https://www.safaribooksonline.com/library/view/80211-wireless-networks/0596100523/ch04.html
This provides some nice graphics for different topologies. For bridging
to work (aka the simh), the WDS case is required. Most consumer wifi
for mobile devices figure that they only have to support their own MAC
address and do not support WDS correctly. That is why SimH on your
laptop cannot tunnel packets sent over WiFi. It's out going packets go
out to the AP in infrastructure mode (Fig. 4-10). The RA is the Mac
address of the base station, SA is the MAC address of the laptop, and DA
is the dest. address from the tap/tun interfaces ethernet frame. Note
that the source address from the frame has now been dropped. The Reply
packet to SimH will be received by the laptop (Fig. 4-9) with DA set to
the MAC address of the laptop's WiFi interface, TA set to the base
stations Mac Address, and SA set to the source address from the original
ethernet frame. Note that the SimH MAC address is not included, so the
frame will not be bridged to the tap/tun interface and seen by SimH.
If WDS is used, then you can still have issues with base stations not
processing the frames correctly. Higher end WiFi access points
generally work correctly, but it doesn't help you run SimH on your
laptop with a consumer grade WiFi interface. Any easy way to see what
is happening in your configuration is to run tcpdump on the wifi
interface of your laptop. You will likely see the frames that should go
to SImH being received by the laptop, but the destination ethernet
address will be that of the laptop not that of the SimH tap/run
interface. Your host OS won't know what to do with them, and they will
be dropped by your laptop.
On 02/28/2016 05:28 AM, Johnny Billquist wrote:
I must admit that my WiFi knowledge is a bit sketchy, but on the other
hand do I think that I know ethernet...
But reading up on 802.11, I don't see what you mean by 802.1. I'm not
even sure what 802.1 actually says.
However, 802.11 uses 802.2 for packets, which is unlike ethernet,
which is not an 802 protocol at all.
Also, reading up on 802.11, it appears that source and destination MAC
address are always present. However, there are potentially two more
MAC addresses in the packet, which I have not found much good
information about yet. Wikipedia suggest the third is for filtering
purposes, and do not even explain the fourth one.
What I do know, from observation, is that if I have something like
simh setup to communicate over WiFi, packets do get sent out, but my
simh instance will not receive any unicast packets to it, which
suggests that the switch do not send such packets out over WiFi to the
correct destination. I would assume it is because switches knows which
stations actually do exist, but that is a guess on my part.
Johnny
On 2016-02-28 07:29, Peter Svensson wrote:
Hi all,
The answers given last time were not all that accurate. WiFi for
historical reasons conserve bandwidth by assuming that the client side
802.11 mac address is the same as the 802.1 sender mac address and thus
omits the latter. This is the so called 3-address mode. This does not
leave any room for more than one 802.1 mac address on a client.
However, there is also a 4 address mode for WiFi which does support
bridging since the 802.1 frames are transported verbatim. This mode has
many different names from vendors. Most commonly it goes by the name
WDS, but that name is unfortunately also used by a bunch of non
transparent mechanisms from other vendors.
See e.g. http://linuxwireless.org/en/users/Documentation/iw/
802.11 does not care about broadcast one way or another. It is just
transported. It does not care what layer 3 is used (except only one
layer 2 sec address in 3 address mode).
The decision to save 6 bytes is an unfortunate historical artefact. The
(802.11 standard) option to not save these bytes is not always exposed
on wifi equipment. Some does, and most can I'd you run OpenWrt or
similar software on them. Not sure about Windows though.
Peter
On February 27, 2016 11:01:31 PM GMT+01:00, Johnny Billquist
<[email protected]> wrote:
On 2016-02-27 20:14, Andreas Davour wrote:
On Sat, 27 Feb 2016, Johnny Billquist wrote:
On 2016-02-27 17:53, Michael Kerpan wrote:
It's not that multiuser is impossible using KLH10
networking emulation
and native TOPS-20 TCP/IP, it's that it's a lot harder
to set up than a
SIMH virtual multiplexer. Wifi (under Linux at least)
doesn't play all
that well with bridging and virtual networking stuff,
which makes
setting up networking in KLH10 nigh impossible for those
of us stuck on
wireless. I can't even really set up a VM and run KLH10
in a guest,
because Virtualbox's bridged mode doesn't work for me.
Additionally, the
idea of putting anything running a 30 year old TCP/IP
stack onto the
Internet scares me, even if the folks at twenex.org
<http://twenex.org> <http://twenex.org>
have done so.
The WiFi problem is easy to solve. It's called a router.
Your host
routes IP between the WiFi interface and the virtual network
the host
and your virtual machine shares.
Trouble setting up networking on the virtual machine? Maybe.
But this
boils down to - if you want to run that host, you should
learn how to
manage it.
Security issues are mostly non-issues. How many script
kiddies today
even know what a TOPS-20 host is. There are most certainly
vulnerabilities, but they are very different from the ones
presented
by modern machines.
I have an RSX system on the Internet, and it gets constant
probing
over telnet and http, but they are all probing in ways that
just don't
make sense. So I have never felt more secure.
To Johnnys suggestions I might add that (I don't know exactly
what
problems you're having) maybe openvswitch might help out?
The problem is that WiFi is not really like ethernet (I think we
covered
this a month ago, but maybe it was on a different list). Anyway,
if you
have a simh instance using WiFi for the network, it do not work,
since
putting the interface in promiscuous mode, and pretend you have a
second
machine with a different MAC address do not work, since with
WiFi, the
base station actually knows which MAC addresses are connected,
and if a
packet comes in for a device for which the MAC address is not
registered, the packet will not be send out over WiFi, so you
will not
get anything, even though you think you have your
interface in
promiscuous mode, and are sending packets out with a different
source
MAC address, which you might think the WiFi switch would learn,
as it
would had it been ethernet.
Johnny
_______________________________________________
Simh mailing list
[email protected]
http://mailman.trailing-edge.com/mailman/listinfo/simh
