Help needed: CardBus iomem allocation problem with 6 CardBus slots

[Joerg Pommnitz]

--- Dominik Brodowski <linux at dominikbrodowski.net> schrieb:
> Hi,
> 
> This is basically a PCI resource allocation problem. You could try out
> the
> following
> 
> a) use latest -mm kernel (it contains a few fixes for resource handling
> in
>    yenta),
> b) pass "override_bios=1" to the yenta module; then it'll try to
> allocate
>    the CardBus resources on its own. Then you can try to fine-tune it in
>    drivers/pcmcia/yenta_socket.c, search for BRIDGE_MEM_{MIN,MAX} and
>    friends, and/or
> c) try making the parent PCI-PCI bridge "transparent"; then the resource
>    areas can also be taken from outside the I/O or Memory "behind
> bridge".
>    [note that this will only work properly with recent -mm kernels].
> 
> 	Dominik
> 

Thanks for answering my question. To overcome the problem I already
changed the yenta socket driver and added a new module option
bridge_mem_max. With this option I can do "modprobe yenta_socket
override_bios=1 bridge_mem_max=256". This parcels the 3MiB io memory
window into 12 256KiB areas, which is exactly what I want. With this
modification in place everything works fine on SuSE 9.2. Unfortunately my
customer is commited to RedHat EL 4. With this kernel only card events in
the first CardBus adapter work. I will have to undo all the RedHat
specific kernel patches to figure out what breaks CardBus.
BTW: Does this ring a bell with anybody on the list? Remember, the exact
same hardware (except for a different IDE drive) works fine under SuSE.

Regards
  Joerg


	

	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de