[Simh] Problem with "attach xq"

[Jan-Benedict Glaw]

On Thu, 2008-01-24 07:34:37 -0600, Hittner, David T. <david.hittner at ngc.com> wrote:
> If you look at the attach code, you'll see that *some* of the real pcap-selectable adapters are
> 'disallowed' and removed from the list of connectable devices if they do not meet
> certain criteria - this was primarily done to remove WAN adapters, bridges (brX), loopbacks (lo0)
> and similar 'illegal' network objects from the list. And yes, this can make it a bit confusing.

While you mention the bridge devices here... It would make some sense
to use a number of tap devices connected via a bridge. That'll allow
to build "LANs" between SIMH instances.

> The reason for the 'ATT XQ ethX' syntax is that the code was originally designed using a
> simple linux system with just the standard eth0 port, and the pcap map-numbered syntax
> was added later to 'standardize' the attach syntax for Windows systems (which have horrible device names).
> 
> However, for MOST users on systems with one or two ethernet ports, the ethX designation will work -
> if not, you can either
>   1) Get the list of allowed devices with SHOW XQ ETH (as Mark suggested), or
>   2) Connect using the hard device name with ATT XQ <device-name-literal>
> 	(on *ix use the full path of the device, not the just the simple ethX designator)
>       [this also works to 'force' connection to one of the illegal devices]

With "modern" Linux distros, the udev/hotplug stuff will rename
devices based on their MAC address. So if you put in some ethernet
adaptor, boot up (first networked boot ever), that'll be eth0.
Throwing it away and adding another ethernet card, that will be eth1,
even while that's the one and only card in the box.

> The pattern 'ethX' is what triggers the PCAP filtered map correlation - if the device
> string does not match the 'ethX' 4-character pattern, it will assume
> that you are using a device-name-literal. I suppose that we could 
> disambiguate it by making the match pattern 'pcapX', however that
> would *still* confuse people who have run tcpdump and are expecting the full list
> of pcap device numbers, not just the 'allowed' list.

Maybe a completely new syntax (at xq index 3) would disambiguate the
situation?

> We would be happy for anyone familiar with *ix device programming to provide
> the code to fill in the device description string to help make it more obvious
> which adapter you are selecting.

With Linux, I think there's no device description available at all.
However, it could be possible to get at least the driver name and
physical path from /sys/class/net/ethX/uevent .  That could be used as
a pseudo-description, but I guess it would be better to patch libpcap
for that, instead of adding Linux-specific code to SIMH.

MfG, JBG

-- 
      Jan-Benedict Glaw      jbglaw at lug-owl.de              +49-172-7608481
 Signature of:                    Don't believe in miracles: Rely on them!
 the second  :
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://mailman.trailing-edge.com/pipermail/simh/attachments/20080124/c26d577a/attachment.sig>