Re: Xen & VMI?

[Jeremy Fitzhardinge <jeremy@goop.org>]

Ingo Molnar wrote:
> Pretty much the only way to ensure a sane ABI is to do it like we do 
> it with the Linux syscall ABI:
>
> 	_to have only one_
>   

Right, but there's also only one implementation; if you were to try and
coordinate Linux and *BSD to have the same kernel ABI it would be a much
trickier prospect.  From OO or Evo's perspective, there are other
syscall ABIs on other systems.  But it doesn't matter - because there's
libc in usermode which papers over the differences in kernel interfaces,
both between Linux versions and between systems.

> We do not unify their pointlessly diverging ABIs to within the kernel 
> via say office_ops (while we could) because that's crappy on its face. 
> Hypervisors arent in any way different, they just _think_ they are 
> special because they are relatively new. But hey, i dont expect you to 
> concede this point ;)
>   


Well, if you want to draw the analogy like this, then pv_ops is libc,
and it doesn't matter what the kernel/hypervisor ABI is from rest of the
app/kernel's perspective, so long as libc/pv_ops holds up its end of the
agreement.

    J