Hi Geliang,
14 Oct 2023 18:42:28 Geliang Tang <geliang.tang@suse.com>:
> Fix userspace pm errors reported by CI:
>
> [11:10:11.889] # 112 userspace pm add & remove address
> [11:10:29.692] # syn [ ok ]
> [11:10:30.294] # synack [ ok ]
> [11:10:30.686] # ack [ ok ]
> [11:10:31.263] # add [ ok ]
> [11:10:31.653] # echo [ ok ]
> [11:10:33.644] # mptcp_info subflows=1:1 [ ok ]
> [11:10:35.761] # subflows_total 2:2 [ ok ]
> [11:10:36.762] # mptcp_info add_addr_signal=1:1 [ ok ]
> [11:10:44.932] # rm [ ok ]
> [11:10:45.321] # rmsf [ ok ]
> [11:10:45.325] # Info: invert
> [11:10:46.405] # mptcp_info subflows=0:0 [ ok ]
> [11:10:48.048] # subflows_total 1:1 [fail] got subflows 0:0 expected 1:1
> [11:10:48.052] # Server ns stats
> [11:10:48.310] # TcpPassiveOpens 2 0.0
> [11:10:48.315] # TcpInSegs 118 0.0
Thank you for looking at that!
Do you mind explaining why it is fixing the issue and/or why it was not OK before please?
> diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh
> index 01480663c102..4803ba1ab42d 100755
> --- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
> +++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
> @@ -3316,7 +3316,7 @@ userspace_pm_rm_addr()
> tk=$(mptcp_lib_evts_get_info token "$evts")
>
> ip netns exec $1 ./pm_nl_ctl rem token $tk id $2
> - wait_rm_addr $1 1
> + wait_rm_addr $1
From what I understand, this will skip the wait: the number of RM_ADDR will always be different than an empty string.
The second argument represents the number of RM_ADDR before sending the command.
Should it not be 0 issue of 1?
Or better:
local cnt
(...)
cnt=$(rm_addr_count ${1})
ip netns exec $1 ./pm_nl_ctl rem token $tk id $2
wait_rm_addr $1 "${cnt}"
No?
> }
>
> # $1: ns ; $2: addr ; $3: id
> @@ -3352,7 +3352,7 @@ userspace_pm_rm_sf()
>
> ip netns exec $1 ./pm_nl_ctl dsf lip $2 lport $sp \
> rip $da rport $dp token $tk
> - wait_rm_sf $1 1
> + wait_rm_sf $1
Same here, you need to pass a second argument to the function otherwise it does nothing. So same as suggested above but with "rm_sf_count" instead.
But I don't know if this will fix the issue.
Cheers,
Matt
> }
>
> userspace_tests()
> --
> 2.35.3
--
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net