[release-branch.go1.4] runtime: fix deadlock in runtime.Stack

It shouldn't semacquire() inside an acquirem(), the runtime
thinks that means deadlock.  It actually isn't a deadlock, but it
looks like it because acquirem() does m.locks++.

Candidate for inclusion in 1.4.1.  runtime.Stack with all=true
is pretty unuseable in GOMAXPROCS>1 environment.

fixes #9321

Change-Id: Iac6b664217d24763b9878c20e49229a1ecffc805
Reviewed-on: https://go-review.googlesource.com/1600Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
(cherry picked from commit 50bc3d5b)
Reviewed-on: https://go-review.googlesource.com/2807Reviewed-by: Andrew Gerrand <adg@golang.org>

src/runtime/mprof.go

@@ -575,20 +575,16 @@ func saveg(pc, sp uintptr, gp *g, r *StackRecord) {
 // If all is true, Stack formats stack traces of all other goroutines
 // into buf after the trace for the current goroutine.
 func Stack(buf []byte, all bool) int {
-	mp := acquirem()
-	gp := mp.curg
 	if all {
 		semacquire(&worldsema, false)
-		mp.gcing = 1
-		releasem(mp)
+		gp := getg()
+		gp.m.gcing = 1
 		onM(stoptheworld)
-		if mp != acquirem() {
-			gothrow("Stack: rescheduled")
-		}
 	}
 
 	n := 0
 	if len(buf) > 0 {
+		gp := getg()
 		sp := getcallersp(unsafe.Pointer(&buf))
 		pc := getcallerpc(unsafe.Pointer(&buf))
 		onM(func() {
@@ -605,11 +601,11 @@ func Stack(buf []byte, all bool) int {
 	}
 
 	if all {
-		mp.gcing = 0
+		gp := getg()
+		gp.m.gcing = 0
 		semrelease(&worldsema)
 		onM(starttheworld)
 	}
-	releasem(mp)
 	return n
 }
 

test/fixedbugs/issue9321.go

+// run
+
+// Copyright 2014 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 main
+
+import (
+	"bytes"
+	"runtime"
+	"runtime/pprof"
+	"sync"
+)
+
+func test() {
+	var wg sync.WaitGroup
+	wg.Add(2)
+	test := func() {
+		for i := 0; i < 100; i++ {
+			buf := &bytes.Buffer{}
+			pprof.Lookup("goroutine").WriteTo(buf, 2)
+		}
+		wg.Done()
+	}
+
+	go test()
+	go test()
+	wg.Wait()
+}
+
+func main() {
+	runtime.GOMAXPROCS(2)
+	for i := 0; i < 100; i++ {
+		test()
+	}
+}