[LinuxBIOS] Proposed change to amdk8 memory controller code

San Mehat san at google.com
Fri Oct 28 18:19:14 CEST 2005


No, but it adds yet more duplicated code. Since in most cases all memory
init code can figure out what the population map looks like on a channel
bases, pass that info to the board. The board module can swizzle the
population map by channel into population map by physical dimm slot (since
that mapping may be different on a per board level).

-San


On 10/28/05, Eric W. Biederman <ebiederman at lnxi.com> wrote:
>
> San Mehat <san at google.com> writes:
>
> > Reworking the attribute is actually causing a huge pain in the ass with
> > relocations it looks like. I'm going to see if i can have the amdk8 MCT
> code
> > return the dimm_mask information back to the board via
> sdram_initialize().
> >
> > I could either just have sdram_initialize() return a merged version of
> > dimm_mask which would limit the number of dimms supported, or I could
> pass in
> > an additional non-static 'population' structure which i think may be
> better.
> >
> > Any thoughts?.. I'm going to modify my stuff to pass in a new structure
> in the
> > meantime..
>
> Is there any reason not to simply query for the data back at the mainboard
> level?  It isn't like the discovery is hard.
>
> Eric
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.coreboot.org/pipermail/coreboot/attachments/20051028/a20a8283/attachment.html>


More information about the coreboot mailing list