On 04/18/2012 10:44 AM, Roberto Bielli wrote:
> Hi Gilles,
> here find the answers.
> sorry for the delay.
> Il 13/04/2012 15:47, Gilles Chanteperdrix ha scritto:
>> On 04/13/2012 10:44 AM, Roberto Bielli wrote:
>>> Hi Gilles,
>>> you don't feel like to repeat.
>>> I'm trying all the things that you told me and i will do all the missing
>>> test that you told me.
>> So, what is the result of running your test with unpatched Linux, but
>> with CONFIG_HIGH_RES_TIMERS enabled?
> the result is the same.
> With the base linux application and CONFIG_HIGH_RES_TIMERS all it's ok,
> instead with xenomai application the task with less priority halt the
> task with higher priority until it sleeps.
> I send you my .config file and the test with pthread base.
You know as well as me that this description of the problem is
inaccurate, what really happens is that the timer is programmed but does
not tick. What you have to do now is try and understand what is the
difference between the way xenomai and linux handle the timer to
understand what cause it not to work in xenomai case. If I were you, to
understand this difference, I would start by dumping all timer registers
values at the time when the problem is detected, that is, before the
timer is programmed again. But I am afraid I can not give you more help,
ultimately you are the one which is doing the port to imx25, so you are
on your own.
Xenomai-core mailing list