[Webkit-unassigned] [Bug 177577] Crash/regression 2.18 on webkitgtk disposal

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=177577

--- Comment #2 from Leo Ufimtsev <leonidas at redhat.com> ---
(In reply to Tomas Popela from comment #1)
> (In reply to Leo Ufimtsev from comment #0)
> > I noticed that if I do an g_object_ref(webview) somewhere near disposal,
> > then the crash doesn't occur. 
>  
> Just to be sure, are you holding a reference on the web view in your
> Eclipse's/SWT's Browser when you create it?
> 
> > Does the above ring a bell with anyone?
> 
> What I would try is to debug the reference count of the web view (set a
> break point on glibc's g_object_ref()/g_object_unref() one you will know the
> address of the newly created web view).
> 
> > Was there a mechanism added to webkit/webkitgtk to somehow auto-cleanup
> > somewhere that didn't exist before?
> 
> No, at least I'm not aware of anything..
> 
> > Btw, what's the correct way to destroy/dispose webkit/webview?
> 
> Just o_object_unref() it.

Ah, thank you.

> > System:
> > Linux Fedora 25
typo, I'm on F26 btw.

> > Gtk3.22
> > WebkitGtk 2.18 (New from upstream/Rawhide)
> 
> Try to please use
> https://bodhi.fedoraproject.org/updates/FEDORA-2017-368b385c04 and don't use
> the rawhide.
> 
> > Please advise, thank you.

Hello Tomas,

Thank you for your suggestions. They were very helpful.

I've removed rawhide packages and installed 2.18 from regular Fedora repos. The crash still occurs (But I guess that was expected).

I found that I only have to g_object_ref(webview) the first instance that is created, then the crash no longer occurs.
This first instance is then cleaned up when JVM shuts down.

I haven't gotten around doing a deeper debug session on what g_object_(ref|unrefs) webview. For the time being we'll use the workaround mentioned above [1]. Later, once we've completed the Webkit2 port[2], we will come back to revisit this crash issue. (may take some months).

I can't reproduce the issue in a snippet, it only occurs in the context of a JVM.

Since I won't get around to doing deeper research for some time to come, should we close the issue for the time being ?(works for me?) and then once I'll get back to this crash issue and have more news re-open it?


[1]
Bug 522733 – Crash on close with webkitgtk 2.18 
https://bugs.eclipse.org/bugs/show_bug.cgi?id=522733

[2] 
Bug 516838 – [GTK3][webkit] Port SWT Browser to webkit2gtk (4.8) 
https://bugs.eclipse.org/bugs/show_bug.cgi?id=516838

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20171003/dd72cd12/attachment-0001.html>