ron minnich wrote: > .. but I'd like to start getting via bits into v3 asap.
I would really like us to do a design review round before substantial amounts of code supporting non-K8 systems are added to v3. I realize everyone is eager to add as much as possible as quickly as possible to v3, but I believe that this kind of frenzy and not much forethought is the major reason for the mess that is v2. I am not saying that we should stop working on v3 for two years. I'm asking that we stick to the Hamburg plan, i.e. to step back and evaluate (maybe tidy too) the code that is in v3 when simple (geodelx) and complex (k8) are both working. I would be surprised if it took more than a week or two even if we are working distributed. A day or two if we were in the same room. But I do believe it is important. There are several things to address. (ACPI/resume, dts, device model, interrupt routing to name just a few...) This would affect all of stage2 - I don't think stage0 and stage1 can change much, and even if they do they are quite small codebases so not much work is needed to update even five or six of them. Those who are not so interested in following/doing the review could instead work on porting more CAR and RAM init code to v3. Comments and flames, as always, most welcome. //Peter -- coreboot mailing list: [email protected] http://www.coreboot.org/mailman/listinfo/coreboot
