RES: rtems_clock_set @ sparc v7 does not work!

[Fabrício de Novaes Kucinskis]

Does the SPARC BSP for RTEMS 4.6.X have mkprom?

I know the 4.0.0 version had, but when I built the 4.6.4 version, there was
no mkprom!
(so, I had to learn how to create the boot image editing the linker script
and using objcopy)

Or are you using the mkprom that comes with RTEMS 4.0.0 to generate an image
to a "4.6.4" application. If this is the case, I don't think this will work.


Fabrício.


-----Mensagem original-----
De: Grigori Khmyrov [mailto:grigori_khmyrov at uml.edu]
Enviada em: quinta-feira, 1 de junho de 2006 10:56
Para: Joel Sherrill
Cc: rtems-users at rtems.com
Assunto: Re: rtems_clock_set @ sparc v7 does not work!


I can compile my project in 4.6.2 SPARC BSP and use MKPROM to get image,
but the binary I create doesn't boot.
I think one parameter missing in the new MKPROM "-Ttext 0x1000000" to
prepare
image for EEPROM.

Thanks for comments,
Grigori

Joel Sherrill wrote:

> Grigori Khmyrov wrote:
>
>> Yes, it would be reasonable explanation!
>> I use RTEMS 4.0.0 for the sparc BSP
>> And 4.6.2 for i386 BSP (which works)
>>
>> How can I patch "memory barrier" for version 4.0.0?
>> I can not switch easy to the new rtems version because of mkprom
>> support.
>>
> If you are using 4.0.0 and having problems, then I really can't begin to
> evaluate what is wrong.  That is a very very old version and the gcc you
> are using is very likely too old to even trip the memory barrier problem.
>
> Why is mkprom an issue?  There has to be a way to do the equivalent
> functionality after a tool upgrade.  Even if you have to use objcopy and
> then run the old mkprom.
>
> --joel
>
>> Thank you,
>> Grigori
>>
>>
>> Joel Sherrill wrote:
>>
>>> Grigori Khmyrov wrote:
>>>
>>>> When I set new time at sparc v7 board using rtems_clock_set(
>>>> &newTime ) function,
>>>> at some point ticks jump back to previous value!?
>>>>
>>>> It does not happen if I use rtems_clock_set twice:
>>>> rtems_clock_set
>>>> rtems_clock_set
>>>>
>>> What version of RTEMS are you using?  Does it have the "memory
>>> barrier" patch?
>>>
>>> This sounds like the interrupt critical section issue that was
>>> addressed by that patch.
>>>
>>>> I have same code running at i386 processor and rtems_clock_set is
>>>> working OK.
>>>>
>>>> Any ideas?
>>>>
>>>> Thank you,
>>>> Grigori
>>>>
>>>
>>
>