Re: [Xen-devel] [PATCH v2 REPOST 03/12] x86/mm: add HYPERVISOR_memory_op to acquire guest resources

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 29.08.17 at 11:31, <Paul.Durrant@xxxxxxxxxx> wrote:
>>  -----Original Message-----
>> From: Jan Beulich [mailto:JBeulich@xxxxxxxx]
>> Sent: 29 August 2017 10:28
>> To: Paul Durrant <Paul.Durrant@xxxxxxxxxx>
>> Cc: Andrew Cooper <Andrew.Cooper3@xxxxxxxxxx>; George Dunlap
>> <George.Dunlap@xxxxxxxxxx>; Wei Liu <wei.liu2@xxxxxxxxxx>; xen-
>> devel@xxxxxxxxxxxxxxxxxxxx 
>> Subject: RE: [Xen-devel] [PATCH v2 REPOST 03/12] x86/mm: add
>> HYPERVISOR_memory_op to acquire guest resources
>> 
>> >>> On 29.08.17 at 11:13, <Paul.Durrant@xxxxxxxxxx> wrote:
>> >>  -----Original Message-----
>> >> From: Jan Beulich [mailto:JBeulich@xxxxxxxx]
>> >> Sent: 29 August 2017 10:00
>> >> To: Paul Durrant <Paul.Durrant@xxxxxxxxxx>
>> >> Cc: Andrew Cooper <Andrew.Cooper3@xxxxxxxxxx>; George Dunlap
>> >> <George.Dunlap@xxxxxxxxxx>; Wei Liu <wei.liu2@xxxxxxxxxx>; xen-
>> >> devel@xxxxxxxxxxxxxxxxxxxx 
>> >> Subject: RE: [Xen-devel] [PATCH v2 REPOST 03/12] x86/mm: add
>> >> HYPERVISOR_memory_op to acquire guest resources
>> >>
>> >> >>> On 29.08.17 at 10:32, <Paul.Durrant@xxxxxxxxxx> wrote:
>> >> >> From: Wei Liu [mailto:wei.liu2@xxxxxxxxxx]
>> >> >> Sent: 28 August 2017 16:01
>> >> >> On Tue, Aug 22, 2017 at 03:50:57PM +0100, Paul Durrant wrote:
>> >> >> > +
>> >> >> > +/*
>> >> >> > + * Get the pages for a particular guest resource, so that they can 
>> >> >> > be
>> >> >> > + * mapped directly by a tools domain.
>> >> >> > + */
>> >> >> > +#define XENMEM_acquire_resource 28
>> >> >> > +struct xen_mem_acquire_resource {
>> >> >> > +    /* IN - the domain whose resource is to be mapped */
>> >> >> > +    domid_t domid;
>> >> >> > +    /* IN - the type of resource (defined below) */
>> >> >> > +    uint16_t type;
>> >> >> > +
>> >> >> > +#define XENMEM_resource_grant_table 0
>> >> >> > +
>> >> >> > +    /*
>> >> >> > +     * IN - a type-specific resource identifier, which must be zero
>> >> >> > +     *      unless stated otherwise.
>> >> >> > +     */
>> >> >> > +    uint32_t id;
>> >> >> > +    /* IN - number of (4K) frames of the resource to be mapped */
>> >> >> > +    uint32_t nr_frames;
>> >> >> > +    /* IN - the index of the initial frame to be mapped */
>> >> >> > +    uint64_aligned_t frame;
>> >> >> > +    /* IN/OUT - If the tools domain is PV then, upon return, 
>> >> >> > gmfn_list
>> >> >> > +     *          will be populated with the MFNs of the resource.
>> >> >> > +     *          If the tools domain is HVM then it is expected 
>> >> >> > that, on
>> >> >> > +     *          entry, gmfn_list will be populated with a list of 
>> >> >> > GFNs
>> >> >> > +     *          that will be mapped to the MFNs of the resource.
>> >> >> > +     */
>> >> >> > +    XEN_GUEST_HANDLE(xen_pfn_t) gmfn_list;
>> >> >>
>> >> >> Why is it not possible to make PV does the same thing as HVM?
>> >> >
>> >> > Because PV guests don't use a P2M as such.
>> >>
>> >> They certainly do, just Xen can't rely on (and hence use) it.
>> >
>> > Oh I know they have one but, as you say, Xen can't use it do put resources
>> > at a particular guest location.
>> >
>> >>
>> >> > An HVM guest can pass GFNs in and
>> >> > say 'I want the resource mapped here'. A PV guest can't do that since 
>> >> > it's
>> >> > using MFNs directly... it has to deal with the resource wherever it may
>> be.
>> >>
>> >> Xen does, however, maintain the M2P, so it would not be impossible
>> >> to return GFNs here for PV guests, requiring the caller to translate
>> >> them back to MFNs if so desired.
>> >
>> > That's possible, but still different to and HVM caller, which will pass 
>> > GFNs
>> > in rather than using any values returned. So I don't really see any
>> advantage
>> > in that.
>> 
>> What's wrong with PV passing in PFNs, and Xen installing the
>> resulting translations into the M2P right away (leaving it to the
>> caller to just fix up its P2M)? That would sufficiently
>> parallel XENMEM_exchange, for example.
> 
> How would that work when the mfns are assigned to a different domain? E.g. 
> when I acquire domU's grant table for mapping in dom0, I don't want to take 
> ownership of the mfns... they still belong to the domU.

Oh, right, these are foreign pages.

Jan

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