5g, 6g, 8g: registerize variables again

My previous CL:

changeset:   9645:ce2e5f44b310
user:        Russ Cox <rsc@golang.org>
date:        Tue Sep 06 10:24:21 2011 -0400
summary:     gc: unify stack frame layout

introduced a bug wherein no variables were
being registerized, making Go programs 2-3x
slower than they had been before.

This CL fixes that bug (along with some others
it was hiding) and adds a test that optimization
makes at least one test case faster.

R=ken2
CC=golang-dev
http://codereview.appspot.com/5174045

2 files changed, 72 insertions, 0 deletions

diff --git a/test/fixedbugs/bug369.dir/pkg.go b/test/fixedbugs/bug369.dir/pkg.go
new file mode 100644
index 000000000..cf5704192
--- /dev/null
+++ b/test/fixedbugs/bug369.dir/pkg.go
@@ -0,0 +1,15 @@
+// Copyright 2011 The Go Authors.  All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package pkg
+
+func NonASCII(b []byte, i int) int {
+	for i = 0; i < len(b); i++ {
+		if b[i] >= 0x80 {
+			break
+		}
+	}
+	return i
+}
+
diff --git a/test/fixedbugs/bug369.go b/test/fixedbugs/bug369.go
new file mode 100644
index 000000000..fbcdf28f3
--- /dev/null
+++ b/test/fixedbugs/bug369.go
@@ -0,0 +1,57 @@
+// $G -N -o slow.$A $D/bug369.dir/pkg.go &&
+// $G -o fast.$A $D/bug369.dir/pkg.go &&
+// $G $D/$F.go && $L $F.$A && ./$A.out
+
+// Copyright 2011 The Go Authors.  All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Test that compiling with optimization turned on produces faster code.
+
+package main
+
+import (
+	"flag"
+	"os"
+	"runtime"
+	"testing"
+
+	fast "./fast"
+	slow "./slow"
+)
+
+var buf = make([]byte, 1048576)
+
+func BenchmarkFastNonASCII(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		fast.NonASCII(buf, 0)
+	}
+}
+
+func BenchmarkSlowNonASCII(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		slow.NonASCII(buf, 0)
+	}
+}
+
+func main() {
+	os.Args = []string{os.Args[0], "-test.benchtime=0.1"}
+	flag.Parse()
+	
+	rslow := testing.Benchmark(BenchmarkSlowNonASCII)
+	rfast := testing.Benchmark(BenchmarkFastNonASCII)
+	tslow := rslow.NsPerOp()
+	tfast := rfast.NsPerOp()
+
+	// Optimization should be good for at least 2x, but be forgiving.
+	// On the ARM simulator we see closer to 1.5x.
+	speedup := float64(tslow)/float64(tfast)
+	want := 1.8
+	if runtime.GOARCH == "arm" {
+		want = 1.3
+	}
+	if speedup < want {
+		println("fast:", tfast, "slow:", tslow, "speedup:", speedup, "want:", want)
+		println("not fast enough")
+	}
+}