[fedora-arm] Pandaboard ES and Fedora

[Peter Robinson]

On Fri, Dec 20, 2013 at 12:12 PM, Peter TB Brett <peter at peter-b.co.uk> wrote:
> On Fri, 20 Dec 2013 12:21:14 +0000, Peter Robinson <pbrobinson at gmail.com>
> wrote:
>
>>> I know that the topic was previously discussed briefly on this list back
>>> in
>>> August, but I was wondering whether anyone's aware of any work being
> done
>>> on Fedora on Pandaboard or any progress being made since then.  I'd
>>> especially be interested in any info that's been gathered that points to
>>> where, exactly, the problem lies...
>>
>> Oh I wish there was news or an answer, believe me if there was I would
>> be screaming about it here and elsewhere.
>>
>> omap4/panda in general is currently is a terrible state since 3.11 as
>> it was converted from a traditional board file to DT without any
>> change over period and the later IMO was poorly tested by those doing
>> it. Panda went from hero to zero in Fedora overnight.
>>
>> I've got two different Panda ES boards. One is prototype and one is
>> the last rev and neither work, nor does the original Panda and ES that
>> pwhalen has. I keep poking at it randomly with test kernels of random
>> bits as time permits but I'm yet to get any working combination but I
>> have limited time. Rest assured if I get it working I'll be posting
>> the details of what kernel is needed here.
>>
>
> Okay, thanks for the info -- although I was quite hoping to receive
> slightly better news than that.

You and me too!

> Have you by any chance collated any information on what exactly "not
> working" entails during your experiments?  I'd really like to avoid
> duplicating any work that other people have already done.  At the moment
> I'd be happy just to get the kernel to boot to a shell on a serial
> console...!

Well the current status for 3.11+ is that there's no output on the
serial console, it doesn't boot, or at least appear to boot with
anything on any console.

> I haven't played with DT before.  I was going to suggest paring down the DT
> to the bare minimum (power, memory, SD card) and seeing if the "core
> system" can be got working, and then worrying about peripherals later, but
> that might be something that (a) wouldn't work and (b) has already been
> tried.

Not that simple unfortunately. There's clocks and all sorts of other
bits that are needed to boot the system with DT, ultimately paring it
down in that regard will achieve nothing, it either boots or it
doesn't and extra things like audio or wifi or whatever in the DT are
going to do little to affect the outcome there.

> Are you still stuck on the memory chip issues, or have those now been
> fixed?

Well at least for me on the ES I never had that working on the
3.7-3.10 releases that would vaguely sort of attempt to boot on my
devices. In theory the DT should possibly resolve those issues but as
we're not getting even to the point I had with 3.7-3.10 it's not under
consideration. From my PoV we need to get to the point where we're at
least getting console output.

I poke at it randomly but my time is limited and there's a lot newer
platforms that are also needing my time.

Peter