[prev in list] [next in list] [prev in thread] [next in thread]
List: kde-core-devel
Subject: Re: Review Request: Fix a VLC crash by delaying object deletion to avoid invalid access by QtDBus
From: "David Faure" <faure () kde ! org>
Date: 2012-02-02 20:07:48
Message-ID: 20120202200748.2187.4400 () vidsolbach ! de
[Download RAW message or body]
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://git.reviewboard.kde.org/r/100577/#review10296
-----------------------------------------------------------
A very general rule to software development is: don't commit code or fixes that you \
don't understand.
So I don't like this, because it's a workaround for an un-identified issue. If it \
crashes, where's the valgrind log that explains exactly why it crashes? If it's a Qt \
bug, then it should be investigated and fixed in Qt. Then we can talk about \
short-term workarounds to make kde users happier, but not before (which would just \
hide and postpone the issue until later).
What's the harm in committing this anyway, you might ask? Well 1) this could happen \
again anywhere else since this looks like a fix at the wrong level, 2) I don't know \
if deleteLater are processed at app quitting time, so this might lead to false \
positives in future memory-leak debugging. 3) in general, any of these non-running \
destructors could be doing things that we actually need to happen.
- David Faure
On Jan. 30, 2012, 6:35 a.m., Dawit Alemayehu wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://git.reviewboard.kde.org/r/100577/
> -----------------------------------------------------------
>
> (Updated Jan. 30, 2012, 6:35 a.m.)
>
>
> Review request for kdelibs and Thiago Macieira.
>
>
> Description
> -------
>
> As described in the bug report when opening the file dialog, canceling it and \
> exiting VLC causes the application to seg fault. The backtraces posted in the bug \
> report shows that the crash happens somewhere in QtDBus and I get a similar \
> backtrace when using v1.1.7 of this application.
> When I looked into this issue what I discovered was that only KIO classes that \
> register themselves with QtDBus cause the crash. I only found that out because \
> fixing the crash in KIO::Scheduler by defering its deletion using deleteLater() did \
> not prevent the application from still crashing on exit. However, the crash had \
> moved to another location, KBookmarkManager. The only common thing between the \
> crash at KBookmarkmanager and KIO::Scheduler was that in both cases the backtrace \
> shows QtDBus attempting to invoke or access some method or slot in the object that \
> was just destroyed. At this point I decided to check how many other classes in KIO \
> register themselves with QtDBus and found only one more. I then applied the same \
> changes to all three, defering the deletion of any object registered with QtDBus \
> using QObject::deleteLater, and the crashing went away.
> Anyhow, I have no idea why the crashes are only seen with this application. I was \
> not able to duplicate the crash using the same sequences in another Qt only app, \
> Arora or any of the other KDE applications I tried.
>
> This addresses bug 234484.
> http://bugs.kde.org/show_bug.cgi?id=234484
>
>
> Diffs
> -----
>
> kio/bookmarks/kbookmarkmanager.cc d8a9cb7
> kio/kio/fileundomanager.cpp f580c29
> kio/kio/scheduler.cpp 6b36d9d
>
> Diff: http://git.reviewboard.kde.org/r/100577/diff/diff
>
>
> Testing
> -------
>
>
> Thanks,
>
> Dawit Alemayehu
>
>
[Attachment #3 (text/html)]
<html>
<body>
<div style="font-family: Verdana, Arial, Helvetica, Sans-Serif;">
<table bgcolor="#f9f3c9" width="100%" cellpadding="8" style="border: 1px #c9c399 \
solid;"> <tr>
<td>
This is an automatically generated e-mail. To reply, visit:
<a href="http://git.reviewboard.kde.org/r/100577/">http://git.reviewboard.kde.org/r/100577/</a>
</td>
</tr>
</table>
<br />
<pre style="white-space: pre-wrap; white-space: -moz-pre-wrap; white-space: \
-pre-wrap; white-space: -o-pre-wrap; word-wrap: break-word;">A very general rule to \
software development is: don't commit code or fixes that you don't \
understand.
