}
	} else {
		v, ok := s.userArenaChunkFree.takeFromFront(size, typ.Align_)
		if ok {
			ptr = unsafe.Pointer(v)
		}
	}
	if ptr == nil {
		// Failed to allocate.
		mp.mallocing = 0
		releasem(mp)
		return nil
	}
	if s.needzero != 0 {
		throw("arena chunk needs zeroing, but should already be zeroed")
	}
	// Set up heap bitmap and do extra accounting.
	if typ.Pointers() {
		if cap >= 0 {

			"funcspdelta",
			"getm",
			"getMCache",
			"isDirectIface",
			"itabHashFunc",
			"nextslicecap",
			"noescape",
			"pcvalueCacheKey",
			"rand32",
			"readUnaligned32",
			"readUnaligned64",
			"releasem",
			"roundupsize",
			"stackmapdata",
			"stringStructOf",
			"subtract1",
			"subtractb",
			"tophash",
			"(*bmap).keys",
			"(*bmap).overflow",
			"(*waitq).enqueue",
			"funcInfo.entry",

						break
					} else {
						cache.inUse--
						releasem(mp)
						return val, pc
					}
				}
			}
		} else if debugCheckCache && (cache.inUse < 1 || cache.inUse > 2) {
			// Catch accounting errors or deeply reentrant use. In principle
			// "inUse" should never exceed 2.
			throw("cache.inUse out of range")
		}
		cache.inUse--
		releasem(mp)
	}

	if !f.valid() {

		mp := acquirem()
		for _, pp := range allp[:cap(allp)] {
			pp.trace.inSweep = false
			pp.trace.maySweep = false
			pp.trace.swept = 0
			pp.trace.reclaimed = 0
		}
		releasem(mp)
	}

	// Release the advance semaphore. If stopTrace is true we're still holding onto
	// traceShutdownSema.
	//
	// Do a direct handoff. Don't let one caller of traceAdvance starve
	// other calls to traceAdvance.

	switch {
	case p == 0:
		return nil, 0
	case p&(align-1) == 0:
		return unsafe.Pointer(p), size + align
	case GOOS == "windows":
		// On Windows we can't release pieces of a
		// reservation, so we release the whole thing and
		// re-reserve the aligned sub-region. This may race,
		// so we may have to try again.
		sysFreeOS(unsafe.Pointer(p), size+align)
		p = alignUp(p, align)

	// might end up being disabled when we load it. In that case we need to undo
	// what we did and bail.
	gen := trace.gen.Load()
	if gen == 0 {
		mp.trace.seqlock.Add(1)
		releasem(mp)
		return traceLocker{}
	}
	return traceLocker{mp, gen}
}

// traceTryAcquireEnabled is like traceAcquireEnabled but may return an invalid

	r1, r2, errno := syscall.Syscall6(trap, a1, a2, a3, a4, a5, a6)
	if GOARCH == "ppc64" || GOARCH == "ppc64le" {
		// TODO(https://go.dev/issue/51192 ): ppc64 doesn't use r2.
		r2 = 0
	}
	if errno != 0 {
		releasem(getg().m)
		allocmLock.unlock()
		startTheWorld(stw)
		return r1, r2, errno
	}

	perThreadSyscall = perThreadSyscallArgs{
		trap: trap,
		a1:   a1,
		a2:   a2,
		a3:   a3,
		a4:   a4,

// Helpers for Go. Must be NOSPLIT, must only call NOSPLIT functions, and must not block.

//go:nosplit
func acquirem() *m {
	gp := getg()
	gp.m.locks++
	return gp.m
}

//go:nosplit
func releasem(mp *m) {
	gp := getg()
	mp.locks--
	if mp.locks == 0 && gp.preempt {
		// restore the preemption request in case we've cleared it in newstack
		gp.stackguard0 = stackPreempt
	}
}

	l.gcEnabled = enableGC
	l.transitioning = true
	// N.B. finishGCTransition releases the lock.
	//
	// We don't release here to increase the chance that if there's a failure
	// to finish the transition, that we throw on failing to acquire the lock.
}

// finishGCTransition notifies the limiter that the GC transition is complete
// and releases ownership of it. It also accumulates STW time in the bucket.

		} else {
			mp.profStack[1] = gp.m.curg.sched.pc
			nstk += 1 + fpTracebackPCs(unsafe.Pointer(gp.m.curg.sched.bp), mp.profStack[2:])
		}
	}

	saveBlockEventStack(cycles, rate, mp.profStack[:nstk], which)
	releasem(mp)
}

// lockTimer assists with profiling contention on runtime-internal locks.
//
// There are several steps between the time that an M experiences contention and