OOPIF: Allow OOP frames to print console messages in layout tests.

content/common/swapped_out_messages.cc

Index: content/common/swapped_out_messages.cc
diff --git a/content/common/swapped_out_messages.cc b/content/common/swapped_out_messages.cc
index 3cb20a571aaa89fe58e2888c07f02cb0c85fe517..3bb622908159d7ca1fcef03d689045c84d705d8f 100644
--- a/content/common/swapped_out_messages.cc
+++ b/content/common/swapped_out_messages.cc
@@ -9,6 +9,7 @@
 #include "content/common/input_messages.h"
 #include "content/common/view_messages.h"
 #include "content/public/common/content_client.h"
+#include "content/shell/common/shell_messages.h"
 
 namespace content {
 
@@ -47,6 +48,8 @@ bool SwappedOutMessages::CanSendWhileSwappedOut(const IPC::Message* msg) {
     // The browser should always have an accurate mirror of the renderer's
     // notion of the current page id.
     case FrameHostMsg_DidAssignPageId::ID:
+    // Used in layout tests; handled in WebKitTestController.
+    case ShellViewHostMsg_PrintMessage::ID:

[dcheng, 2015/04/09 01:08:55]

Would it make sense to tie this to the --site-per-process flag?

[Charlie Reis, 2015/04/09 17:32:25]

There's nothing in Chrome that processes this message, so it's not an issue for
Chrome builds.

We could tie it to --site-per-process if you're concerned that existing tests
might fail due to swapped out RVHs sending console messages, but that seems
unlikely to me.  Layout tests don't really do many (any?) cross-process
navigations.

I think this is fine until we move WebKitTestRunner from RenderView to
RenderFrame.

[dcheng, 2015/04/09 17:37:31]

sgtm

       return true;
     default:
       break;