On 5/2/19 7:58 AM, Jan Beulich wrote: > This is what both callers of guest_physmap_add_page() in memory.c want > (for the !paging_mode_translate() case), and it is also what both > callers in gnttab_transfer() need (but have been lacking). The other > (x86-specific) callers are all HVM-only, and hence unaffected by this > change. > > Sadly this isn't enough yet to drop Arm's dummy macro, as there's one > more use in page_alloc.c. > > Signed-off-by: Jan Beulich <[email protected]> > > --- a/xen/arch/x86/mm/p2m.c > +++ b/xen/arch/x86/mm/p2m.c > @@ -841,15 +841,19 @@ guest_physmap_add_entry(struct domain *d > * any guest-requested type changes succeed and remove the IOMMU > * entry). > */ > - if ( !need_iommu_pt_sync(d) || t != p2m_ram_rw ) > + if ( t != p2m_ram_rw ) > return 0;
So, you seem to be claiming that the only way to get here is via guest_physmap_add_page(), which will always call this function with p2m_ram_rw. So then what's the point of this conditional at all anymore? Would it be better to add an ASSERT(t == p2m_ram_rw) here? And if we ever *do* get here with t == p2m_ram_rw, do we really not want to call set_gpfn_from_mfn()? -George _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
