp) racerelease(qp) } // 拷贝数据到ep指针 if ep != nil { typedmemmove(c.elemtype, ep, qp) } // 清除qp typedmemclr(c.elemtype, qp) c.recvx++ if c.recvx == c.dataqsiz { c.recvx = 0 } c.qcount-- unlock(&c.lock) return true, true }
-
阻塞的情况,缓冲区没有数据,且没有writer
// 阻塞 gp := getg() //拿到当前的goroutine mysg := acquireSudog() // 获取一个sudog mysg.releasetime = 0 if t0 != 0 { mysg.releasetime = -1 } //sudog 关联 mysg.elem = ep mysg.waitlink = nil gp.waiting = mysg mysg.g = gp mysg.isSelect = false mysg.c = c gp.param = nil c.recvq.enqueue(mysg) //入队 // Signal to anyone trying to shrink our stack that we're about // to park on a channel. The window between when this G's status // changes and when we set gp.activeStackChans is not safe for // stack shrinking. atomic.Store8(&gp.parkingOnChan, 1) // 挂起当前goroutine,等待writer唤醒 gopark(chanparkcommit, unsafe.Pointer(&c.lock), waitReasonChanReceive, traceEvGoBlockRecv, 2) // 唤醒后 if mysg != gp.waiting { throw("G waiting list is corrupted") } gp.waiting = nil gp.activeStackChans = false if mysg.releasetime > 0 { blockevent(mysg.releasetime-t0, 2) } closed := gp.param == nil gp.param = nil // sudog解除关联 mysg.c = nil // 释放sudog releaseSudog(mysg)
close 关闭操作
当close一个chan时,编译器会替换成对closechan函数的调用,将closed字段置为1,并将recvq和sendq中的goroutine释放唤醒,对sendq中未写入的数据做清除,且writer会发生panic异常。
func closechan(c *hchan) {
if c == nil {
panic(plainError("close of nil channel"))
}
// 加锁
lock(&c.lock)
// 不可重复close
if c.closed != 0 {
unlock(&c.lock)
panic(plainError("close of closed channel"))
}
if raceenabled {
callerpc := getcallerpc()
racewritepc(c.raceaddr(), callerpc, funcPC(closechan))
racerelease(c.raceaddr())
}
c.closed = 1
var glist gList
// 释放所有的
for {
// 出队
sg := c.recvq.dequeue()
if sg == nil {
break
}
// 清零
if sg.elem != nil {
typedmemclr(c.elemtype, sg.elem)
sg.elem = nil
}
if sg.releasetime != 0 {
sg.releasetime = cputicks()
}
gp := sg.g
gp.param = nil
if raceenabled {
raceacquireg(gp, c.raceaddr())
}
glist.push(gp)
}
// 释放所有writer
for {
// 出队
sg := c.sendq.dequeue()
if sg == nil {
break
}
// 丢弃数据
sg.elem = nil
if sg.releasetime != 0 {
sg.releasetime = cputicks()
}
gp := sg.g
gp.param = nil
if raceenabled {
raceacquireg(gp, c.raceaddr())
}
glist.push(gp)
}
unlock(&c.lock)
// 唤醒所有g
for !glist.empty() {
gp := glist.pop()
gp.schedlink = 0
goready(gp, 3)
}
}
chan使用小技巧
-
避免read、write一个nil chan
func main() { ch := make(chan int,1) go func() { time.Sleep(1*time.Second) ch = nil }() ch<-1 ch<-1 // 协程直接挂起 }
-
从chan中read时,使用带指示的访问方式,读取的时候无法感知到close的关闭
func main() { ch := make(chan int) go func() { ch <- 10 close(ch) }() for { select { // case i, ok := <-ch: // if ok { // break //} case i := <-ch: fmt.Println(i) time.Sleep(100 * time.Millisecond) } } }
-
从chan中read时,不要使用已存在变量接收, chan close之后,缓冲区没有数据的话,使用存在变量读取时,会将变量清零
func main() { a := 10 ch := make(chan int,1) fmt.Println("before close a is: ", a) // a is 10 close(ch) a = <-ch fmt.Println("after close a is: ", a) // a is 0 }
-
使用select+default可以实现 chan的无阻塞读取
// 使用select反射包实现无阻塞读写 func tryRead(ch chan int) (int, bool) { var cases []reflect.SelectCase caseRead := reflect.SelectCase{ Dir: reflect.SelectRecv, Chan: reflect.ValueOf(ch), } cases = append(cases, caseRead) cases = append(cases, reflect.SelectCase{ Dir: reflect.SelectDefault, }) _, v, ok := reflect.Select(cases) if ok { return (v.Interface()).(int), ok } return 0, ok } func tryWrite(ch chan int, data int) bool { var cases []reflect.SelectCase caseWrite := reflect.SelectCase{ Dir: reflect.SelectSend, Chan: reflect.ValueOf(ch), Send: reflect.ValueOf(data), } cases = append(cases, caseWrite) cases = append(cases, reflect.SelectCase{ Dir: reflect.SelectDefault, }) chosen, _, _ := reflect.Select(cases) return chosen == 0 } // 使用select + default实现无阻塞读写 func tryRead2(ch chan int) (int, bool) { select { case v, ok := <-ch: return v, ok default: return 0, false } } func tryWrite2(ch chan int, data int) bool { select { case ch <- data: return true default: return false } }
原因是如果select的case中存在default,对chan的读写会使用无阻塞的方法
func selectnbsend(c *hchan, elem unsafe.Pointer) (selected bool) { return chansend(c, elem, false, getcallerpc()) } func selectnbrecv(elem unsafe.Pointer, c *hchan) (selected bool) { selected, _ = chanrecv(c, elem, false) return }