Re: [PATCH v1 13/14] xen/arm: Fixed error when PCI device is assigned to guest

[Julien Grall <julien@xen.org>]

Hi,

On 19/08/2021 13:12, Jan Beulich wrote:
> On 19.08.2021 14:02, Rahul Singh wrote:
>> --- a/xen/arch/arm/domctl.c
>> +++ b/xen/arch/arm/domctl.c
>> @@ -173,6 +173,8 @@ long arch_do_domctl(struct xen_domctl *domctl, struct domain *d,
>>   
>>           return rc;
>>       }
>> +    case XEN_DOMCTL_ioport_permission:
>> +        return 0;
> 
> I don't think returning success for something that doesn't make
> much sense in the first place (there aren't truly "I/O ports" on
> Arm afaik) is a good idea.

They are memory mapped. IIRC, there is a region that is reserved to fake 
them (for PCI cards that still using them).

> Instead I think the tool stack should
> avoid making arch-specific calls in an arch-independent way.

+1. The current approach means that it will be impossible to implement 
those sub operations without breaking Xen if we ever needs it.

Cheers,

-- 
Julien Grall