[Windows Sandbox Tests] Process Mitigations.

sandbox/win/tests/integration_tests/hooking_win_proc.h

Index: sandbox/win/tests/integration_tests/hooking_win_proc.h
diff --git a/sandbox/win/tests/integration_tests/hooking_win_proc.h b/sandbox/win/tests/integration_tests/hooking_win_proc.h
new file mode 100644
index 0000000000000000000000000000000000000000..eb5e5e1e00f122fc439807aa16989905c07ed0a8
--- /dev/null
+++ b/sandbox/win/tests/integration_tests/hooking_win_proc.h
@@ -0,0 +1,17 @@
+// Copyright 2016 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef SANDBOX_TESTS_INTEGRATION_TESTS_HOOKING_WIN_PROC_H_
+#define SANDBOX_TESTS_INTEGRATION_TESTS_HOOKING_WIN_PROC_H_
+
+#include <windows.h>
+
+namespace hooking_win_proc {
+
+constexpr wchar_t g_winproc_file[] = L"sbox_integration_test_win_proc.exe ";

[Will Harris, 2017/06/28 12:27:55]

I still don't quite get the point of this g_winproc_file declaration, it seems
to only be used in one file so why not just declare it locally there...?

Is this because you feel their declaration is 'owned' by hooking_win_proc? I can
understand them being declared here if they are also used in hooking_win_proc.cc
but I don't think there is any precedent for not just declaring these locally
where they are used in other files if they are only used in one place... we make
same assumptions like this about target files and exported procedures elsewhere.

(this comment applies elsewhere, so refer to here)

[penny, 2017/06/28 22:33:33]

HookingWinProc (exe): 
  - g_winproc_event name is required in 2 places - the test and the exe. 
They're both accessing a named os resource during testing.
  - g_winproc_file name is indeed currently only used from inside the test
itself (to launch the exe).  however, since I need a .h file for the event name,
I figured it made sense to put the exe filename in here too.  exactly as you
mentioned, it kind of does "belong" to the test exe project itself.  if any
other projects want to use it - they should include this new .h file and refer
to the global instead of hard-coding the filename.

HookingDll (dll):
  - g_hook_event name is required in 2 places - the test and the exe.  They're
both accessing a named os resource during testing.
  - g_hook_dll_file same as above, except a perfect example of one that is used
by more than one test.  CL2 (coming right after this reorg CL) with the new
mitigation tests also use the hooking_dll, and access this global.
  - WRT the three exported function name strings
(SetHook/WasHookCalled/HookProc), I say two things:
1) I'm happy to move them into the extensionpoints .cc file as they're only used
here for now
2) I had put them in the .h file just because the names of these exports are
basically "owned" by the test DLL project, but I realize we don't tend to do
this now for dynamic function look-ups (though I would argue that's just because
we usually lookup from MS system dlls which don't provide the function name
strings as a service).  I would argue with you if I knew any other test files
would be doing the dynamic lookups - but they're not for now. 

Sorry - these defines were a bit confusing, and more so now because I split up
my mega CL.  In the next half, the new tests use some of these defines.

+constexpr wchar_t g_winproc_event[] = L"ChromeExtensionTestEvent";
+
+}  // namespace hooking_win_proc
+
+#endif  // SANDBOX_TESTS_INTEGRATION_TESTS_HOOKING_WIN_PROC_H_