Perf test to measure loading CDM library.

chrome/chrome_tests.gypi

Index: chrome/chrome_tests.gypi
diff --git a/chrome/chrome_tests.gypi b/chrome/chrome_tests.gypi
index c3db22264c47512bcc162df61eb39babec079986..ad02f56b65d6100c51a276cca97e3ca613800b53 100644
--- a/chrome/chrome_tests.gypi
+++ b/chrome/chrome_tests.gypi
@@ -2807,6 +2807,24 @@
         '../third_party/webrtc/tools/tools.gyp:rgba_to_i420_converter',
       ],
     },  # target 'webrtc_test_tools'
+    {
+      # Executable to measure time to load libraries.
+      'target_name': 'load_library_perf_test',

[M-A Ruel, 2014/02/12 02:32:19]

I'm generally against the "add at the bottom of the list" strategy but this file
is already completely unsorted. Just noting this fact, nothing to do.

[shadi1, 2014/02/12 18:34:06]

Yeah, I was not sure if there is any order to follow...

+      'type': '<(gtest_target_type)',
+      'dependencies': [
+        '../base/base.gyp:test_support_perf',
+        '../testing/gtest.gyp:gtest',
+        '../testing/perf/perf_test.gyp:*',
+        '../third_party/widevine/cdm/widevine_cdm.gyp:widevine_cdm_version_h',

[M-A Ruel, 2014/02/12 02:32:19]

Then this target should not even exist on pure open source checkout.

[ddorwin, 2014/02/12 18:12:52]

The target exists in an external checkout. There is a dummy
widevine_cdm_version_h that is used when branding != "Chrome". This allows us to
control the presence solely with branding = "Chrome" and not need to spread the
logic throughout the GYP files.

+      ],
+      'sources': [
+        'browser/load_library_perf_test.cc',
+      ],
+      'include_dirs': [
+        '..',
+        '<(SHARED_INTERMEDIATE_DIR)',
+      ],
+    },  # target 'load_library_perf_test'
   ],
   'conditions': [
     ['OS=="mac"', {