Bug 133010 - [WebKit2] Wake up threads blocked in waitForAndDispatchImmediately() if we lose our connection
Summary: [WebKit2] Wake up threads blocked in waitForAndDispatchImmediately() if we lo...
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Andy Estes
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-05-16 14:33 PDT by Andy Estes
Modified: 2014-05-19 11:00 PDT (History)
0 users

See Also:

Attachments
[WebKit2] Wake up threads blocked in waitForAndDispatchImmediately() if we lose our connection (4.08 KB, patch)
2014-05-16 14:36 PDT, Andy Estes 		ggaren: review+
Details | Formatted Diff | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Andy Estes 2014-05-16 14:33:26 PDT 
[WebKit2] Wake up threads blocked in waitForAndDispatchImmediately() if we lose our connection
Comment 1 Andy Estes 2014-05-16 14:36:16 PDT 
Created attachment 231591 [details]
[WebKit2] Wake up threads blocked in waitForAndDispatchImmediately() if we lose our connection
Comment 2 Geoffrey Garen 2014-05-16 14:47:27 PDT 
Comment on attachment 231591 [details]
[WebKit2] Wake up threads blocked in waitForAndDispatchImmediately() if we lose our connection

View in context: https://bugs.webkit.org/attachment.cgi?id=231591&action=review

r=me

> Source/WebKit2/ChangeLog:27
> +2014-05-16  Andy Estes  <aestes@apple.com>
> +
> +        Need a short description (OOPS!).
> +        Need the bug URL (OOPS!).
> +
> +        Reviewed by NOBODY (OOPS!).
> +
> +        * Platform/IPC/Connection.cpp:
> +        (IPC::Connection::Connection):
> +        (IPC::Connection::waitForMessage):
> +        (IPC::Connection::connectionDidClose):
> +        * Platform/IPC/Connection.h:

ChangeLog conflict-o.

> Source/WebKit2/Platform/IPC/Connection.cpp:682
> +        m_waitForMessageCondition.notify_all();
> +    }

It is an optimization to drop the lock *before* calling notify_all(). Otherwise, you will wake a thread and it will promptly sleep again due to failing to acquire the locked mutex.

Please move the call to notify_all() out of the lock scope.
Comment 3 Andy Estes 2014-05-19 11:00:54 PDT 
Committed r169051: <http://trac.webkit.org/changeset/169051>