Bug 52156 - REGRESSION (r75289): Assertion failure in JSC::Heap::destroy() when running JS tests
Summary: REGRESSION (r75289): Assertion failure in JSC::Heap::destroy() when running J...
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: JavaScriptCore (show other bugs)
Version: 528+ (Nightly build)
Hardware: PC Windows XP
: P2 Normal
Assignee: Nobody
URL: http://build.webkit.org/builders/Wind...
Keywords: InRadar, LayoutTestFailure, PlatformOnly
Depends on:
Blocks:
 
Reported: 2011-01-10 11:01 PST by Adam Roben (:aroben)
Modified: 2011-01-10 12:17 PST (History)
4 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Adam Roben (:aroben) 2011-01-10 11:01:22 PST
To reproduce:

1. run-javascriptcore-tests

Three tests will hit an assertion. The tests are:

	js1_5/Regress/regress-152646.js
	js1_5/Regress/regress-192414.js
	js1_5/Regress/regress-96526-002.js

The assertion is in Heap::destroy. It is:

    ASSERT(!m_globalData->dynamicGlobalObject);

m_globalData->dynamicGlobalObject seems always to be 0x01500000.

Here's the backtrace:

>	JavaScriptCore.dll!JSC::Heap::destroy()  Line 90 + 0x28 bytes	C++
 	jsc.exe!cleanupGlobalData(JSC::JSGlobalData * globalData=0x00f2be88)  Line 364	C++
 	jsc.exe!main(int argc=6, char * * argv=0x00cfed00)  Line 356 + 0x9 bytes	C++
 	jsc.exe!__tmainCRTStartup()  Line 597 + 0x17 bytes	C
 	kernel32.dll!_BaseProcessStart@4()  + 0x23 bytes
Comment 1 Adam Roben (:aroben) 2011-01-10 11:01:38 PST
Is there any way to skip JS tests?
Comment 2 Adam Roben (:aroben) 2011-01-10 11:02:11 PST
<rdar://problem/8842022>
Comment 3 Adam Roben (:aroben) 2011-01-10 11:03:32 PST
Looks like run-javascriptcore-tests has support for skipping tests.
Comment 4 Adam Roben (:aroben) 2011-01-10 11:11:12 PST
I skipped these tests in r75392.
Comment 5 Adam Roben (:aroben) 2011-01-10 11:12:03 PST
Maybe this was caused by r75289?
Comment 6 Oliver Hunt 2011-01-10 11:19:09 PST
(In reply to comment #5)
> Maybe this was caused by r75289?

I would guess so, rather than disabling those tests, i think it would be better to rollout that patch.
Comment 7 Geoffrey Garen 2011-01-10 11:24:03 PST
Yes, please roll out the offending patch.
Comment 8 Geoffrey Garen 2011-01-10 11:24:23 PST
(...and re-enable the tests.)
Comment 9 Adam Roben (:aroben) 2011-01-10 12:03:26 PST
I don't know what the offending patch is yet. I'm testing r75289 now.
Comment 10 Adam Roben (:aroben) 2011-01-10 12:06:29 PST
Rolling out r75289 fixes the assertions.
Comment 11 Gavin Barraclough 2011-01-10 12:08:25 PST
:-(
Comment 12 Adam Roben (:aroben) 2011-01-10 12:17:09 PST
Rolled out r75289 and r75392. I commented in bug 26276.