ac100 team mailing list archive

[Jani Monoses <jani@xxxxxxxxxx>]

Marc,
your observation of toolchain changing was spot on.

I built the latest 3.0 for armel and armhf (so those with precise
armhf can test too) using gcc 4.5

http://startx.ro/~jani/

Others please test :)

And it boots fine.
Rebuilt same tree with 4.6 and locks up with Toshiba logo.

I am not sure if it is an unrelated toolchain issue, or only triggered
along with the RAM timing change, but I suspect
it is toolchain only. I rebuilt a 2.6.38 too with the 4.6 toolchain
and it did not boot.

I'll need to back that change up and rebuild again to make sure.

So I had the deb working locally as I did not yet upgrade to 4.6.2 but
when built in the archive it resulted in an unbootable image.

Jani

On Thu, Dec 15, 2011 at 9:20 AM, Marc Dietrich <marvin24@xxxxxx> wrote:
> On Thursday 15 December 2011 01:33:14 Jani Monoses wrote:
>> I made a new deb from latest git branch but now it no longer boots on
>> my machine.
>> I also have hynix RAM, so it is even more strange that the Dec 6
>> snapshot booted.
>
> also fails here. Kernel boots with my config (paz00_defconfig) and your configs
> didn't changed as it seems. On the other hand, you seem to have changed the
> toolchain: gcc 4.6.1 -> 4.6.2. Do you still have the old compiler around? Is
> this compiled on a hf system?
>
> Marc
>
>>
>> Jani
>>
>> http://startx.ro/~jani/linux-image-3.0.8-2-ac100_3.0.8-2.1_armel.deb
>>
>> On Wed, Dec 14, 2011 at 8:02 PM, Jani Monoses <jani@xxxxxxxxxx> wrote:
>> > Thanks for the feedback, good to know.
>> > Others have reported it not booting at all on armhf builds, but if it
>> > is memory related they'll likely hit
>> > the same issue on armel.
>> >
>> > I'll roll a new package for the more recent tree these days then, once
>> > your change is in.
>> >
>> > Jani
>> >
>> > On Wed, Dec 14, 2011 at 7:54 PM, Marc Dietrich <marvin24@xxxxxx> wrote:
>> >> Hi Jani,
>> >>
>> >> thanks for the new kernel!
>> >>
>> >> On Wednesday 14 December 2011 01:38:51 Jani Monoses wrote:
>> >>> >> However we should aim for stability and no churn towards the
>> >>> >> end of
>> >>> >> the cycle(March/April)
>> >>> >> Are there any known regressions vs 2.6.38? Is sound not
>> >>> >> working yet?
>> >>> >
>> >>> > sound works as on 2.6.38 (maybe better, no mic support yet). No
>> >>> > regressions I know of by now (but I'm a bad tester). It is
>> >>> > possible
>> >>> > that people with hynix memory may not be able to boot. Some IRC
>> >>> > user
>> >>> > wanted to test it this evening. If it doesn't work, I'll disable
>> >>> > the
>> >>> > memory scaling feature again.
>> >>>
>> >>> Here's a 3.0 deb of the snapshot of a week ago
>> >>>
>> >>>
>> >>> http://startx.ro/~jani/linux-image-3.0.8-1-ac100_3.0.8-1.1_armel.deb
>> >>>
>> >>>
>> >>> For me resume does not work on a stock Ubuntu Precise install, which
>> >>> worked with the default 2.6.38 kernel.
>> >>> I get noise on the screen, the machine does not answer key presses
>> >>> or pings.>>
>> >> I tested it on Oneiric (armel). It boots and suspends/resumes without
>> >> problems. One possible issue is that the kernel likely fails on hynix
>> >> ram based system as I only fixed this issue on Dec. 9th  and the
>> >> kernel is form Dec. 6th.
>> >>
>> >> The resume problems you see may also be RAM clocking related (the
>> >> noice on the screen). So I better disable the stuff alltogether as it
>> >> seems to create a lot of problems. I will push a fix in a minute ...
>> >>
>> >> Marc
>> >>
>> >>> commit 735f939f421c6c9c5db79250955e7b25ee272e0f
>> >>> Author: Marc Dietrich <marvin24@xxxxxx>
>> >>> Date:   Tue Dec 6 20:35:30 2011 +0100
>> >>>
>> >>>     ARM: tegra: paz00: make sound modular
>> >>>
>> >>>     Switch the static sound driver from chromeos to modular version
>> >>> used by 2.6.38 kernels.
>> >>>
>> >>>     Signed-off-by: Marc Dietrich <marvin24@xxxxxx>