Re: [PATCH bpf-next 1/4] selftests/bpf: Handle SIGINT when creating netns

[Matthieu Baerts <matttbe@kernel.org>]

Hi Alexei,

Thank you for the review!

On 07/05/2024 16:43, Alexei Starovoitov wrote:
> On Tue, May 7, 2024 at 3:53 AM Matthieu Baerts (NGI0)
> <matttbe@kernel.org> wrote:
>>
>> From: Geliang Tang <tanggeliang@kylinos.cn>
>>
>> It's necessary to delete netns during the MPTCP bpf tests interrupt,
>> otherwise the next tests run will fail due to unable to create netns.
>>
>> This patch adds a new SIGINT handle sig_int, and deletes NS_TEST in it.
>>
>> Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
>> Reviewed-by: Mat Martineau <martineau@kernel.org>
>> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
>> ---
>>  tools/testing/selftests/bpf/prog_tests/mptcp.c | 7 +++++++
>>  1 file changed, 7 insertions(+)
>>
>> diff --git a/tools/testing/selftests/bpf/prog_tests/mptcp.c b/tools/testing/selftests/bpf/prog_tests/mptcp.c
>> index 274d2e033e39..baf976a7a1dd 100644
>> --- a/tools/testing/selftests/bpf/prog_tests/mptcp.c
>> +++ b/tools/testing/selftests/bpf/prog_tests/mptcp.c
>> @@ -64,11 +64,18 @@ struct mptcp_storage {
>>         char ca_name[TCP_CA_NAME_MAX];
>>  };
>>
>> +static void sig_int(int sig)
>> +{
>> +       signal(sig, SIG_IGN);
>> +       SYS_NOFAIL("ip netns del %s", NS_TEST);
>> +}
>> +
>>  static struct nstoken *create_netns(void)
>>  {
>>         SYS(fail, "ip netns add %s", NS_TEST);
>>         SYS(fail, "ip -net %s link set dev lo up", NS_TEST);
>>
>> +       signal(SIGINT, sig_int);
>>         return open_netns(NS_TEST);
> 
> That's a drop in the bucket.
> ctrl-c of test_progs doesn't really work.
> Such clean up needs to be generic as part of network_helpers.c

It makes sense. I can drop this patch and ask Geliang to add a similar
'create_netns()' helper in network_helpers.c creating the netns, and
handling SIGINT. This helper will no longer be specific to MPTCP BPF
selftests then.

Cheers,
Matt
-- 
Sponsored by the NGI0 Core fund.