[Webkit-unassigned] [Bug 29344] REGRESSION: fast/workers/dedicated-worker-lifecycle.html failing intermittently on leopard bot

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Sun Oct 11 15:33:54 PDT 2009


https://bugs.webkit.org/show_bug.cgi?id=29344


Alexey Proskuryakov <ap at webkit.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|atwilson at chromium.org       |webkit-unassigned at lists.web
                   |                            |kit.org
           Platform|PC                          |Macintosh Intel
          Component|Tools / Tests               |JavaScriptCore
                 CC|                            |ggaren at apple.com,
                   |                            |oliver at apple.com




--- Comment #19 from Alexey Proskuryakov <ap at webkit.org>  2009-10-11 15:33:53 PDT ---
On my home machine (and a debug build), this test isn't just flaky - it's
always failing. Worker activity management appears to be working well, so this
looks like a general GC failure. I see the worker wrapper marked by
Heap::markConservatively(), as if there were references to it remaining on the
stack.

The test already attempts black magic to fight this. As far as I know, it
shouldn't be necessary, and it doesn't work anyway.

        // Orphan our worker (no more references to it) and wait for it to
exit.
        worker.onmessage = 0;
        worker = 0;
        // For some reason, the worker object does not get GC'd unless we
allocate a new object here.
        // The conjecture is that there's a value on the stack that appears to
point to the worker which this clobbers.
        new Date();
        waitUntilWorkerThreadsExit(orphanedTimeoutWorkerExited);

Here, waitUntilWorkerThreadsExit invokes garbage collector, each 10 ms, waiting
for GC to collect the unused Worker object, which never happens.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.



More information about the webkit-unassigned mailing list