context: skip the Done receive in Err when the channel is closedchan

src/context/benchmark_test.go

@@ -208,3 +208,42 @@ func BenchmarkErrCanceled(b *testing.B) {
 		}
 	}
 }
+
+func BenchmarkErrOKParallel(b *testing.B) {
+	ctx, cancel := WithCancel(Background())
+	defer cancel()
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if err := ctx.Err(); err != nil {
+				b.Fatalf("ctx.Err() = %v", err)
+			}
+		}
+	})
+}
+
+func BenchmarkErrCanceledParallel(b *testing.B) {
+	ctx, cancel := WithCancel(Background())
+	cancel()
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if err := ctx.Err(); err == nil {
+				b.Fatalf("ctx.Err() = %v", err)
+			}
+		}
+	})
+}
+
+// BenchmarkErrCanceledParallelDistinct gives each goroutine its own canceled
+// context, so the contexts share nothing with each other. Any contention here
+// is contention on state shared by every canceled context in the program.
+func BenchmarkErrCanceledParallelDistinct(b *testing.B) {
+	b.RunParallel(func(pb *testing.PB) {
+		ctx, cancel := WithCancel(Background())
+		cancel()
+		for pb.Next() {
+			if err := ctx.Err(); err == nil {
+				b.Fatalf("ctx.Err() = %v", err)
+			}
+		}
+	})
+}

src/context/context.go

@@ -465,7 +465,11 @@ func (c *cancelCtx) Err() error {
 	// An atomic load is ~5x faster than a mutex, which can matter in tight loops.
 	if err := c.err.Load(); err != nil {
 		// Ensure the done channel has been closed before returning a non-nil error.
-		<-c.Done()
+		// closedchan is closed at init and shared by every canceled context, so
+		// receiving from it would take one process-global lock for nothing.
+		if done := c.Done(); done != closedchan {
+			<-done
+		}
 		return err.(error)
 	}
 	return nil