Juergen Beisert wrote:
Jonathan,

On Tuesday 15 August 2006 22:25, Jonathan Sturges wrote:
  
So I'm back to working on getting LB on my Geode-based MaxTerm 230 thin
client again.  I've built a test image from the Eaglelion 5bcm target
that boots this machine.  However, IRQ mapping is not working.  I have
no doubt that is because the MaxTerm differs from the 5bcm.  :-)
    

What exactly does not work (error messages and so on)? I also started with the 
Eaglelion 5bcm for my AXUS-TC320 (with WinCE-Loader, no BIOS). On my first 
tries with LinuxBIOS Linux complained about a broken IRQ map, but I played 
around with the settings and now it works.

Juergen

  
I was aware of the issues with LB failing to copy the IRQ map, due to the GX1's cache not being enabled.  However, I applied the two northbridge.c patches for enabling the cache, so now that much works.  (By the way, can someone on the development team commit those cache patches?  As of revision 2360, they're not included.  I'll send links to the posts if needed.)  But what I get now is that there is no IRQ defined for my on-board USB and Ethernet adapters.  So 'insmod'-ing a driver for either results in this message:

PCI: No IRQ known for interrupt pin A of device 00:13.0. Please try using pci=biosirq.
usb-ohci.c: found OHCI device with no IRQ assigned. check BIOS settings!

...which basically means the entries for these devices in the IRQ map are either wrong, or missing.  What's needed now is a way to determine those interrupts, and put them in the map.  Did you have to go through a similar process on your board, and if so, can you detail what you did?

Thanks,
Jonathan
-- 
linuxbios mailing list
linuxbios@linuxbios.org
http://www.openbios.org/mailman/listinfo/linuxbios

Reply via email to