Re: [Xen-devel] [PATCH 2/2] x86/xen: make mfn_to_pfn() work with MFNs that are 1:1 in the p2m

[Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>]

On Fri, 3 Jan 2014, David Vrabel wrote:
> From: David Vrabel <david.vrabel@xxxxxxxxxx>
> 
> The _PAGE_IOMAP PTE flag is used to indicate that the PTE contains an
> MFN that is an identify frame (1:1) in the p2m.  This is so the
> correct conversion of MFN to PFN can be done reading a PTE.
> 
> If mfn_to_pfn() returned the correct PFN instead the _PAGE_IOMAP flag
> is not required and can be removed.
> 
> In mfn_to_pfn() the PFN found in the M2P is checked in P2M.  If the
> two MFNs differ then the MFN is one of three possibilities:
> 
>   a) its a foreign MFN with an m2p override.
> 
>   b) it's a foreign MFN with /no/ m2p override.
> 
>   c) it's a identity MFN.
> 
> It is not permitted to call mfn_to_pfn() no a foreign MFN without an
> override was the resulting PFN will not incorrect.  We can therefore
> assume case (c) and return PFN == MFN.
> 
> [ This patch should probably be split into the Xen-specific parts and
> an x86 patch to remove _PAGE_IOMAP. ]
> 
> Signed-off-by: David Vrabel <david.vrabel@xxxxxxxxxx>
> ---
>  arch/x86/include/asm/pgtable_types.h |   12 +++---
>  arch/x86/include/asm/xen/page.h      |   18 +++++++---
>  arch/x86/mm/init_32.c                |    2 +-
>  arch/x86/mm/init_64.c                |    2 +-
>  arch/x86/pci/i386.c                  |    2 -
>  arch/x86/xen/enlighten.c             |    2 -
>  arch/x86/xen/mmu.c                   |   58 
> +++++-----------------------------
>  7 files changed, 28 insertions(+), 68 deletions(-)
> 
> diff --git a/arch/x86/include/asm/pgtable_types.h 
> b/arch/x86/include/asm/pgtable_types.h
> index 0ecac25..0b12657 100644
> diff --git a/arch/x86/include/asm/xen/page.h b/arch/x86/include/asm/xen/page.h
> index b913915..eb11963 100644
> --- a/arch/x86/include/asm/xen/page.h
> +++ b/arch/x86/include/asm/xen/page.h
> @@ -112,11 +112,18 @@ static inline unsigned long mfn_to_pfn(unsigned long 
> mfn)
>       pfn = mfn_to_pfn_no_overrides(mfn);
>       if (get_phys_to_machine(pfn) != mfn) {
>               /*
> -              * If this appears to be a foreign mfn (because the pfn
> -              * doesn't map back to the mfn), then check the local override
> -              * table to see if there's a better pfn to use.
> +              * This is either:
>                *
> -              * m2p_find_override_pfn returns ~0 if it doesn't find anything.
> +              * a) a foreign MFN with an override.
> +              *
> +              * b) a foreign MFN without an override.
> +              *
> +              * c) an identity MFN that is not in the the p2m.

You have missed a case here:

           d) a local MFN with an override

think of blkfront and blkback in the same domain.
However I don't think this case would be affected by the patch.


> +              * For (a), look in the m2p overrides.  For (b) and
> +              * (c) assume identify MFN since mfn_to_pfn() will
> +              * only be called on foreign MFNs iff they have
> +              * overrides.
>                */
>               pfn = m2p_find_override_pfn(mfn, ~0);
>       }
> @@ -126,8 +133,7 @@ static inline unsigned long mfn_to_pfn(unsigned long mfn)
>        * entry doesn't map back to the mfn and m2p_override doesn't have a
>        * valid entry for it.
>        */
> -     if (pfn == ~0 &&
> -                     get_phys_to_machine(mfn) == IDENTITY_FRAME(mfn))
> +     if (pfn == ~0)
>               pfn = mfn;
>  
>       return pfn;


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