USB Networking troubles
Zack Mollusc
zmollusc at blueyonder.co.uk
Sat Aug 9 10:54:45 CEST 2008
michael cole wrote:
> Andrew,
>
> I'll make the netmask changes you suggest.
>
> Further to my comments above re: the USB networking page @ openmoko.
> My issue could have been avoided if the networking page made clear
> that the Freerunner--Desktop connection is on a different network than
> the LAN the desktop uses to access the Internet. Likewise a
> troubleshooting note when one cannot ping past the desktop machine.
> The document, as it is now written, focuses on just the
> Freerunner--Desktop connection and makes an assumption that the LAN is
> not 192.168.0.X Your description placing the task of setting up USB
> networking in the context of the whole LAN makes clear how the
> FR--Desktop network is situated. It probably seems obvious for most of
> the users here.
>
> Perhaps someone with editing rights to the networking page can add a
> couple of lines to make clear the need for distinct networks in
> specifying the setup. Also, it would help to extend the worked example
> to show (at each stage) what output one should see from "route -n " on
> each machine. This is also a point to check in troubleshooting cases
> where one can ping the desktop but no other machines on the LAN.
>
> Once again I appreciate the time everyone has taken to help.
>
> Thanks,
>
> Michael
>
> On Fri, Aug 8, 2008 at 8:40 AM, Andrew C. Dingman
> <andrew+openmoko at dingman.org <mailto:andrew%2Bopenmoko at dingman.org>>
> wrote:
>
> On Fri, 2008-08-08 at 06:32 -0400, michael cole wrote:
> > First thanks to all for replying.
> >
> > These changes worked.
>
> Glad it helped. I hope the IP routing mini-lecture made clear *why* :)
>
> > So on my Freerunner /etc/network/interfaces includes:
> >
> > # Ethernet/RNDIS gadget (g_ether)
> > # ... or on host side, usbnet and random hwaddr
> > auto usb0
> > iface usb0 inet static
> > address 192.168.2.202 <http://192.168.2.202>
> > netmask 255.255.255.0 <http://255.255.255.0>
> > network 192.168.2.0 <http://192.168.2.0>
> > gateway 192.168.2.200 <http://192.168.2.200>
> > # orig:
> > # up echo nameserver 192.168.0.200 <http://192.168.0.200>
> >/etc/resolv.conf
> > # using home (nimiq.net <http://nimiq.net>)
> > up echo nameserver 207.235.125.4 <http://207.235.125.4>
> >/etc/resolv.conf
> >
> > on the desktop /etc/network/interfaces includes:
> > allow-hotplug usb0
> > iface usb0 inet static
> > address 192.168.2.200 <http://192.168.2.200>
> > netmask 255.255.255.192 <http://255.255.255.192>
> > post-up iptables -A POSTROUTING -t nat -j MASQUERADE -s
> 192.168.2.192/26 <http://192.168.2.192/26>
> > post-up echo 1 > /proc/sys/net/ipv4/ip_forward
> > post-up iptables -P FORWARD ACCEPT
> > pre-down iptables -D POSTROUTING -t nat -j MASQUERADE -s
> > 192.168.2.192/26 <http://192.168.2.192/26>
>
> Almost right, and in fact this configuration won't break in practice
> because you can't plug additional devices into your USB network.
> However, the FreeRunner and the desktop really *should* agree
> about the
> netmask. You have the FreeRunner thinking its local network
> includes 256
> addresses (192.168.2.0-255), and the desktop thinking that the network
> it and the FreeRunner live in includes only 64 (192.168.2.192-255). If
> you choose to fix this on the desktop by using the 255.255.255.0
> <http://255.255.255.0>
> netmask, you'll also want to fix the masquerade rule to specify
> 192.168.2.192/24 <http://192.168.2.192/24>, as the /n part
> indicates a netmask with binary 1 in
> the n most significant bits. (255.255.255.0 <http://255.255.255.0>
> == binary IP 11111111
> 11111111 11111111 00000000. 255.255.255.192
> <http://255.255.255.192> == binary 11111111 11111111
> 11111111 11000000)
>
> If you want to play with this stuff a bit more, you might be
> interested
> in a tool called 'gip' that you can install on your desktop with
> 'apt-get install gip'. It lets you enter IPs and netmasks in any
> of the
> normal forms and shows you the resulting network sizes and binary,
> hexadecimal, and decimal representations of the results.
> --
>
>
> _______________________________________________
> support mailing list
> support at lists.openmoko.org <mailto:support at lists.openmoko.org>
> https://lists.openmoko.org/mailman/listinfo/support
>
>
>
>
> --
> Michael Cole
> mjcole137 at gmail.com <mailto:mjcole137 at gmail.com>
> ------------------------------------------------------------------------
>
> _______________________________________________
> support mailing list
> support at lists.openmoko.org
> https://lists.openmoko.org/mailman/listinfo/support
>
I put a line in yesterday pointing out that the networks need to be
distinctly different, but it seems to have gone again :-)
More information about the support
mailing list