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

[James Cameron]

On Sat, May 19, 2012 at 07:13:27PM +1200, Tom Parker wrote:
> On 19/05/12 17:31, James Cameron wrote:
> 
> > The audio driver in the kernel could be changed, yes, but "flag that
> > the microphone is mono" is ambiguous.  I don't know whether the kernel
> > should duplicate the samples from one channel to another when an
> > application asks for stereo input, or refuse to set up a two channel
> > input stream.
> 
> My guess is applications either understand the mono flag and request
> a mono input stream or don't understand and will behave best if it
> has largely the same signal on both channels. So I guess my naive
> approach would be to tell the applications that the hardware only
> supports mono and simulate stereo by sending both channels the same
> data. Of course I've never done any audio related development so you
> probably shouldn't listen to me.

Sounds like something to be discussed between kernel and activity
people.

> On Moa with Q4D10 and 
> 
> > - with the device connected, does the laptop freeze if the escape
> >    key is pressed to reach the ok prompt during the countdown?
> 
> No, the OK prompt works.
>  
> > - is there any serial output that corresponds with the laptop
> >    freezing?
> 
> I haven't hooked up the serial port yet.
>  
> > - if you can get to the ok prompt, try "p2" or "dir u:\" or "boot"
> >    and tell me which of these freeze the laptop.
> 
> p2 reports:
> 
> USB devices:
> /usb at d420800/hub at 0,0
> /usb at d420800/hub at 0,0/device at 4,1
> /usb at d420800/hub at 0,0/scsi at 4,0
> /usb at d420800/hub at 0,0/scsi at 4,0/disk
> 
> dir u:\ freezes

Ah, good data, thanks.

Please reflash to Q4D14, connect a serial console, then run through
this test sequence:

a.  use "no-page" to turn off the output pager,

b.  use "dir u:\" to freeze again,

c.  press the interrupt key (this is the top right key on the
keyboard, for all types of keyboard), and the ok prompt should appear,

d.  type "ftrace" and press enter.  A stack trace at the point of the
hang should be printed.

If the ok prompt does not appear on keyboard interrupt, we must dig
deeper.

> This behaviour was the same on Q4D13 on Kiwi and Q3C06 on Ivy (p2
> was different but still returned 3 entries associated with the
> phone) and Q2E48 on Moodle (p2 different again but still 3 entries
> associated with the phone).
> 
> The phone isn't presenting a disk, it's presenting an ADB connection
> and maybe some other services, but I haven't enabled it's mass
> storage option.
> 
> lsusb -v on Linux reports the following:
> 
> Bus 001 Device 100: ID 12d1:1038 Huawei Technologies Co., Ltd. Ideos (debug mode)
> Device Descriptor:

[...]

> The serial port that hangs on boot is:
> 
> ID 067b:2303 Prolific Technology, Inc. PL2303 Serial Port
> 
> and it appears to hang in the kernel:
> https://plus.google.com/photos/118146064923365210751/albums/5468706263336090945/5744135759942268450?banner=pwa

Thanks, I can reproduce this, I'll look into it.  #11871.  There's a
"Data Abort" reported on the serial console, and ftrace output looks
useful.

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