1108271 - [System] Only start the idle timer until the boot is complete

Status: RESOLVED WONTFIX

(Firefox OS Graveyard :: Gaia::System, defect)

Description

[Thomas Zimmermann [:tzimmermann] [:tdz]]

There's a timeout of 60 sec after which the System app switches off the screen by releasing a wakelock. The emulator-{jb,kk} requires five to six minutes for startup, two of them in Gaia (no kidding).

Consequently, the system suspends during startup. Automated testing fails in this case with a timeout.

Comment 1

[Thomas Zimmermann [:tzimmermann] [:tdz]]

I think this should be solved by keeping suspending disabled unconditionally until startup has been finished.

Comment 2

[Gregor Wagner [:gwagner]]

Tim, can your team take this one?

Comment 3

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

Yeah I can put this in my queue. Is this urgent?

Comment 4

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

Update the title to make it more descriptive.

Comment 5

[Thomas Zimmermann [:tzimmermann] [:tdz]]

(In reply to Tim Guan-tin Chien [:timdream] (MoCo-TPE) (please ni?) from comment #3)
> Yeah I can put this in my queue. Is this urgent?

It's not urgent, but should have priority over regular features. We want to get emulator-kk ready and this is currently the one unresolved blocker.

Comment 6

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

(In reply to Thomas Zimmermann (on PTO) [:tzimmermann] [:tdz] from comment #5)
> (In reply to Tim Guan-tin Chien [:timdream] (MoCo-TPE) (please ni?) from
> comment #3)
> > Yeah I can put this in my queue. Is this urgent?
> 
> It's not urgent, but should have priority over regular features. We want to
> get emulator-kk ready and this is currently the one unresolved blocker.

The system app start-up patch is current a mess, it would be hard for the part of the system (specifically, ScreenManager) to realize the time the phone actually boots.

The quick workaround is to set mozSettings value "screen.timeout" of the profile-to-test to 0, which will make the screen never sleeps. We should do that first to unblock your work.

Comment 7

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

(In reply to Tim Guan-tin Chien [:timdream] (MoCo-TPE) (please ni?) from comment #6)
> The system app start-up patch is current a mess, it would be hard for the
> part of the system (specifically, ScreenManager) to realize the time the
> phone actually boots.
> 
> The quick workaround is to set mozSettings value "screen.timeout" of the
> profile-to-test to 0, which will make the screen never sleeps. We should do
> that first to unblock your work.

Greg,

Does lock screen / lock screen window manager currently emits events to the System app scope saying either

"Hey, the main panel is shown."

or

"Hey, I am loaded but lock screen is not enabled."

during System app boot-up?

I think these are the two events needed for this bug.

Comment 8

[Greg Weng [:snowmantw][:gweng][:λ]]

Main panel is a part of the index.html of System app, if your 'shown' means HTML is ready, it's implicit done by loading the page at bootstraping. If you mean the initialization code of LockScreen has been executed, we need to create it. However, since there is a 'lockscreen-appopened' which means the app window animation is done, I think it should be 'lockscreen-appready' or something like that, which is similar to the idea of allowing app to dispatch an event through API or IAC that I've sent a mail to dev-gaia and then been totally ignored.

Comment 9

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

Please read comment 6.

Comment 10

[Thomas Zimmermann [:tzimmermann] [:tdz]]

Hi

This bug blocks any progress with automated tests on emulator-kk. Can we find a solution in the foreseeable future?

Comment 11

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

Hi Thomas,

I thought changing the "screen.timeout" value will unblock? Is that not the case?

Comment 12

[Thomas Zimmermann [:tzimmermann] [:tdz]]

(In reply to Tim Guan-tin Chien [:timdream] (MoCo-TPE) (please ni?) from comment #11)
> Hi Thomas,
> 
> I thought changing the "screen.timeout" value will unblock? Is that not the
> case?

Hi

Is there a way to make this the default on emulators? We'd like to run automated tests on emulator-kk.

Comment 13

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

You can do something like this, pass a SCREEN_TIMEOUT=0 from BoardConfig.mk:

https://github.com/mozilla-b2g/device-rpi/blob/ab404c/BoardConfig.mk#L54

If DEVICE_DEBUG=1 (I don't know where is it being set), the timeout will be changed to 300 sec

https://github.com/mozilla-b2g/gaia/blob/05f8670/Makefile#L125

Comment 14

[Hsin-Yi Tsai (she/her) [:hsinyi]]

Hi all,

This bug is focusing on a long-term solution for system app. I just filed bug 1124567 for a quick workaround for emulator-kk. Let's move there for kk discussion :)

I will adjust bug dependency accordingly.

Comment 15

[Thomas Zimmermann [:tzimmermann] [:tdz]]

Don't know why ni

Comment 16

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

Probably something doable once bug 1094759 lands.

Comment 17

[(inactive after 6/18) Alive Kuo [:alive]]

Write down my idea here:

We could only set the idle timer in ScreenManager when it knows the system is fully loaded.
Now there's no this event, but after bug 1094759, we could make system/js/app.js to dispatch an event, or even register a service and in ScreenManager.start() we could:

Service.request("fullyLoaded").then(() => {
  this.setIdleTimeout();
});

Comment 18

[GitHub Autolander Bot]

Attachment: [gaia] alivedise:bugzilla/1108271/set-idle-timeout-once-fullyloaded > mozilla-b2g:master

Comment 19

[(inactive after 6/18) Alive Kuo [:alive]]

Comment on attachment 8613440 [details] [review]
[gaia] alivedise:bugzilla/1108271/set-idle-timeout-once-fullyloaded > mozilla-b2g:master

Hi Thomas, could you make sure this work for you or not? Thanks.

Comment 20

[(inactive after 6/18) Alive Kuo [:alive]]

(In reply to Alive Kuo [:alive][NEEDINFO!] from comment #17)
> Write down my idea here:
> 
> We could only set the idle timer in ScreenManager when it knows the system
> is fully loaded.
> Now there's no this event, but after bug 1094759, we could make
> system/js/app.js to dispatch an event, or even register a service and in
> ScreenManager.start() we could:
> 
> Service.request("fullyLoaded").then(() => {
>   this.setIdleTimeout();
> });

Insteadly, I am using visuallyLoaded (Service.request('schedule')) in the WIP. If it's not working for emulator I will implement the fullyLoaded version.

Comment 21

[(inactive after 6/18) Alive Kuo [:alive]]

Note: I am working on reland bug 1094759. Will back to this later

Comment 22

[Thomas Zimmermann [:tzimmermann] [:tdz]]

Thanks for all your work! There's no need to rush in a patch for this bug, as we currently have a workaround in place.

Comment 23

[Thomas Zimmermann [:tzimmermann] [:tdz]]

Comment on attachment 8613440 [details] [review]
[gaia] alivedise:bugzilla/1108271/set-idle-timeout-once-fullyloaded > mozilla-b2g:master

I don't know enough about the code to give good feedback on the patch itself.

What we need is to keep the emulator running until the automated tests have been started. We currently have a workaround in the emulator's board config, [1] but a solution in gecko/gaia would certainly be better.

I'd guess that the automated tests only start after the system sent 'fullyLoaded' and a timeout of 60 seconds might not be enough to reach this state on the emulator. (Booting takes 5+ minutes.) Using 'fullyLoaded' might be safer.

There's no good way of testing, because failures only show up as connection timeouts in the test logs. In any case, I'd give a try to any solution you propose.

[1] https://github.com/mozilla-b2g/platform_build/pull/97/files

Comment 24

[BMO Automation]

Firefox OS is not being worked on