[coreboot] [PATCH] Kontron KT690 mainboard

[Carl-Daniel Hailfinger]

On 05.10.2009 19:59, Stefan Reinauer wrote:
> Carl-Daniel Hailfinger wrote:
>   
>> On 05.10.2009 16:07, Stefan Reinauer wrote:
>>   
>>     
>>> Hi,
>>>
>>> we're happy to release coreboot® support for another Kontron embedded
>>> mainboard: The Kontron KT690/mITX
>>>
>>> http://us.kontron.com/products/boards+and+mezzanines/embedded+motherboards/miniitx+motherboards/kt690mitx.html
>>>
>>> This patch adds (initial) support for the Kontron KT690 mainboard. 
>>> It's an embedded AMD 690/SB600 mainboard with a Mobile Sempron CPU.
>>>
>>> Issues with this port:
>>> - hangs early during "Starting Windows" with Windows 7, after loading all the
>>>   drivers
>>>   
>>>     
>>>       
>> Probably because of the SATA ACPI code which only works if MMCONF is
>> enabled, but the 690 code explicitly disables MMCONF during late init.
>>   
>>     
>
> Yes, it dies in disk.sys
>   

That is what I expected. I have a workaround for this specific SATA
issue (the workaround only works if the OS doesn't perform resource
reallocation), but other PCI config space accesses in the ASL will fail
as well, causing all sorts of weird issues (hang in USB etc).
Look for enable_pcie_bar3() and disable_pcie_bar3() to find out what's
being done. The real fix would be to either mark the MMCONFIG area as
fixed-location and remove calls to disable_pcie_bar3() or make sure it
gets reenabled after resource reallocation. Linux (and probably also
Windows) expects that the memory for the MMCONFIG BAR is reserved and
specially designated in ACPI (the latter is sort of done in the board
specific ACPI code, but I recommend to try the code Rudolf wrote for
MMCONFIG instead).


>>> - sound is untested and probably not working  
>>> - powernow seems to be not working  
>>>
>>> Signed-off-by: Stefan Reinauer <stepan at coresystems.de>
>>>   
>>>     
>>>       
>> Nice! Do you also have a diff against dbm690t for easier comparison/review?
>>
>>   
>>     
> Unfortunately not.
>   

I ran the diff. Except for interrupt routing and superio config, this is
pretty much a copy of amd/dbm690t. That also means you will need some
ACPI adjustments, but for now it's great to have another port in the tree.

Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006 at gmx.net>

Regards,
Carl-Daniel

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