[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#800698: libqtgui4: GUI applications are unusable on amd64 systems



Hi,

>>  control: retitle 800698 libqtgui4: GUI applications become unusable when
>>  shared memory is exhausted
>
> With this title I'm even more convinced that this is not a bug ;) Please read
> below.

This is only an assumption which is based on found bug reports in other projects.

>>  >> but hey, why Qt5 based and Gtk+
>>  >> based programs were working without any issues in the same conditions?
>>  >
>>  > I do have an idea. Where those java applications creating an icon in the
>>  > taskbar?
>>  >
>>  > The only change that *might* be causing the problem here is the addition
>>  > of a patch that allows the systray to use sni-qt.
>>
>>  At first look I do not see anything related to shared memory usage in patch
>>  [1].
>>
>>  [1]
>>  http://anonscm.debian.org/cgit/pkg-kde/qt/qt4-x11.git/tree/debian/patches/plugin_system_for_systemtray.patch?h=debian/4.8.7%2bdfsg-3
>>  > Did any of the Java created a taskbar icon?
>>
>>  If you had in mind a notification area, then yes, the most complex of these
>>  Java based programs creates an icon in system tray. And this icon is
>>  blinking when some events inside program trigger it.
>
> The only source change between the versions you mentioned in the original bug
> report is that patch. And I know Qt4 is using (X-11-related?) shared memory to
> draw things because I remember having to fix some permissions issues in past
> releases (for Wheezy IIRC).
>
> I don't know why GTK doesn't fails, and I *guess* that Qt5 might be using some
> OpenGL-related stuff that might make things go without problems.

Yes, it looks that Qt4 uses X11 rendering by default. I have checked [1] and [2].
In the first case GUI rendering is broken and unusable and there are a lot of 
repeated X11 error messages in console [3]. But in the second case program looks
more or less fine (only font in the main menu and in context menus looks ugly
because of font hinting does not work here).

[1] okular --graphicssystem raster
[2] okular --graphicssystem opengl
[3] The beginning of log from stderr:
QNativeImage: Unable to attach to shared memory segment. 
X Error: BadDrawable (invalid Pixmap or Window parameter) 9
  Major opcode: 62 (X_CopyArea)
  Resource id:  0x0
X Error: BadPixmap (invalid Pixmap parameter) 4
  Major opcode: 56 (X_ChangeGC)
  Resource id:  0x0
X Error: BadDrawable (invalid Pixmap or Window parameter) 9
  Major opcode: 62 (X_CopyArea)
  Resource id:  0x0
...

> Are you running sni-qt? If so maybe the bug is there, or at least related.

I have removed sni-qt package and after reboot bug is still present.

>>  The main question is what to do with Qt4 libraries?
>>
>>  Other Debian users may face to this bug because package default-jdk depends
>>  from openjdk-7-jdk in Debian Stretch.
>
> Well, if the memory gets exhausted is no wonder that Qt4 fails. Now I hardly
> doubt that this is a bug in Qt4 due to the amount of testing the current
> version has, after all it's just bug fixes from stable, and not many linux-
> related precisely. And the behavior started with just a patch for sni-qt.
>
> If you have sni-qt installed I would definitely take a look there first.

sni-qt does not cause to this bug. The problem is somewhere else.

Unfortunately, I do not have cowbuilder environment for amd64 now thus I cannot
rebuild qt4-x11 package without discussed patch for testing, but I have force
installed libqtgui4/4:4.8.7+dfsg-1 from Debian snapshots archive [4] (yes, only
this one package and nothing more) and after reboot I can reproduce the bug.
So the problem is definitely not in the plugin_system_for_systemtray.patch.

Probably an update in completely anoter package have lead to this bug.

[4] http://snapshot.debian.org/archive/debian/20150526T221958Z/pool/main/q/qt4-x11/libqtgui4_4.8.7%2Bdfsg-1_amd64.deb

> I'll leave the bug open tagged as moreinfo in case you want to reassign it,
> else I'll close it in a couple of weeks, as nothing shows this is really a Qt4
> bug.

I do not have any experience in Java and I do not have enough time and interest
for investigating the bug in openjdk-7 dipper. Also we can do nothing from Qt4
side. And there is a known workaround. So this bug report might be closed now.

Best regards,
Boris


Reply to: