2 files changed, 45 insertions, 4 deletions
diff --git a/src/runtime/mgcmark.go b/src/runtime/mgcmark.go
index 47456857e9..d05ad6549f 100644
--- a/src/runtime/mgcmark.go
+++ b/src/runtime/mgcmark.go
@@ -831,7 +831,10 @@ func gcDrain(gcw *gcWork, flags gcDrainFlags) {
 		if blocking {
 			b = gcw.get()
 		} else {
-			b = gcw.tryGet()
+			b = gcw.tryGetFast()
+			if b == 0 {
+				b = gcw.tryGet()
+			}
 		}
 		if b == 0 {
 			// work barrier reached or tryGet failed.
@@ -894,7 +897,11 @@ func gcDrainN(gcw *gcWork, scanWork int64) int64 {
 		//         PREFETCH(wbuf->obj[wbuf.nobj - 3];
 		//  }
 		//
-		b := gcw.tryGet()
+		b := gcw.tryGetFast()
+		if b == 0 {
+			b = gcw.tryGet()
+		}
+
 		if b == 0 {
 			break
 		}
@@ -1087,8 +1094,9 @@ func greyobject(obj, base, off uintptr, hbits heapBits, span *mspan, gcw *gcWork
 	// Previously we put the obj in an 8 element buffer that is drained at a rate
 	// to give the PREFETCH time to do its work.
 	// Use of PREFETCHNTA might be more appropriate than PREFETCH
-
-	gcw.put(obj)
+	if !gcw.putFast(obj) {
+		gcw.put(obj)
+	}
 }
 
 // gcDumpObject dumps the contents of obj for debugging and marks the
diff --git a/src/runtime/mgcwork.go b/src/runtime/mgcwork.go
index 63a3ade3a6..d04840b686 100644
--- a/src/runtime/mgcwork.go
+++ b/src/runtime/mgcwork.go
@@ -116,6 +116,22 @@ func (w *gcWork) put(obj uintptr) {
 	wbuf.nobj++
 }
 
+// putFast does a put and returns true if it can be done quickly
+// otherwise it returns false and the caller needs to call put.
+//go:nowritebarrier
+func (w *gcWork) putFast(obj uintptr) bool {
+	wbuf := w.wbuf1.ptr()
+	if wbuf == nil {
+		return false
+	} else if wbuf.nobj == len(wbuf.obj) {
+		return false
+	}
+
+	wbuf.obj[wbuf.nobj] = obj
+	wbuf.nobj++
+	return true
+}
+
 // tryGet dequeues a pointer for the garbage collector to trace.
 //
 // If there are no pointers remaining in this gcWork or in the global
@@ -147,6 +163,23 @@ func (w *gcWork) tryGet() uintptr {
 	return wbuf.obj[wbuf.nobj]
 }
 
+// tryGetFast dequeues a pointer for the garbage collector to trace
+// if one is readily available. Otherwise it returns 0 and
+// the caller is expected to call tryGet().
+//go:nowritebarrier
+func (w *gcWork) tryGetFast() uintptr {
+	wbuf := w.wbuf1.ptr()
+	if wbuf == nil {
+		return 0
+	}
+	if wbuf.nobj == 0 {
+		return 0
+	}
+
+	wbuf.nobj--
+	return wbuf.obj[wbuf.nobj]
+}
+
 // get dequeues a pointer for the garbage collector to trace, blocking
 // if necessary to ensure all pointers from all queues and caches have
 // been retrieved.  get returns 0 if there are no pointers remaining.