[coreboot] Moved 'pci_set_subsystem' to a weak function: Support for Siemens Mainboard

[Georgi, Patrick]

Am Dienstag, den 25.01.2011, 10:44 +0000 schrieb Joseph Kellermann:
> Also I've found an issue:
> The default 'pci_set_subsystem' sets the subvendor/device id register#
> 0x2c. However, for a few devices in the RS690 the register is 0x4c
> respectively 0x50. 
> 
> Moved 'pci_set_subsystem' to a weak function in 'pci_device.c'. So
> boards can add an override for the default function in
> mainboard/../override_df.c (or mainboard.c) and set it up too.
Given that these implement chipset specific behaviour (either the device
requires writing to 0x4c instead of 0x2c or it doesn't), and that
pci_dev_set_subsystem is always called indirectly (via function pointers
in the respective chipset drivers), couldn't you just modify the chipset
drivers to point to a different set_subsystem implementation where
necessary?

It seems like the GFX function of RS690 is affected (just skimmed
through the AMD docs, so others might also require such a change). In
this case, it should work to change src/southbridge/amd/rs690/gfx.c to
point lops.set_subsystem to your own implementation which does whatever
their GFX device needs (write to 0x4c in extended config space in that
case, if I understand the docs), no?

> Hoping to receive a 'ack'.
> Signed-off-by: Josef Kellermann joseph.kellermann at heitec.de 

If this change is actually required, I'll ack and commit it, but I'd
like to understand your use case better first. We already have a layer
of indirection in place, introducing another one will only create
confusion down the road.


Regards,
Patrick
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://www.coreboot.org/pipermail/coreboot/attachments/20110125/4ba1c4cd/attachment.sig>