RE: [Xen-ia64-devel] machine ITM = nearest guest ITM vs. full guesttime virtualization

["Magenheimer, Dan (HP Labs Fort Collins)" <dan.magenheimer@xxxxxx>]

> > Yes, that's the difference.  I use ac_timer only for
> > scheduling (ITC==XenITM).  I don't even need to use it
> > there but it is convenient for interfacing to the Xen
> > scheduling code.
> Yes it can work, but we are deviating from X86 XEN more. Any 
> strategy in deviation from X86? I think this deviation is not 
> necessary.

The timer mechanisms are sufficiently different between X86
and ia64 (and other architectures) that it may make sense
for ac_timer.c to be moved to archdep code.

> > So, yes, when switching from domainA to domainB, if a significant
> > amount of time has transpired since domainB ran, it's highly
> > likely that the first instruction that will be executed by
> > domainB will be in the IVT at the interrupt vector to handle
> > a clock tick.
> The problem is how you know domain B transpired a significant
> amount of time without extra data structure? I mean HV needs 
> to track last guest ITC and ITM to know there is condition 
> for timer IRQ to fire. If you would like to track the last 
> guest ITC, the design will be much like same. Maybe you can 
> give more comments after you saw the code.

If you think of the "Xen ITC" as the official ITC, domain switch
from A to B requires adding the offset of domainA to return to
Xen ITC, then subtracting the offset of domainB to obtain domainB's
ITC.

I think this works and is very simple... but as I said, this hasn't
been tested because Linux doesn't do it.

_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel