Bug 187794

Summary: screen garbled with i810 driver
Product: [openSUSE] SUSE Linux 10.1 Reporter: michel munnix <michel.munnix>
Component: X.OrgAssignee: Marcus Schaefer <ms>
Status: RESOLVED DUPLICATE QA Contact: Stefan Dirsch <sndirsch>
Severity: Normal    
Priority: P5 - None CC: sndirsch, suse-beta
Version: Final   
Target Milestone: ---   
Hardware: i686   
OS: Other   
Whiteboard:
Found By: Other Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: /etc/X11/xorg.conf
/var/log/Xorg.0.log
/etc/X11/xorg.conf driver modified to i810xorg71
/var/log/Xorg.0.log for driver i810xorg71
i810xorg71.tar.bz2
/var/log/Xorg.0.log for driver i810xorg71 and IBM6737
/etc/X11/xorg.conf driver i810xorg71 and IBM6737
/var/log/boot.msg

Description michel munnix 2006-06-23 14:36:25 UTC
when using the configuration created with sax2, the screen missbehaves out of sync, moving the mouse left or right varies the speed and direction of the moving picture (horizontaly). Moving the mouse up and down moves the baseline of the screen up and down but interfers also with horizontal movements.
This is an IBM netvista desktop
00:02.0 VGA compatible controller: Intel Corporation 82845G/GL[Brookdale-G]/GE Chipset Integrated Graphics Device (rev 01)
The fbdev driver works ok
Comment 1 Stefan Dirsch 2006-06-23 14:39:20 UTC
Could you attach /etc/X11/xorg.conf and /var/log/Xorg.0.log? Thanks.
Comment 2 michel munnix 2006-06-26 06:37:42 UTC
Created attachment 91524 [details]
/etc/X11/xorg.conf
Comment 3 michel munnix 2006-06-26 06:47:36 UTC
Created attachment 91526 [details]
/var/log/Xorg.0.log

The graphical login screen was displayed with the described symptoms, then when I switched back to console mode, the log file had 1886 lines. When I switched back to VT7, I did not get the image back
Comment 4 Stefan Dirsch 2006-06-26 07:18:14 UTC
Nothing obvious I can see in the config/logfile. Could you also try "i810xorg71" driver. Simply replace "i810" with "i810xorg71" in your config file.
Comment 5 michel munnix 2006-06-26 07:55:08 UTC
Created attachment 91531 [details]
/etc/X11/xorg.conf driver modified to i810xorg71
Comment 6 Stefan Dirsch 2006-06-26 07:58:13 UTC
Ok. Does it help? I suggest to reboot before testing the Xserver again.
Comment 7 michel munnix 2006-06-26 08:02:04 UTC
Created attachment 91532 [details]
/var/log/Xorg.0.log for driver i810xorg71

the driver i810xorg71 was not found.
But I see a message in the first logfile which seems worth to be analysed, perhaps the driver cannot set the display modes in the graphics adapter:
(II) I810(0): Will use BIOS call 0x5f05 to set refresh rates for CRTs.
and
(WW) I810(0): Successfully set original devices
(WW) I810(0): Setting the original video mode instead of restoring
        the saved state
(II) I810(0): BIOS call 0x5f05 not supported, setting refresh with VBE 3 method.
Comment 8 Stefan Dirsch 2006-06-26 08:05:34 UTC
Sorry, the i810xorg71 driver has been integrated after 10.1 release. I'll attach it.
Comment 9 Stefan Dirsch 2006-06-26 08:08:13 UTC
Created attachment 91533 [details]
i810xorg71.tar.bz2
Comment 10 michel munnix 2006-06-26 09:34:11 UTC
I tried the driver i810xorg71 : At first it seems to work, but if switch to a text console and then back to VT7, the display shows "power saving mode" or "input signal out of range" (in a green box, generated inside the monitor).
After retrying several times, it seems to me it works just after a fresh boot, afterwards the screen gets the garbling when going from init 3 to init 5, and "power saving mode" when switching VTs.
I suspect that the original i810 driver has the same behaviour.
I then tried selecting the exact definition for my monitor (IBM 6737). Here, I'll get "input signal out of range" message at the first switch to VT7 after rebooting. I'am attaching xorg.conf and Xorg.O.log for this last test case.
Comment 11 michel munnix 2006-06-26 09:35:22 UTC
Created attachment 91541 [details]
/var/log/Xorg.0.log for driver i810xorg71 and IBM6737
Comment 12 michel munnix 2006-06-26 09:36:13 UTC
Created attachment 91542 [details]
/etc/X11/xorg.conf driver i810xorg71 and IBM6737
Comment 13 Stefan Dirsch 2006-06-26 09:43:09 UTC
Ok. Does adding (with using i810xorg71)

  Option "Noaccel"
  Option "SWcursor"

to the section "Device" of your config file help? Does the signal also get lost during logout of a user session (when a new xserver is started)?
Comment 14 michel munnix 2006-06-26 09:47:19 UTC
Created attachment 91545 [details]
/var/log/boot.msg

Just an other idea, in boot.log there is a message about a patch:
Patching video bios 855resolution version 0.4, by Alain Poirier

Chipset: 845G (id=0x25608086)
VBIOS type: 3
VBIOS Version: 2603

** Patch mode 3c to resolution 1024x768 complete
done
Comment 15 Stefan Dirsch 2006-06-26 09:54:12 UTC
Sure, this is for broken BIOSes, which do not support higher resolutions. If you think this might be a problem set VIDEOBIOS_PATCH to "no" in /etc/sysconfig/videobios.
Comment 16 michel munnix 2006-06-26 11:07:35 UTC
Options "Noaccel" and "SWcursor" did not fix the problem.

But setting VIDEOBIOS_PATCH to "no" was the solution.
Comment 17 Stefan Dirsch 2006-06-26 11:24:13 UTC
Wow! You've been the first with a problem with 855resolution. So do you still need the i810xorg71 driver with 855resolution disabled?
Comment 18 michel munnix 2006-06-26 15:09:18 UTC
No, if 855resolution is disabled, i810 is working fine.

I made some checks to get a better monitor definition from YaST, but everytime I changed something 855resolution was reenabled in /etc/sysconfig/videobios and I had to disable it, then reboot to check the result.

Now, I have a working config with i810 at 1024x768 / 107Hz
Comment 19 Stefan Dirsch 2006-06-26 15:24:23 UTC
Yes, that's a problem. SaX2 always reenables this. The easiest fix/workaround for you is to uninstall the 855resolution package. Setting to NORMAL since you've a fix/workaround for now.
Comment 20 Stefan Dirsch 2006-09-01 06:18:44 UTC
Marcus, here's another one for the 855resolution blacklist.
Comment 21 Marcus Schaefer 2006-09-01 09:01:24 UTC
yes indeed. The information from the command

   dmidecode

is needed here to build up the blacklist. This problem
will be tracked in bug: #201338

Thanks

*** This bug has been marked as a duplicate of bug 201338 ***