every 2nd echo-request malformed when ping -s >4067

[Harald Schmalzbauer]

 schrieb Jeremy Chadwick am 24.10.2012 20:12 (localtime):
> ...
> root at icarus:/root # ping -D -s 4068 127.0.0.1
> PING 127.0.0.1 (127.0.0.1): 4068 data bytes
> 4076 bytes from 127.0.0.1: icmp_seq=0 ttl=64 time=0.030 ms
> 4076 bytes from 127.0.0.1: icmp_seq=1 ttl=64 time=0.032 ms
> 4076 bytes from 127.0.0.1: icmp_seq=2 ttl=64 time=0.024 ms
> ^C
> --- 127.0.0.1 ping statistics ---
> 3 packets transmitted, 3 packets received, 0.0% packet loss
> round-trip min/avg/max/stddev = 0.024/0.029/0.032/0.003 ms
>
> I also ran tcpdump for this too; no anomalies -- all 3 packets showed up
> correctly (decoded correctly).  My uname -a is below, with csup run
> about 20 minutes before the kernel build date.
>
> FreeBSD icarus.home.lan 9.1-PRERELEASE FreeBSD 9.1-PRERELEASE #0: Sun Oct 21 05:24:09 PDT 2012     root at icarus.home.lan:/usr/obj/usr/src/sys/X7SBA_RELENG_9_amd64  amd64
>
> This is on bare-metal hardware, BTW.  I mention that because I've seen
> some of your other threads talking about NIC driver ordeals under VMs (I
> think).

Oh, I thought it's not specific to my system because I remembered this
issue well on a completely different HW; real HW in that case.

A quick test showed that the big-ping to localhost works on all my other
machines... Should have checked that before, sorry.

I have zero_copy_sockets in all my kernels, so I think that can't be the
cause.
But I don't have another machine where real nics have MTU > 1500.
I'll try to find out why this only affects one machine here at the
moment, but tonight's time to have dinner->sleep.

Thanks a lot!

-Harry



-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 196 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20121024/fa536fa4/attachment.sig>