[76] in linux-net channel archive
plip hw address problem?
daemon@ATHENA.MIT.EDU (John Paul Morrison)
Fri Feb 10 19:49:00 1995
From: jmorriso@bogomips.ee.ubc.ca (John Paul Morrison)
To: linux-net@vger.rutgers.edu
Date: Thu, 9 Feb 1995 21:48:53 -0800 (PST)
I tried PLIP in 1.1.90 and found it worked with one program and not
with others. If it matters, I used plip as a loadable module. I
haven't used plip is a long time, so this may not be new. PLIP worked
with KA9Q and NCSA telnet, but not JNOS (1.08c and 1.10g). I used
the Crynwr packet driver for all test cases. This is surprising, since
JNOS is derived from KA9Q source code.
I turned on packet tracing in KA9Q and JNOS, they both say the same
thing, but only KA9Q responds to pings, while JNOS ignores them.
The packet traces say:
Thu Feb 09 21:26:46 1995 - plip recv:
Ether: len 98 fc:fc:c0:09:d5:09->fc:fc:c0:09:d5:0a type IP
IP: etc.
this shows that linux is transmitting, and KA9Q and JNOS are both
receiving the packet correctly. But the destination ethernet hardware
address is wrong.
fc:fc:c0:09:d5:09 is the pseudo ethernet address of the linux plip side
the plip.com packet driver says it's ethernet address is 02:00:2C:6A:15:00
linux arp shows the HW address of the DOS side correctly: 02:00:2C:6A:15:00.
so at least the arp layer is behaving.
Why is linux plip sending packets to the wrong HW address? Obviously
PLIP is point-to-point, it seems that some software doesn't care about
receiving the correct HW address, but it seems others do. So is there
a good reason for Linux plip to deliberately send to the wrong
ethernet address? Or is it the plip packet driver?
---------------------------------------------------------------------------
BogoMIPS Research Labs -- bogosity research & simulation -- VE7JPM --
jmorriso@bogomips.ee.ubc.ca ve7jpm@ve7jpm.ampr.org jmorriso@ve7ubc.ampr.org
---------------------------------------------------------------------------
