Re: [PATCH 3 of 3] x86/mm: Teach paging to page table-based p2m

[Tim Deegan <tim@xen.org>]

Hi, 

The last version of this patch had the beginnings of an interlock to
avoid iommu-pt-sharing and p2m-fu happening at the same time.  I
suggested taht it wasn't complete enough, but it seems to have gone away
entirely!

Also: 

At 15:23 -0400 on 21 Mar (1332343380), Andres Lagar-Cavilla wrote:
> @@ -615,11 +618,12 @@ pod_retry_l1:
>                             sizeof(l1e));
>              
>      if ( ret == 0 ) {
> +        unsigned long l1e_mfn = l1e_get_pfn(l1e);
>          p2mt = p2m_flags_to_type(l1e_get_flags(l1e));
> -        ASSERT(l1e_get_pfn(l1e) != INVALID_MFN || !p2m_is_ram(p2mt));
> +        ASSERT( (l1e_mfn != INVALID_MFN || !p2m_is_ram(p2mt)) ||
> +                (l1e_mfn == INVALID_MFN && p2m_is_paging(p2mt)) );

I guess, given the discussion in the other subthread, that this ASSERT
always passes, and should be using mfn_valid() instead?

Cheers,

Tim.