Bug 115307

Summary: X installation oddness with intel i810 on Cappucino mini-pc
Product: [openSUSE] SUSE LINUX 10.0 Reporter: Vince Negri <vnegri>
Component: SaX2Assignee: Marcus Schaefer <ms>
Status: RESOLVED WONTFIX QA Contact: E-mail List <qa-bugs>
Severity: Minor    
Priority: P5 - None    
Version: Beta 4   
Target Milestone: ---   
Hardware: 32bit   
OS: SUSE Other   
Whiteboard:
Found By: Other Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---

Description Vince Negri 2005-09-05 13:25:07 UTC
Tried the installation of beta 4 on a cappucion mini-pc I had lying around. This
is based on the VIA Ezra C3 cpu, with Intel i810 video and 128MB ram. This pc
has had SUSE Linux 9.0 and 9.1 running on it previously.

First thing noticed - no graphical bootscreen (no framebuffer) but YaST ran in
800x600 mode - presumably falling back to VESA.

Installation proceeded smoothly. Final hardware detection stage correctly
detected  graphics chipset and attached LCD monitor - set up for 16-bit 1024x768.

However, on finishing setup, kdm failed to start. Logged in as root and tried to
run bare X - didn't start either. (Idiot that I am, forgot to save log file.
Next chance I have I'll re-run the install from scratch)

Shutdown and rebooted the pc. This time X starts! 

Is the probing of sax2 possibly putting the i810 into a funny state?
Comment 1 Marcus Schaefer 2005-09-05 15:31:49 UTC
The probing if only one card is installed doesn't soft-boot the 
card so there is no hardware access from within sax which reinforces 
the later X-server startup. 
 
I don't know why your X-Server didn't start without a log, as you 
already said :-) So if the problem occurs again please save the log 
Comment 2 Vince Negri 2005-09-06 08:13:21 UTC
I am going to re-install from scratch on this pc tonight anyway for a different
reason (testing installation with ACPI enabled, which didn't work with earlier
SuSE kernels) so will capture log if problem re-occurs.
Comment 3 Vince Negri 2005-09-07 12:53:59 UTC
Curiouser and curiouser..

Performed a "Minimum Graphical" VESA-gui installation with ACPI enabled. At the
graphics card probe stage, screen went blank and system locked up.

Tried again, this time doing a text-mode  "Minimum Graphical" install, ACPI
disabled. Graphics card probe appeared to work, card & monitor detected ok. At
end of install, start of X caused system lock-up. Hit reset button to reboot
machine -  system comes up fine, X starts up with correct resolution.

Nothing in the logs (probably because system lock-up stops anything useful
making it to disk...)

Will try again today - I need to establish if the above is repeatable, since a
hardware problem can't be discounted.
Comment 4 Vince Negri 2005-09-08 15:47:46 UTC
Unable to get a repeatable test case - last night X did start up after
installation. It seems the problem is either not really an X/sax problem (e.g. a
general chipset stability issue with the 2.6.13.x kernel) or the hardware has
developed a fault.

Will re-test with the next release candidate. Can we close this bug and I will
re-raise if a genuine stability issue seems to be present.
Comment 5 Vince Negri 2005-10-12 07:44:15 UTC
Tested again with 10.0 GM release.

Installation was default KDE with acpi disabled. Installation proceeded ok, and
on finishing setup kdm did start. However, system froze halfway through loading kde.

Taking a tip from a discussion I'd seen last week on EPIA motherboards (which
also use C3 processors), I disabled powersaved (via runlevel editor.) So far
system is now stable - I am running soak tests.

If it turns out that this is indeed a problem, maybe we need a "blacklist" for
powersaved (probably the longhaul module in this case) along the lines of the
acpi blacklist?

Comment 6 Stephan Kulow 2008-06-25 09:33:06 UTC
mass reopening all SuSE Linux bugs that are set to REMIND+LATER to change the resolution to WONTFIX (adapting to new policy)
Comment 7 Stephan Kulow 2008-06-25 09:34:37 UTC
mass reopening all SuSE Linux bugs that are set to REMIND+LATER to change the resolution to WONTFIX (adapting to new policy)
Comment 8 Marcus Schaefer 2008-06-25 09:35:55 UTC
there is the new intel driver

===> wontfix
Comment 9 Stephan Kulow 2008-06-25 09:40:58 UTC
mass reopening all SuSE Linux bugs that are set to REMIND+LATER to change the resolution to WONTFIX (adapting to new policy)
Comment 10 Marcus Schaefer 2008-06-25 09:42:00 UTC
there is the new intel driver

===> wontfix