[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [v3][PATCH 04/16] xen/passthrough: extend hypercall to support rdm reservation policy



>>> On 12.06.15 at 11:20, <tiejun.chen@xxxxxxxxx> wrote:
> On 2015/6/12 16:45, Jan Beulich wrote:
>>>>> On 12.06.15 at 08:31, <tiejun.chen@xxxxxxxxx> wrote:
>>> On 2015/6/11 17:28, Tian, Kevin wrote:
>>>>> From: Chen, Tiejun
>>>>> Sent: Thursday, June 11, 2015 9:15 AM
>>>>> @@ -1940,7 +1942,8 @@ static int intel_iommu_remove_device(u8 devfn, 
>>>>> struct 
> pci_dev
>>>>> *pdev)
>>>>>                 PCI_DEVFN2(bdf) != devfn )
>>>>>                continue;
>>>>>
>>>>> -        rmrr_identity_mapping(pdev->domain, 0, rmrr);
>>>>> +        rmrr_identity_mapping(pdev->domain, 0, rmrr,
>>>>> +                              XEN_DOMCTL_DEV_RDM_RELAXED);
>>>>
>>>> ditto
>>>
>>> It doesn't matter when we're trying to remove a device since we don't
>>> care this flag.
>>
>> In such a case it helps to add a brief comment saying that the precise
>> value passed is irrelevant. Or maybe this could be expressed by
> 
> Okay.
> 
>> folding this and the "map" parameters of the function (in which case it
>> might become self-documenting)?
>>
> 
> Sorry, I don't know exactly how to implement this idea. Have we any 
> similar example on Xen side?

No idea what you're after. What we have with your change are
tuples like
(map, relaxed)
(map, strict)
(unmap, <ignored>)
Clearly these can be represented with three distinct numbers. I.e.
along with using XEN_DOMCTL_DEV_RDM_* without the "map"
boolean, (unmap, <ignored>) could e.g. be expressed by passing
-1.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.