diff options
| author | shess@chromium.org <shess@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2010-09-13 19:35:52 +0000 |
|---|---|---|
| committer | shess@chromium.org <shess@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2010-09-13 19:35:52 +0000 |
| commit | 72a1623f05cdfec636a5554baef322e32c3517bd (patch) | |
| tree | e5e7e36f05d536b0a4c194cfcdc73d638aa5af93 /chrome/browser/automation | |
| parent | b904feb7878fa9a67881758b5b24e9a5b9252aa5 (diff) | |
| download | chromium_src-72a1623f05cdfec636a5554baef322e32c3517bd.zip chromium_src-72a1623f05cdfec636a5554baef322e32c3517bd.tar.gz chromium_src-72a1623f05cdfec636a5554baef322e32c3517bd.tar.bz2 | |
Revert "Make AutomationProvider fail fast on messages it doesn't understand."
Original review: http://codereview.chromium.org/3224001
The mentioned bug involves a crash on chromebot on Mac. My working
hypothesis, based on some of the log messages seen, is that some file
descriptors are being trashed. The CL being reverted closes the
channel while handling a message, which seems like it has potential to
be a problem.
Since this should not have user-facing impact, I am reverting the
change for merging to the 517 branch, after which I will re-instate
the change. We have been trying to verify that this change is in the
crashing set, or out, but our data turned out to be incorrect, so I'm
doing this in parallel with a new run.
BUG=54453
TEST=chromebot
Review URL: http://codereview.chromium.org/3317023
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@59260 0039d316-1c4b-4281-b951-d872f2087c98
Diffstat (limited to 'chrome/browser/automation')
| -rw-r--r-- | chrome/browser/automation/automation_provider.cc | 15 | ||||
| -rw-r--r-- | chrome/browser/automation/automation_provider.h | 2 |
2 files changed, 0 insertions, 17 deletions
diff --git a/chrome/browser/automation/automation_provider.cc b/chrome/browser/automation/automation_provider.cc index 361cdd2..6642a2a 100644 --- a/chrome/browser/automation/automation_provider.cc +++ b/chrome/browser/automation/automation_provider.cc @@ -388,24 +388,9 @@ void AutomationProvider::OnMessageReceived(const IPC::Message& message) { IPC_MESSAGE_HANDLER_DELAY_REPLY(AutomationMsg_LoginWithUserAndPass, LoginWithUserAndPass) #endif // defined(OS_CHROMEOS) - - IPC_MESSAGE_UNHANDLED(OnUnhandledMessage()); IPC_END_MESSAGE_MAP() } -void AutomationProvider::OnUnhandledMessage() { - // We should not hang here. Print a message to indicate what's going on, - // and disconnect the channel to notify the caller about the error - // in a way it can't ignore, and make any further attempts to send - // messages fail fast. - LOG(ERROR) << "AutomationProvider received a message it can't handle. " - << "Please make sure that you use switches::kTestingChannelID " - << "for test code (TestingAutomationProvider), and " - << "switches::kAutomationClientChannelID for everything else " - << "(like ChromeFrame). Closing the automation channel."; - channel_->Close(); -} - // This task just adds another task to the event queue. This is useful if // you want to ensure that any tasks added to the event queue after this one // have already been processed by the time |task| is run. diff --git a/chrome/browser/automation/automation_provider.h b/chrome/browser/automation/automation_provider.h index 2ad6dea..82af88c 100644 --- a/chrome/browser/automation/automation_provider.h +++ b/chrome/browser/automation/automation_provider.h @@ -192,8 +192,6 @@ class AutomationProvider : public base::RefCounted<AutomationProvider>, CancelableRequestConsumer consumer_; private: - void OnUnhandledMessage(); - // IPC Message callbacks. void WindowSimulateDrag(int handle, std::vector<gfx::Point> drag_path, |