On Fri, Apr 25, 2008 at 04:51:12PM -0600, Jordan Crouse wrote:
My point is that buildrom is integrated fairly tightly with coreboot already and it should not need to duplicate this mapping between e.g. ALIX1C and geodelx.
For v3 read the dts. For v2, hm, maybe look for "cpu/amd/model_lx" in mainboard/v/b/Config.lb.
No - this is actually stuff external to coreboot itself.
I know.
In this case its the code that appends VSA to the ROM.
Yes.
it is named geodelx.mk because thats a handy name, not because there is any sort of mapping.
All boards in coreboot using cpu/amd/model_lx need VSA, so rather than teaching buildrom one coreboot target at a time, I am suggesting to once and for all teach buildrom that all coreboot targets using cpu/amd/model_lx need VSA, by reusing a mapping (between $target_name and lx) that can be found in coreboot instead of duplicating the mapping in buildrom.
This can be extrapolated to work for other things as well, but right now I don't think there are any other uses.
Put another way: buildrom should not need it's own list of boards that use model_lx when coreboot already has one. (Though the index is backwards.)
//Peter