1.3.91 and ARP

Manfred Petz (pm@radawana.cg.tuwien.ac.at)
Fri, 19 Apr 1996 14:55:02 +0200 (MET DST)


Sorry if I bother you with a silly mistake, but it seems to be a kernel
problem:

When connecting with my SVR4.2 (not connected to any network) to my
1.3.91 with SLIP (IP forwarding enabled), it seems that I cause serious
network troubles. My SLIP connection works perfectly but some Windoze PCs
on our 192.56.14 net (ether) with TCP/IP software refuse to work. They
say "IP address already in use". My SLIP configuration seems to be ok on
both sides.

During the SLIP session the kernel doesn't log any related messages.

Below is an excerpt of my syslog immediately after boot. Could somebody give
me a hint, what the 'arp_cache_bind() backlogged', 'arp_run_bh() reinstalled'
and 'was born dead' messages mean? If ARP does something wierd, this could be
the cause. I dunno which device belongs to the hardware address shown (it's
not the Linux box).

Oh yes: I use the same kernel on two Linux machines on the same net. The
dial-in SLIP machine writes these arp messages to syslog immediately after
boot (network interface card is 3c509), the other machine (ne2000 clone)
doesn't. Hmmmmm??

If you need more info, tell me. I could check this out with Linux
on both sides.

Apr 19 12:27:16 hf kernel: Adding Swap: 16504k swap-space
Apr 19 12:27:16 hf kernel: eth0: Setting Rx mode to 1 addresses.
Apr 19 12:28:07 hf kernel: arp_cache_bind: 060e38c0 backlogged
Apr 19 12:28:07 hf kernel: arp_run_bh: 060e38c0 reinstalled
Apr 19 12:28:07 hf kernel: arp_run_bh: entry 060e38c0 was born dead
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Apr 19 12:32:07 hf xntpd[54]: time error 3564.444317 is way too large (set clock manually)
Apr 19 13:36:30 hf kernel: Socket destroy delayed (r=0 w=536)
Apr 19 13:36:40 hf kernel: Socket destroy delayed (r=0 w=536)

Thanks
pm