diff --git a/src/runtime/mgcmark.go b/src/runtime/mgcmark.go
index 7850f86bb2..d6ee7ff6fa 100644
--- a/src/runtime/mgcmark.go
+++ b/src/runtime/mgcmark.go
@@ -302,26 +302,27 @@ func markrootBlock(b0, n0 uintptr, ptrmask0 *uint8, gcw *gcWork, shard int) {
 //TODO go:nowritebarrier
 func markrootFreeGStacks() {
 	// Take list of dead Gs with stacks.
-	lock(&sched.gflock)
-	list := sched.gfreeStack
-	sched.gfreeStack = nil
-	unlock(&sched.gflock)
-	if list == nil {
+	lock(&sched.gFree.lock)
+	list := sched.gFree.stack
+	sched.gFree.stack = gList{}
+	unlock(&sched.gFree.lock)
+	if list.empty() {
 		return
 	}
 
 	// Free stacks.
-	tail := list
-	for gp := list; gp != nil; gp = gp.schedlink.ptr() {
+	q := gQueue{list.head, list.head}
+	for gp := list.head.ptr(); gp != nil; gp = gp.schedlink.ptr() {
 		shrinkstack(gp)
-		tail = gp
+		// Manipulate the queue directly since the Gs are
+		// already all linked the right way.
+		q.tail.set(gp)
 	}
 
 	// Put Gs back on the free list.
-	lock(&sched.gflock)
-	tail.schedlink.set(sched.gfreeNoStack)
-	sched.gfreeNoStack = list
-	unlock(&sched.gflock)
+	lock(&sched.gFree.lock)
+	sched.gFree.noStack.pushAll(q)
+	unlock(&sched.gFree.lock)
 }
 
 // markrootSpans marks roots for one shard of work.spans.
diff --git a/src/runtime/proc.go b/src/runtime/proc.go
index 2a780e49ee..5cb7f13016 100644
--- a/src/runtime/proc.go
+++ b/src/runtime/proc.go
@@ -3471,25 +3471,21 @@ func gfput(_p_ *p, gp *g) {
 		gp.stackguard0 = 0
 	}
 
-	gp.schedlink.set(_p_.gfree)
-	_p_.gfree = gp
-	_p_.gfreecnt++
-	if _p_.gfreecnt >= 64 {
-		lock(&sched.gflock)
-		for _p_.gfreecnt >= 32 {
-			_p_.gfreecnt--
-			gp = _p_.gfree
-			_p_.gfree = gp.schedlink.ptr()
+	_p_.gFree.push(gp)
+	_p_.gFree.n++
+	if _p_.gFree.n >= 64 {
+		lock(&sched.gFree.lock)
+		for _p_.gFree.n >= 32 {
+			_p_.gFree.n--
+			gp = _p_.gFree.pop()
 			if gp.stack.lo == 0 {
-				gp.schedlink.set(sched.gfreeNoStack)
-				sched.gfreeNoStack = gp
+				sched.gFree.noStack.push(gp)
 			} else {
-				gp.schedlink.set(sched.gfreeStack)
-				sched.gfreeStack = gp
+				sched.gFree.stack.push(gp)
 			}
-			sched.ngfree++
+			sched.gFree.n++
 		}
-		unlock(&sched.gflock)
+		unlock(&sched.gFree.lock)
 	}
 }
 
