Re: [PATCH mptcp-next v9 5/5] selftests: mptcp: add mp_fail testcases

From: Matthieu Baerts <matthieu.baerts@tessares.net>
To: Mat Martineau <mathew.j.martineau@linux.intel.com>,
	Geliang Tang <geliang.tang@suse.com>
Cc: mptcp@lists.linux.dev, Davide Caratti <dcaratti@redhat.com>
Subject: Re: [PATCH mptcp-next v9 5/5] selftests: mptcp: add mp_fail testcases
Date: Thu, 13 Jan 2022 11:36:41 +0100	[thread overview]
Message-ID: <8d549123-341e-c037-0313-8c1380807b21@tessares.net> (raw)
In-Reply-To: <ea44a519-11b9-175-cceb-abb69232abbd@linux.intel.com>

Hi Mat,

On 13/01/2022 02:06, Mat Martineau wrote:
> On Mon, 10 Jan 2022, Geliang Tang wrote:
> 
>> Added the test cases for MP_FAIL, use 'tc' command to trigger the
>> checksum failure.
>>
>> Suggested-by: Davide Caratti <dcaratti@redhat.com>
>> Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
>> Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
>> Signed-off-by: Geliang Tang <geliang.tang@suse.com>
>> ---
>> tools/testing/selftests/net/mptcp/config      |   5 +
>> .../testing/selftests/net/mptcp/mptcp_join.sh | 111 ++++++++++++++++--
>> 2 files changed, 107 insertions(+), 9 deletions(-)
>>
> 
> Does anyone else see the "leaked reference" errors when running these
> fail tests with the latest export branch? Looks like the reference
> tracker finds a tc-related error:
> 
> [  238.871372] leaked reference.
> [  238.872393]  fl_change+0x2db/0x2520
> [  238.873148]  tc_new_tfilter+0x6c4/0x11f0
> [  238.873959]  rtnetlink_rcv_msg+0x49f/0x640
> [  238.874798]  netlink_rcv_skb+0xc4/0x1f0
> [  238.875570]  netlink_unicast+0x2d5/0x410
> [  238.876364]  netlink_sendmsg+0x3b3/0x6c0
> [  238.877155]  sock_sendmsg+0x91/0xa0
> [  238.877890]  ____sys_sendmsg+0x3ad/0x3f0
> [  238.878684]  ___sys_sendmsg+0xdb/0x140
> [  238.879354]  __sys_sendmsg+0xae/0x120
> [  238.879947]  do_syscall_64+0x3b/0x90
> [  238.880683]  entry_SYSCALL_64_after_hwframe+0x44/0xae
> 
> It doesn't look like MPTCP is to blame, but I'm curious if others see
> the same.

The public CI didn't report that:

  https://api.cirrus-ci.com/v1/task/5851145250799616/logs/test.log

We can see a lot of TC warnings that still need to be fixed:

  tc action pedit offset 162 out of bounds

I guess we need to change the test to send data mainly in one direction
to avoid all these pure ACKs causing issues with the TC filter and
probably causing the corresponding test to fail.

For more details about the results from the CI:

===== 8< =====
Our CI did some validations and here is its report:

- KVM Validation: normal:
  - Unstable: 1 failed test(s): selftest_mptcp_join 🔴:
  - Task: https://cirrus-ci.com/task/4725245343956992
  - Summary:
https://api.cirrus-ci.com/v1/artifact/task/4725245343956992/summary/summary.txt

- KVM Validation: debug:
  - Unstable: 2 failed test(s): selftest_diag selftest_mptcp_join 🔴:
  - Task: https://cirrus-ci.com/task/5851145250799616
  - Summary:
https://api.cirrus-ci.com/v1/artifact/task/5851145250799616/summary/summary.txt

Initiator: Patchew Applier
Commits:
https://github.com/multipath-tcp/mptcp_net-next/commits/274e87003fe5
===== 8< =====

Cheers,
Matt
-- 
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net