diff --git a/src/runtime/preempt.go b/src/runtime/preempt.go
index e1091cfd68427e..71c30898307a0f 100644
--- a/src/runtime/preempt.go
+++ b/src/runtime/preempt.go
@@ -282,7 +282,11 @@ func asyncPreempt()
 func asyncPreempt2() {
 	gp := getg()
 	gp.asyncSafePoint = true
-	mcall(preemptPark)
+	if gp.preemptStop {
+		mcall(preemptPark)
+	} else {
+		mcall(gopreempt_m)
+	}
 	gp.asyncSafePoint = false
 }
 
@@ -316,7 +320,8 @@ func init() {
 // wantAsyncPreempt returns whether an asynchronous preemption is
 // queued for gp.
 func wantAsyncPreempt(gp *g) bool {
-	return gp.preemptStop && readgstatus(gp)&^_Gscan == _Grunning
+	// Check both the G and the P.
+	return (gp.preempt || gp.m.p != 0 && gp.m.p.ptr().preempt) && readgstatus(gp)&^_Gscan == _Grunning
 }
 
 // isAsyncSafePoint reports whether gp at instruction PC is an
diff --git a/src/runtime/proc.go b/src/runtime/proc.go
index 6740169cf8ed2b..5ef9b3241768c3 100644
--- a/src/runtime/proc.go
+++ b/src/runtime/proc.go
@@ -2487,11 +2487,13 @@ func schedule() {
 	}
 
 top:
+	pp := _g_.m.p.ptr()
+	pp.preempt = false
+
 	if sched.gcwaiting != 0 {
 		gcstopm()
 		goto top
 	}
-	pp := _g_.m.p.ptr()
 	if pp.runSafePointFn != 0 {
 		runSafePointFn()
 	}
@@ -4654,6 +4656,13 @@ func preemptone(_p_ *p) bool {
 	// Setting gp->stackguard0 to StackPreempt folds
 	// preemption into the normal stack overflow check.
 	gp.stackguard0 = stackPreempt
+
+	// Request an async preemption of this P.
+	if preemptMSupported && debug.asyncpreemptoff == 0 {
+		_p_.preempt = true
+		preemptM(mp)
+	}
+
 	return true
 }
 
diff --git a/src/runtime/runtime2.go b/src/runtime/runtime2.go
index aba62930d41f20..4ee075a36a3d19 100644
--- a/src/runtime/runtime2.go
+++ b/src/runtime/runtime2.go
@@ -642,6 +642,10 @@ type p struct {
 	// Race context used while executing timer functions.
 	timerRaceCtx uintptr
 
+	// preempt is set to indicate that this P should be enter the
+	// scheduler ASAP (regardless of what G is running on it).
+	preempt bool
+
 	pad cpu.CacheLinePad
 }