Packets from host -> virt with broken ip checksum.
Nikolay Denev
ndenev at gmail.com
Mon Aug 25 18:34:31 UTC 2014
On Mon, Aug 25, 2014 at 7:25 PM, Allan Jude <allanjude at freebsd.org> wrote:
> On 2014-08-25 05:32, Nikolay Denev wrote:
>> On Mon, Aug 25, 2014 at 11:01 AM, Nikolay Denev <ndenev at gmail.com> wrote:
>>> I've started playing around with Bhyve and I've noticed that I can't
>>> talk from the host to the virt and vice versa.
>>> At the same time the virt can talk to the bridged subnet.
>>>
>>> My setup is pretty simple, FreeBSD 10-STABLE (about two week old
>>> build), with two bce(4) interfaces in lagg0 interface.
>>> Then I have bridge0 and the virt's tap0 interface joined in the bridge
>>> with lagg0.
>>> IP address is on lagg0.
>>>
>>> When I try to ping the virt from the host, tcpdump on vtnet0 shows
>>> incoming icmp echo requests,
>>> but no response. tcpdump shows bad checksums, netstat -s shows "bad
>>> header checksums" incrementing.
>>> Packes from the virt to the host seem ok, as I can try to ping from
>>> the virt, and I see the icmp echo req, and the reply coming back, but
>>> the reply is broken in this case so ping does not succeed.
>>>
>>>
>>> --Nikolay
>>
>> So removing bridge/lagg out of the equation and configuring IP on tap0
>> on the host and
>> another IP from the same subnet ot vtnet0 on the virt it works.
>>
>> --Nikolay
>> _______________________________________________
>> freebsd-virtualization at freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-virtualization
>> To unsubscribe, send any mail to "freebsd-virtualization-unsubscribe at freebsd.org"
>>
>
> You might try just disabling hardware checksum offload (ifconfig bce0
> -hwcsum) (on both adapters) and see if that solves the problem too
>
> --
> Allan Jude
>
It's actually bge(4), and disabling hardware tx checksums (ifconfig
bge[0-1] -txcsum) fixes this.
--Nikolay
More information about the freebsd-virtualization
mailing list