container/dispatcher: start goroutine in dispatcher

This allows instrumentation of calls from goroutine without relying on finalizers.

Signed-off-by: Ophestra <cat@gensokyo.uk>

container/dispatcher.go

@@ -22,10 +22,10 @@ type osFile interface {
 
 // syscallDispatcher provides methods that make state-dependent system calls as part of their behaviour.
 type syscallDispatcher interface {
-	// new returns a new instance of syscallDispatcher for use in another goroutine.
+	// new starts a goroutine with a new instance of syscallDispatcher.
 	// A syscallDispatcher must never be used in any goroutine other than the one owning it,
 	// just synchronising access is not enough, as this is for test instrumentation.
-	new() syscallDispatcher
+	new(f func(k syscallDispatcher))
 
 	// lockOSThread provides [runtime.LockOSThread].
 	lockOSThread()
@@ -145,7 +145,7 @@ type syscallDispatcher interface {
 // direct implements syscallDispatcher on the current kernel.
 type direct struct{}
 
-func (k direct) new() syscallDispatcher { return k }
+func (k direct) new(f func(k syscallDispatcher)) { go f(k) }
 
 func (direct) lockOSThread() { runtime.LockOSThread() }
 

container/dispatcher_test.go

@@ -12,6 +12,7 @@ import (
 	"runtime"
 	"slices"
 	"strings"
+	"sync"
 	"syscall"
 	"testing"
 	"time"
@@ -113,7 +114,7 @@ type simpleTestCase struct {
 func checkSimple(t *testing.T, fname string, testCases []simpleTestCase) {
 	for _, tc := range testCases {
 		t.Run(tc.name, func(t *testing.T) {
-			k := &kstub{t: t, want: tc.want}
+			k := &kstub{t: t, want: tc.want, wg: new(sync.WaitGroup)}
 			if err := tc.f(k); !errors.Is(err, tc.wantErr) {
 				t.Errorf("%s: error = %v, want %v", fname, err, tc.wantErr)
 			}
@@ -141,7 +142,7 @@ func checkOpBehaviour(t *testing.T, testCases []opBehaviourTestCase) {
 		for _, tc := range testCases {
 			t.Run(tc.name, func(t *testing.T) {
 				state := &setupState{Params: tc.params}
-				k := &kstub{t: t, want: [][]kexpect{slices.Concat(tc.early, []kexpect{{name: "\x00"}}, tc.apply)}}
+				k := &kstub{t: t, want: [][]kexpect{slices.Concat(tc.early, []kexpect{{name: "\x00"}}, tc.apply)}, wg: new(sync.WaitGroup)}
 				errEarly := tc.op.early(state, k)
 				k.expect("\x00")
 				if !errors.Is(errEarly, tc.wantErrEarly) {
@@ -272,10 +273,14 @@ type kstub struct {
 	track int
 	// sub stores addresses of kstub created by new.
 	sub []*kstub
+	// wg waits for all descendants to complete.
+	wg *sync.WaitGroup
 }
 
 // handleIncomplete calls f on an incomplete k and all its descendants.
 func (k *kstub) handleIncomplete(f func(k *kstub)) {
+	k.wg.Wait()
+
 	if k.want != nil && len(k.want[k.track]) != k.pos {
 		f(k)
 	}
@@ -331,13 +336,15 @@ func checkArgReflect(k *kstub, arg string, got any, n int) bool {
 	return true
 }
 
-func (k *kstub) new() syscallDispatcher {
+func (k *kstub) new(f func(k syscallDispatcher)) {
 	k.expect("new")
 	if len(k.want) <= k.track+1 {
 		k.t.Fatalf("new: track overrun")
 	}
-	k.sub = append(k.sub, &kstub{t: k.t, want: k.want, track: k.track + 1})
-	return k.sub[len(k.sub)-1]
+	sk := &kstub{t: k.t, want: k.want, track: len(k.sub) + 1, wg: k.wg}
+	k.sub = append(k.sub, sk)
+	k.wg.Add(1)
+	go func() { defer k.wg.Done(); f(sk) }()
 }
 
 func (k *kstub) lockOSThread() { k.expect("lockOSThread") }

container/init.go

@@ -178,6 +178,7 @@ func initEntrypoint(k syscallDispatcher, prepareLogger func(prefix string), setV
 				fmt.Sprintf("cannot prepare op at index %d:", i))
 			k.beforeExit()
 			k.exit(1)
+			return
 		}
 	}
 
@@ -218,6 +219,7 @@ func initEntrypoint(k syscallDispatcher, prepareLogger func(prefix string), setV
 				fmt.Sprintf("cannot apply op at index %d:", i))
 			k.beforeExit()
 			k.exit(1)
+			return
 		}
 	}
 
@@ -333,7 +335,7 @@ func initEntrypoint(k syscallDispatcher, prepareLogger func(prefix string), setV
 	info := make(chan winfo, 1)
 	done := make(chan struct{})
 
-	go func(k syscallDispatcher) {
+	k.new(func(k syscallDispatcher) {
 		var (
 			err     error
 			wpid    = -2
@@ -360,7 +362,7 @@ func initEntrypoint(k syscallDispatcher, prepareLogger func(prefix string), setV
 		}
 
 		close(done)
-	}(k.new())
+	})
 
 	// handle signals to dump withheld messages
 	sig := make(chan os.Signal, 2)
@@ -385,7 +387,9 @@ func initEntrypoint(k syscallDispatcher, prepareLogger func(prefix string), setV
 				}
 				continue
 			}
+			k.beforeExit()
 			k.exit(0)
+			return
 
 		case w := <-info:
 			if w.wpid == cmd.Process.Pid {
@@ -396,9 +400,11 @@ func initEntrypoint(k syscallDispatcher, prepareLogger func(prefix string), setV
 				case w.wstatus.Exited():
 					r = w.wstatus.ExitStatus()
 					k.verbosef("initial process exited with code %d", w.wstatus.ExitStatus())
+
 				case w.wstatus.Signaled():
 					r = 128 + int(w.wstatus.Signal())
 					k.verbosef("initial process exited with signal %s", w.wstatus.Signal())
+
 				default:
 					r = 255
 					k.verbosef("initial process exited with status %#x", w.wstatus)
@@ -410,11 +416,13 @@ func initEntrypoint(k syscallDispatcher, prepareLogger func(prefix string), setV
 		case <-done:
 			k.beforeExit()
 			k.exit(r)
+			return
 
 		case <-timeout:
 			k.printf("timeout exceeded waiting for lingering processes")
 			k.beforeExit()
 			k.exit(r)
+			return
 		}
 	}
 }