|
ru.nethack
- RU.NETHACK -------------------------------------------------------------------
From : Pavel Pokrovsky 2:5020/400 28 Apr 2003 07:20:53
To : Dmitry Tcvetcov
Subject : Как скpыть МАС-адpес сетевой платы
--------------------------------------------------------------------------------
Hello, Dmitry
Мыло до тебя не проходит (smtp пишет rejected), ну и хрен с ним. кидаю чего
обещал в эху. наверное не одному тебе будет интересно :-)
русскими буковками в схеме помечены цвета проводов.
Clipping the UTP (10-BASE-T, 100-BASE-TX)
Creating UTP receive-only adapters is harder. Ethernet hubs check the "link
integrity" with the adapter, which is done by sending a regular voltage
pulse down the wire. On 100BaseTX, the problem is even more difficult
because both sides send a steady stream of empty traffic down the wire.
Therefore, you can't simply cut the transmit wires, because the hub will
think the adapter is no longer there, and disconnect it.
However, you can "denature" the wire, introducing CRC errors that disrupt
outgoing frames, but which still allows the stream of symbols from the card
and/or link integrity pulses. This is actually a problem you may have
encountered before: the link lights on the hub and adapter show a
connection, but you cannot communicate on the network. Ethernet is actually
very robust, so creating such a cable is difficult.
The high-speed integrity is maintained by using two wires for each signal,
and twisting the wires around each other. You can think of it as one wire
shielding the other, or that any electrical disturbance affects both
equally, and since the difference in voltage is measured, it all evens out
(Geek note: differential SCSI or Ultra-DMA/66 cables are based on the same
principle).
As a consequence, simply "untwisting" the transmit pair will degrade the
signal to the point that outgoing traffic will be corrupted with CRC erros,
yet the hub will still get enough of a signal to know the adapter is still
there.
__
+-~~ ~~-+
|........|
-++++++++-
12345678
------------------------------------------------
Pin MDI signal MDI-X signal
------------------------------------------------
бо 1 TD+ xmit to UTP RD+ rcv from UTP
о 2 TD- xmit RD- rcv from UTP
бз 3 RD+ rcv from UTP TD+ xmit to UTP
с 4 Not used Not used
бс 5 Not used Not used
з 6 RD- rcv from UTP TD- xmit to UTP
бк 7 Not used Not used
к 8 Not used Not used
In order to denature it, you must use a 4-pair cable where normally two of
the pairs are unused. Each twisted-pair is color coded, with COLOR-stripes
on white for one wire, and white-stripes on COLOR for the other wire. In
the above diagram, the transmit pair is normally orange/white, whereas the
receive pair is normally green/white (for the AT&T or EIA/TIA 568B
standard). However, the EIA/TIA 568A standard reverses these colors
(aargg). Therefore, all you are worried about is that pins 3 and 6 must
share the same color, but in order to denature the transmit pair, pins 1
and 2 must NOT share a color. Pick any of the others, it doesn't matter.
Both ends of the cable must be wired the same.
You should make this a long cable, as long as the standard will allow. The
problem is that it doesn't always degrade the signal enough, so the longer
the better (I've used 40-feet, but still about 5% of the transmitted frames
still get through). Furthermore, you should use 100-mbps (100-BASE-TX) for
this, as the signal degrades easier. I've never tried with 10-mbps.
However, this approach still leaves the possibility that somebody might
"fix" this by replacing the bad cable with a good one. An alternate
technique involves altering the NIC itself, inside the box where nobody can
accidentally fix it. In this case, instead of clipping the wire you should
add something to it. Most adapters leave the traces exposed (though some
are beginning to shield the RJ45 connector, but you can easily peel back
the shielding). Solder a paperclip or wire onto the transmit pin #1 on the
backside of the connector. Adjust the length as needed until the
transmission is screwed up. I've never tried this, so if you get it to
work, please drop me e-mail.
Using either of these techniques may cause the FCC to come after you
because of radio wave transmission. The computer case should shield the
altered adapter, but I make no guarantees.
Another proposal is to create a circuit to generate the symbols. This may
be easy for hardware engineers, but I'm a software engineer and have never
done it.
-P
--- ifmail v.2.15dev5
* Origin: FidoNet Online - http://www.fido-online.com (2:5020/400)
Вернуться к списку тем, сортированных по: возрастание даты уменьшение даты тема автор
|
