On Tue, Nov 06, 2018 at 07:45:19PM +0100, Borislav Petkov wrote: > Ok, having swapped the whole thing back into my brain, forget what I > said earlier today. > > Didn't we talk about passing info with setup_data to the later kernel > stage? You even had a patch: > > https://lkml.kernel.org/r/20181022154204.kagmdb55jtoez4ca@gabell > > So what is that "idea" again about adding SRAT parsing code to > arch/x86/mm/kaslr.c?!?! > > The intent of passing info with setup_data is to *avoid* parsing SRAT > yet another time and duplicating that code one more time. > > IOW: > > * The first place that needs SRAT parsing, does the parsing - i.e., the > compressed stage. > > * Later stages get information passed to them with setup_data. No second > parsing. > > Ok? Yes, I think I can see what you are saying. However, I'm not sure how I use the setup_data in legacy BIOS environment. So, I said the another idea which adding the SRAT parsing code to arch/x86/mm/kaslr.c as well. Yes, as you said, that is not so good... I would appreciate if you could help to use setup_data or something to pass the information to later kernel stage in BIOS environment. Thanks, Masa
