Re: [Xenomai] [PATCH] rtcan: driver for the Advantech PCI cards

[Thierry Bultel <thierry.bultel@wanadoo.fr>]

Le 11/10/2012 09:08, Gilles Chanteperdrix a écrit :
> On 10/11/2012 08:43 AM, Thierry Bultel wrote:
>
>> Why v3.2 from Freescale? For the SabreLite or Nitrohen6x I know about
>>>> 3.0.36 from Freescale (rel_imx_3.0.35_12.09.02 from
>>>> git.freecale.com/linux-2.6) or, much better, mainline Linux 3.4.
>>>> Concerning the iPipe patches I think you have currently no other choice.
>>
>> In both cases, (latest one from Freescale, or 3.4 from mainline) there is currently
>> no available matching ipipe patch.
>> I do not know if the latest kernel from mainline will fit for the reference board that I will receive.
>>
>> I do not realize the amount of work for porting either
>> - ipipe-core-3.2.21-arm.patch to 3.4 (mainline)
>> or
>> - adeos-ipipe-3.0.43-* to the latest (which one ?) kernel from Freescale
>>
>> About that latest option I am a little bit puzzled.
>> Namely from what I know, Freescale kernels are on
>>
>> http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/refs/heads
>> and the latest published 3.x kernel is ... a 3.0.15_12 that has not moved for months.
>> Thanks for clarifying
>
> The 3.4 patch is working for arm, I still have some issues, but it is
> already usable:
>
> git://git.xenomai.org/ipipe-gch.git branch for-core-3.4
>
> For now, you simply have to avoid THUMB user-space binaries (but you can
> compile the kernel with CONFIG_THUMB2_KERNEL, and it results in lower
> latencies from what I measured).
>
> It works on the sabre board "out-of-the-box". I suspect, starting with
> core-3.4, we have no additional work to do to work with any cortex a9
> based processor (except, as I already said, for some old cortex a9
> revisions without a global timer, but that is not the case of the imx6q
> I tested).
>

AFAIK, my DevKit should finally come with a 3.0.35 kernel
How difficult would it be to apply the adeos-ipipe-3.0.43- patches on it ?
It is feasible ?

Thanks
Thierry