On 04/18/2012 12:28 PM, Roberto Bielli wrote:
> Hi Gilles,
>
> i try to give only another accurate information.
>
> i made another test:
> i write a xenomai driver for reading the timer and avic register and i
> see a strange behaviour.
> when there is the problem the interrupt is NOT masqueraded in avic but
> the timer has the interrupt not acked.
Ah, that is good news, at least now we have a clue. Note that it could
also mean that the timer has ticked, but the irq is not generated for
another reason. I think you should dump all the timer registers, and see
if anything is wrong by decoding their values with the datasheet.
A quick test also, assuming that the write to the MX3_TSTAT regiter in
gpt_irq_acknowledge may be posted, try rereading the register after
writing it.
>
> After this i understand that now are my problems.
> Do you know somebody that can help me understanding the problem only for
> xenomai side?
If you need someone's help, you need to give access to your board to
that person or company. Because indirect debugging is really
inefficient: look at how many mails I had to send you over how much time
to get you to do what anybody with xenomai experience would have done in
just a few hours.
--
Gilles.
_______________________________________________
Xenomai-core mailing list
Xenomai-core@gna.org
https://mail.gna.org/listinfo/xenomai-core
