diff --git a/src/runtime/mgcmark.go b/src/runtime/mgcmark.go
index a079358e706..66d61bae1e8 100644
--- a/src/runtime/mgcmark.go
+++ b/src/runtime/mgcmark.go
@@ -144,11 +144,10 @@ func markroot(gcw *gcWork, i uint32) {
 			gp.waitsince = work.tstart
 		}
 
-		// Shrink a stack if not much of it is being used but not in the scan phase.
-		if gcphase == _GCmarktermination {
-			// Shrink during STW GCmarktermination phase thus avoiding
-			// complications introduced by shrinking during
-			// non-STW phases.
+		if gcphase == _GCmarktermination && status == _Gdead {
+			// Free gp's stack if necessary. Only do this
+			// during mark termination because otherwise
+			// _Gdead may be transient.
 			shrinkstack(gp)
 		}
 
@@ -599,6 +598,13 @@ func scanstack(gp *g) {
 		throw("can't scan gchelper stack")
 	}
 
+	// Shrink the stack if not much of it is being used. During
+	// concurrent GC, we can do this during concurrent mark.
+	if !work.markrootDone {
+		shrinkstack(gp)
+	}
+
+	// Prepare for stack barrier insertion/removal.
 	var sp, barrierOffset, nextBarrier uintptr
 	if gp.syscallsp != 0 {
 		sp = gp.syscallsp
@@ -647,6 +653,7 @@ func scanstack(gp *g) {
 		throw("scanstack in wrong phase")
 	}
 
+	// Scan the stack.
 	var cache pcvalueCache
 	gcw := &getg().m.p.ptr().gcw
 	n := 0
diff --git a/src/runtime/stack.go b/src/runtime/stack.go
index 06e64166176..fdd6710bad8 100644
--- a/src/runtime/stack.go
+++ b/src/runtime/stack.go
@@ -1069,7 +1069,8 @@ func gostartcallfn(gobuf *gobuf, fv *funcval) {
 // Called at garbage collection time.
 // gp must be stopped, but the world need not be.
 func shrinkstack(gp *g) {
-	if readgstatus(gp) == _Gdead {
+	gstatus := readgstatus(gp)
+	if gstatus&^_Gscan == _Gdead {
 		if gp.stack.lo != 0 {
 			// Free whole stack - it will get reallocated
 			// if G is used again.
@@ -1084,6 +1085,9 @@ func shrinkstack(gp *g) {
 	if gp.stack.lo == 0 {
 		throw("missing stack in shrinkstack")
 	}
+	if gstatus&_Gscan == 0 {
+		throw("bad status in shrinkstack")
+	}
 
 	if debug.gcshrinkstackoff > 0 {
 		return
@@ -1119,9 +1123,7 @@ func shrinkstack(gp *g) {
 		print("shrinking stack ", oldsize, "->", newsize, "\n")
 	}
 
-	oldstatus := casgcopystack(gp)
 	copystack(gp, newsize, false)
-	casgstatus(gp, _Gcopystack, oldstatus)
 }
 
 // freeStackSpans frees unused stack spans at the end of GC.