On 09/08/09 12:34, Konrad Rzeszutek Wilk wrote: > On Tue, Sep 08, 2009 at 10:22:46PM +0300, Pasi Kärkkäinen wrote: > >> On Tue, Sep 08, 2009 at 12:14:54PM -0700, Boris Derzhavets wrote: >> >>> Console drops into the same stack trace in both cases. >>> >>> >> Jeremy: Any ideas could some recent change in pv_ops dom0 kernel cause >> usb stuff to stop working with >4096M of dom0 memory? >> > The last set of changes in Jeremy's kernel are a set of patches I sent, > so the problem is most likely caused by my work. > Yes, I suspect so. Last time around, USB was good for highlighting problems in the swiotlb code. If there's a problem at 4G then generally its some physical address being accidentally truncated by an implicit cast in the middle of an expression. Or something else. (Why, yes, I am helpful today.) J > A dmesg from both Xen and Linux kernel would be most helpfull. Also > a lspci -vvv would be nice. > > And as Pasi suggested, booting the same kernel without Xen would > be most helpful. > > > >> dom0_mem=4096M works >> dom0_mem=4097M fails with stacktrace >> >> Boris: What was the last kernel that worked without backtraces? >> Also, does that same kernel work as baremetal (without Xen) ? >> > _______________________________________________ > Xen-devel mailing list > Xen-devel@xxxxxxxxxxxxxxxxxxx > http://lists.xensource.com/xen-devel > > -- Fedora-xen mailing list Fedora-xen@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-xen