Andi Kleen wrote:

On Monday 07 August 2006 06:47, Rusty Russell wrote:
  

quoted

This patch does the dumbest possible replacement of paravirtualized
instructions: calls through a "paravirt_ops" structure.  Currently
these are function implementations of native hardware: hypervisors
will override the ops structure with their own variants.
    

You should call it HAL - that would make it clearer what it is.
  

I've always found the term "HAL" to be vague to the point of 
meaningless.  What would it mean in this case:  "hypervisor abstraction 
layer"?  It certainly doesn't attempt abstract all hardware.

I think I would prefer to patch always. Is there a particular
reason you can't do that?
  

Some calls just don't need patching; an indirect call is fast enough, 
and simple.  But I can't think of a good reason to not patch patchable 
calls, other than for debugging perhaps (easier to place one breakpoint 
than one per inline site).

    J