Bug 159731 - installation on Compaq PL6400R does not complete
Summary: installation on Compaq PL6400R does not complete
Status: RESOLVED FIXED
Alias: None
Product: SUSE Linux 10.1
Classification: openSUSE
Component: Installation (show other bugs)
Version: Beta 8
Hardware: i686 Other
: P5 - None : Blocker (vote)
Target Milestone: ---
Assignee: Stefan Dirsch
QA Contact: Klaus Kämpf
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-03-21 12:23 UTC by Per Jessen
Modified: 2006-04-24 10:17 UTC (History)
0 users

See Also:
Found By: Other
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments
Output of "hwinfo" (198.86 KB, text/plain)
2006-03-21 17:56 UTC, Per Jessen
Details
List of loaded modules (1.38 KB, text/plain)
2006-03-21 17:56 UTC, Per Jessen
Details
lspci output (1.84 KB, text/plain)
2006-03-21 17:57 UTC, Per Jessen
Details
Output of "lspci -vv" (11.37 KB, text/plain)
2006-03-21 17:57 UTC, Per Jessen
Details
/var/log/YaST2 - tarred and gzipped (3.88 MB, application/octet-stream)
2006-03-21 18:00 UTC, Per Jessen
Details
/var/log/X* /var/log/SaX.log tarred and gzipped (6.69 KB, application/x-gzip)
2006-04-24 08:49 UTC, Per Jessen
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Per Jessen 2006-03-21 12:23:16 UTC
Compaq PL6400R - 4 x 500MHz Xeons with 4Gb RAM.  Single SCSI install-disk, plus fiber-attached disk array.  The installation appears to work just fine, right up until the last step - "Hardware Configuration", at which point the screen blanks, and I only see a solid cursor in the upper left hand corner.  However, the system does not appear to be frozen - the Num-Lock key responds, but other e.g. Ctrl-Alt-F1, Ctrl-Alt-Backspace, Crtl-Alt-Delete are ignored.
Comment 1 Michael Gross 2006-03-21 13:45:34 UTC
Can you change to a console and check the messages? Start a rescue system and look what the run of `hwinfo' produces and at which point it hangs. Try salvaging the yast logfiles of the installation. Did this machine work fine with older releases? Be more verbosive about the used hardware, specially the loaded kernel modules in the installation system, `lspci'. Change to tty2 and execute these commands - you can surely do that before the actual hang.
Comment 2 Michael Gross 2006-03-21 13:50:08 UTC
I forgot: try booting with ACPI=off, pci=routeirq, pci=noacpi and/or the `Safe setting' and see if it still results in a hang.
Comment 3 Per Jessen 2006-03-21 17:53:47 UTC
OK, the system is actually able to boot and complete runlevel 5, so collecting the data isn't too difficult :-)
running "hwinfo" works fine - see output attached.  However, once I've got KDE running, I log in, then start YaST2 and select "Keyboard Layout" - after this, the machine enters the same state as described earlier (single solid cursor in upper left corner), but it responds to keyboard input and a Ctrl-Alt-Delete does a normal shutdown (although I can't see it) and reboot. 
This is the first try with SUSE 10.x - this system previously ran 9.3 with no problems.
 
