Add command-line parameter to trigger panic on signal

This is to troubleshoot problems with a hung process that is
not responding to 'runsc debug --stack' command.

PiperOrigin-RevId: 210483513
Change-Id: I4377b210b4e51bc8a281ad34fd94f3df13d9187d

4 files changed, 31 insertions, 1 deletions

diff --git a/runsc/boot/config.go b/runsc/boot/config.go
index bc392deb3..efb8563ea 100644
--- a/runsc/boot/config.go
+++ b/runsc/boot/config.go
@@ -204,7 +204,12 @@ type Config struct {
 	// TODO: Remove this when multiple container is fully supported.
 	MultiContainer bool
 
+	// WatchdogAction sets what action the watchdog takes when triggered.
 	WatchdogAction watchdog.Action
+
+	// PanicSignal register signal handling that panics. Usually set to
+	// SIGUSR2(12) to troubleshoot hangs. -1 disables it.
+	PanicSignal int
 }
 
 // ToFlags returns a slice of flags that correspond to the given Config.
@@ -225,5 +230,6 @@ func (c *Config) ToFlags() []string {
 		"--strace-syscalls=" + strings.Join(c.StraceSyscalls, ","),
 		"--strace-log-size=" + strconv.Itoa(int(c.StraceLogSize)),
 		"--watchdog-action=" + c.WatchdogAction.String(),
+		"--panic-signal=" + strconv.Itoa(c.PanicSignal),
 	}
 }
diff --git a/runsc/boot/loader.go b/runsc/boot/loader.go
index 3963ed55d..0ad830a6b 100644
--- a/runsc/boot/loader.go
+++ b/runsc/boot/loader.go
@@ -20,6 +20,7 @@ import (
 	"fmt"
 	"math/rand"
 	"os"
+	"os/signal"
 	"runtime"
 	"sync"
 	"sync/atomic"
@@ -229,7 +230,18 @@ func New(spec *specs.Spec, conf *Config, controllerFD int, ioFDs []int, console
 		return nil, fmt.Errorf("failed to ignore child stop signals: %v", err)
 	}
 	// Ensure that signals received are forwarded to the emulated kernel.
-	stopSignalForwarding := sighandling.PrepareForwarding(k, false)()
+	ps := syscall.Signal(conf.PanicSignal)
+	stopSignalForwarding := sighandling.PrepareForwarding(k, ps)()
+	if conf.PanicSignal != -1 {
+		// Panics if the sentry receives 'conf.PanicSignal'.
+		panicChan := make(chan os.Signal, 1)
+		signal.Notify(panicChan, ps)
+		go func() { // S/R-SAFE: causes sentry panic.
+			<-panicChan
+			panic("Signal-induced panic")
+		}()
+		log.Infof("Panic signal set to %v(%d)", ps, conf.PanicSignal)
+	}
 
 	procArgs, err := newProcess(spec, creds, utsns, ipcns, k)
 	if err != nil {
diff --git a/runsc/cmd/debug.go b/runsc/cmd/debug.go
index 7952489de..b20987b2c 100644
--- a/runsc/cmd/debug.go
+++ b/runsc/cmd/debug.go
@@ -15,6 +15,8 @@
 package cmd
 
 import (
+	"syscall"
+
 	"context"
 	"flag"
 	"github.com/google/subcommands"
@@ -27,6 +29,7 @@ import (
 type Debug struct {
 	pid    int
 	stacks bool
+	signal int
 }
 
 // Name implements subcommands.Command.
@@ -48,6 +51,7 @@ func (*Debug) Usage() string {
 func (d *Debug) SetFlags(f *flag.FlagSet) {
 	f.IntVar(&d.pid, "pid", 0, "sandbox process ID. Container ID is not necessary if this is set")
 	f.BoolVar(&d.stacks, "stacks", false, "if true, dumps all sandbox stacks to the log")
+	f.IntVar(&d.signal, "signal", -1, "sends signal to the sandbox")
 }
 
 // Execute implements subcommands.Command.Execute.
@@ -96,6 +100,12 @@ func (d *Debug) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 		Fatalf("sandbox %q is not running", c.Sandbox.ID)
 	}
 
+	if d.signal > 0 {
+		log.Infof("Sending signal %d to process: %d", d.signal, c.Sandbox.Pid)
+		if err := syscall.Kill(c.Sandbox.Pid, syscall.Signal(d.signal)); err != nil {
+			Fatalf("failed to send signal %d to processs %d", d.signal, c.Sandbox.Pid)
+		}
+	}
 	if d.stacks {
 		log.Infof("Retrieving sandbox stacks")
 		stacks, err := c.Sandbox.Stacks()
diff --git a/runsc/main.go b/runsc/main.go
index 0a2cbca6c..773ec6486 100644
--- a/runsc/main.go
+++ b/runsc/main.go
@@ -61,6 +61,7 @@ var (
 	overlay        = flag.Bool("overlay", false, "wrap filesystem mounts with writable overlay. All modifications are stored in memory inside the sandbox.")
 	multiContainer = flag.Bool("multi-container", false, "enable *experimental* multi-container support.")
 	watchdogAction = flag.String("watchdog-action", "log", "sets what action the watchdog takes when triggered: log (default), panic.")
+	panicSignal    = flag.Int("panic-signal", -1, "register signal handling that panics. Usually set to SIGUSR2(12) to troubleshoot hangs. -1 disables it.")
 )
 
 var gitRevision = ""
@@ -139,6 +140,7 @@ func main() {
 		StraceLogSize:  *straceLogSize,
 		MultiContainer: *multiContainer,
 		WatchdogAction: wa,
+		PanicSignal:    *panicSignal,
 	}
 	if len(*straceSyscalls) != 0 {
 		conf.StraceSyscalls = strings.Split(*straceSyscalls, ",")