Get rid of SharedMemory::GiveToProcess.

content/browser/renderer_host/render_message_filter.cc

Index: content/browser/renderer_host/render_message_filter.cc
diff --git a/content/browser/renderer_host/render_message_filter.cc b/content/browser/renderer_host/render_message_filter.cc
index 216d089790e5075ba8da1a59b095a9ca392e5ca4..fc05196e1d408f13cd427d9dcd2d25b6656f9b36 100644
--- a/content/browser/renderer_host/render_message_filter.cc
+++ b/content/browser/renderer_host/render_message_filter.cc
@@ -1,4 +1,3 @@
-// Copyright (c) 2012 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.
 
@@ -240,7 +239,9 @@ void RenderMessageFilter::SendLoadFontReply(IPC::Message* reply,
     result->font_data_size = 0;
     result->font_id = 0;
   } else {
-    result->font_data.GiveToProcess(base::GetCurrentProcessHandle(), &handle);
+    handle = result->font_data.handle().Duplicate();
+    result->font_data.Unmap();
+    result->font_data.Close();

[Nico, 2017/05/02 15:58:45]

This code looks kind of weird to me. Don't you really want something like

handle = result->font_data.Release()

or

handle = std::move(result->font_data)

?

[erikchen, 2017/05/02 18:45:57]

[Duplicate changes the ownership_passes_to_ipc member of the underlying handle].


We could add a new public method to base::SharedMemory, but there are only two
consumers and honestly this *is* a really weird use case. The original semantics
of GiveToProcess were to synchronously "transfer" a handle from one process to
another. 

I don't want to make it easy for consumers of base::SharedMemory to do this
really weird thing.

   }
   RenderProcessHostMsg_LoadFont::WriteReplyParams(
       reply, result->font_data_size, handle, result->font_id);