Re: [coreboot] [PATCH] AMD F10h: set MMCONF bus count according toconfigured value

[Myles Watson]

On Tue, Oct 19, 2010 at 12:34 PM, Scott Duplichan  wrote:
> -----Original Message-----
> From: [e-mail] [mailto:[e-mail]] On Behalf Of Carl-Daniel Hailfinger
> Sent: Tuesday, October 19, 2010 01:25 PM
> To: Scott Duplichan
> Cc: 'Arne Georg Gleditsch'; 'Coreboot'
> Subject: Re: [coreboot] [PATCH] AMD F10h: set MMCONF bus count according toconfigured value
>
> ]Hi Scott,
> ]
> ]On 19.10.2010 19:47, Scott Duplichan wrote:
> ]> Last night, I thought I would just learn about gas macros and do it
> ]> that way. It was more difficult than I thought. I cannot even find
> ]> examples of gas macros that use arguments.
> ]
> ]Take a look at src/cpu/amd/car/cache_as_ram.inc, specifically
> ]extractmask and simplemask.
> ]
> ]Regards,
> ]Carl-Daniel
> ]
> ]--
> ]http://www.hailfinger.org/
>
> Thanks Carl-Daniel.
>
> Hello Myles,
>
> I am not so happy with the new one either. What I really wanted was a
> macro called highestSetBit. That way, the reader can guess what macro
> invocation highestSetBit (CONFIG_MMCONF_BUS_NUMBER) does without
> actually digging it out. But I could not find a way to make that work.
> Another idea was to find highestSetBit at runtime, if it could be done
> a short and readable code sequence. This would be possible with family
> 15h, where instructions such as lzcnt are supported. But family 10h
> does not have this instruction.
>
> Now that Carl-Daniel has showed how to do gas macros, how about I add a
> highestSetBit macro to amd/mtrr.h? Then everything else can be inlined?

I would say it isn't worth the effort, but it's up to you.  The
problem with macros in assembly is that it's unclear which registers
they clobber.  That was why I suggested a pre-processor macro, but
it's not that important.

Thanks,
Myles