Use a stable device-user identifier for CryptAuth enrollments.

components/proximity_auth/cryptauth/cryptauth_enrollment_utils.cc

Index: components/proximity_auth/cryptauth/cryptauth_enrollment_utils.cc
diff --git a/components/proximity_auth/cryptauth/cryptauth_enrollment_utils.cc b/components/proximity_auth/cryptauth/cryptauth_enrollment_utils.cc
new file mode 100644
index 0000000000000000000000000000000000000000..5f31b762040a8f7a5aee0b217a89ed1eac27080c
--- /dev/null
+++ b/components/proximity_auth/cryptauth/cryptauth_enrollment_utils.cc
@@ -0,0 +1,22 @@
+// Copyright 2015 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 "base/md5.h"
+#include "components/proximity_auth/cryptauth/cryptauth_enrollment_utils.h"

[Ilya Sherman, 2015/02/10 01:31:05]

nit: This include should be first.

[Tim Song, 2015/02/10 22:20:33]

Done.

+
+namespace proximity_auth {
+
+std::string CalculateDeviceUserId(const std::string& device_id,
+                                  const std::string& user_id) {
+  base::MD5Context context;
+  base::MD5Init(&context);
+  base::MD5Update(&context, device_id);
+  base::MD5Update(&context, user_id);
+
+  base::MD5Digest digest;
+  base::MD5Final(&digest, &context);
+  return base::MD5DigestToBase16(digest);

[Ilya Sherman, 2015/02/10 01:31:05]

MD5 is somewhat frowned upon these days, as secure hash functions go.  Any
reason not to use something more secure?  I realize it doesn't super matter for
our use-case, but it seems like it wouldn't hurt.

[Tim Song, 2015/02/10 22:20:33]

I mainly did it because MD5 is fast, but I switched to SHA1 as speed isn't
really important here either.

[Ilya Sherman, 2015/02/10 23:20:14]

As long as we're using SHA, why not use SHA-256, via //crypto/sha2.h?

+}
+
+}  // namespace proximity_auth