[Webkit-unassigned] [Bug 62628] [EFL] Crash when running EWebLauncher

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=62628





--- Comment #17 from Lucas De Marchi <demarchi at webkit.org>  2011-06-14 15:46:10 PST ---
(In reply to comment #16)
> (From update of attachment 97101 [details])
> This patch causes crashes.
> 
> Can you write a test that demonstrates the issue?  It seems likely that this is a bug in the EFL WebKit Layer.  You're probably calling some sort of end-ish function too many times.

(In reply to comment #15)
> (In reply to comment #14)
> > @abarth: looks like the cr-linux-ews is just confused.
> 
> It's not confused.  Those test failures are real.  The patch is wrong.

The patch may be wrong, but this piece of code is wrong as well. m_parser cannot disappear after calling m_parser->flush(this), so why checking if m_parser is 0 after that(In reply to comment #16)
> (From update of attachment 97101 [details])
> This patch causes crashes.
> 
> Can you write a test that demonstrates the issue?  It seems likely that this is a bug in the EFL WebKit Layer.  You're probably calling some sort of end-ish function too many times.

Looking the backtrace Michal provided, I can say for sure it's not specific to EFL. If DocumentWriter::endIfNotLoadingMainResource() cannot be called with m_parser == 0, it seems a problem in soup and therefore this can be reproduced in GTK port too. Michal, could you test this on GTK as well?

Anyway, the patch might not be right but this piece of code does look wrong. It's not possible that m_parser becomes 0 after calling m_parser->flush(this) and the check should be removed or moved up like the patch Michal sent.

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