Re: [pok-devel] ARINC - partition

[Matias Hastaran <matias.hastaran@xxxxxxxxxxx>]

> > Now, I'm working on ARINC partition's and I would like some
> > enlightenment:
> >  - Can you confirm me POK doesn't use partition period,
> >  start_condition and lock_level values ?
> 
> Can you be more precise ? What do you mean exactly ?


I need to find this 3 values for GET_PARTITION_STATUS function.
For the period, in a comment it said "not used for the moment", is it always relevant ? I don't find the period value in AADL model ... How can I found this value?
Same problem for the start_condition value... I don't find this value elsewhere in the libpok/arinc653... 
For lock_level, my was stupid, you said me POK can't disable preemption and the lock_level value is used to lock preemption in one partition thus I suppose this value doesn't exist..



> >  - Why did you say only error thread can stop or restart a
> >  partition ? If I understand well the ARINC documentation, all
> >  partition processes can change the partition mode.
> 
> As far as I remember, in the ARINC standard:
> - The partition goes from INIT to ACTIVE only by a call from the main
>   thread (the one activated when starting a partition).

I have re-read ARINC doc, I agree with you for this point.


> - The partition go from ACTIVE to INIT or STOP only by the
> error-handler
>   thread. No other thread can change the partition mode.

ARINC said : "The SET_PARTITION_MODE service allows the application associated with a partition to request a change to its operating mode." Thus, for me any partition processes can change the mode.


> I do not really understand what means that POK is more strict than
> ARINC. Would it be possible to be more accurate and explain exactly
> what
> is meant by "more strict" ?

My fault, the question was asked really badly. After studying the question again, the only point bothering me is the COLD_START. You don't use the COLD_START in ARINC. If I understand well, the difference between the WARM_START and the COLD start is the hardware context (e.g., no need for copying code from Non Volatile Memory to RAM). Can POK support the COLD_START ?

 
> Ok, before commiting, would it be possible to review the patch in
> order
> to check the consistency with the initial design ?

Of course, I will recontact you at this moment.

> Thanks for your questions,

Thanks for your answer,
Matias