Jon Fairbairn wrote: > Brian Jackson <i...@theiggy.com> writes: > > > On Friday 28 August 2009 01:14:42 pm Jon Fairbairn wrote: > >> I'm experimenting with a virtual router. I did this a few years ago with > >> Xen and it worked well enough, but then fedora changed and it stopped > >> working, so I gave up for a while. Now I have a machine that supports > >> hardware virtualisation, I thought I'd try again. > >> > >> The setup was done through virt-manager. The network between the host > >> and guest is a virtual bridge. What I've been trying to do is to assign > >> a USB cable modem to the guest, > > > > > > This is probably your problem here. KVM only emulates a usb1.1 > > controller, > > that shouldn't be a problem as the cable modem is only a usb 1.1 device. > > > and from all reports, it doesn't really do that very well. > > Not very well? We're talking about a factor of *FIFTY* slow down here. > If I'd implemented the driver in a 1980s lazy functional language I > would have only expected a factor of ten :-P ;-). > > > There have been numerous reports of poor performance even for a usb1.1 > > device. You should check the archives to see if there was ever any > > kind of tips or resolution to some of those problems. > > I had a look, but I couldn't find anything helpful. Maybe I'm searching > for the wrong things?
I don't have any specific answers, but you might be able to get some more info by uncomenting the "//#define DEBUG" in kvm's usb-linux.c. You might also try capturing the USB traffic (e.g. with wireshark) both in the host with the modem connected directly, and in the guest with the modem passed through, to see if there are any differences other than speed (packet sizes, etc). Also try comparing the "lsusb -vvv" output in both guest and host? -jim -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
