Bug 60101 – ASSERT(m_state == ParsingState) fires @ www.canalplus.fr

• Component: WebCore Misc.
• Status: RESOLVED FIXED

[Geoffrey Garen]

1. Go to http://www.canalplus.fr/c-sport/c-football/italie/pid3517-videos-calcio.html
--> ASSERT

[Geoffrey Garen]

The ASSERT was added in http://trac.webkit.org/changeset/66841 with the rationale "It is better to ensure it is only called while parsing." There doesn't seem to be any explanation of what guarantees this invariant -- just an aspiration toward a world where this invariant was true.

[Adam Barth]

We'll need to look a why the ASSERT is firing to determine whether the ASSERT is incorrect or whether the caller is incorrect.

[Alexey Proskuryakov]

See also: bug 50253.

[Tony Gentilcore]

I'm preparing a reduction.

[Tony Gentilcore]

I'm pretty sure we don't want to remove the ASSERT. If it hits, we'd go through http://www.whatwg.org/specs/web-apps/current-work/#the-end twice.

The page is doing some tricky combinations of writing into an iframe then calling document.close() which is incorrectly leading to the second prepareToStopParsing() call for the document. It's turning out to be hard to reduce.

One potential fix is to change the "if (isStopped())" in prepareToStopParsing() to "if (!isParsing())", but until I get a proper reduction, I'm not sure I understand the issue fully. I'll keep working on it, but just wanted to give a quick update.

[Tony Gentilcore]

Created attachment 92428 [details]
Testcase

[Tony Gentilcore]

Created attachment 92431 [details]
Patch

[Adam Barth]

The code for stopping/canceling/ending parsing really needs to be beaten with a stick.

[Tony Gentilcore]

(In reply to comment #8)
> The code for stopping/canceling/ending parsing really needs to be beaten with a stick.

I couldn't agree more. It doesn't line up with the spec well, has known and unknown bugs and several fixmes. The problem is that it is so fragile, and the stopping/canceling sometimes bleeds over into port-specific logic. :(

[WebKit Commit Bot]

Comment on attachment 92431 [details]
Patch

Rejecting attachment 92431 [details] from commit-queue.

Failed to run "['./Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '--bot-id=cr-jail-4', 'build-..." exit_code: 2

Last 500 characters of output:
....
http/tests/xmlhttprequest/cross-origin-authorization.html -> failed
............................................................................................................................
http/tests/xmlhttprequest/web-apps ...............
http/tests/xmlhttprequest/workers ...........
http/tests/xmlviewer .
http/tests/xmlviewer/dumpAsText ...........
797.38s total testing time

23459 test cases (99%) succeeded
1 test case (<1%) had incorrect layout
15 test cases (<1%) had stderr output

Full output: http://queues.webkit.org/results/8571536

[WebKit Commit Bot]

Created attachment 92465 [details]
Archive of layout-test-results from cr-jail-4

The attached test failures were seen while running run-webkit-tests on the commit-queue.
Bot: cr-jail-4  Port: Mac  Platform: Mac OS X 10.6.7

[Eric Seidel (no email)]

Comment on attachment 92431 [details]
Patch

That's a known flaky test.  I expect it just flaked twice in a row when we tried to land. :(

[WebKit Commit Bot]

The commit-queue encountered the following flaky tests while processing attachment 92431 [details]:

http/tests/appcache/non-html.xhtml bug 60308 (author: [e-mail])
The commit-queue is continuing to process your patch.

[WebKit Commit Bot]

Comment on attachment 92431 [details]
Patch

Clearing flags on attachment: 92431

Committed r85894: <http://trac.webkit.org/changeset/85894>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Ademar Reis]

Revision r85894 cherry-picked into qtwebkit-2.2 with commit c88e89d <http://gitorious.org/webkit/qtwebkit/commit/c88e89d>