[Chromoting] Retrieve process resource usage (ProcessStats) and its tests

remoting/host/process_stats_sender.cc

Index: remoting/host/process_stats_sender.cc
diff --git a/remoting/host/process_stats_sender.cc b/remoting/host/process_stats_sender.cc
new file mode 100644
index 0000000000000000000000000000000000000000..29b05317e0e055648a06f1d15fda0e448ee7a7bd
--- /dev/null
+++ b/remoting/host/process_stats_sender.cc
@@ -0,0 +1,55 @@
+// Copyright 2017 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 "remoting/host/process_stats_sender.h"
+
+#include <utility>
+
+#include "base/location.h"
+#include "base/logging.h"
+#include "remoting/host/process_stats_agent.h"
+#include "remoting/host/process_stats_util.h"
+
+namespace remoting {
+
+ProcessStatsSender::ProcessStatsSender(
+    protocol::ProcessStatsStub* host_stats_stub,
+    base::TimeDelta interval)
+    : host_stats_stub_(host_stats_stub), interval_(interval) {
+  DCHECK(thread_checker_.CalledOnValidThread());
+  DCHECK(host_stats_stub_);
+  DCHECK(!interval_.is_zero());
+
+  timer_.Start(FROM_HERE, interval_, this, &ProcessStatsSender::ReportUsage);
+}
+
+ProcessStatsSender::~ProcessStatsSender() {
+  DCHECK(thread_checker_.CalledOnValidThread());
+  timer_.Stop();
+}
+
+void ProcessStatsSender::AddProcessStatsAgent(
+    std::unique_ptr<ProcessStatsAgent> agent) {
+  DCHECK(thread_checker_.CalledOnValidThread());
+  agents_.push_back(std::move(agent));
+}
+
+void ProcessStatsSender::ReportUsage() {
+  DCHECK(thread_checker_.CalledOnValidThread());
+  if (agents_.empty()) {

[Sergey Ulanov, 2017/04/26 05:42:44]

Maybe replace this with a check that usages is not empty below? I don't think we
want to send an empty AggregateProcessResourceUsage.

[Hzj_jie, 2017/04/26 20:47:19]

If so, I would also update the comment in header file to explain the expected
behavior: users should call AddProcessStatsAgent immediately after the
constructor.
Done.

[joedow, 2017/04/26 23:48:01]

That seems like a dangerous assumption.  Maybe you should not start the timer
until an agent is added.  That kind of makes sense as there is no reason to fire
a timer if there are no agents registered.

[Hzj_jie, 2017/04/27 02:21:33]

Emmm, looks like once the if-check has been changed into a DCHECK, it's more
reasonable to move the functionality of AddProcessStatsAgent() into the
constructor directly.
Meanwhile taking ownership of ProcessStatsAgent is also not a good idea,
especially for ForwardProcessStatsAgent, it needs to work with both
ProcessStatsSender and IPC channel.

[joedow, 2017/04/27 15:01:55]

I don't think you need to fire the timer if no one registers an agent.  We can
see how it feels once you have a working implementation.

[Hzj_jie, 2017/04/27 18:28:49]

As far as I can imagine, it's definitely wrong if one creates a
ProcessStatsSender but does not register an agent.
Yes, let me finish the next change so we can discuss this with more context.

+    return;
+  }
+
+  std::vector<protocol::ProcessResourceUsage> usages;
+  for (const auto& agent : agents_) {
+    protocol::ProcessResourceUsage usage = agent->GetResourceUsage();
+    if (!IsEmptyProcessResourceUsage(usage)) {
+      usages.push_back(std::move(usage));
+    }
+  }
+
+  host_stats_stub_->OnProcessStats(AggregateProcessResourceUsage(usages));
+}
+
+}  // namespace remoting