at91 sama5d3 "regressions"

[Ahmad Fatoum]

Hi,

On 19.09.21 08:32, Peter Rosin wrote:
> On 2021-09-18 01:57, Peter Rosin wrote:
>> On 2021-09-18 00:37, Peter Rosin wrote:
>>> Hi!
>>>
>>> I have old boards that I'm trying to get up to current
>>> revisions of various pieces of software, and I'm running
>>> into a couple of problems with barebox. After bisecting
>>> and sorting though git history for a couple of days,
>>> these are my findings...
>>
>> I meant to mention that I'm using this extra patch for my board.
>>
>> As you can see, it's been a couple of years, and it is not at
>> all unlikely that the root cause is in these files, and that
>> there has been changes in the other parts of the sama5d3 code
>> that I need to pick up.
>>
>> I will dig into updating these files, and check if that help.
> 
> Ok, it turns out I don't need to do that update, because the
> only thing that's actually used from the extra patch are the
> tse850_defconfig file and the environment. But since the
> tse850_defconfig file fails to set CONFIG_MACH_TSE850, the
> actual code in the extra patch is not used (not even compiled).
> Instead, the board code for sama5d3x is what I have been using
> all along. Big blush about that, and sorry for the confusion!

No worries.

> However, the above also means that the quotes around the
> "regressions" word in the subject can be removed. The problems
> are squarely in the upstream code base, and I do need fixes
> for region overlap, eXecute Never and RO_DATA_SECTION.

I am interested in understanding how that breakage came about.
I have a sama5d3_xplained here with NAND, which I will try to
find some time to reproduce this on.

For your immediate issue: I'd like to use the occasion to get
all sama5d3 to probe from device tree. We already have one
sama5d3 DT-enabled board upstream (microchip-ksz9477-evb) and
I just CC'd you on a series to migrate the sama5d3_xplained
from legacy board code to DT. I tested this boots correctly
after at91bootstrap from NAND without your series here.

The relevant defconfig is at91_multi_defconfig. It will generate
a barebox-sama5d3-xplained.img, which should boot to shell
on your board.

If that works, we could create a new barebox-sama5d3xek.img image
that does the same as barebox-sama5d3-xplained.img, but with
different device tree and see if all peripherals you use
still work.


With everything nice and dandy, we could finally remove non-DT
code for sama5d3 altogether and your board would reuse the same
early bootstrap code basically all other platforms use and test.

Sounds good?

Cheers,
Ahmad

> 
> I have this hunch that the region overlap is what triggers
> the other problems, but that's a pretty wild guess...
> 
> Cheers,
> Peter
> 
> _______________________________________________
> barebox mailing list
> barebox at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/barebox
> 


-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |