Re: [2.0.x/2.1.x] Lots of ICMP error msgs

Ricky Beam (
Tue, 30 Sep 1997 11:12:28 -0400 (EDT)

Letting the chips fall where they may, I quote Andi Kleen:
>I have seen them with broken Modems (especially ELSAs who got too hot)
>or terminal servers (Ascend MAX). Strictly speaking it's against the RFC
>anyways to log checksum errors. It's only enabled in 2.1. to catch bugs,
>that it is enabled in the 2.0.x masquerading code is a bug IMHO. There are
>many legitimate reasons why a packet might get corrupted.

Oh, like the kernel mis-sequencing a packet. I know it's doing it, but I
haven't had time to collect enough data to really prove it beyond the
hardware (Intel has been known to report data incorrectly, but ...)

I have data to indicate the kernel getting or assembling the headers and
data segments exactly TWO frames out of sequence. That has explained every
csum (TCP csum) failure I have captured. I suspect it may be a wild race
condition -- an interrupt at an odd time (those are kinda hard to localize.)