tcpdump mailing list archives

Re: hardware loop and frame length increase


From: Fulko Hew <fulko.hew () gmail com>
Date: Mon, 6 Jun 2011 18:22:34 -0400

On Mon, Jun 6, 2011 at 6:02 PM, Martin T <m4rtntns () gmail com> wrote:

I made a RJ45 hardware loopback connector(connected Tx- with Rx+ and
Tx+ with Rx-), connected this to my eth2 port, configured
192.168.88.0/24 to eth2 and executed:

"ping -i0.1 -c3 192.168.88.88"

..while running tcpdump like this:


... snip ...


As you can see, every second I sent and received one frame. The
question is, why is the frame, which I receive, 18 bytes longer than
the one I sent? I mean what are those 144 0-bits at the end of the
each frame back from the hardware loop?


You are capturing the 'outgoing' information that is being given to the
driver,
which is a 'short' packet.  The hardware padded it to the
minimum length that a packet is allowed to be, and then sent it on the wire.

The hardware is receiving that incoming padded packet

That would be on 'full duplex' interface.

(If your hardware is half-duplex, the padding must have been accomplished
with a different technique.)

See: http://wiki.wireshark.org/Ethernet
-
This is the tcpdump-workers list.
Visit https://cod.sandelman.ca/ to unsubscribe.


Current thread: