Bug 63595 - [845g uxa 2.19] missing text in firefox on first run
Summary: [845g uxa 2.19] missing text in firefox on first run
Status: NEW
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: 7.7 (2012.06)
Hardware: x86 (IA32) Linux (All)
: medium normal
Assignee: Chris Wilson
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2013-04-16 09:58 UTC by sauron99
Modified: 2015-07-26 19:56 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
image (85.44 KB, image/png)
2013-04-16 11:16 UTC, sauron99
no flags Details
Xorg.log (39.83 KB, text/plain)
2013-04-16 11:17 UTC, sauron99
no flags Details
syslog (128.65 KB, text/plain)
2013-04-16 11:17 UTC, sauron99
no flags Details
screenshot of KTip + Konsole 3.5.10 automatically resumed from previous session (63.08 KB, image/png)
2013-09-05 04:26 UTC, Felix Miata
no flags Details
dmesg excerpt from the system from which comment 12 screenshot taken (3.51 KB, text/plain)
2013-09-05 04:26 UTC, Felix Miata
no flags Details
Xorg.0.log from session with corrupted text shown in comment 12 screenshot (31.20 KB, text/plain)
2013-09-05 04:26 UTC, Felix Miata
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description sauron99 2013-04-16 09:58:33 UTC
Hi there,
i have a problem with the latest version of Webconverger and Xorg 7.7 which comes with it. 

Here you can see all installed packages:

http://build.webconverger.org/webc-19.1.txt

and there you can download Webconverger:

http://build.webconverger.org/webc-19.1.txt


The old version´s worked quite well, but the current one doesn´t.

After booting up all characters are chaotic and it looks broken. After closing Firefox with ALT + F4 and starting it up again, everything looks ok.

So this is just a fault right after booting up Webconverger with my onboard Intel 82845G/GL. 

Im very unexperienced with Xorg, so please give me a hint how to give you the right informations. 

Thanks!
Comment 1 Chris Wilson 2013-04-16 10:12:16 UTC
Start by attaching your Xorg.0.log please.
Comment 2 Chris Wilson 2013-04-16 10:13:44 UTC
If you can capture the corruption using a screenshot or a photograph that would be very useful.
Comment 3 sauron99 2013-04-16 11:16:57 UTC
Created attachment 78064 [details]
image
Comment 4 sauron99 2013-04-16 11:17:17 UTC
Created attachment 78065 [details]
Xorg.log
Comment 5 sauron99 2013-04-16 11:17:34 UTC
Created attachment 78066 [details]
syslog
Comment 6 Chris Wilson 2013-04-16 13:12:24 UTC
How easy is it for you to install a more recent xserver-xorg-video-intel driver?
Comment 7 sauron99 2013-04-16 13:15:55 UTC
Depends on how easy it is ;). Webcovnerger is a live-cd so every change goes away after rebooting.
Comment 8 Chris Wilson 2013-04-16 13:20:02 UTC
You can try 'sudo apt-get -t experimental install xserver-xorg-video-intel' which is still a few months old (Debian you fail me!) but should have at least one relevant fix. Whilst the change will be volatile, it should be enough for testing. :)
Comment 9 sauron99 2013-04-16 13:59:31 UTC
I will try to get a new version of WEbconverger to try this out, because it´s useless to install the new driver and restart X. AFter restarting Firefox in normal mode it works anyway. 

So maybe I will get a version of it with this package from experimental. Otherwise i have to build it by myself.
Comment 10 Chris Wilson 2013-04-16 14:19:36 UTC
I have to admit being very puzzled by the behaviour. Strongly suggests that some GPU memory is incoherent on the first run (invalid TLB or bad caching etc). I would be very interested in seeing if you can build the package for the iso yourself with the latest drivers and then trying SNA (either by building --with-default-accel=sna or by adding 

Section "Device"
  Identifier "Device0"
  Driver "intel"
  Option "AccelMethod" "sna"
EndSection

to your /etc/X11/xorg.conf -- actually probably easier to rebuild ;-)
Comment 11 sauron99 2013-04-18 14:12:45 UTC
The sna thing does work. 
The experimental intel-driver doesn´t.

Strange about this is the fact that it only works with the Live CD. I used to put the Webcovnerger onto a hdd with dd. With this version he tries to boot but just before he sets the correct resolution it crashs. Its very similar to the behavior before the SNA trick.
Comment 12 Felix Miata 2013-09-05 04:26:34 UTC
Created attachment 85219 [details]
screenshot of KTip + Konsole 3.5.10 automatically resumed from previous session

Screenshot shows what happens running 3.11.rcX, but not when running 3.10.x, on first session startup after booting openSUSE running KDE3, but not on session restarts on the same boot. It looks much like what comment 0 describes. It only happens on my i845G system with KDE3, not another with i845G and KDE4, nor with several other systems with either newer Intel video (i945G) or systems needing Radeon or Nouveau drivers.

VERSION="13.1 Milestone 4 (Bottle)"
kdebase3-kdm-3.5.10.1-342.11.i586
kernel-desktop-3.10.1-3.5.g0cd5432.i686
kernel-desktop-3.11.rc7-1.2.g99e1318.i686
xf86-video-intel-2.21.14-2.2.i586
xorg-x11-7.6_1-10.1.noarch
xorg-x11-driver-video-7.6_1-10.3.i586
xorg-x11-server-7.6_1.14.2-2.2.i586
Comment 13 Felix Miata 2013-09-05 04:26:46 UTC
Created attachment 85220 [details]
dmesg excerpt from the system from which comment 12 screenshot taken

Segfault info from dmesg similar/identical? to what appears on tty1 during boot of 3.11.rc7 openSUSE kernel.
Comment 14 Felix Miata 2013-09-05 04:26:54 UTC
Created attachment 85221 [details]
Xorg.0.log from session with corrupted text shown in comment 12 screenshot
Comment 15 Chris Wilson 2013-09-05 10:31:51 UTC
Hmm, that's pretty severe, but it may just be the first character in every run is lost. I guess it also only affects CA glyphs. You could try turning off sub-pixel font rendering in your system settings to see if that makes any difference.
Comment 16 Felix Miata 2013-09-05 15:02:36 UTC
Your comment suggested to me to try SXGA LCD instead of CRT, but it didn't help. I know not what is making the deselected sub-pixed setting to be unavailable regardless of display connected. Changing hinting from full to slight does not help. Removing /etc/X11/xorg.conf.d/50-monitor.conf, which reenables forced 96 DPI, changed the corruption pattern a bit, enabling complete titlebars to render with no corruption apparent, but disabled all text within the KTip window and Kicker. Pale gray text appears in the tooltips when Kicker is hovered, and no text appears when App Starter is opened. Switching to tty2 and back also changes nothing.
Comment 17 Chris Wilson 2014-09-07 15:29:58 UTC
I've just spent some time fixing up an issue affecting igt, and seems a reasonable suspect for this bug:

http://patchwork.freedesktop.org/patch/33192/

If you could apply that to your kernel and test, that would be very useful.
Comment 18 Felix Miata 2014-09-11 06:27:51 UTC
I am only able to test whatever binaries others are able to make available.
Comment 19 Felix Miata 2015-07-06 10:15:24 UTC
The comment 12 problem on i845G host gx260 does appear to have gone away by kernel 3.12.39, intel driver 2.99.906 and X server 1.14.3.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.