[coreboot] AMD 690 board with dual 88e8056

[Carl-Daniel Hailfinger]

Hi,

On 16.03.2009 17:48, Daniel Toussaint wrote:
> I am very very sorry to have provided the wrong info at first :  the PCIe x8
> DID work anyway - I had messed up my Config.lb to a stage where it was to
> far of from the original. After a recompile and reflash the PCIe x8 (raid)
> card came up just fine....
>   

No problem. Great to hear PCIe works.

> So yes, I am getting a x1 card tomorrow, with the same marvell chipset , and
> will test it. If it works properly, I am just going to have to wait for the
> schematic (will get it only a few weeks later unfortunately.....) and
> determine the gpio that is keepint the marvell's from showing up.

The lspci output I requested earlier would still be very interesting. It
is possible that subordinate buses (containing the Marvell ethernet)
don't show up as expected.


> It is
> definately not GPM3 as in the reference schematic - I have tried this
> already.
> The board is an ODM which my company outsourced to a board manufacturer -
> will send all the info and pics ASAP.

Cool! Will these boards be available for purchase? Most 690G/SB600
boards are out of production and porting coreboot to them is somewhat
pointless. I repeatedly get requests for coreboot-supported mainboards
with AMD chipsets.


> It has a x16 slot and a mini pci slot.
> The two 88e8056's are soldered on board. The rest of it is pretty much equal
> to the dbm690T. It also has on on board touch panel controller.
> So far I got the VGA , USB , SATA, IDE, audio Codec , all working fine.
> I guess what is still left for me to do is solve the problem with the lan
> chips and write proper ACPI tables - so that I can upload it your source
> tree ?
>   

It should be possible to reuse most parts of the DBM690T ACPI tables.
Since a few revisions, some parts of ACPI are autogenerated, so
hardcoding P-States etc. should not be needed anymore.

There's one catch, though: My Asus M2A-VM (690G/SB600) won't work with
more than 4 GB RAM and coreboot. Proprietary BIOS handles even 6 GB just
fine. The problem seems to be ACPI related. I have a patch to allow
64-bit addresses in the DSDT and that patch fixes parts of the problem.
Even with that bugfix, the Linux kernel still won't get past SATA init.
With 2 GB RAM, everything is OK.
I assume that problem applies to your board as well. It would be great
if you could verify that, though.


Regards,
Carl-Daniel

-- 
http://www.hailfinger.org/