Change id that identifies client in crash reports. Whenever metrics service recording is enabled, it sets the client id for crash reporting.

- On Windows this id gets stored in the registry so that we can read it pretty early regardless of the process type. If the id has not been generated (like in the case of first run) we initialize with empty string but the real id gets inserted once metrics service gets initialized.
- On Linux we were creating a hash and storing it in 'Consent to Send Stats'. This change replaces that hash with the metrics id. Unlike before calling SetConsentToSendStats doesn't generate a new id, if an id already exists.
- On Mac there was no id set. Now we use metrics id as guid for the browser process. For other process types a change is still required to pass that id as command line param to renderers/plugins (like Linux).

BUG=23658
TEST=Cause a deliberate crash in Chrome renderer/browser/plugin and make sure the clientID reported to the crash server is the right GUID.

Review URL: http://codereview.chromium.org/346007

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@31143 0039d316-1c4b-4281-b951-d872f2087c98

1 files changed, 3 insertions, 0 deletions

diff --git a/chrome/common/child_process_logging.h b/chrome/common/child_process_logging.h
index 7d1f16f..aa336af 100644
--- a/chrome/common/child_process_logging.h
+++ b/chrome/common/child_process_logging.h
@@ -14,6 +14,9 @@ namespace child_process_logging {
 // the URL.
 void SetActiveURL(const GURL& url);
 
+// Sets the Client ID that is used as GUID if a Chrome process crashes.
+void SetClientId(const std::string& client_id);
+
 // Simple wrapper class that sets the active URL in it's constructor and clears
 // the active URL in the destructor.
 class ScopedActiveURLSetter {