Re: [PATCH] testsuite/smokey/alchemytests: Relax mq-2 test over VM

On 08.05.26 09:43, Florian Bezdeka wrote:
> On Thu, 2026-05-07 at 14:53 +0200, Jan Kiszka wrote:
>> On 07.05.26 14:35, Florian Bezdeka wrote:
>>> On Thu, 2026-05-07 at 14:23 +0200, Jan Kiszka wrote:
>>>> On 07.05.26 13:03, Florian Bezdeka wrote:
>>>>> On Wed, 2026-05-06 at 12:26 +0200, Jan Kiszka wrote:
>>>>>> From: Jan Kiszka <[email protected]>
>>>>>>
>>>>>> VM environment may delay or reorder the execution of tasks, so relax the
>>>>>> checks.
>>>>>>
>>>>>> Part of #52.
>>>>>>
>>>>>> Signed-off-by: Jan Kiszka <[email protected]>
>>>>>> ---
>>>>>>  testsuite/smokey/alchemytests/mq-2.c | 8 ++++++--
>>>>>>  1 file changed, 6 insertions(+), 2 deletions(-)
>>>>>>
>>>>>> diff --git a/testsuite/smokey/alchemytests/mq-2.c b/testsuite/smokey/alchemytests/mq-2.c
>>>>>> index a389e46db0..09b26b2135 100644
>>>>>> --- a/testsuite/smokey/alchemytests/mq-2.c
>>>>>> +++ b/testsuite/smokey/alchemytests/mq-2.c
>>>>>> @@ -7,6 +7,7 @@
>>>>>>  #include <alchemy/queue.h>
>>>>>>  
>>>>>>  static struct traceobj trobj;
>>>>>> +static bool on_vm;
>>>>>>  
>>>>>>  static int tseq[] = {
>>>>>>  	3, 4, 5, 6,
>>>>>> @@ -87,9 +88,10 @@ static void main_task(void *arg)
>>>>>>  
>>>>>>  	ret = rt_queue_inquire(&q, &info);
>>>>>>  	traceobj_check(&trobj, ret, 0);
>>>>>> -	traceobj_assert(&trobj, info.nmessages == 0);
>>>>>> +	traceobj_assert(&trobj, info.nmessages == 0 || on_vm);
>>>>>>  
>>>>>> -	traceobj_verify(&trobj, tseq, sizeof(tseq) / sizeof(int));
>>>>>> +	if (info.nmessages == 0)
>>>>>> +		traceobj_verify(&trobj, tseq, sizeof(tseq) / sizeof(int));
>>>>>>  
>>>>>>  	traceobj_exit(&trobj);
>>>>>>  }
>>>>>> @@ -99,6 +101,8 @@ int main(int argc, char *const argv[])
>>>>>>  	RT_TASK t_main;
>>>>>>  	int ret;
>>>>>>  
>>>>>> +	on_vm = argc > 1 && strcmp(argv[1], "--vm") == 0;
>>>>>> +
>>>>>
>>>>> Seems a bit risky to expect --vm as first argument only.
>>>>>
>>>>
>>>> It's not a new pattern, and I thought (please prove me wrong) that this
>>>> is already a filtered argv, freed from generic switches.
>>>
>>> Yes, that seems to be the case, but the filtering seems questionable:
>>>
>>> Taken from run_alchemytests():
>>>
>>> 	tmp = smokey_run_extprog(XENO_TEST_DIR, tests[i],
>>> 				 smokey_on_vm ? args : args + 5,
>>> 				 &test_ret);
>>>
>>> Where args is
>>>
>>> 	const char *args = "--vm";
>>>
>>> args + 5 is pointing to garbage in case smokey_on_vm is false. Right?
>>
>> Yeah, went unnoticed in via 61fc9ff4b7214fb1d9802aa7d177a555f0ad47e3.
>> Please fix and also check the other tests at that chance.
>>
> 
> testsuite/smokey/psostests/mq-3.c has also a check for --vm in argv[1]
> but never got this cmdline parameter...
> 
> Should I remove the check or fix the cmdline arg invocation? Were there
> any issues seen with this specific test (in CI) in the past?

That's a fail of my recent patch, you already commented on it. Still in
next, let me bake a v2.

Jan

-- 
Siemens AG, Foundational Technologies
Linux Expert Center