[openib-general] IPoIB odd loopback packet from arp

[Todd Bowman]

Thanks Eli.

So the switch is incrementing the rcvswrelay counter when it sends the
broadcast back through the original port.  This doesn't seem to be correct
behavior, it makes that counter unreliable.



On 10/24/06, Eli Cohen <eli at dev.mellanox.co.il> wrote:
>
> Todd,
> This does not look like an error. The first arp is a broadcast
> (qpn=ffffff) so it is received in at the sending interface and is
> dropped. The second on is a unicast arp (qpn=0x000404) so it is not
> received at the local interface.
>
>
> On Mon, 2006-10-23 at 13:48 -0600, Todd Bowman wrote:
> > Using the OFED 1.0 and OFED 1.1 stack I have notice some rcvswrelay
> > errors.  I have tracked it down to the arp request.  I can reproduce
> > the problem with the following steps:
> >
> > ( I have used both 2.6.14.14 and 2.6.18.1 kernels)
> >
> > ib109> arp -d ib110
> > ib109> ping ib110 -c 2
> >
> > # ib_ipoib module debug
> > 13:15:46 ib109 kernel: ib0: sending packet, length=60 address=f6187200
> > qpn=0xffffff
> > 13:15:46 ib109 kernel: ib0: called: id 34, op 0, status: 0
> > 13:15:46 ib109 kernel: ib0: send complete, wrid 34
> > 13:15:46 ib109 kernel: ib0: called: id -2147483623, op 128, status: 0
> > 13:15:46 ib109 kernel: ib0: received 100 bytes, SLID 0x0369
> > 13:15:46 ib109 kernel: ib0: dropping loopback packet
> > 13:15:46 ib109 kernel: ib0: called: id -2147483622, op 128, status: 0
> > 13:15:46 ib109 kernel: ib0: received 100 bytes, SLID 0x016d
> > 13:15:46 ib109 kernel: ib0: sending packet, length=88 address=f6e57520
> > qpn=0x000404
> > 13:15:46 ib109 kernel: ib0: called: id 35, op 0, status: 0
> > 13:15:46 ib109 kernel: ib0: send complete, wrid 35
> > 13:15:46 ib109 kernel: ib0: called: id -2147483621, op 128, status: 0
> > 13:15:46 ib109 kernel: ib0: received 128 bytes, SLID 0x016d
> > 13:15:47 ib109 kernel: ib0: sending packet, length=88 address=f6e57520
> > qpn=0x000404
> > 13:15:47 ib109 kernel: ib0: called: id 36, op 0, status: 0
> > 13:15:47 ib109 kernel: ib0: send complete, wrid 36
> > 13:15:47 ib109 kernel: ib0: called: id -2147483620, op 128, status: 0
> > 13:15:47 ib109 kernel: ib0: received 128 bytes, SLID 0x016d
> > 13:15:51 ib109 kernel: ib0: called: id -2147483619, op 128, status: 0
> > 13:15:51 ib109 kernel: ib0: received 100 bytes, SLID 0x016d
> > 13:15:51 ib109 kernel: ib0: sending packet, length=60 address=f6e57520
> > qpn=0x000404
> > 13:15:51 ib109 kernel: ib0: called: id 37, op 0, status: 0
> > 13:15:51 ib109 kernel: ib0: send complete, wrid 37
> >
> > # tcpdump -i ib0
> > 13:15:46.977578 arp who-has ib110 tell ib109 hardware #32
> > 13:15:46.977682 arp reply ib110 is-at
> > 00:00:04:04:fe:80:00:00:00:00:00:00:00:08:f1:04:03:96:11:59 hardware
> > #32
> > 13:15:46.977710 IP ib109 > ib110: icmp 64: echo request seq 0
> > 13:15:46.977790 IP ib110 > ib109: icmp 64: echo reply seq 0
> > 13:15:47.977772 IP ib109 > ib110: icmp 64: echo request seq 1
> > 13:15:47.977892 IP ib110 > ib109: icmp 64: echo reply seq 1
> > 13:15:51.977076 arp who-has ib109 tell ib110 hardware #32
> > 13:15:51.977094 arp reply ib109 is-at
> > 00:02:00:14:fe:80:00:00:00:00:00:00:00:02:c9:02:00:00:3b:31 hardware
> > #32
> >
> > # error dump
> > rcvswrelayerrors:1 MT47396 Infiniscale-III 0x2c9010b022090[1]
> > <--------> ib109 HCA-1 0x2c90200003b30[1]
> >
> > 1) The ping is successful and the arp table is populated so Is this
> > really a problem or a false positive?
> > 2) The second arp does not generate an error (the error dump reports
> > all new errors in switches). Why?
> >
> > Any ideas?
> >
> > Thanks in advance.
> >
> > Todd
> >
> > _______________________________________________
> > openib-general mailing list
> > openib-general at openib.org
> > http://openib.org/mailman/listinfo/openib-general
> >
> > To unsubscribe, please visit
> http://openib.org/mailman/listinfo/openib-general
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/general/attachments/20061024/0721848a/attachment.html>