Corrupted layouts on Qt4/Kde4 applications using Qt-4.5 on Intel graphics

March 10, 2009 · Posted in Gentoo 

Qt-4.5 upgrade on my laptop wasn’t that smooth. Every qt4/kde4 application had a corrupted layout and fonts were unreadable.

The fact that I didnt have such issues on any other machine made me think that there was something seriously wrong with my laptop.
KDE+QT were compile with the same CFLAGS+USE flags on every machine and he kde+qt4 versions were exactly the same. So, I figured out that it has to be something related to graphics card ( or drivers ).

KDE4 ( with qt-4.5 ) was running smoothly on every pc with Nvidia graphics card. But my laptop has Intel graphics chipset and Qt-4.5 sucked big time.

This is my hardware

0:02.0 VGA compatible controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (rev 03)
00:02.1 Display controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (rev 03)

So , after some googling , I did the below changes

1) Xorg.conf

Added the following on Device Section

Option “DRI” “true”
Option “AccelMethod” “XAA”
Option “XAANoOffscreenPixmaps” “True”
Option “XaaNoPixmapCache”

2) .bashrc
Added the following on my bashrc file

export INTEL_BATCH=1

Since then, my kde4 runs perfectly with Qt-4.5 . No corrupted layouts or unreadable fonts anymore , plus it is so much faster.

So, give it a try if you are having a hard time with Qt-4.5 + Intel graphics card

Comments

9 Responses to “Corrupted layouts on Qt4/Kde4 applications using Qt-4.5 on Intel graphics”

  1. RealNC on March 10th, 2009 4:45 am

    XAA is the slowest render path of all. I think Intel supports UXA, which is the fastest. Also EXA, which also much faster than XAA.

    So try these, in this order, and keep the one that works:

    Option “AccelMethod” “UXA”

    Option “AccelMethod” “EXA”

    Option “AccelMethod” “XAA” # Slow, slow, slow.

  2. Markos on March 10th, 2009 9:41 am

    Yeah but UXA gives me a nice system lock up :p and EXA results on corrupted applications again :(

    So…

  3. Markos on March 10th, 2009 12:10 pm

    And also I must say that I am using quite old drivers (2.5.X) series cause the latest 2.6.X triggers this bug

    https://bugs.freedesktop.org/show_bug.cgi?id=20542

    So until then , 2.5.X + XAA is the only solution for me

  4. Rahul on March 26th, 2009 12:52 pm

    I am using this on qt-4.5, intel-2.6.3-r1 and kde-4.2

    Section “Device”
    Identifier “Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller”
    Driver “intel”
    BusID “PCI:0:2:0″
    Option “AccelMethod” “UXA”
    Option “Tiling” “false”
    EndSection

    Works very well, and no lockups on resume or anything like that. Only problem is that after 4-5 hours of uptime, the swap space gets full. However, compositing runs very smoothly – better than ever before (compared to compiz and all that).
    Would use your config and see if I still get the swap problems.

  5. Markos on March 26th, 2009 12:59 pm

    Actually my config fits on intel-2.5.X drivers. I use UXA as well on 2.6.3.

    Are you using xorg-1.6? Cause since I upgraded to xorg-1.6 I kinda have some small(?) memory leaks :\

  6. Rahul on March 26th, 2009 1:06 pm

    Yeah, your config does not work with 2.6.x series – got a matrix screen with that. :)
    Yup, I am on xorg-1.6 and since then I have these annoying memory leaks. Nothing major, but it requires me to zap X after every 5 hours. Maybe 2.6.29 will fix this but I am waiting for tuxonice-2.6.29 to make it to portage. (I never had these X lockups on getting back from suspend/hibernate) but I have always been on tuxonice so maybe tuxonice works better.

  7. Markos on March 26th, 2009 1:20 pm

    Actually I am getting the same behavior with 2.6.29 gentoo-sources. I need to find some time to google about that :\

  8. Rahul on April 12th, 2009 9:13 pm

    Yeah, I still have the same problem with a full swap after running X for 4-5 hours.

    2.6.29-zen3
    x11-drivers/xf86-video-intel-2.6.99.902
    x11-base/xorg-server-1.6.0

    Plus, a new warning:
    (WW) intel(0): libpciaccess reported 0 rom size, guessing 64kB

  9. Markos on April 13th, 2009 2:07 am

    Actually it is not that bad over here

    Eternity ~ # free -m
    total used free shared buffers cached
    Mem: 1986 1949 37 0 227 969
    -/+ buffers/cache: 752 1234
    Swap: 1913 132 1781

    I am using 2.6.29-gentoo-r1 sources, with xf86-video-intel-2.6.99.902 and all the unstable thingie for x11 overlay

    I get that warning about libpciaccess as well. It doesnt sound good to me though :P

Leave a Reply




GreekBloggers.com
Patras Wireless Metropolitan Network
Planet Hellug
iloog
forum hellug