Comment 4 Per Jessen 2006-03-21 17:56:08 UTC
Created attachment 74263 [details]
Output of "hwinfo"
Comment 5 Per Jessen 2006-03-21 17:56:46 UTC
Created attachment 74264 [details]
List of loaded modules
Comment 6 Per Jessen 2006-03-21 17:57:14 UTC
Created attachment 74265 [details]
lspci output
Comment 7 Per Jessen 2006-03-21 17:57:42 UTC
Created attachment 74266 [details]
Output of "lspci -vv"
Comment 8 Per Jessen 2006-03-21 18:00:33 UTC
Created attachment 74270 [details]
/var/log/YaST2 - tarred and gzipped
Comment 9 Michael Gross 2006-03-22 12:39:25 UTC
This sounds like a hardware problem. Unfortunately you didn't answear to comment #2. Is this triggered only the the yast kbd-module? It's not likely because at the screen you mentioned, no kbd settings are applied. Does the system run stable that way? What gfx-card are you using?
Comment 10 Per Jessen 2006-03-22 13:28:21 UTC
1) Graphicscard - see the lspci listing.  It's an onboard ATI 3D Rage IIC.
2) I can't say if it's only triggered by the kbd module - which other modules do you suspect? 
3) Booting with ACPI=off, pci=routeirq, pci=noacpi - I'll get back to you.
Comment 11 Per Jessen 2006-03-22 19:06:29 UTC
Booting with acpi=off,pci=routeirq,pci=noacpi makes the system freeze up after a little while and it does complete the startup. 
Booting with the Failsafe option completes, but it cannot start kdm - on tty10 I see:  "kdm :0 IO error in XOpenDisplay" followed by 2-3 other messages saying kdm couldn't be started.
Comment 12 Per Jessen 2006-03-22 19:07:36 UTC
First sentence  should of course have read: "... a little while and it does NOT complete the startup." 
Comment 13 Michael Gross 2006-03-23 08:57:29 UTC
Is this a shared-memory card you are using? If so, apply the `mem=<...>M' append line to your kernel, whereas <...> is the amount of memory minus the amount of memory used by the gfxcard.
Comment 14 Per Jessen 2006-03-23 10:31:44 UTC
It's an onboard/builtin graphics chip - I don't know if it uses shared memory - I doubt it.  The box has 4Gb memory - I'll try booting with mem=3072M and see what happens.
Comment 15 Per Jessen 2006-03-23 11:24:54 UTC
OK, tried booting with mem=3072M - then Yast2->Hardware->Keyboard Layout - still ended up with a blank screen with a solid cursor in the upper left hand corner. 
Comment 16 Michael Gross 2006-03-23 13:19:43 UTC
Maby this is a SMP problem. Try booting with `numcpus=1' and check whether the system runs stable afterwards. Furthermore, if you encounter a crash, fetch the syslog messages before the hang. You can also check the syslog for oopses manually.
Comment 17 Per Jessen 2006-03-23 18:42:44 UTC
I tried booting with numcpus=1 - didn't seem to make any difference; in fact, the system still seems to run on four CPUs.  top shows 4, /proc/cpuinfo shows 4. 
The system overall appears to run fine - I ran a 24+ hour stress-test/burn-in of a disk-array with no problems.  
Only when I start YaST->Hardware->Keyboard Layout do I see this problem - isn't that worth looking into?  
Comment 18 Michael Gross 2006-03-24 11:53:41 UTC
OK then, Jiri?
Comment 19 Stanislav Visnovsky 2006-03-24 15:04:03 UTC

*** This bug has been marked as a duplicate of 155018 ***
Comment 20 Per Jessen 2006-04-20 18:51:14 UTC
Reopening - bug 155018 has been closed, but it is not at all clear if and when a fix was created or applied.
Comment 21 Stanislav Visnovsky 2006-04-21 05:49:00 UTC
You are probably right, this does not seem to be a duplicate - bug 155010 did not exhibit a solid cursor.

Marcus, the YaST log ends up in initializing X11 library.
Comment 22 Marcus Schaefer 2006-04-24 07:47:47 UTC
This system contains an "ATI Rage IIC" card which is configured using
the ati driver. If this is a X11 problem you should be able to reproduce
it with the following workflow:

   - init 3
   - login as root
   - rm -f /var/log/X*   

   - start an X-Server with the fbdev based config from the installation system
  
     X -xf86config /etc/X11/xorg.conf.install &
     Ctrl-Alt-F1
     export DISPLAY=:0
     xterm
     Ctrl-Alt-F7

   - type the following command into the xterm

     sax2 -r

If the problem appears send the log files
  
   /var/log/Xorg*.log
   /var/log/SaX.log
  
if not it's not an X11 bug :-)

Thanks
Comment 23 Per Jessen 2006-04-24 08:46:33 UTC
After "sax2 -r", the screen blanks and I have a solid white cursor in the upper left hand corner.
Trying to leave with Ctrl-Alt-F1 doesn't work, killing the Xserver with Ctrl-Alt-Backspace does'nt work. However, Ctrl-Alt-Delete makes the system reboot.
Comment 24 Per Jessen 2006-04-24 08:49:22 UTC
Created attachment 79626 [details]
/var/log/X* /var/log/SaX.log tarred and gzipped
Comment 25 Marcus Schaefer 2006-04-24 09:12:34 UTC
ok, so this seems to be an ati driver bug.
Stefan logs has been attached
Comment 26 Stefan Dirsch 2006-04-24 10:08:57 UTC
  Model: "ATI 3D Rage IIC 215IIC [Mach64 GT IIC]"
  Vendor: pci 0x1002 "ATI Technologies Inc"
  Device: pci 0x4756 "3D Rage IIC 215IIC [Mach64 GT IIC]"

Looks like we need "fbdev" for this GPU. :-(
Comment 27 Stefan Dirsch 2006-04-24 10:17:51 UTC
fixed for RC3.