Split the unit level from the integration level in cmd, builder, link, catalogue, inventory; cheap fixtures for the archive-entry bound, the registry waits and the catalogue copy (issue 895)

internal/builder/kill_integration_test.go

@@ -0,0 +1,58 @@
+//go:build integration
+
+package builder
+
+import (
+	"context"
+	"errors"
+	"os"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"syscall"
+	"testing"
+	"time"
+)
+
+func TestAKilledBuildEndsItsWholeProcessGroup(t *testing.T) {
+	dir := t.TempDir()
+	child := filepath.Join(dir, "child")
+	ctx, cancel := context.WithCancel(context.Background())
+	done := make(chan error, 1)
+	began := time.Now()
+	go func() {
+		// A shell that starts a grandchild and waits on it: killing the shell alone would leave the
+		// grandchild running, holding the output open.
+		_, err := Command(ctx, dir, "sh", "-c", `sleep 60 & echo $! > child; wait`)
+		done <- err
+	}()
+	var pid int
+	for deadline := time.Now().Add(5 * time.Second); time.Now().Before(deadline); time.Sleep(20 * time.Millisecond) {
+		if raw, err := os.ReadFile(child); err == nil && strings.TrimSpace(string(raw)) != "" {
+			pid, _ = strconv.Atoi(strings.TrimSpace(string(raw)))
+			break
+		}
+	}
+	if pid == 0 {
+		t.Fatal("the command never started its child")
+	}
+	cancel()
+	select {
+	case err := <-done:
+		if err == nil {
+			t.Fatal("a killed command reported success")
+		}
+	case <-time.After(KillWait + 5*time.Second):
+		t.Fatal("the killed command never returned")
+	}
+	if took := time.Since(began); took > KillWait {
+		t.Errorf("ending the command took %s: the group was not killed, the wait ran out", took)
+	}
+	for deadline := time.Now().Add(2 * time.Second); time.Now().Before(deadline); time.Sleep(20 * time.Millisecond) {
+		if err := syscall.Kill(pid, 0); errors.Is(err, syscall.ESRCH) {
+			return
+		}
+	}
+	_ = syscall.Kill(pid, syscall.SIGKILL)
+	t.Fatalf("the grandchild %d outlived the kill", pid)
+}