[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Problems running XFree86 4.2.1 (unstable) on a Motorola Powerstack



Hi,

I recently obtained a Motorola Powerstack (PPC 604e, 128 MB Ram, 
Network/Sound/SCSI (wide) onboard) that came equipped with a Cirrus Logic GLD 
5446 PCI graphics card. The BIOS identifies itself as Motorola Open Firmware 
Version 1.2 RM11. I did not find any documentation or newer firmware on the 
web besides some "Motorola Firmware" that is some ARC firmware intended to 
boot Windows NT for PowerPC or IBM AIX and refuses to boot linux :(

Debian/PPC (unstable) installed without much problems (the installer kernel 
did not recognize the scsi adapter, but that was easily changed by preparing 
a newer kernel) and can use the kernel framebuffer, but I have not managed to 
run XFree86. I use the latest .debs from unstable (4.2.1-5)

I tried the installed Cirrus Logic card and encountered 2 problems: 
First, the debian packages come without the cirrus.o module neccessary to 
drive the card, even though the changelog has entries for this card that 
refer to bug fixes for the ppc architecture. 
Second, trying to use the vga.o module leads to a complete crash of the 
system. The console stays blank, a network login is no longer possible and 
users who are already logged in only get "illegal instruction" error 
messages, whatever they try to do. No logs are written, and so I have no idea 
what happens. I tried with the "UseFBDev" and "NoAccel" on and off.

Next I got a Matrox Millenium, and tried the mga.o and vga.o module -> Same 
crash. The kernel framebuffer worked flawlessly (tested with fbi and mplayer)

Finally I tried an ATI Rage 3D II+DVD (some Mach64 variant), but the Motorola 
Open Firmware would not recognize the card (blank screen instead of the 
BIOS). The Powerstack booted nonetheless, and the linux kernel framebuffer 
found and used the card. XFree86 with the ati.o driver did not crash the box, 
but instead complained that the "card is not initialized" and refused to 
start. But it still hosed the framebuffer, which was unusable after starting 
X. 

I searched google and this mailing list for similar problems, but this kind of 
machine seems to be rare, and I found no further hints. A friend of mine 
bought the same machine and claims that suse linux runs out-of-the-box which 
makes me think that a hardware problem in this specific model could be less 
likely.

I found -dbg packages for the XFree86 servers. Is there some way I could use 
those to pinpoint the problem? As I wrote earlier, attempts to create logs 
from the crashing X server always failed with 0-byte-sized log files or 
ext2fs errors in the log file. Makes me think the kernel might have panicked.

Any ideas? 

Magnus



Reply to: