Arjan van de Ven wrote: > I didn't see much of this last sentence in the actual patch, which is > good, because I just don't see any reason to do that at all; if they > features are there anyway, why not preserve their detection, it's not > hurting and it means you need to be less different "just because"... > mach-xen/head.S is simpler than kernel/head.S, partly because it doesn't contain the CPU identification and feature tests (cpuid is assumed, though ironically, we don't actually use it). J