[Gta04-owner] Modem reenummerating (was: Re: Crowdfunding an Ubuntu smartphone)

[Neil Jerram]

On 2013-08-28 16:48, Dr. H. Nikolaus Schaller wrote:
> Am 28.08.2013 um 17:30 schrieb Radek Polak:
> 
>> On Wednesday, August 28, 2013 05:15:48 PM Dr. H. Nikolaus Schaller
>> wrote:
>> 
>>> Am 28.08.2013 um 16:50 schrieb Radek Polak:
>>>> On Wednesday, August 28, 2013 04:31:42 PM Dr. H. Nikolaus
>> Schaller wrote:
>>>> > One more question: does it happen only when an incoming call
>> starts
>>>> > (RING) or SMS is received. Or does it happen during a call? Or
>> does it
>>>> > depend on a call being answered (ATA)?
>>>> 
>>>> I am not using SMS very much, so cant tell. For me it happens
>> when the
>>>> other side hangs up - so no need to answer the call.
>>> 
>>> Ok, good to know.
>>> 
>>> So it appears to be tied to the RING mechanism since:
>>> * outgoing calls are not affected
>>> * no need to answer the call
>>> * may or may not happen on SMS

I've seen many reenumerations - for example, whenever travelling by 
train and with 3G enabled - that I strongly doubt were caused by 
incoming calls, so I don't think RING can be the only possible cause.

To be precise, I mean whatever it is that is detected by QtMoko's 
devices/gta04/patches/restart-when-modem-stops-working.patch.  I think 
that's the same as what you (Nikolaus and Radek) mean by reenumeration, 
but I'm not sure about that.

>>> Does it happen at the moment the other side hangs up or at the
>> moment the
>>> ring starts? Or is it not yet possible to decide between these
>> cases (e.g.
>>> because of lack of precision to time the events)?
>> 
>> I'll keep running my dial scripts and focus on this to confirm or
>> deny it. Unfortunately the phone now somehow started to behave much
>> better and reenumerate is not well reproducible. Could it be that my
>> battery is now charged full?
> 
> Hm. If it also depends on battery charge it might be the initial
> power-on current peak when the modem starts talking to the network.
> There are big capacitors but that might still not be enough if the
> battery voltage is low.

I believe it's a common observation - for modern smartphones in general 
- that travelling fast with 3G will eat battery quite fast.  (I guess, 
owing to the power usage for constantly establishing contact with new 
base stations.)  A power problem could be the common factor for both the 
RING and travelling-fast scenarios.

I also wonder how much impact the user space could have when a RING 
occurs?  Could user space actions at that time cause a significant spike 
in the CPU's power requirement?

Regards,
      Neil