Golang atomic.Xadd函数代码示例

func RunSchedLocalQueueEmptyTest(iters int) {
	// Test that runq is not spuriously reported as empty.
	// Runq emptiness affects scheduling decisions and spurious emptiness
	// can lead to underutilization (both runnable Gs and idle Ps coexist
	// for arbitrary long time).
	done := make(chan bool, 1)
	p := new(p)
	gs := make([]g, 2)
	ready := new(uint32)
	for i := 0; i < iters; i++ {
		*ready = 0
		next0 := (i & 1) == 0
		next1 := (i & 2) == 0
		runqput(p, &gs[0], next0)
		go func() {
			for atomic.Xadd(ready, 1); atomic.Load(ready) != 2; {
			}
			if runqempty(p) {
				println("next:", next0, next1)
				throw("queue is empty")
			}
			done <- true
		}()
		for atomic.Xadd(ready, 1); atomic.Load(ready) != 2; {
		}
		runqput(p, &gs[1], next1)
		runqget(p)
		<-done
		runqget(p)
	}
}

func semrelease(addr *uint32) {
	root := semroot(addr)
	atomic.Xadd(addr, 1)

	// Easy case: no waiters?
	// This check must happen after the xadd, to avoid a missed wakeup
	// (see loop in semacquire).
	if atomic.Load(&root.nwait) == 0 {
		return
	}

	// Harder case: search for a waiter and wake it.
	lock(&root.lock)
	if atomic.Load(&root.nwait) == 0 {
		// The count is already consumed by another goroutine,
		// so no need to wake up another goroutine.
		unlock(&root.lock)
		return
	}
	s := root.head
	for ; s != nil; s = s.next {
		if s.elem == unsafe.Pointer(addr) {
			atomic.Xadd(&root.nwait, -1)
			root.dequeue(s)
			break
		}
	}
	unlock(&root.lock)
	if s != nil {
		if s.releasetime != 0 {
			s.releasetime = cputicks()
		}
		goready(s.g, 4)
	}
}

// Called from runtime.
func semacquire(addr *uint32, profile semaProfileFlags) {
	gp := getg()
	if gp != gp.m.curg {
		throw("semacquire not on the G stack")
	}

	// Easy case.
	if cansemacquire(addr) {
		return
	}

	// Harder case:
	//	increment waiter count
	//	try cansemacquire one more time, return if succeeded
	//	enqueue itself as a waiter
	//	sleep
	//	(waiter descriptor is dequeued by signaler)
	s := acquireSudog()
	root := semroot(addr)
	t0 := int64(0)
	s.releasetime = 0
	s.acquiretime = 0
	if profile&semaBlockProfile != 0 && blockprofilerate > 0 {
		t0 = cputicks()
		s.releasetime = -1
	}
	if profile&semaMutexProfile != 0 && mutexprofilerate > 0 {
		if t0 == 0 {
			t0 = cputicks()
		}
		s.acquiretime = t0
	}
	for {
		lock(&root.lock)
		// Add ourselves to nwait to disable "easy case" in semrelease.
		atomic.Xadd(&root.nwait, 1)
		// Check cansemacquire to avoid missed wakeup.
		if cansemacquire(addr) {
			atomic.Xadd(&root.nwait, -1)
			unlock(&root.lock)
			break
		}
		// Any semrelease after the cansemacquire knows we're waiting
		// (we set nwait above), so go to sleep.
		root.queue(addr, s)
		goparkunlock(&root.lock, "semacquire", traceEvGoBlockSync, 4)
		if cansemacquire(addr) {
			break
		}
	}
	if s.releasetime > 0 {
		blockevent(s.releasetime-t0, 3)
	}
	releaseSudog(s)
}

// Get a full work buffer off the work.full list.
// If nothing is available wait until all the other gc helpers have
// finished and then return nil.
// getfull acts as a barrier for work.nproc helpers. As long as one
// gchelper is actively marking objects it
// may create a workbuffer that the other helpers can work on.
// The for loop either exits when a work buffer is found
// or when _all_ of the work.nproc GC helpers are in the loop
// looking for work and thus not capable of creating new work.
// This is in fact the termination condition for the STW mark
// phase.
//go:nowritebarrier
func getfull(entry int) *workbuf {
	b := (*workbuf)(lfstackpop(&work.full))
	if b != nil {
		b.logget(entry)
		b.checknonempty()
		return b
	}

	incnwait := atomic.Xadd(&work.nwait, +1)
	if incnwait > work.nproc {
		println("runtime: work.nwait=", incnwait, "work.nproc=", work.nproc)
		throw("work.nwait > work.nproc")
	}
	for i := 0; ; i++ {
		if work.full != 0 {
			decnwait := atomic.Xadd(&work.nwait, -1)
			if decnwait == work.nproc {
				println("runtime: work.nwait=", decnwait, "work.nproc=", work.nproc)
				throw("work.nwait > work.nproc")
			}
			b = (*workbuf)(lfstackpop(&work.full))
			if b != nil {
				b.logget(entry)
				b.checknonempty()
				return b
			}
			incnwait := atomic.Xadd(&work.nwait, +1)
			if incnwait > work.nproc {
				println("runtime: work.nwait=", incnwait, "work.nproc=", work.nproc)
				throw("work.nwait > work.nproc")
			}
		}
		if work.nwait == work.nproc && work.markrootNext >= work.markrootJobs {
			return nil
		}
		_g_ := getg()
		if i < 10 {
			_g_.m.gcstats.nprocyield++
			procyield(20)
		} else if i < 20 {
			_g_.m.gcstats.nosyield++
			osyield()
		} else {
			_g_.m.gcstats.nsleep++
			usleep(100)
		}
	}
}

// sweeps one span
// returns number of pages returned to heap, or ^uintptr(0) if there is nothing to sweep
//go:nowritebarrier
func sweepone() uintptr {
	_g_ := getg()

	// increment locks to ensure that the goroutine is not preempted
	// in the middle of sweep thus leaving the span in an inconsistent state for next GC
	_g_.m.locks++
	sg := mheap_.sweepgen
	for {
		idx := atomic.Xadd(&sweep.spanidx, 1) - 1
		if idx >= uint32(len(work.spans)) {
			mheap_.sweepdone = 1
			_g_.m.locks--
			return ^uintptr(0)
		}
		s := work.spans[idx]
		if s.state != mSpanInUse {
			s.sweepgen = sg
			continue
		}
		if s.sweepgen != sg-2 || !atomic.Cas(&s.sweepgen, sg-2, sg-1) {
			continue
		}
		npages := s.npages
		if !s.sweep(false) {
			npages = 0
		}
		_g_.m.locks--
		return npages
	}
}

// sweeps one span
// returns number of pages returned to heap, or ^uintptr(0) if there is nothing to sweep
//go:nowritebarrier
func sweepone() uintptr {
	_g_ := getg()

	// increment locks to ensure that the goroutine is not preempted
	// in the middle of sweep thus leaving the span in an inconsistent state for next GC
	_g_.m.locks++
	sg := mheap_.sweepgen
	for {
		idx := atomic.Xadd(&sweep.spanidx, 1) - 1
		if idx >= uint32(len(work.spans)) {
			mheap_.sweepdone = 1
			_g_.m.locks--
			if debug.gcpacertrace > 0 && idx == uint32(len(work.spans)) {
				print("pacer: sweep done at heap size ", memstats.heap_live>>20, "MB; allocated ", mheap_.spanBytesAlloc>>20, "MB of spans; swept ", mheap_.pagesSwept, " pages at ", mheap_.sweepPagesPerByte, " pages/byte\n")
			}
			return ^uintptr(0)
		}
		s := work.spans[idx]
		if s.state != mSpanInUse {
			s.sweepgen = sg
			continue
		}
		if s.sweepgen != sg-2 || !atomic.Cas(&s.sweepgen, sg-2, sg-1) {
			continue
		}
		npages := s.npages
		if !s.sweep(false) {
			npages = 0
		}
		_g_.m.locks--
		return npages
	}
}

// lockextra locks the extra list and returns the list head.
// The caller must unlock the list by storing a new list head
// to extram. If nilokay is true, then lockextra will
// return a nil list head if that's what it finds. If nilokay is false,
// lockextra will keep waiting until the list head is no longer nil.
//go:nosplit
func lockextra(nilokay bool) *m {
	const locked = 1

	incr := false
	for {
		old := atomic.Loaduintptr(&extram)
		if old == locked {
			yield := osyield
			yield()
			continue
		}
		if old == 0 && !nilokay {
			if !incr {
				// Add 1 to the number of threads
				// waiting for an M.
				// This is cleared by newextram.
				atomic.Xadd(&extraMWaiters, 1)
				incr = true
			}
			usleep(1)
			continue
		}
		if atomic.Casuintptr(&extram, old, locked) {
			return (*m)(unsafe.Pointer(old))
		}
		yield := osyield
		yield()
		continue
	}
}

// pop removes and returns a span from buffer b, or nil if b is empty.
// pop is safe to call concurrently with other pop operations, but NOT
// to call concurrently with push.
func (b *gcSweepBuf) pop() *mspan {
	cursor := atomic.Xadd(&b.index, -1)
	if int32(cursor) < 0 {
		atomic.Xadd(&b.index, +1)
		return nil
	}

	// There are no concurrent spine or block modifications during
	// pop, so we can omit the atomics.
	top, bottom := cursor/gcSweepBlockEntries, cursor%gcSweepBlockEntries
	blockp := (**gcSweepBlock)(add(b.spine, sys.PtrSize*uintptr(top)))
	block := *blockp
	s := block.spans[bottom]
	// Clear the pointer for block(i).
	block.spans[bottom] = nil
	return s
}

// gcDrainN blackens grey objects until it has performed roughly
// scanWork units of scan work or the G is preempted. This is
// best-effort, so it may perform less work if it fails to get a work
// buffer. Otherwise, it will perform at least n units of work, but
// may perform more because scanning is always done in whole object
// increments. It returns the amount of scan work performed.
//
// The caller goroutine must be in a preemptible state (e.g.,
// _Gwaiting) to prevent deadlocks during stack scanning. As a
// consequence, this must be called on the system stack.
//
//go:nowritebarrier
//go:systemstack
func gcDrainN(gcw *gcWork, scanWork int64) int64 {
	if !writeBarrier.needed {
		throw("gcDrainN phase incorrect")
	}

	// There may already be scan work on the gcw, which we don't
	// want to claim was done by this call.
	workFlushed := -gcw.scanWork

	gp := getg().m.curg
	for !gp.preempt && workFlushed+gcw.scanWork < scanWork {
		// See gcDrain comment.
		if work.full == 0 {
			gcw.balance()
		}

		// This might be a good place to add prefetch code...
		// if(wbuf.nobj > 4) {
		//         PREFETCH(wbuf->obj[wbuf.nobj - 3];
		//  }
		//
		b := gcw.tryGetFast()
		if b == 0 {
			b = gcw.tryGet()
		}

		if b == 0 {
			// Try to do a root job.
			//
			// TODO: Assists should get credit for this
			// work.
			if work.markrootNext < work.markrootJobs {
				job := atomic.Xadd(&work.markrootNext, +1) - 1
				if job < work.markrootJobs {
					markroot(gcw, job)
					continue
				}
			}
			// No heap or root jobs.
			break
		}
		scanobject(b, gcw)

		// Flush background scan work credit.
		if gcw.scanWork >= gcCreditSlack {
			atomic.Xaddint64(&gcController.scanWork, gcw.scanWork)
			workFlushed += gcw.scanWork
			gcw.scanWork = 0
		}
	}

	// Unlike gcDrain, there's no need to flush remaining work
	// here because this never flushes to bgScanCredit and
	// gcw.dispose will flush any remaining work to scanWork.

	return workFlushed + gcw.scanWork
}

func semrelease(addr *uint32) {
	root := semroot(addr)
	atomic.Xadd(addr, 1)

	// Easy case: no waiters?
	// This check must happen after the xadd, to avoid a missed wakeup
	// (see loop in semacquire).
	if atomic.Load(&root.nwait) == 0 {
		return
	}

	// Harder case: search for a waiter and wake it.
	lock(&root.lock)
	if atomic.Load(&root.nwait) == 0 {
		// The count is already consumed by another goroutine,
		// so no need to wake up another goroutine.
		unlock(&root.lock)
		return
	}
	s := root.head
	for ; s != nil; s = s.next {
		if s.elem == unsafe.Pointer(addr) {
			atomic.Xadd(&root.nwait, -1)
			root.dequeue(s)
			break
		}
	}
	if s != nil {
		if s.acquiretime != 0 {
			t0 := cputicks()
			for x := root.head; x != nil; x = x.next {
				if x.elem == unsafe.Pointer(addr) {
					x.acquiretime = t0
				}
			}
			mutexevent(t0-s.acquiretime, 3)
		}
	}
	unlock(&root.lock)
	if s != nil { // May be slow, so unlock first
		readyWithTime(s, 5)
	}
}

//go:nosplit
func semawakeup(mp *m) {
	atomic.Xadd(&mp.waitsemacount, 1)
	ret := thrwakeup(uintptr(unsafe.Pointer(&mp.waitsemacount)), 1)
	if ret != 0 && ret != _ESRCH {
		// semawakeup can be called on signal stack.
		systemstack(func() {
			print("thrwakeup addr=", &mp.waitsemacount, " sem=", mp.waitsemacount, " ret=", ret, "\n")
		})
	}
}

//go:nosplit
func semawakeup(mp *m) {
	atomic.Xadd(&mp.waitsemacount, 1)
	// From NetBSD's _lwp_unpark(2) manual:
	// "If the target LWP is not currently waiting, it will return
	// immediately upon the next call to _lwp_park()."
	ret := lwp_unpark(int32(mp.procid), unsafe.Pointer(&mp.waitsemacount))
	if ret != 0 && ret != _ESRCH {
		// semawakeup can be called on signal stack.
		systemstack(func() {
			print("thrwakeup addr=", &mp.waitsemacount, " sem=", mp.waitsemacount, " ret=", ret, "\n")
		})
	}
}

func dopanic(unused int) {
	gp := getg()
	if gp.sig != 0 {
		signame := signame(gp.sig)
		if signame != "" {
			print("[signal ", signame)
		} else {
			print("[signal ", hex(gp.sig))
		}
		print(" code=", hex(gp.sigcode0), " addr=", hex(gp.sigcode1), " pc=", hex(gp.sigpc), "]\n")
	}

	level, all, docrash := gotraceback()
	_g_ := getg()
	if level > 0 {
		if gp != gp.m.curg {
			all = true
		}
		if gp != gp.m.g0 {
			print("\n")
			goroutineheader(gp)
			traceback(0)
		} else if level >= 2 || _g_.m.throwing > 0 {
			print("\nruntime stack:\n")
			traceback(0)
		}
		if !didothers && all {
			didothers = true
			tracebackothers(gp)
		}
	}
	unlock(&paniclk)

	if atomic.Xadd(&panicking, -1) != 0 {
		// Some other m is panicking too.
		// Let it print what it needs to print.
		// Wait forever without chewing up cpu.
		// It will exit when it's done.
		lock(&deadlock)
		lock(&deadlock)
	}

	if docrash {
		crash()
	}

	exit(2)
}

func startpanic() {
	_g_ := getg()
	// Uncomment when mheap_ is in Go.
	// if mheap_.cachealloc.size == 0 { // very early
	//	print("runtime: panic before malloc heap initialized\n")
	//	_g_.m.mallocing = 1 // tell rest of panic not to try to malloc
	// } else
	if _g_.m.mcache == nil { // can happen if called from signal handler or throw
		_g_.m.mcache = allocmcache()
	}

	switch _g_.m.dying {
	case 0:
		_g_.m.dying = 1
		_g_.writebuf = nil
		atomic.Xadd(&panicking, 1)
		lock(&paniclk)
		if debug.schedtrace > 0 || debug.scheddetail > 0 {
			schedtrace(true)
		}
		freezetheworld()
		return
	case 1:
		// Something failed while panicing, probably the print of the
		// argument to panic().  Just print a stack trace and exit.
		_g_.m.dying = 2
		print("panic during panic\n")
		dopanic(0)
		exit(3)
		fallthrough
	case 2:
		// This is a genuine bug in the runtime, we couldn't even
		// print the stack trace successfully.
		_g_.m.dying = 3
		print("stack trace unavailable\n")
		exit(4)
		fallthrough
	default:
		// Can't even print!  Just exit.
		exit(5)
	}
}

func queuefinalizer(p unsafe.Pointer, fn *funcval, nret uintptr, fint *_type, ot *ptrtype) {
	lock(&finlock)
	if finq == nil || finq.cnt == uint32(len(finq.fin)) {
		if finc == nil {
			finc = (*finblock)(persistentalloc(_FinBlockSize, 0, &memstats.gc_sys))
			finc.alllink = allfin
			allfin = finc
			if finptrmask[0] == 0 {
				// Build pointer mask for Finalizer array in block.
				// Check assumptions made in finalizer1 array above.
				if (unsafe.Sizeof(finalizer{}) != 5*sys.PtrSize ||
					unsafe.Offsetof(finalizer{}.fn) != 0 ||
					unsafe.Offsetof(finalizer{}.arg) != sys.PtrSize ||
					unsafe.Offsetof(finalizer{}.nret) != 2*sys.PtrSize ||
					unsafe.Offsetof(finalizer{}.fint) != 3*sys.PtrSize ||
					unsafe.Offsetof(finalizer{}.ot) != 4*sys.PtrSize) {
					throw("finalizer out of sync")
				}
				for i := range finptrmask {
					finptrmask[i] = finalizer1[i%len(finalizer1)]
				}
			}
		}
		block := finc
		finc = block.next
		block.next = finq
		finq = block
	}
	f := &finq.fin[finq.cnt]
	atomic.Xadd(&finq.cnt, +1) // Sync with markroots
	f.fn = fn
	f.nret = nret
	f.fint = fint
	f.ot = ot
	f.arg = p
	fingwake = true
	unlock(&finlock)
}