On XO-1.5 with 11.3.0/11.3.1 -- hang during shutdown?

[Martin Langhoff]

Hi folks,

where is Deepak Muddhaa based? Any reason his failing XO and SD card
can't be traded for good ones, and the failing units shipped to James,
Miami or Boston, where we can look at things at a lower level?

We'll gladly provide a replacement unit.

I appreciate all the analysis, but it' is apparent that it is being
done on rather poor data. Hands-on debugging wins.

cheers,




m

On Sun, Jun 24, 2012 at 11:32 PM, James Cameron <quozl at laptop.org> wrote:
> Thanks for your reply!
>
> On Mon, Jun 25, 2012 at 11:16:26AM +1000, Sridhar Dhanapalan wrote:
>> On 21 June 2012 16:14, James Cameron <quozl at laptop.org> wrote:
>> > On Thu, Jun 21, 2012 at 02:37:35PM +1000, Sridhar Dhanapalan wrote:
>> >> On 16 June 2012 17:08, James Cameron <quozl at laptop.org> wrote:
>> >> > That means the hang should not exceed 15 seconds. ?Is this what you
>> >> > find? ?If not, then this casts doubt on your solution.
>> >>
>> >> I'm going to propose something extremely hackish: [...]
>> >
>> > Just to remind you that I'm still interested to know if the hang you
>> > observe exceeds 15 seconds or not. ?I've not had the time to reproduce
>> > this hang yet. ?Building a mental model of the problem is important to
>> > me, because I can sometimes resolve a problem if I have a good model.
>>
>> Yes; we have left it for several minutes and no shutdown has
>> occurred.
>
> Ooh, I'm surprised.
>
> This observation, and the statistical results from your temporary
> solution (a delay), implies a combination effect, of both the
> processes not yet terminated, and the umount, leading to a process
> hang of umount.
>
> I can't think of a hack that would meet the requirements:
>
> - survive the process deletion steps, and
>
> - detect the stalled umount process.
>
> I guess you might try remounting the filesystem -o sync, just to
> further shift the timing.
>
> The problem needs a kernel developer to reproduce it.
>
> Do you have a way to encourage the problem to occur?  If it can be
> made to occur on a higher percentage of shutdowns, it becomes easier
> to debug.  For instance, there is a two second delay in the code, so
> does the hang occur more frequently if this is reduced to zero?
>
>> > The XO-1.75 CPU has a hardware watchdog that could be used for this,
>> > but you aren't likely to ever have a heat problem with XO-1.75.
>>
>> That is interesting. Why is that?
>
> I take it you mean why won't you have a heat problem with XO-1.75.
> There are two new characteristics of the XO-1.75 over the XO-1.5:
>
>
> 1.  the maximum power draw of the XO-1.75 at full utilisation is a
> long way below that of the XO-1.5.  In a scenario where the laptop is
> powered on and insulated from cooling air flow, this means:
>
> 1.a. the temperature rise toward equilibrium will be slower,
>
> 1.b. the equilibrium temperature will be lower for a given level of
> insulation, (stacking, or cloth covers, or both),
>
> 1.c. the insulation will have to be far greater to achieve the same
> equilibrium temperature.
>
>
> 2.  the XO-1.75 has a thermal protection feature that forces the power
> off if the temperature of the CPU exceeds 85 degrees C, rather than
> slowing or stopping the CPU as on XO-1.5.  In a scenario where the
> laptop is powered on and insulated from cooling air flow, this means:
>
> 2.a. the temperature rise will be interrupted by a sudden loss of
> input heat, rather than be slowed by a gradual loss of input heat,
>
> 2.b. the insulation will have to be far far greater to achieve the
> same equilibrium temperature.
>
>
> In this scenario, the heat spreader has very little bearing on the
> matter.  The heat spreader relies on cooling air flow to the top of
> the case.  If there is no air flow, the heat spreader is ineffective.
>
> The new thermal protection feature isn't a perfect protection; the
> battery charge circuit remains powered.  So a laptop held between very
> good insulation (e.g. thick polystyrene with sealed edges) with a flat
> battery will still heat up, but not nearly as much as one with an
> active CPU.
>
> (Please, test this yourselves with an IR thermometer.  If you don't
> have one, the closest in Sydney to you would be at the Jaycar store
> at 127 York St.)
>
> --
> James Cameron
> http://quozl.linux.org.au/



-- 
 martin at laptop.org -- Software Architect - OLPC
 - ask interesting questions
 - don't get distracted with shiny stuff  - working code first
 - http://wiki.laptop.org/go/User:Martinlanghoff