[OLPC New Zealand] [Testing] Testing Summary, Auckland - 5 May 2012

[James Cameron]

Thanks for testing!

On Sun, May 13, 2012 at 07:02:05PM +1200, Tom Parker wrote:
> Testing Summary, Auckland - 5 May 2012
> Who: Fabiana, John, Tom
> 
> We tested 12.1.0 for XO-1.75, customized (build 9) and 12.1.0 for
> XO-1.5, customized (build 9)
> 
> nine, an XO-1.75:
> Measure:
> On Nine, an XO-1.75: Seems to work ok, except that every time that I
> click on voltage sensor and then I go back to microphone it swaps
> which of the two coloured channels displays the input from
> environment sound. Eg, I start with a flat orange line, and a wiggly
> blue line, and then after hitting voltage sensor and back to the
> microphone icon then the orange is wiggly and the blue is
> flat. Eventually it froze. 

Thanks.  I saw this last week in Audacity, nice to see it confirmed in
Measure.  See ticket #11842.  I've made a note there.

> On Hoiho, an XO-1.75: I don?t see the swapping between the two
> colour channels. Managed to freeze it by clicking on the different
> subitems 

With #11842 my tests showed the swapping, if it was going to occur,
would only occur on the first use of the input audio path since boot.
So you might be able to reproduce on Hoiho by rebooting first.  If it
does not reproduce, this will be interesting for #11842, because it
may mean the problem is laptop specific.

> On Rosella an XO-1.5: Both channels are connected to the microphone
> - whereas in Hoiho and Nine both channels are visible but only one
> seems to have data.

On the XO-1.5, the internal microphone is fed to both audio inputs on
the codec chip.

On the XO-1.75, the internal microphone is only fed to the left
channel input on the codec chip.  The right channel should have no
audio detected when there is no external microphone plugged in.

Do you think that our kernel should behave as if both channels have
the same data on XO-1.75?

> However, it is odd that in the FFT the trace for yellow and orange are
> slightly different - I assume that this is because both channels are
> alternating in data capture? I also had to start it twice - the first
> time I started I just got heaps of noise. Managed to crash this too.

I have no explanation for either of these two observations.

> On Kiwi, an XO-1.75: Froze this by swapping between the voltage and
> microphone and fft modes repeatedly and quickly. See attached for
> GDB backtrace and log.

I'd like the activity team to look at this one first.

> All the XO-1.75s are very slow to shut down, usually taking more
> than a minute. The last message on alt f2 was ?store sound card
> state [ok]?

I see this too.  It shows up also when I'm switching run levels.  I
have no explanation.

> Hoiho does not boot if an android phone (with cyanogen mode and USB
> Debugging turned on) is plugged in, holding down checkmark we see
> the normal openfirmware banner and type the esc key to interrupt
> message and the 2 - 1 and then the laptop freees. I?ve also seem
> something similar with a USB Serial adaptor but haven?t explored
> this.

This sounds interesting.  Can you work with me on that during the
week?  Will need to know firmware version you have on Hoiho.

> Poppy Copy, an XO-1.5 B2 doesn?t boot -- it complains about read
> only filesystems in the initramfs. We have reported this before, but
> still haven?t looked at exactly what is going on.

The host name is set from the wireless card MAC address.  Poppy Copy
had a dodgy wireless card connect, check to see if it has gone dodgy
again?  The method of anchoring the wireless card was changed after
this point in development, so if it is caused by the mounting the
problem is slightly uninteresting.

-- 
James Cameron
http://quozl.linux.org.au/