Golang GMP模型 关于空闲P的G偷取顺序

golang GMP   图片截取知乎

 

但本地队列没有可运行的goroutine，会有steal机制，从从其他P里获取可运行的G ，本地队列运行顺序为 先从本地队列查询，全局队列查询，然后再从其他P里偷取，具体源码在runtime的proc.go里

// Finds a runnable goroutine to execute.
// Tries to steal from other P's, get g from local or global queue, poll network.
func findrunnable() (gp *g, inheritTime bool) {
	_g_ := getg()

	// The conditions here and in handoffp must agree: if
	// findrunnable would return a G to run, handoffp must start
	// an M.

top:
	_p_ := _g_.m.p.ptr()
	if sched.gcwaiting != 0 {
		gcstopm()
		goto top
	}
	if _p_.runSafePointFn != 0 {
		runSafePointFn()
	}

	now, pollUntil, _ := checkTimers(_p_, 0)

	if fingwait && fingwake {
		if gp := wakefing(); gp != nil {
			ready(gp, 0, true)
		}
	}
	if *cgo_yield != nil {
		asmcgocall(*cgo_yield, nil)
	}

	// local runq                  先查询本地队列
	if gp, inheritTime := runqget(_p_); gp != nil {
		return gp, inheritTime
	}

	// global runq                 再查询全局队列
	if sched.runqsize != 0 {
		lock(&sched.lock)
		gp := globrunqget(_p_, 0)
		unlock(&sched.lock)
		if gp != nil {
			return gp, false
		}
	}

	// Poll network.
	// This netpoll is only an optimization before we resort to stealing.
	// We can safely skip it if there are no waiters or a thread is blocked
	// in netpoll already. If there is any kind of logical race with that
	// blocked thread (e.g. it has already returned from netpoll, but does
	// not set lastpoll yet), this thread will do blocking netpoll below
	// anyway.
	if netpollinited() && atomic.Load(&netpollWaiters) > 0 && atomic.Load64(&sched.lastpoll) != 0 {
		if list := netpoll(0); !list.empty() { // non-blocking
			gp := list.pop()
			injectglist(&list)
			casgstatus(gp, _Gwaiting, _Grunnable)
			if trace.enabled {
				traceGoUnpark(gp, 0)
			}
			return gp, false
		}
	}

	// Steal work from other P's.         最后从其他P里偷取
	procs := uint32(gomaxprocs)
	ranTimer := false
	// If number of spinning M's >= number of busy P's, block.
	// This is necessary to prevent excessive CPU consumption
	// when GOMAXPROCS>>1 but the program parallelism is low.
	if !_g_.m.spinning && 2*atomic.Load(&sched.nmspinning) >= procs-atomic.Load(&sched.npidle) {
		goto stop
	}
	if !_g_.m.spinning {
		_g_.m.spinning = true
		atomic.Xadd(&sched.nmspinning, 1)
	}
	for i := 0; i < 4; i++ {
		for enum := stealOrder.start(fastrand()); !enum.done(); enum.next() {
			if sched.gcwaiting != 0 {
				goto top
			}
			stealRunNextG := i > 2 // first look for ready queues with more than 1 g
			p2 := allp[enum.position()]
			if _p_ == p2 {
				continue
			}
			if gp := runqsteal(_p_, p2, stealRunNextG); gp != nil {
				return gp, false
			}

			// Consider stealing timers from p2.
			// This call to checkTimers is the only place where
			// we hold a lock on a different P's timers.
			// Lock contention can be a problem here, so
			// initially avoid grabbing the lock if p2 is running
			// and is not marked for preemption. If p2 is running
			// and not being preempted we assume it will handle its
			// own timers.
			// If we're still looking for work after checking all
			// the P's, then go ahead and steal from an active P.
			if i > 2 || (i > 1 && shouldStealTimers(p2)) {
				tnow, w, ran := checkTimers(p2, now)
				now = tnow
				if w != 0 && (pollUntil == 0 || w < pollUntil) {
					pollUntil = w
				}
				if ran {
					// Running the timers may have
					// made an arbitrary number of G's
					// ready and added them to this P's
					// local run queue. That invalidates
					// the assumption of runqsteal
					// that is always has room to add
					// stolen G's. So check now if there
					// is a local G to run.
					if gp, inheritTime := runqget(_p_); gp != nil {
						return gp, inheritTime
					}
					ranTimer = true
				}
			}
		}
	}
	if ranTimer {
		// Running a timer may have made some goroutine ready.
		goto top
	}