diff --git a/src/runtime/malloc.go b/src/runtime/malloc.go
index 77a5a387687ef..eaf8db72208cd 100644
--- a/src/runtime/malloc.go
+++ b/src/runtime/malloc.go
@@ -976,6 +976,7 @@ func mallocgc(size uintptr, typ *_type, needzero bool) unsafe.Pointer {
 			throw("malloc called with no P")
 		}
 	}
+	var span *mspan
 	var x unsafe.Pointer
 	noscan := typ == nil || typ.ptrdata == 0
 	if size <= maxSmallSize {
@@ -1028,10 +1029,10 @@ func mallocgc(size uintptr, typ *_type, needzero bool) unsafe.Pointer {
 				return x
 			}
 			// Allocate a new maxTinySize block.
-			span := c.alloc[tinySpanClass]
+			span = c.alloc[tinySpanClass]
 			v := nextFreeFast(span)
 			if v == 0 {
-				v, _, shouldhelpgc = c.nextFree(tinySpanClass)
+				v, span, shouldhelpgc = c.nextFree(tinySpanClass)
 			}
 			x = unsafe.Pointer(v)
 			(*[2]uint64)(x)[0] = 0
@@ -1052,7 +1053,7 @@ func mallocgc(size uintptr, typ *_type, needzero bool) unsafe.Pointer {
 			}
 			size = uintptr(class_to_size[sizeclass])
 			spc := makeSpanClass(sizeclass, noscan)
-			span := c.alloc[spc]
+			span = c.alloc[spc]
 			v := nextFreeFast(span)
 			if v == 0 {
 				v, span, shouldhelpgc = c.nextFree(spc)
@@ -1063,15 +1064,14 @@ func mallocgc(size uintptr, typ *_type, needzero bool) unsafe.Pointer {
 			}
 		}
 	} else {
-		var s *mspan
 		shouldhelpgc = true
 		systemstack(func() {
-			s = largeAlloc(size, needzero, noscan)
+			span = largeAlloc(size, needzero, noscan)
 		})
-		s.freeindex = 1
-		s.allocCount = 1
-		x = unsafe.Pointer(s.base())
-		size = s.elemsize
+		span.freeindex = 1
+		span.allocCount = 1
+		x = unsafe.Pointer(span.base())
+		size = span.elemsize
 	}
 
 	var scanSize uintptr
@@ -1112,7 +1112,7 @@ func mallocgc(size uintptr, typ *_type, needzero bool) unsafe.Pointer {
 	// This may be racing with GC so do it atomically if there can be
 	// a race marking the bit.
 	if gcphase != _GCoff {
-		gcmarknewobject(uintptr(x), size, scanSize)
+		gcmarknewobject(span, uintptr(x), size, scanSize)
 	}
 
 	if raceenabled {
diff --git a/src/runtime/mgcmark.go b/src/runtime/mgcmark.go
index dafb4634b4204..fe988c46d9c46 100644
--- a/src/runtime/mgcmark.go
+++ b/src/runtime/mgcmark.go
@@ -1627,11 +1627,21 @@ func gcDumpObject(label string, obj, off uintptr) {
 //
 //go:nowritebarrier
 //go:nosplit
-func gcmarknewobject(obj, size, scanSize uintptr) {
+func gcmarknewobject(span *mspan, obj, size, scanSize uintptr) {
 	if useCheckmark { // The world should be stopped so this should not happen.
 		throw("gcmarknewobject called while doing checkmark")
 	}
-	markBitsForAddr(obj).setMarked()
+
+	// Mark object.
+	objIndex := span.objIndex(obj)
+	span.markBitsForIndex(objIndex).setMarked()
+
+	// Mark span.
+	arena, pageIdx, pageMask := pageIndexOf(span.base())
+	if arena.pageMarks[pageIdx]&pageMask == 0 {
+		atomic.Or8(&arena.pageMarks[pageIdx], pageMask)
+	}
+
 	gcw := &getg().m.p.ptr().gcw
 	gcw.bytesMarked += uint64(size)
 	gcw.scanWork += int64(scanSize)
diff --git a/src/runtime/mwbbuf.go b/src/runtime/mwbbuf.go
index f444452bab502..632769c11450d 100644
--- a/src/runtime/mwbbuf.go
+++ b/src/runtime/mwbbuf.go
@@ -296,6 +296,13 @@ func wbBufFlush1(_p_ *p) {
 			continue
 		}
 		mbits.setMarked()
+
+		// Mark span.
+		arena, pageIdx, pageMask := pageIndexOf(span.base())
+		if arena.pageMarks[pageIdx]&pageMask == 0 {
+			atomic.Or8(&arena.pageMarks[pageIdx], pageMask)
+		}
+
 		if span.spanclass.noscan() {
 			gcw.bytesMarked += uint64(span.elemsize)
 			continue