[PATCH 3/4] ARM: AT91: Add AT91RM9200 support to DT board

Nicolas Ferre nicolas.ferre at atmel.com
Mon Oct 15 04:28:36 EDT 2012


On 10/14/2012 06:39 PM, Joachim Eastwood :
> On Sun, Oct 14, 2012 at 4:54 PM, Jean-Christophe PLAGNIOL-VILLARD
> <plagnioj at jcrosoft.com> wrote:
>> On 19:08 Fri 12 Oct     , Joachim Eastwood wrote:
>>> On Fri, Oct 12, 2012 at 6:27 PM, Jean-Christophe PLAGNIOL-VILLARD
>>> <plagnioj at jcrosoft.com> wrote:
>>>> On 17:28 Fri 12 Oct     , ludovic.desroches wrote:
>>>>> Le 10/12/2012 04:22 PM, Jean-Christophe PLAGNIOL-VILLARD a écrit :
>>>>>> On 00:05 Fri 12 Oct     , Joachim Eastwood wrote:
>>>>>>> Signed-off-by: Joachim Eastwood <manabian at gmail.com>
>>>>>>> ---
>>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> This patch has some potential issues.
>>>>>>> Before this patch board-dt would fail building when only AT91RM9200 was enabled because at91sam926x_timer symbol would be missing. This patch uses the at91rm9200_timer which
>>>>>>> will fail if AT91RM9200 is not enabled.
>>>>>> this need work with ot wtihout rm9200
>>>
>>> btw, to solve the build issue with board-dt in mainline now we need to
>>> add a select CONFIG_SOC_AT91SAM9 to config MACH_AT91SAM_DT.
>>>
>>>>>>>
>>>>>>> Any thoughts on solving this? As mention above this bug exists in mainline now.
>>>>>> duplicate the board-dt with one for rm9200 only
>>>>>> as rm9200 ans sam9 are 2 distict familly
>>>>>
>>>>> Why not adding a new machine descriptor for rm9200 in order to
>>>>> prevent file duplication?
>>>> because the soc are different and can only be compile if the timer is enable
>>>> and I do not want to enable the rm9200 timer on sam9 so instead of a ifdef i
>>>> the board-dt create a new board is better as we have a 50 lines file
>>>>
>>>> with different board_compat and different machine descriptor
>>>
>>> I am okey with either approach, but I would like to hear what Nicolas
>>> Ferre has to say since he is the on the one that added board-dt. It
>>> would be nice to have everything in one board DT file, but I
>>> understand your concern with the RM9200 timer.
>>>
>>> We will also bump into this again on AT91X40 I guess.
>> simple on x40 forget about it the x40 is no MMU only SoC
>>
>> so you can not enable  it by default as the all other at91 are use with MMU
>>
>> I did the necessary to make the board-dt nearly empty and the same for all the
>> sam9 but the board-dt is sam9 only and need to be keeped this way
>>
>> Nico will tell you the same
> 
> okay. I'll make the necessary changes to the patch set.

Well, even if did not see the advantage of a new board-dt specifically
for rm9200 (and armv4 and armv5 may share the same zImage in the
future...), I understand the fact that compiling the rm9200 timer for a
sam9 machine is a bit surprising.

So I have a pretty mixed feelings about that, but I will not go against
Jean-Christophe nor you Joachim: so continue in this path, we will be
able to merge all board-*-dt later if needed anyway...

Bye,
-- 
Nicolas Ferre



More information about the linux-arm-kernel mailing list