[ovs-dev] [PATCH 1/1] [branch-1.4] [ofproto-dpif] Memory leak at specific PACKET_INs

[Zoltan Kiss]

Hi,

Sorry, I've forgot to answer this question. There aren't any other ovs 
messages in the log, just these. And they only appear if I register a 
controller (DVSC in my case) and send every traffic to the controller 
with this rule:

ovs-ofctl add-flow xenbr0 priority=65535,actions=controller

Even without any particular load these appear after a while, but quite 
rarely, and in these short batches.

Zoli

On 03/05/13 21:09, Ben Pfaff wrote:
> Thanks.  Do these messages appear in the log just as you show them, or
> are they interspersed with other messages?
>
> On Fri, May 03, 2013 at 05:40:57PM +0100, Zoltan Kiss wrote:
>> Now I've tried with the latest 1.4 code, but it still produces these
>> message intermittently. Here are some examples:
>>
>> May  3 16:32:27 localhost ovs-vswitchd:
>> 00093|ofproto_dpif|WARN|unexpected flow from datapath in_port(1),eth(src=00:21:1b:f3:63:45,dst=00:e0:ed:26:a9:bc),eth_type(0x0800),ipv4(src=10.80.2.155,dst=10.80.227.83,proto=6,tos=0x10,ttl=62,frag=no),tcp(src=53495,dst=22)
>> May  3 16:32:28 localhost ovs-vswitchd:
>> 00094|ofproto_dpif|WARN|unexpected flow from datapath in_port(1),eth(src=00:21:1b:f3:63:45,dst=00:e0:ed:26:a9:bc),eth_type(0x0800),ipv4(src=10.80.2.155,dst=10.80.227.83,proto=6,tos=0x10,ttl=62,frag=no),tcp(src=53495,dst=22)
>> May  3 16:32:29 localhost ovs-vswitchd:
>> 00095|ofproto_dpif|WARN|unexpected flow from datapath in_port(1),eth(src=00:21:1b:f3:63:45,dst=00:e0:ed:26:a9:bc),eth_type(0x0800),ipv4(src=10.80.2.155,dst=10.80.227.83,proto=6,tos=0x10,ttl=62,frag=no),tcp(src=53495,dst=22)
>> May  3 16:32:30 localhost ovs-vswitchd:
>> 00096|ofproto_dpif|WARN|unexpected flow from datapath in_port(1),eth(src=00:21:1b:f3:63:45,dst=00:e0:ed:26:a9:bc),eth_type(0x0800),ipv4(src=10.80.2.155,dst=10.80.227.83,proto=6,tos=0x10,ttl=62,frag=no),tcp(src=53495,dst=22)
>>
>> Regards,
>>
>> Zoli
>>
>> On 30/04/13 23:16, Ben Pfaff wrote:
>>> Commit fdd80c39b1822db6c8b5c (datapath: Check gso_type for correct
>>> sk_buff in queue_gso_packets().) could be related, but I am not
>>> certain.  It's the mostly likely known-fixed problem I see in a quick
>>> scan of branch-1.4 commits following cc8ccb5.
>>>
>>> Could you provide a log excerpt that includes some of these messages?
>>>
>>> On Tue, Apr 30, 2013 at 10:08:20AM +0100, Zoltan Kiss wrote:
>>>> Hi,
>>>>
>>>> The latest commit we included in XS 6.1 is cc8ccb5 (26 Jun 2012
>>>> 14:43:54: lib: Do not assume sig_atomic_t is int)
>>>>
>>>> Regards,
>>>>
>>>> Zoli
>>>>
>>>> On 25/04/13 19:25, Ben Pfaff wrote:
>>>>> It's not "normal".  It indicates a bug, although if they are only seen
>>>>> occasionally it is not a big deal.  What commit is the original XS 6.1
>>>>> OVS based on?
>>>>>
>>>>> On Wed, Apr 24, 2013 at 09:48:32PM +0100, Zoltan Kiss wrote:
>>>>>> Hi,
>>>>>>
>>>>>> Thanks, I've made my comments on that thread. Another related
>>>>>> question came to my mind: I've seen from time to time some
>>>>>> "ofproto_dpif|WARN|unexpected flow from datapath" messages, not just
>>>>>> with these patches but with the original XS 6.1 ovs. I couldn't see
>>>>>> a pattern in the type of flows. Is this normal?
>>>>
>>