Terminate child process if sandbox::TargetProcess is destroyed.

sandbox/win/src/target_process.cc

Index: sandbox/win/src/target_process.cc
diff --git a/sandbox/win/src/target_process.cc b/sandbox/win/src/target_process.cc
index 72e2780c8c3fff520c93704b4ca4b3c66e1a4283..fffdb79c4faacbfce2a8a9b84c874d6f24562896 100644
--- a/sandbox/win/src/target_process.cc
+++ b/sandbox/win/src/target_process.cc
@@ -63,30 +63,17 @@ TargetProcess::TargetProcess(base::win::ScopedHandle initial_token,
       base_address_(NULL) {}
 
 TargetProcess::~TargetProcess() {
-  DWORD exit_code = 0;
   // Give a chance to the process to die. In most cases the JOB_KILL_ON_CLOSE
   // will take effect only when the context changes. As far as the testing went,
   // this wait was enough to switch context and kill the processes in the job.
   // If this process is already dead, the function will return without waiting.
-  // TODO(nsylvain):  If the process is still alive at the end, we should kill
-  // it. http://b/893891
   // For now, this wait is there only to do a best effort to prevent some leaks
   // from showing up in purify.
   if (sandbox_process_info_.IsValid()) {
     ::WaitForSingleObject(sandbox_process_info_.process_handle(), 50);
-    // At this point, the target process should have been killed.  Check.
-    if (!::GetExitCodeProcess(sandbox_process_info_.process_handle(),
-                              &exit_code) || (STILL_ACTIVE == exit_code)) {
-      // Something went wrong.  We don't know if the target is in a state where
-      // it can manage to do another IPC call.  If it can, and we've destroyed
-      // the |ipc_server_|, it will crash the broker.  So we intentionally leak
-      // that.
-      if (shared_section_.IsValid())
-        shared_section_.Take();
-      ignore_result(ipc_server_.release());
-      sandbox_process_info_.TakeProcessHandle();
-      return;
-    }
+    // Terminate the process if it's still alive, as its IPC server is going
+    // away.
+    ::TerminateProcess(sandbox_process_info_.process_handle(), 2);

[Will Harris, 2016/11/23 16:44:23]

why 2?

I think that's RESULT_CODE_HUNG. If it's meant to be RESULT_CODE_HUNG then
please use that code - do you mean to use RESULT_CODE_KILLED?

   }
 
   // ipc_server_ references our process handle, so make sure the former is shut