On 08/24/2017 02:14 PM, Andrew Cooper wrote: > This avoids the explicit boxing/unboxing of mfn_t in relevant codepaths. > > Signed-off-by: Andrew Cooper <andrew.coop...@citrix.com>
[snip] > diff --git a/xen/include/asm-x86/page.h b/xen/include/asm-x86/page.h > index 242903f..8463d71 100644 > --- a/xen/include/asm-x86/page.h > +++ b/xen/include/asm-x86/page.h > @@ -71,6 +71,12 @@ > #define l4e_get_pfn(x) \ > ((unsigned long)(((x).l4 & (PADDR_MASK&PAGE_MASK)) >> PAGE_SHIFT)) > > +/* Get mfn mapped by pte (mfn_t). */ > +#define l1e_get_mfn(x) _mfn(l1e_get_pfn(x)) > +#define l2e_get_mfn(x) _mfn(l2e_get_pfn(x)) > +#define l3e_get_mfn(x) _mfn(l3e_get_pfn(x)) > +#define l4e_get_mfn(x) _mfn(l4e_get_pfn(x)) Hmm, "get" and "put" have specific meanings elsewhere in the code that don't apply here, but the context of which is confusing enough that people might think they apply. What if we did "mfn_from_l1e" instead, to be symmetric with l1e_from_mfn()? -George _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org https://lists.xen.org/xen-devel