Bug 120424

Matthias, we should try the latest X.Org CVS radeon driver first. STABLE is 
already updated to current X.Org CVS. 

Shall I install an Open Suse Alpha 1 snapshot and update the Xorg package to
stable, or shall I just leave the machine as it is?

Alpha1 won't help. Updating to current X.Org packages from STABLE is not trivial. 
Matthias will take care of this ASAP.

Assigning to Matthias. IIRC  
 
  Option "MonitorLayout" "LVDS" 
 
needs to specified, because "CRT" seems to be the default on this machine. :-( 
Matthias knows the details. 

Right now there is no possibility to detect the default monitor. We cannot even
detect whether the system has an LVDS panel, if it is *not* forced by MonitorLayout.

Maybe the radeon Superpatch changes something WRT this problem (if it works at all).

We might set this option only for exactly this chipset (by using
SubVendor/Subdevice information).

I doubt very much this is a *chipset* issue. It's a problem of this particular
laptop.

But there is hope that this chipset (Vendor ^ Devic ^ Subvendor ^ Subdevice) is
only used in this laptop.

AFAIK we have contact to ACER/AMD for this laptop (Ferrari F-4000).
What do they need to provide or modify?

According to Bodo we won't be able to convience them to do any changes to their
BIOS or whatever, right?
Could we then at least do it as supposed by Stefan in comment #8?

> Could we then at least do it as supposed by Stefan in comment #8? 
At least we should add this information to this bugreport, before we return 
this notebook! 

Created attachment 53603 [details]
hwinfo --gfxcard of Ferrari F-4000 X700 graphics card

Thanks. BTW, when do we need to return the notebook again?

(In reply to comment #10)
> According to Bodo we won't be able to convience them to do any changes to their
> BIOS or whatever, right?

I doubt that this is the BIOS' fault, anyway. It is some combination of output
drivers connected to the radeon chip the current driver cannot handle.
I'd like to test the radeon superpatch today, and decide afterwards.

Alan wrote in Xorg bugzilla #3621:

> I've also got a laptop (Acer Ferrari 4000 with Radeon Mobility X700) which can't
> find it's display unless you force it to use LVDS, and then it crashes.   The
> superpatch doesn't seem to make any difference with it. I've put log files and 
> a simple fix for the crash in bug #4001.


So the superpatch won't help.
Stefan, the only possibility I see so far is to add

  Option "MonitorLayout" "LVDS,CRT"

for this Laptop only. This works for both configurations, single head and dual head.

Ok. I've added this information now to to our x11DB. Should work with 10.1 (Alpha2).

Tested Alpha2plus: Does not run out of the box.
Coping modified xorg.conf from mhopf that worked, also makes it work on 10.1 Alpha2plus.
Any special options I have to set?

Did you do a 'sax2 -r'? The information from the database is only used when sax is creating a fresh configuration.

I moved /etc/X11/xorg.conf to /etc/X11/xorg.conf.bak
Then I did a init 3, sax2 -r  --> machine hangs hard (no keystrokes - NUM Leds, nothing). Last output:
SaX: no X-Server is running
SaX: will start own server if needed

Why is this sax2 -r needed, I hoped it could work by just clicking through the installation dialogs, of course with a bad resolution, but at least it should come up with something?

Wow.
sax2 -r is needed because only in this case /etc/X11/xorg.conf is *not* read in and discarded.
Can you open the machine's sshd in the firewall?

No feedback since a long time --> WONTFIX.

-> three weeks holidays.
The machine is now up as hammer18.

The option:
 Option "MonitorLayout" "LVDS,CRT"
still is not set automatically in NLD10...

Please post the Device section of our xorg.conf.

s/our/your

Section "Device"
  BoardName    "Ferrari F-4000 X700"
  BusID        "PCI:1@0:0:0"
  Driver       "radeon"
  Identifier   "Device[0]"
  Option       "SWcursor"
  VendorName   "ATI"
  Option       "MonitorLayout" "LVDS"
EndSection

Some strange things happen in CDB (export). I've added 

  Option "MonitorLayout" "LVDS,CRT"

but 
  
  Option       "MonitorLayout" "LVDS"

is the result of th export. :-(

# ATI Technologies Inc: Ferrari F-4000 X700
1002 5653 1025 007e

I don't understand...

The SAX generated xorg.conf contained either value for MonitorLayout. The Option line in Comment #26 was manually added to make this beast work.

But then the export is broken as well or SaX2 ignores the options. :-(

Just checked, the export looks ok.

NAME=ATI§DEVICE= Ferrari F-4000 X700\
§VID=0x1002§DID=0x5653§SERVER=radeon\
§EXT=§OPT=§RAW=Option "MonitorLayout" "LVDS,CRT"\
§PROFILE=§SCRIPT3D=§PACKAGE3D=§FLAG=DEFAULT§SUBVENDOR=0x1025§SUBDEVICE=0x007e

cdb/Cards has the correct export option as well.
Investigating.

Are you sure this static option isn't overwritten by some kind
of profile mechanism ? It would be usefull if someone could call

  sax2 -r -a

and have a look into the automatically written file /etc/X11/xorg.conf
The output on the console as well as the log file /var/log/SaX.log
is needed as well

Thanks

Marcus, the MonitorLayout option from the database gets nuked as soon as you try to configure something in SaX2. Additionally, I have to check what actually happens to the option so its value is truncated (and the closing " is missing).

Also the configured resolution got switched back to 1024x768 when I read in an old configuration with higher resolution.

One additional problem:
If *only* Monitorlayout is set to "LVDS,CRT" and *no* secondary monitor configuration is issued, the driver defaults its external maximum resolution to 640x480, and as no Metamodes are given, it creates an internal one for 640x480,640x480, disabling all higher resolutions.
-> We are only allowed to set MonitorLayout with two output devices for radeon if we have a valid configuration.

One sidenote: all options in xorg.conf are case insensitive. Except for the contents of MonitorLayout of the radeon driver. Tried lvds,crt once, it complained about 'crt' (not about 'lvds'...) and - no signal. "LVDS,CRT" works like a charme.

Gee, this is getting crapier with every single test :-(((

Created attachment 63639 [details]
xorg.conf from sax2 -r -a

This is still NLD10 preview. If I should test newer sax packages, please tell me.

Created attachment 63640 [details]
SaX.log from sax2 -r -a

Am I able to login to the machine ?

Created attachment 63641 [details]
xorg.conf from sax2 -r and setting the correct resolution

This config works fine after adding
Option "MonitorLayout" "LVDS"

Created attachment 63642 [details]
SaX.log from sax2 -r and setting the correct resolution

fixed and tested. I hate to say but for this device a special handling
within the GUI is needed. I added that into a function named fixBrokenCards()

I testes NLD-10 x86_64 Beta1, and it's still going wrong. Looks like something with the database export went wrong:
...
  Identifier   "Device[0]"
  Option       "CRT2HSync" "31-48"
  OptionCRT"   ,Option  "MonitorLayout" "LVDS
  Option       "SaXDualMonitorVendor" "--> VESA"
  Option       "SaXDualOrientation" "LeftOf"
...

Full xorg.conf is attached...

Created attachment 64731 [details]
xorg.conf as created by YaST during installation

It is fixed ! This patch was checked in by the autobuild team after beta1

-------------------------------------------------------------------
Tue Jan 17 16:45:59 CET 2006 - ms@suse.de
...
- fixed RawDef option handling from Identity map RAW entries