On Tue, Oct 9, 2012 at 5:24 AM, <fred.konrad@xxxxxxxxxxxxx> wrote: > From: Peter Maydell <peter.maydell@xxxxxxxxxx> > > Add support for the generic MMIO based virtio transport. > > This patch is a modyfied patch of > Peter Maydell <peter.maydell@xxxxxxxxxx>. Changes are to have > virtio-mmio bridge device which provides virtio-mmio bus. To this bus > virtio-mmio-transport device is connected and in turn provides > virtio-transport bus. Then virtio backends could be connected to this > bus. > > Also this patch includes some fixes for bugs spotted by > Ying-Shiuan Pan <yspan@xxxxxxxxxxx>. I'm trying to figure out how this is intended to work for both a virtio block device and virtio net device. Can I use a single transport device for both devices (with these patches this seems to generate a segfault on qemu at startup) or am I supposed to define two virtio-mmio-transport thingies, but then how does that get mapped in terms of addresses? Do I pass this as a parameter to the second mmio transport (then why didn't we do it for the first?) or do I have to modify the vexpress.c code for a second transport layer. If I don't need to create two transport channels, then why do we segfault? (I'll be happy to dig in and try to figure out the problem, but pointers to get going would be appreciated). > > Signed-off-by: Evgeny Voevodin <e.voevodin@xxxxxxxxxxx> > > Conflicts: > > Makefile.objs nit: you probably don't want to include merge conflict notices in patches for review, but make sure they rebase cleanly on some defined base Thanks, -Christoffer _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
