On 05/01/18 14:11, Jan Beulich wrote: >>>> On 04.01.18 at 14:05, <[email protected]> wrote: >> --- a/xen/arch/x86/guest/xen.c >> +++ b/xen/arch/x86/guest/xen.c >> @@ -72,6 +72,30 @@ void __init probe_hypervisor(void) >> xen_guest = true; >> } >> >> +static void map_shared_info(struct e820map *e820) >> +{ >> + paddr_t frame = 0xff000000; /* TODO: Hardcoded beside magic frames. */ > What are the plans here?
Nothing immediately. This is compatible with all versions of libxc in existance, but we need to start a thread discussing HVM guest physical address space. We've also just found a passive performance hole, where enabling any kind of PCI Passthrough causes Windows and Linux's grant table mappings to turn uncached because they are allocated inside what the OS thinks is an MMIO BAR. I'll start a thread when I'm a little less busy. ~Andrew _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
