Mojo: Use new message APIs to reduce copying

mojo/public/cpp/bindings/lib/message_buffer.cc

+// 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.
+
+#include "mojo/public/cpp/bindings/lib/message_buffer.h"
+
+#include <limits>
+
+#include "mojo/public/cpp/bindings/lib/bindings_serialization.h"
+
+namespace mojo {
+namespace internal {
+
+MessageBuffer::MessageBuffer(size_t capacity, bool zero_initialized) {
+  DCHECK_LE(capacity, std::numeric_limits<uint32_t>::max());
+  data_num_bytes_ = static_cast<uint32_t>(capacity);
+
+  MojoResult rv = AllocMessage(capacity, nullptr, 0,
+                               MOJO_ALLOC_MESSAGE_FLAG_NONE, &message_);
+  CHECK_EQ(rv, MOJO_RESULT_OK);
+
+  if (capacity == 0) {
+    buffer_ = nullptr;
+  } else {
+    rv = GetMessageBuffer(message_.get(), &buffer_);
+    CHECK_EQ(rv, MOJO_RESULT_OK);
+
+    if (zero_initialized)
+      memset(buffer_, 0, capacity);
+  }
+}
+
+MessageBuffer::MessageBuffer(ScopedMessageHandle message, uint32_t num_bytes) {
+  message_ = std::move(message);
+  data_num_bytes_ = num_bytes;
+
+  MojoResult rv = GetMessageBuffer(message_.get(), &buffer_);

[yzshen1, 2016/04/29 16:50:20]

I noticed that the previous method explicitly set buffer_ to null if
data_num_bytes_ is 0.
Please make them consistent. (Either way seems safe.)

[Ken Rockot(use gerrit already), 2016/04/29 17:37:13]

done

+  CHECK_EQ(rv, MOJO_RESULT_OK);
+}
+
+MessageBuffer::~MessageBuffer() {}
+
+void* MessageBuffer::Allocate(size_t delta) {
+  delta = internal::Align(delta);
+
+  DCHECK_LE(delta, static_cast<size_t>(data_num_bytes_));
+  DCHECK_GT(bytes_claimed_ + static_cast<uint32_t>(delta), bytes_claimed_);
+
+  uint32_t new_bytes_claimed = bytes_claimed_ + static_cast<uint32_t>(delta);
+  if (new_bytes_claimed > data_num_bytes_) {
+    NOTREACHED();
+    return nullptr;
+  }
+
+  char* start = static_cast<char*>(buffer_) + bytes_claimed_;

[yzshen1, 2016/04/29 16:50:20]

It relies on the fact that |buffer_| itself is 8-byte aligned.
Is that guaranteed by the edk implementation? (I guess yes.) If yes, please
comment it explicitly in the API documentation.

[Ken Rockot(use gerrit already), 2016/04/29 17:37:13]

done

+  bytes_claimed_ = new_bytes_claimed;
+  return static_cast<void*>(start);
+}
+
+}  // namespace internal
+}  // namespace mojo