Here I go replying to my own email....

But whilst running tcpdump, I am seeing that any packet I send out, I'm 
sending out twice!!!  For example....

01:11:59.417082 eldorado.rdu.pimpworks.org.1018 > 
eclipse.pimpworks.org.ssh: . ack 364 win 16464 <nop,nop,timestamp 41716 
1237602>
01:11:59.417083 eldorado.rdu.pimpworks.org.1018 > 
eclipse.pimpworks.org.ssh: . ack 364 win 16464 <nop,nop,timestamp 41716 
1237602>
01:12:14.214685 eldorado.rdu.pimpworks.org.1018 > 
eclipse.pimpworks.org.ssh: P 240:260(20) ack 364 win 16464 
<nop,nop,timestamp 41745 1237602>
01:12:14.214686 eldorado.rdu.pimpworks.org.1018 > 
eclipse.pimpworks.org.ssh: P 240:260(20) ack 364 win 16464 
<nop,nop,timestamp 41745 1237602>
01:12:14.258850 eclipse.pimpworks.org.ssh > 
eldorado.rdu.pimpworks.org.1018: P 364:376(12) ack 260 win 17520 
<nop,nop,timestamp 1237632 41745> (DF)
01:12:14.418026 eldorado.rdu.pimpworks.org.1018 > 
eclipse.pimpworks.org.ssh: . ack 376 win 16464 <nop,nop,timestamp 41746 
1237632>
01:12:14.418027 eldorado.rdu.pimpworks.org.1018 > 
eclipse.pimpworks.org.ssh: . ack 376 win 16464 <nop,nop,timestamp 41746 
1237632>
01:12:16.024796 eldorado.rdu.pimpworks.org.1018 > 
eclipse.pimpworks.org.ssh: F 260:260(0) ack 376 win 16464 
<nop,nop,timestamp 41749 1237632>
01:12:16.024797 eldorado.rdu.pimpworks.org.1018 > 
eclipse.pimpworks.org.ssh: F 260:260(0) ack 376 win 16464 
<nop,nop,timestamp 41749 1237632>

What the hell?

Jeff

Stoned koala bears drooled eucalyptus spit in awe as Jeff Workman exclaimed:

> All,
>
> I just compiled 1.5.3_ALPHA on my AlphaServer 4100, and it was causing
> networking problems. My LAN is setup kinda weird, due to the problems I
> asked about in...
>
> http://mail-index.netbsd.org/netbsd-help/2001/12/29/0003.html
>
> As it turned out, this seemed to not only be a problem with Linux, but
> also Solaris and NetBSD machines that were behind the Linksys.  Seems
> like it's some sort of idiosyncracy with my DSL line.  The workaround,
> was to set the mtu on all of my machines behind the Linksys to 1412.  For
> some reason, setting the mtu on the Linksys itself to 1412 just broke
> more stuff.
>
> This worked fine with the kernel I had been running on the AlphaServer
> (1.5W), but after booting the 1.5.3_ALPHA kernel, things begin to break.
> The kernel started discarding all frames that were < mtu, which caused
> most connections to timeout.
>
> The workaround I have put in place was to comment out this block of code
> from /usr/src/sys/net/if_ethersubr.c:
>
> /*      if (m->m_pkthdr.len >
>             ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
>                 printf("%s: discarding oversize frame (len=%d)\n",
>                     ifp->if_xname, m->m_pkthdr.len);
>                 m_freem(m);
>                 return;
>         }
> */
>
> This works, but I'm sure that there's going to be a kernel panic if
> somebody manages to get a very large frame across the wire to this
> machine. Being the non-programmer that I am, I don't know what else to
> do, though. Any suggestions? Maybe something I can set with sysctl?
>
> Jeff
>
> --
> Jeff Workman | jworkman@pimpworks.org | http://www.pimpworks.org



--
Jeff Workman | jworkman@pimpworks.org | http://www.pimpworks.org