Re: [PATCH v15 14/19] pvh: Use PV handlers for PIO

[George Dunlap <george.dunlap@eu.citrix.com>]

On 11/12/2013 02:33 PM, Jan Beulich wrote:
>>>> On 11.11.13 at 15:57, George Dunlap <george.dunlap@eu.citrix.com> wrote:
>> @@ -566,7 +581,10 @@ int hvm_domain_initialise(struct domain *d)
>>       d->arch.hvm_domain.io_handler->num_slot = 0;
>>
>>       if ( is_pvh_domain(d) )
>> +    {
>> +        register_portio_handler(d, 0, 0xffffffff, handle_pvh_io);
>
> Nice idea, but I'm afraid it'll need a little more than this long term
> (i.e. I'm fine for this to go is as is, with the second paragraph below
> address in some way): hvm_io_intercept() first tries
> dpci_ioport_intercept(), and without device model I think that's not
> correct. If indeed it isn't, a simple solution might be to just disallow
> the registration of translated I/O port ranges for PVH guests.
>
> The other thing here is the use of 0xffffffff: The parameter is a
> size one, not an end of range. And considering that we're talking
> about I/O ports, 0x10000 would seem to be the right value (or,
> if taking the potential wrapping into account, 0x10003).

Oh, right -- I was looking at a trace of the guest, but confusing the 
values written with the address.  Do you just want to change it to 
0x10000? (Or 0x10003?)

  -George