796670 - [System] Wifi sleep timer doesn't actually works

Status: RESOLVED FIXED

(Firefox OS Graveyard :: Gaia, defect)

Description

[GH to BZ]

[GitHub issue by timdream on 2012-09-18T10:07:56Z, https://github.com/mozilla-b2g/gaia/issues/4857]
/blame me, I should have find out earlier.

The problem lies at Line 107 of system/js/wifi.js. At the moment, upon the conditions are met,`setTimeout` was executed and should turn off wifi after 60 sec. But that `setTimeout` timer is never executed, presumably because the CPU is in fact asleep.

So, questions:

1. If CPU is let to sleep, what's the point to turn off wifi?
2. Does it make sense to set a 60 sec CPU wake lock just for turning off wifi if we should turn it off?

CCing @kanru @vingtetun @michaelwu 


Related issue: [Bug 784733] Otoro device drains overnight when left untethered with wifi enabled
https://bugzilla.mozilla.org/show_bug.cgi?id=784733

Comment 1

[GH to BZ]

[GitHub comment by kanru on 2012-09-20T10:47:40Z]
 1. If the CPU is unable to sleep for some reason, then the timer will go off.
 2. No. However you should reset the timer after CPU wakeup.

Comment 2

[GH to BZ]

[GitHub comment by autonome on 2012-09-24T15:51:46Z]
Blocking for battery drain. If the CPU is asleep it doesn't make sense to have much else running.

Comment 3

[GH to BZ]

[GitHub comment by timdream on 2012-09-25T04:31:45Z]
Let's use Alarm API for that 60 sec timeout ...

Comment 4

[GH to BZ]

[GitHub comment by timdream on 2012-09-26T11:08:54Z]
I cannot make System app to successfully access Alarms API, which is very bad. Gonna try to keep the CPU awake for 60 sec instead.

Comment 5

[GH to BZ]

[GitHub comment by timdream on 2012-09-27T11:21:13Z]
Hum ... wake lock seems useless even it does set `mozPower.cpuSleepAllowed` to false.

Comment 6

[Tim Guan-tin Chien [:timdream] (please needinfo)]

Gene, I need your help on this one.

https://github.com/timdream/gaia/compare/c8983e2f...timdream:a2b73b8

I've written a patch that I believe it should work, however upon "wiki timer set" message, I didn't receive any callback on line 75. What can possible go wrong? Is there a DEBUG param I can set in System Message script that I can see what's going on internally?

Comment 7

[Gene Lian (I already quit Mozilla)]

I can support to take a look on that. I'll come back with some testing results. Please stay tuned. :)

Comment 8

[Gene Lian (I already quit Mozilla)]

I've already narrow down the root cause (please refer to bug 797803) but need to discuss more with Fabrice to work out an optimal solution to solving it.

Comment 9

[Tim Guan-tin Chien [:timdream] (please needinfo)]

Attachment: Patch v1

Problem with mozSetMessageHandler is being workarounded with idea described in bug 797803 comment 6

Comment 10

[Tim Guan-tin Chien [:timdream] (please needinfo)]

https://github.com/mozilla-b2g/gaia/commit/340f75fcaee6e86acfdd158b983004375b4b05b2

Comment 11

[Gene Lian (I already quit Mozilla)]

(In reply to Gene Lian [:gene] from comment #8)
> I've already narrow down the root cause (please refer to bug 797803) but
> need to discuss more with Fabrice to work out an optimal solution to solving
> it.

Btw, bug 797809 has already been r+'ed and is going to be checked in sooner or later in a couple of days. That would help you in the future if you'd still prefer receiving system messages within the System App. ;)