@@ -3497,44 +3493,42 @@ func gfput(_p_ *p, gp *g) {
 // If local list is empty, grab a batch from global list.
 func gfget(_p_ *p) *g {
 retry:
-	gp := _p_.gfree
-	if gp == nil && (sched.gfreeStack != nil || sched.gfreeNoStack != nil) {
-		lock(&sched.gflock)
-		for _p_.gfreecnt < 32 {
-			if sched.gfreeStack != nil {
-				// Prefer Gs with stacks.
-				gp = sched.gfreeStack
-				sched.gfreeStack = gp.schedlink.ptr()
-			} else if sched.gfreeNoStack != nil {
-				gp = sched.gfreeNoStack
-				sched.gfreeNoStack = gp.schedlink.ptr()
-			} else {
-				break
+	if _p_.gFree.empty() && (!sched.gFree.stack.empty() || !sched.gFree.noStack.empty()) {
+		lock(&sched.gFree.lock)
+		// Move a batch of free Gs to the P.
+		for _p_.gFree.n < 32 {
+			// Prefer Gs with stacks.
+			gp := sched.gFree.stack.pop()
+			if gp == nil {
+				gp = sched.gFree.noStack.pop()
+				if gp == nil {
+					break
+				}
 			}
-			_p_.gfreecnt++
-			sched.ngfree--
-			gp.schedlink.set(_p_.gfree)
-			_p_.gfree = gp
+			sched.gFree.n--
+			_p_.gFree.push(gp)
+			_p_.gFree.n++
 		}
-		unlock(&sched.gflock)
+		unlock(&sched.gFree.lock)
 		goto retry
 	}
-	if gp != nil {
-		_p_.gfree = gp.schedlink.ptr()
-		_p_.gfreecnt--
-		if gp.stack.lo == 0 {
-			// Stack was deallocated in gfput. Allocate a new one.
-			systemstack(func() {
-				gp.stack = stackalloc(_FixedStack)
-			})
-			gp.stackguard0 = gp.stack.lo + _StackGuard
-		} else {
-			if raceenabled {
-				racemalloc(unsafe.Pointer(gp.stack.lo), gp.stack.hi-gp.stack.lo)
-			}
-			if msanenabled {
-				msanmalloc(unsafe.Pointer(gp.stack.lo), gp.stack.hi-gp.stack.lo)
-			}
+	gp := _p_.gFree.pop()
+	if gp == nil {
+		return nil
+	}
+	_p_.gFree.n--
+	if gp.stack.lo == 0 {
+		// Stack was deallocated in gfput. Allocate a new one.
+		systemstack(func() {
+			gp.stack = stackalloc(_FixedStack)
+		})
+		gp.stackguard0 = gp.stack.lo + _StackGuard
+	} else {
+		if raceenabled {
+			racemalloc(unsafe.Pointer(gp.stack.lo), gp.stack.hi-gp.stack.lo)
+		}
+		if msanenabled {
+			msanmalloc(unsafe.Pointer(gp.stack.lo), gp.stack.hi-gp.stack.lo)
 		}
 	}
 	return gp
@@ -3542,21 +3536,18 @@ retry:
 
 // Purge all cached G's from gfree list to the global list.
 func gfpurge(_p_ *p) {
-	lock(&sched.gflock)
-	for _p_.gfreecnt != 0 {
-		_p_.gfreecnt--
-		gp := _p_.gfree
-		_p_.gfree = gp.schedlink.ptr()
+	lock(&sched.gFree.lock)
+	for !_p_.gFree.empty() {
+		gp := _p_.gFree.pop()
+		_p_.gFree.n--
 		if gp.stack.lo == 0 {
-			gp.schedlink.set(sched.gfreeNoStack)
-			sched.gfreeNoStack = gp
+			sched.gFree.noStack.push(gp)
 		} else {
-			gp.schedlink.set(sched.gfreeStack)
-			sched.gfreeStack = gp
+			sched.gFree.stack.push(gp)
 		}
-		sched.ngfree++
+		sched.gFree.n++
 	}
-	unlock(&sched.gflock)
+	unlock(&sched.gFree.lock)
 }
 
 // Breakpoint executes a breakpoint trap.
@@ -3669,9 +3660,9 @@ func badunlockosthread() {
 }
 
 func gcount() int32 {
-	n := int32(allglen) - sched.ngfree - int32(atomic.Load(&sched.ngsys))
+	n := int32(allglen) - sched.gFree.n - int32(atomic.Load(&sched.ngsys))
 	for _, _p_ := range allp {
-		n -= _p_.gfreecnt
+		n -= _p_.gFree.n
 	}
 
 	// All these variables can be changed concurrently, so the result can be inconsistent.
@@ -4581,7 +4572,7 @@ func schedtrace(detailed bool) {
 			if mp != nil {
 				id = mp.id
 			}
-			print("  P", i, ": status=", _p_.status, " schedtick=", _p_.schedtick, " syscalltick=", _p_.syscalltick, " m=", id, " runqsize=", t-h, " gfreecnt=", _p_.gfreecnt, "\n")
+			print("  P", i, ": status=", _p_.status, " schedtick=", _p_.schedtick, " syscalltick=", _p_.syscalltick, " m=", id, " runqsize=", t-h, " gfreecnt=", _p_.gFree.n, "\n")
 		} else {
 			// In non-detailed mode format lengths of per-P run queues as:
 			// [len1 len2 len3 len4]
diff --git a/src/runtime/runtime2.go b/src/runtime/runtime2.go
index 5bd37e49be..bbbe1ee852 100644
--- a/src/runtime/runtime2.go
+++ b/src/runtime/runtime2.go
@@ -506,8 +506,10 @@ type p struct {
 	runnext guintptr
 
 	// Available G's (status == Gdead)
-	gfree    *g
-	gfreecnt int32
+	gFree struct {
+		gList
+		n int32
+	}
 
 	sudogcache []*sudog
 	sudogbuf   [128]*sudog
@@ -578,10 +580,12 @@ type schedt struct {
 	runqsize int32
 
 	// Global cache of dead G's.
-	gflock       mutex
-	gfreeStack   *g
-	gfreeNoStack *g
-	ngfree       int32
+	gFree struct {
+		lock    mutex
+		stack   gList // Gs with stacks
+		noStack gList // Gs without stacks
+		n       int32
+	}
 
 	// Central cache of sudog structs.
 	sudoglock  mutex