>No, mostly.  Welcome to wormful land of "firmware" world.  Although most 
>computer users
>do not recognize, there is the whale of work/sweat accomplished prior to 
>load
>and exec OS kernel.  Firmware (BIOS in PC term) aware of PCI scheme has to
>enumulate PCI devices on system and allocate and assign address space in 
>complicated
>manner, and further, do many register initialization work for each 
>particular PCI devices.
>Upon OS kernel starts running it expects PCI world is in peace with full 
>order of
>"specification."  This means you can not tweak/skip PCI scheme bypassing 
>shortcommings
>of  less-engineered firmware.

Well I think it is possible in the other way.  Before the 69030 card is up 
the "dmesg" output is displayed in the serial console. At some time the 
69030 card gets detected and comes into the match and attach function call. 
I can access the PCI strucutres that provides the offset of the mem map of 
the 69030 controller. This way I will get access to the registers of the 
card. So in the match function of the driver I would call an init_card 
routine which initilaizes and brings up the 69030 card, then if required I 
would call the xxxcnattach of the driver. This way I belive it is possible 
though it is not in a proper way.

If you want to do the inital testing of the Video card this would be better. 
Later the same code could be inserted in PMON if it supports the PCI layer.

>Engineer PMON, preferably replace it with PMON2000 to start, and realize 
>how much
>BIOS programmers pump their sweat to make things work...

May be later some time I would do that for sure, but at present my priority 
is to support 69030 Card on my Target Board.

PS: The earlier question were put up by my colleague.

- Anand

_________________________________________________________________
Get Married!  http://www.bharatmatrimony.com/cgi-bin/bmclicks1.cgi?74 Search 
from 7 lakh Brides & Grooms.