Bug 187356

Created attachment 32172 [details]
graphical corruption screenshot

I should note this was not and issue using qt-4.4

Please find out how your Qt4.5 was built (ask your distribution packagers), if they used the configure switch -graphicssystem raster or -graphicssystem opengl then file a bugreport with your distribution. Thats simply wrong, enabling anything but the native graphicssystem on X11 will cause random applications to break in random ways - from painting errors as you see to crashes.

It's gentoo packages... wouldn't it be built from source?

Yes, so what configure options where used when Qt was built?

http://bugs.gentoo.org/show_bug.cgi?id=262844 I've opened a downstream bug, to be sure. but I do not believe either were enabled... given useflag options.

Created attachment 32246 [details]
new screenshot

I recompiled qt, kdelibs,and okular making sure that both the opengl and raster use flags were disabled. meaning it should be at it's default... although I've still not gotten a confirmation from gentoo...

the problem seems to be less frequent and a bit different from this newly attached screenshot ...

*** Bug 187684 has been marked as a duplicate of this bug. ***

I have noticed this problem in dragon player, amarok. dolphin, kmail.
In dragon player under full screen mode. the progress bar turn green but gets redrawn fast.
I am attaching the kmail and dolphin ones.

Created attachment 32314 [details]
kmail

Created attachment 32315 [details]
dolphin

I forgot to mention I am using kubuntu jaunty alpha 6 with kde 4.2.1 packages and qt 4.5

no, no special neither opengl, nor rasterization were used in my build of qt.

Why no one is looking in to this bug. This is the most annoying issue I encounter with kde. What ever kde application I use say quassel or amarok or dolphin or kate they have like huge holes through which we can see desktop.

Well it's hard to pin down "Qt version changes, now KDE looks different".  What would be useful for troubleshooting is going through some of the following things:

* Does this happen with every style, or just Oxygen (or are there styles where this problem is not noted, etc.)
* Does this happen if KWin's Desktop Effects are disabled? (Configure Window Behavior in the right-click menu from any title bar)
* Does this happen if XOrg's Composite and Damage extensions are disabled?

If a KDE dev was running across this behavior there would be more feedback to the bug, so presumably it is not a common occurrence.  So lacking that it is important to make the bug as easy to reproduce and isolate as possible.  Screenshots are good to confirm the issue but all we know at this point is that there is an issue, which may even be a Qt and not a KDE bug for all we know.

issues do exist when compositing in kwin are disabled (compositing crashes more frequently these days). I can check my xorg.log at some point to see the status of compositing and damage as I don't actually configure an xorg.conf on my system.

I believe the problem is with oxygen. These are my observations.
1. With oxygen style , with or without kwin desktop effects. I have the graphical corruptions.
2. With plastique style, with or without kwin desktop effects , I do not have this problem.

Sorry to say but this bug should be reopened: it also affects intel video cards, the bug has been forwarded to intel who answered that the problem is in the application, not in X or in the drivers. This is somehow related to the update to qt 4.5, since there were no problems before (and no problems with gtk apps too)...
see a discussion about it here:
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-ati/+bug/338669
from my experience, it seems to happen mostly with oxygen style. Since I use qtcurve, it only happens in skype,  from time ti time...don't know why...

I see the problem affecting the window manager at times too... I get a weird bluish overlay hiding part of konversation when taskbar stuff pops up, it goes away and leaves it hidden... it's possible it's a bug related to intel, I have an i915. this info should help narrow it down, it's also possible it's a qt bug, but as a non developer it's hard for me to say, if you can discern that, it'd be nice for a dev to explain the issue (or perceived) to the qt devs and link back here.

Giovanni, this bug is still open, but I'll mark it as NEW rather than UNCONFIRMED based on the number of users confirming it.

a) this is precisely why we cautioned against shipping Qt 4.5 with KDE 4.2. once again, distros didn't listen, and once again, we get to deal with the bug reports and our users get to deal with bugs they shouldn't have to.

b) this is probably some odd bug in Qt 4.5, though it may well be triggered by oxygen. it could also be a fun interaction between Qt 4.5 and the graphics driver ... but i haven't seen this problem and don't know how to duplicate it. but then, i'm using svn trunk/ (what will be KDE 4.3), know that everything is built against Qt 4.5, etc. and yes, i also have an an intel chipset in this laptop.

i haven't heard about this problem from others who are running trunk, either, and we are _all_ using Qt 4.5 there, so that's even more suspicious.

so ... when the first betas of KDE 4.3 come out, the reporters should install it and test against that. if the problems magically go away then we can chalk this up once again to distros ignoring upstream's recommendations.

if it still exists in the betas, then we will need to find a hardware/driver/application/use case combination that we can reproduce this on and then get it fixed before KDE 4.3.0 is released.

in the case of our distro, qt-4.5 is available but not stable, 4.4 is... so one could suggest it's the users fault for using it. if it was a binary/non-rolling I'd say we shouldn't have packaged it at all... I would like to say... that kde and distro's need to start working together, for binary distro's this would be the second time they've screwed up when kde said don't do it.

i just found a way to reproduce it, i think! if i put Dragon Player into full screen mode the toolbar is painted at first with graphical corruption similar to the screenshots.

now to figure out why it would happen in that case ... hmmm.

The issue with Dragon player seems different to the issues stated above: even though Dragon player's progress bar has corruption regardless of which computer I'm using or which graphics card I'm using, I can't re-produce the corruption in the Oxygen theme... with Intel or nVidia graphics cards.

I'll be honest... I'm not sure what the corruption I see in some cases would have to do with oxygen, such as that rendering in okular, or sometimes when the window manager leaves glyphs on an open window after something over top of it hides.

Is there any work around for this bug other than using qt 4.4.

I have the same problem, using the latest xf-video-ati driver from GIT with EXA.
Changing the style from Oxygen to QtCurve seems to work for me. Probably we have
to report this bug to the trolls?

I experience the same issue on openSuse 11.1 with KDE4 UNSTABLE packages and both the Qt and the Qt45 (which are a little bit newer) repositories enabled.

I'm using the radeon driver for my Radeon Mobility X1400 on a T60.
The problem seems to occur only after some time or after an suspend to ram -> resume cycle. When fresh booted, the issue isn't this frequent.

I experience the problem especially when using kontact/akregator and opening/closing yakuake. After yakuake is closed, the corruption remains in some of the parts that were hidden by it before.

b) this is probably some odd bug in Qt 4.5, though it may well be triggered by
oxygen.

I think this is the case. My KDE provider KDEmod http://chakra-project.org/ compiled 4.2 against qt 4.5 and I can get rid of this bug simply by switching to the qt-curve theme. If this doesn't appear in trunk it may have been a bug in Oxygen? Maybe it was already fixed by the trolls? It's definitely a weird case. Anyway I will fix a bug with KDEmod so that they listen to KDE devs!

I'm not using oxygen for anything, I don't get this bug except in text canvases - ie. in kate's editing pane.

full description and screenshots here: https://bugs.launchpad.net/ubuntu/+source/qt4-x11/+bug/371114

subscribing

Hmm, switched to QtCurve yesterday and the corruption hasn't occured since. What I suspect is, that it is a problem when painting gradients or how plasma does it since it mostly affects parts of widgets where there are gradients used (like the upper third of a window).

Sorry, meant of course how oxygen does paint gradients, not plasma.

I haven't noticed this in 4.2.3 w/ qt 4.5.1 yet... doesn't mean it's not there...

anyone else seeing this still with qt 4.5.1?

Created attachment 33507 [details]
Konversation corruption

Not sure if it's the same kind of corruption but I often get this in Konversation.
Using Qt 4.5.1 and KDE 4.3 from svn.

I'm still seeing it in 4.2.3 w/ qt 4.5.1.

Still seeing it with KDE 4.2.85 / Qt 4.5.1 when Oxygen is used as widget theme.
Doesn't happen e.g. with Qt Curve. 
Experiencing it mainly in kontact.
It doesn't matter if effects are on or not.

Created attachment 33604 [details]
Screenshot of kontact with Akregator open

As already said, it happens only with Oxygen and only in the upper third / quarter of the app where Oxygen is painting a gradient.

There is only users commenting here I see this ot I dont see this issue. Can an y KDE devs comment on what's actually happening?

qt-4.5.x tickles a few X/video driver bugs it seems, comment #38 is one, by the looks of it.

I also have the same problem, with Qt 4.5.1, my video card is Intel 945.

This happens to both KDE 4.2.2 and KDE 4.3 trunk (4.2.85), but only with oxygen style.

For cleanlooks style, sometimes the artifacts seen but much less compared to oxygen style. As a workaround now I use polyester style.

The artifacts were gone after I upgraded xserver-xorg-core from version 1.6.1 to 1.6.2. So seems that this is an upstream bug and has been resolved upstream as well.

Can anyone else confirm this who previously experienced this bug after upgrading xserver-xorg-core to 1.6.2?  (Also known as xorg-server on at least Gentoo)

(In reply to comment #43)
> Can anyone else confirm this who previously experienced this bug after
> upgrading xserver-xorg-core to 1.6.2?  (Also known as xorg-server on at least
> Gentoo)

Yep, I haven't seen this corruption for a while now (I've been using 1.6.2 RCs too).

Yeah, one of the last upgrades to xorg-x11-server (from openSUSE XOrg repo) seems to have fixed the issue. Can't see these corruptions anymore. Thank you very much!

Same here, on both, Mandriva and Gentoo, Xorg Server 1.6.2 fixed the problem

I still see the corruption with 1.6.2-rc1. So it must have been fixed in rc2.
I will test it and report back.

Well up to now no corruption with 1.6.2!

Sounds like we have a fix then, install xorg server 1.6.2 and things seem fixed.