w := tl.writer()
	if pp := tl.mp.p.ptr(); pp != nil && !pp.trace.statusWasTraced(tl.gen) && pp.trace.acquireStatus(tl.gen) {
		w = w.writeProcStatus(uint64(pp.id), procStatus, pp.trace.inSweep)
	}
	if gp := tl.mp.curg; gp != nil && !gp.trace.statusWasTraced(tl.gen) && gp.trace.acquireStatus(tl.gen) {
		w = w.writeGoStatus(uint64(gp.goid), int64(tl.mp.procid), goStatus, gp.inMarkAssist, 0 /* no stack */)
	}
	return traceEventWriter{w}
}

// func switchToCrashStack0(fn func())
TEXT runtime·switchToCrashStack0(SB), NOSPLIT, $0-8
	MOVD fn+0(FP), CTXT	// context register
	MOVD	g_m(g), R2	// curm

	// set g to gcrash
	MOVD	$runtime·gcrash(SB), g	// g = &gcrash
	MOVD	R2, g_m(g)	// g.m = curm
	MOVD	g, m_g0(R2)	// curm.g0 = g

	// switch to crashstack
	I64Load (g_stack+stack_hi)(g)
	I64Const $(-4*8)
	I64Add
	I32WrapI64
	Set SP

		// for that memory to get freed.
		KeepAlive(ep)
		return
	}

	// stack
	if gp := getg(); gp.m.curg.stack.lo <= uintptr(p) && uintptr(p) < gp.m.curg.stack.hi {
		found := false
		var u unwinder
		for u.initAt(gp.m.curg.sched.pc, gp.m.curg.sched.sp, 0, gp.m.curg, 0); u.valid(); u.next() {
			if u.frame.sp <= uintptr(p) && uintptr(p) < u.frame.varp {
				found = true
				break
			}
		}

	lock(&work.wbufSpans.lock)
	if gcphase != _GCoff || work.wbufSpans.free.isEmpty() {
		unlock(&work.wbufSpans.lock)
		return false
	}
	systemstack(func() {
		gp := getg().m.curg
		for i := 0; i < batchSize && !(preemptible && gp.preempt); i++ {
			span := work.wbufSpans.free.first
			if span == nil {
				break
			}
			work.wbufSpans.free.remove(span)

			// If this is a self-scan, put the user G in
			// _Gwaiting to prevent self-deadlock. It may
			// already be in _Gwaiting if this is a mark
			// worker or we're in mark termination.
			userG := getg().m.curg
			selfScan := gp == userG && readgstatus(userG) == _Grunning
			if selfScan {
				casGToWaitingForGC(userG, _Grunning, waitReasonGarbageCollectionScan)
			}

			// TODO: suspendG blocks (and spins) until gp

		// Examine the execution tracer's view of the CPU profile samples. Filter it
		// to only include samples from the single test goroutine. Use the goroutine
		// ID that was recorded in the events: that should reflect getg().m.curg,
		// same as the profiler's labels (even when the M is using its g0 stack).
		totalTraceSamples := 0
		traceSamples := 0
		traceStacks := make(map[string]int)
		r, err := trace.NewReader(bytes.NewReader(tb))

// See racecallback for command codes.
TEXT	runtime·racecallbackthunk(SB), NOSPLIT|NOFRAME, $0
	// Handle command raceGetProcCmd (0) here.
	// First, code below assumes that we are on curg, while raceGetProcCmd
	// can be executed on g0. Second, it is called frequently, so will
	// benefit from this fast path.
	MOVD	$0, R0		// clear R0 since we came from C code
	CMP	R3, $0
	BNE	rest

		if gp.trace.statusWasTraced(gen) {
			return
		}
		// Scribble down information about this goroutine.
		ug := untracedG{gp: gp, mid: -1}
		systemstack(func() {
			me := getg().m.curg
			// We don't have to handle this G status transition because we
			// already eliminated ourselves from consideration above.
			casGToWaitingForGC(me, _Grunning, waitReasonTraceGoroutineStatus)

	semacquire1(addr, false, 0, 0, waitReasonSemacquire)
}

func semacquire1(addr *uint32, lifo bool, profile semaProfileFlags, skipframes int, reason waitReason) {
	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

	sigmask       sigset            // storage for saved signal mask
	tls           [tlsSlots]uintptr // thread-local storage (for x86 extern register)
	mstartfn      func()
	curg          *g       // current running goroutine
	caughtsig     guintptr // goroutine running during fatal signal
	p             puintptr // attached p for executing go code (nil if not executing go code)
	nextp         puintptr