Re: [PATCH 1/6] xen/pvh: Support ParaVirtualized Hardware extensions.

[Ian Campbell <Ian.Campbell@citrix.com>]

On Tue, 2012-10-23 at 15:07 +0100, Konrad Rzeszutek Wilk wrote:
> > +/*
> > + * Unmaps the page appearing at a particular GPFN from the specified guest's
> > + * pseudophysical address space.
> > + * arg == addr of xen_remove_from_physmap_t.
> > + */
> > +#define XENMEM_remove_from_physmap      15
> > +struct xen_remove_from_physmap {
> > +    /* Which domain to change the mapping for. */
> > +    domid_t domid;
> > +
> > +    /* GPFN of the current mapping of the page. */
> > +    xen_pfn_t gpfn;
> > +};
> 
> I just realized that this a bit of unfortunate. We end up
> with on 64-bit with this layout:
> 
> {
>   0->1 [domid]
>   2->7 [nothing]
>   8->16 [gpfn]
> 
> }
> 
> which if one were to do a 32-bit build you would get:
> {
>  0->1 [domid]
>  2->3 [nothing]
>  4->7 [gpfn]
> }
> which means another compat layer in Xen.

A relatively simple one, but yes.

> So I think it makes sense to modify this to be 32 and 64-bit
> clean, something like this:

That works.
We could just swap the domid and gpfn members around, although that
ordering reads a bit unnaturally.

> 
> {
> 	domid_t	domid;
> 	u8	pad[6];
> 	xen_pfn_t gpfn;
> 	/* xen_pfn_t under 32-bit x86 is 4 bytes, so extend it */
> #ifdef CONFIG_X86_32
> 	__u8	pad1[4];
> #endif
> }
> 
> that way the structure is exactly the same size and the offsets
> align.
> 
> Mukesh, you OK with that?