So I don't like this, because it's a workaround for an un-identified issue. \
If it crashes, where's the valgrind log that explains exactly why it crashes? If \
it's a Qt bug, then it should be investigated and fixed in Qt. Then we can talk \
about short-term workarounds to make kde users happier, but not before (which would \
just hide and postpone the issue until later).
What's the harm in committing this anyway, you might ask? Well 1) this could \
happen again anywhere else since this looks like a fix at the wrong level, 2) I \
don't know if deleteLater are processed at app quitting time, so this might lead \
to false positives in future memory-leak debugging. 3) in general, any of these \
non-running destructors could be doing things that we actually need to happen.</pre> \
<br />
<p>- David</p>
<br />
<p>On January 30th, 2012, 6:35 a.m., Dawit Alemayehu wrote:</p>
<table bgcolor="#fefadf" width="100%" cellspacing="0" cellpadding="8" \
style="background-image: \
url('http://git.reviewboard.kde.org/media/rb/images/review_request_box_top_bg.png'); \
background-position: left top; background-repeat: repeat-x; border: 1px black \
solid;"> <tr>
<td>
<div>Review request for kdelibs and Thiago Macieira.</div>
<div>By Dawit Alemayehu.</div>
<p style="color: grey;"><i>Updated Jan. 30, 2012, 6:35 a.m.</i></p>
<h1 style="color: #575012; font-size: 10pt; margin-top: 1.5em;">Description </h1>
<table width="100%" bgcolor="#ffffff" cellspacing="0" cellpadding="10" \
style="border: 1px solid #b8b5a0"> <tr>
<td>
<pre style="margin: 0; padding: 0; white-space: pre-wrap; white-space: \
-moz-pre-wrap; white-space: -pre-wrap; white-space: -o-pre-wrap; word-wrap: \
break-word;">As described in the bug report when opening the file dialog, canceling \
it and exiting VLC causes the application to seg fault. The backtraces posted in the \
bug report shows that the crash happens somewhere in QtDBus and I get a similar \
backtrace when using v1.1.7 of this application.
When I looked into this issue what I discovered was that only KIO classes that \
register themselves with QtDBus cause the crash. I only found that out because fixing \
the crash in KIO::Scheduler by defering its deletion using deleteLater() did not \
prevent the application from still crashing on exit. However, the crash had moved to \
another location, KBookmarkManager. The only common thing between the crash at \
KBookmarkmanager and KIO::Scheduler was that in both cases the backtrace shows QtDBus \
attempting to invoke or access some method or slot in the object that was just \
destroyed. At this point I decided to check how many other classes in KIO register \
themselves with QtDBus and found only one more. I then applied the same changes to \
all three, defering the deletion of any object registered with QtDBus using \
QObject::deleteLater, and the crashing went away.
Anyhow, I have no idea why the crashes are only seen with this application. I was not \
able to duplicate the crash using the same sequences in another Qt only app, Arora or \
any of the other KDE applications I tried.</pre> </td>
</tr>
</table>
<div style="margin-top: 1.5em;">
<b style="color: #575012; font-size: 10pt; margin-top: 1.5em;">Bugs: </b>
<a href="http://bugs.kde.org/show_bug.cgi?id=234484">234484</a>
</div>
<h1 style="color: #575012; font-size: 10pt; margin-top: 1.5em;">Diffs</b> </h1>
<ul style="margin-left: 3em; padding-left: 0;">
<li>kio/bookmarks/kbookmarkmanager.cc <span style="color: \
grey">(d8a9cb7)</span></li>
<li>kio/kio/fileundomanager.cpp <span style="color: grey">(f580c29)</span></li>
<li>kio/kio/scheduler.cpp <span style="color: grey">(6b36d9d)</span></li>
</ul>
<p><a href="http://git.reviewboard.kde.org/r/100577/diff/" style="margin-left: \
3em;">View Diff</a></p>
</td>
</tr>
</table>
</div>
</body>
</html>
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic