Hi,

Yeah, I'm pretty against CPU_FTR_POWER7.  Every loon is going to
attach anything POWER7 to it.  

I'm keen to see it setup in  __setup_cpu_power7.  Either a function
pointer or use the patch_instruction infrastructure to avoid indirect
function calls on small copies.  

Instruction patching in __setup_cpu_power7 could work. We might want to
have a nop at the start of the base functions and a label at the start
of the next instruction so we can easily override the base function and
jump back to it if things are too hard (like I do in the
copy_tofrom_user patch).

Anton