[coreboot] first output: dbe62 and v3

[Marc Jones]

Carl-Daniel Hailfinger wrote:
> On 05.03.2008 18:13, Marc Jones wrote:
>> ron minnich wrote:
>>   
>>> One thing I plan to change pretty soon: power button control can move
>>> from stage1 to stage2, and power button control will be managed in dts
>>> (another dts setting we would NOT want visible in Kconfig; you could
>>> render a board totally unbootable and unrepairable with the wrong
>>> setting. Kconfig should be for settings that are safe to change).
>>>     
>> This would be great. It should be easy to add a dts item and call 
>> cs5536_setup_power_button or not. Moving it to stage 2 should be fine 
>> BUT until the power button is setup it won't work. Some people might not 
>> like that if you hang or delay somewhere prior to it being setup.
>>   
> 
> Can we set up the power button in a way that makes it work always, then 
> change the setup in stage2?
> 
> 

That is the problem. For systems that don't have a powerbutton, enabling 
the button logic will cause a reset in 4 seconds which would be fine if 
you could gaurnettee that it would be reprogrammed by then but that is 
just the oposite problem described above. What if there is a delay prior 
to rprograming it?



>>>         msr = rdmsr(GLCP_SYS_RSTPLL);
>>>         if (msr.lo & (0x3f << 26)) {
>>>                 /* PLL is already set and we are reboot from PLL reset. */
>>>                 return;
>>>         }
>>>
>>>         cs5536_setup_idsel();
>>>     
>> Need this to do stage2. Without it the 5536 won't decode PCI.
>>   
> 
> Did you mean stage1?
> 

You need to do it in stage1 so you can do stage2 (stage 2 needs to find 
the device before it can program it).

Marc


-- 
Marc Jones
Senior Firmware Engineer
(970) 226-9684 Office
mailto:Marc.Jones at amd.com
http://www.amd.com/embeddedprocessors