Re: Xen BUG in mm / Xen 4.0.1 with 2.6.32.18/21 pvops Kernel?

["Jan Beulich" <JBeulich@novell.com>]

>>> On 08.09.10 at 14:15, Carsten Schiers <carsten@schiers.de> wrote:
> (XEN) mm.c:860:d0 Error getting mfn 80000 (pfn 5555555555555555) from L1 
> entry 8000000080000473 for l1e_owner=0, pg_owner=32753

DOMID_IO seen here generally means that Dom0 tried to map a page
it doesn't own (likely because of your use of dom0_mem=). As the
page really is a RAM one, Xen doesn't allow the access. Given that
this apparently happens in the context of
acpi_ex_system_memory_space_handler() you'd have to look at (or
provide) your DSDT and SSDT(s) to see where this reference comes
from. Very likely this is just a bogus reference, that you get away
with on native, perhaps because this code in ioremap.c

	last_pfn = last_addr >> PAGE_SHIFT;
	for (pfn = phys_addr >> PAGE_SHIFT; pfn <= last_pfn; pfn++) {
		int is_ram = page_is_ram(pfn);

		if (is_ram && pfn_valid(pfn) && !PageReserved(pfn_to_page(pfn)))
			return NULL;
		WARN_ON_ONCE(is_ram);
	}

should result in returning NULL there, while it wouldn't cover the
situation under Xen. (While the code is meaningless under Xen in
its current shape anyway, using dom0_mem= with a value above
2G should get you around the issue, as then PFN 0x80000 would
be considered RAM there too.)

Jan