Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 4 May 2018 06:09:44 +0000
From:      "Khoje, Anand" <Anand.Khoje@cavium.com>
To:        "freebsd-questions@freebsd.org" <freebsd-questions@freebsd.org>
Subject:   Krping server port not getting released after completion of krping session
Message-ID:  <BYAPR07MB48726ED6A7A6EF022113A0B599860@BYAPR07MB4872.namprd07.prod.outlook.com>

Next in thread | Raw E-Mail | Index | Archive | Help
Hi,

I am currently working on RoCE v1 support for Cavium BigBear adapters. Alth=
ough, RoCE traffic seems to be working now, I have noticed that once we est=
ablish krping server at one end, using command :
echo "server,addr=3D192.168.11.11,port=3D9991,verbose" > /dev/krping

and run krping client from the peer system, using command :
echo "client,count=3D10,addr=3D192.168.11.11,port=3D9991,verbose" > /dev/kr=
ping

The session ends after sending 10 RDMA packets.
Here are the logs :
server ping data: rdma-ping-0: ABCDEFGHIJKLMNOPQRSTUVWXYZ[\134]^_`abcdefghi=
jklmnopqr
server ping data: rdma-ping-1: BCDEFGHIJKLMNOPQRSTUVWXYZ[\134]^_`abcdefghij=
klmnopqrs
server ping data: rdma-ping-2: CDEFGHIJKLMNOPQRSTUVWXYZ[\134]^_`abcdefghijk=
lmnopqrst
server ping data: rdma-ping-3: DEFGHIJKLMNOPQRSTUVWXYZ[\134]^_`abcdefghijkl=
mnopqrstu
server ping data: rdma-ping-4: EFGHIJKLMNOPQRSTUVWXYZ[\134]^_`abcdefghijklm=
nopqrstuv
server ping data: rdma-ping-5: FGHIJKLMNOPQRSTUVWXYZ[\134]^_`abcdefghijklmn=
opqrstuvw
server ping data: rdma-ping-6: GHIJKLMNOPQRSTUVWXYZ[\134]^_`abcdefghijklmno=
pqrstuvwx
server ping data: rdma-ping-7: HIJKLMNOPQRSTUVWXYZ[\134]^_`abcdefghijklmnop=
qrstuvwxy
server ping data: rdma-ping-8: IJKLMNOPQRSTUVWXYZ[\134]^_`abcdefghijklmnopq=
rstuvwxyz
server ping data: rdma-ping-9: JKLMNOPQRSTUVWXYZ[\134]^_`abcdefghijklmnopqr=
stuvwxyzA
DISCONNECT EVENT...
wait for RDMA_READ_ADV state 10
cq completion in ERROR state


But, if we attempt to re-establish the krping server in the same system wit=
h the same port number (i.e. 9991), we get an error saying :
rdma_bind_addr error -48 (address already in use?)

Meanwhile, I checked with Mellanox Connect X3 adapters, I see the same issu=
e.
My question is, is this behavior normal? Or is there some issue with my set=
up? Is it due to some ibcore state machine being stuck in some state waitin=
g for an event which has not been taken care of by the driver?

P.S:
Here is the output of netstat when a krping server is established  :
# netstat
Active Internet connections
Proto Recv-Q Send-Q      Local Address                   Foreign Address   =
            (state)
tcp4       0            0             192.168.11.11.9991        *.*        =
                               CLOSED

This remains the same throughout, even after completion of krping session. =
I cannot see this information on the krping client system though.

Thanks,
Anand A. K.



Want to link to this message? Use this URL: <http://docs.FreeBSD.org/cgi/mid.cgi?BYAPR07MB48726ED6A7A6EF022113A0B599860>