[LTP] report a fail about udp6_ipsec*

[Alexey Kodanev]

On 05/19/2018 09:28 AM, Sun Lianwen wrote:
> Hi Petr
> 
>    I have a question about the udp6_ipsec*. the testcase udp6_ipsec03,
> 
> udp6_ipsec05,udp6_ipsec09 ... test fail on RHEL6.10Beta. and report the
> 
> same error message. "send(7, 0x7fca636badb0, 65527, 16384) failed: EMSGSIZE". 
> 
> I'm not sure whether is means the send() do not support send this size 
> 
> message on the rhel6.10beta or the testcase have some problem.
> 
>   Can you give me some advise.
> 

It would be good if you get results for Mainline kernel as well, otherwise
it is difficult to say where is the problem. May be the kernel there missing
some related fixes.

> 
> man send
> 
> EMSGSIZE
>       The  socket type requires that message be sent atomically, and the
>       size of the message to be sent made this impossible
> 
> the rhel6.10beta do not support fastopen.

The test is using UDP, not related to TCP Fastopen.


> 
> the test result is below:
> <<<test_output>>>
> udp_ipsec 1 TINFO: Network config (local -- remote):
> udp_ipsec 1 TINFO: eth1 -- eth1
> udp_ipsec 1 TINFO: 192.168.1.74/24 -- 192.168.1.76/24
> udp_ipsec 1 TINFO: fd00:1:1:1::1/64 -- fd00:1:1:1::2/64
> udp_ipsec 1 TINFO: run server 'netstress -T udp -R 500000 -B /tmp/ltp-t29eH8gszB/udp_ipsec.uAhB4jpE2Y'
> udp_ipsec 1 TINFO: run client 'netstress -l -T udp -H fd00:1:1:1::2 -n 100 -N 100 -a 2 -r 500 -d tst_netload.res -g 42349'
> udp_ipsec 1 TPASS: netstress passed, time spent '31' ms
> udp_ipsec 2 TINFO: run server 'netstress -T udp -R 500000 -B /tmp/ltp-t29eH8gszB/udp_ipsec.uAhB4jpE2Y'
> udp_ipsec 2 TINFO: run client 'netstress -l -T udp -H fd00:1:1:1::2 -n 1000 -N 1000 -a 2 -r 500 -d tst_netload.res -g 59313'
> udp_ipsec 2 TPASS: netstress passed, time spent '43' ms
> udp_ipsec 3 TINFO: run server 'netstress -T udp -R 500000 -B /tmp/ltp-t29eH8gszB/udp_ipsec.uAhB4jpE2Y'
> udp_ipsec 3 TINFO: run client 'netstress -l -T udp -H fd00:1:1:1::2 -n 65527 -N 65527 -a 2 -r 500 -d tst_netload.res -g 37520'


I think we should change the test a little bit here and add some
intermediate points, i.e. use -s "100 1000 2000 10000 65527"


> tst_test.c:1015: INFO: Timeout per run is 0h 50m 00s
> netstress.c:891: INFO: connection: addr 'fd00:1:1:1::2', port '37520'
> netstress.c:892: INFO: client max req: 500
> netstress.c:893: INFO: clients num: 2
> netstress.c:898: INFO: client msg size: 65527
> netstress.c:899: INFO: server msg size: 65527
> netstress.c:912: INFO: max timeout errors 12
> netstress.c:944: INFO: using UDP
> netstress.c:477: INFO: Running the test over IPv6
> safe_net.c:176: BROK: netstress.c:331: send(6, 0x7fca62cb9db0, 65527, 16384) failed: EMSGSIZE
> safe_net.c:176: BROK: netstress.c:331: send(7, 0x7fca636badb0, 65527, 16384) failed: EMSGSIZE
> 
> Summary:
> passed   0
> failed   0
> skipped  0
> warnings 0
> udp_ipsec 3 TFAIL: expected 'pass' but ret: '1'
> 
> Thanks you,
> Lianwen.
> 
> 
>