On Thu, 2007-03-08 at 09:01 +0100, Ingo Molnar wrote:

* Jeremy Fitzhardinge [off-list ref] wrote:

quoted

quoted

Your implementation is almost the perfect prototype, if you move the 
128 bit hackery into the hypervisor and hide it away from the kernel 
:)

The point is to use the tsc to avoid making any hypercalls, so dealing 
with the tsc->ns conversion has to happen on the guest side somehow.

you are obsessed with avoiding a hypercall, but why? Granted it's slow 
especially on things like SVN/VMX, but it's not fundamentally slow. We 
definitely do not want to design our whole APIs and abstractions around 
the temporary notion that 'hypercalls are slow'. I'd expect hypercalls 
to be put into silicon just as much as SYSENTER was put into silicon. 

Indeed, I expect them to fall somewhere between system calls and context
switches.  Perhaps not slow, but definitely worth minimising.

Anyway, in terms of guest time code, a /big/ amount of design junk can 
be avoided by not trying to do sillynesses like 'virtual time'. The TSC 
is awfully unreliable.

You mean stolen time?

I find this whole discussion really irritating, to be honest.  I just
want Thomas to implement the timer code for lguest, because that code
scares me...

I look forward to your patch 8)
Rusty.