[fedora-arm] Announcing Fedora 18 Beta for Allwinner A10 based devices

[Hans de Goede]

Hi,

On 01/17/2013 10:12 AM, Hans de Goede wrote:
> Hi,
>
> On 01/17/2013 09:54 AM, Daniel Veillard wrote:
>> On Thu, Jan 17, 2013 at 08:01:18AM +0000, Peter Robinson wrote:
>>> On Thu, Jan 17, 2013 at 5:51 AM, Daniel Veillard <veillard at redhat.com> wrote:
>>>>    How do I find out the stuck process leading to this extra load ?
>>>
>>> try atop. I find it gives a lot of extra information about processes
>>
>>    Hum, compared to a normal 'desktop' box, it list one process
>> as being in #tslpu state, i.e. sleep ininteruptible, and I wonder if
>> that's what is accounted for it. There is basically only 2 processes
>> active in the system: atop and ksoftirqd/0 which is basically the
>> kernel interrupt handler (well there is also ssh but it usually doesn't
>> even shows up).
>>    The process in D state is apparently a kernel usb thread
>>
>> root    18  0.0  0.0      0     0 ?        D    Jan16   0:00 [usb-hardware-sc]
>>
>> it is present in that state on both boxes (no USB device connected, but
>> if I plug mouse or keyboard they work just fine).
>
> Interesting. I'm seeing the same issue on 1 of my a10 devices, but not
> on another (and I have yet to try the rest). The one which exhibits the problem
> also has issues connecting to wifi (it contains an usb wifi adapter like most
> A10 devices).
>
> I'll investigate this further.

Scrap that turns out to be a different issue, the one a10 system I had showing
similar symptoms had a stuck interrupt caused by the ax20x pmu chip, after a
lot of debugging I decided to just put a multi-meter on the irq pin of the chip,
and indeed it read 0 volt (it is low active). Then I measured again and it was
2.5 volt, after that the system magically started working ...  seems like either
a hw glitch, or pcb issue which applying the multi-meter wires measuring pin /
needle (what is the correct English term for that?) "fixed". Either way the stuck
interrupt is gone now while running the exact same software ...

So back to the cubie board load issue. It seems this is something else which I
cannot reproduce. Luckily I'll have a cubie board myself soon, and then I'll
investigate further.

Regards,

Hans