[coreboot] patch: init gx cache earlier.

Marc Jones marc.jones at amd.com
Tue Feb 5 00:27:36 CET 2008



ron minnich wrote:
> Why can't we shadow?
> 

We could.


> If we never write to the ROM space, at all, why not put the changed
> cache settings in right BEFORE we jump to the payload? i.e. don't
> change it to "slow" mode until we no longer need it cached?


I think that this would be ok. Maybe a generic call pre_payload() before 
the payload is run. I think that this would be a good genreic call for 
mainboard customization, like hardware_stage1(). It does mean that there 
will be duplicate code for each Geode platform.

I was working on a patch but it grows the size of the bootblock. It was 
only a few bytes too large but I grew it from 16KB to 20KB. I know we 
don't expect the bootblock to change much but should is be mainboard 
specific?


I'm still hitting the PCI device problem so I have only tested that it 
builds and the ROM is being cached. I can't get to the cache disable 
portion.


Marc


-- 
Marc Jones
Senior Firmware Engineer
(970) 226-9684 Office
mailto:Marc.Jones at amd.com
http://www.amd.com/embeddedprocessors
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: romcache.patch
URL: <http://www.coreboot.org/pipermail/coreboot/attachments/20080204/c65fcb39/attachment.ksh>


More information about the coreboot mailing list