856946 - Voice call power numbers are regressed

Status: VERIFIED FIXED

(Firefox OS Graveyard :: General, defect)

Description

[vasanth]

The regression caused by bug 828283 is not completely resolved in bug 853454.
voice call power numbers are still not good.

Before regression, during voice call, apps processor was able to power collapse (sleep). That was also broken but not fixed in bug 853454 patch

Below statement in ScreenManager.js blocks cpu sleep during voice call and 
because of that power numbers are high

power.cpuSleepAllowed = (this._cpuState != 'locked-foreground' &&
                             this._cpuState != 'locked-background' &&
                             !this._audioActive);

CPU sleep should be allowed during voice call as it was before.

Comment 1

[Michael Vines [:m1] [:evilmachines]]

(tef? as I suggest triage consider this for v1.0.1)

Comment 2

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Andrea or Alive, would either of you be able to look into this?

Comment 3

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

Andrew, please check if patch of bug 828283 still needs cpu sleep part per this issue.

Comment 4

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

Taken first because Andrew seems busy

Comment 5

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

:mchen is helping to find out cpu backend sleep/idle strategy so ni? from him if he gets progress.
Before we understand the strategy I'm leaving this bug because it seems ir-relavant to bug 828283 per :mchen.

Comment 6

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

After discussion w/ mchen I made a WIP about ignoring telephony channel,
[e-mail] please help to check if this works:
https://github.com/mozilla-b2g/gaia/pull/9061

Comment 7

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

Hi Michael, could you help to find someone to verify the fix? We're not able to estimate power usage by ourself. Thanks.

Comment 8

[Michael Vines [:m1] [:evilmachines]]

The ni on vasanth was sufficient actually, he'll verify tomorrow IST

Comment 9

[vasanth]

It works. With this change apps processor goes to power collapse and also voice call power numbers are back to normal.
Please mainline this change and we can close this issue.

Comment 10

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

(In reply to vasanth from comment #9)
> It works. With this change apps processor goes to power collapse and also
> voice call power numbers are back to normal.
> Please mainline this change and we can close this issue.

Thanks for verifying so now we know what's the root cause, but I am willing to discuss with audio backend guys to see if there's a proper solution because this one is just a shortcut. I'll update on this bug soon, tomorrow or the first day in work week. If we confirmed that no any other quick way I will set the patch into review process and close this one.

:mchen and :baku, we have to either create a follow-up one or figure out a formal solution about this one.

According to our discussion, :baku added this to avoid the app playing audio keeps cpu on the platform other than our current platform. I didn't know this when doing review but now I think it's a little strange to deal with hardware difference in gaia with an obscure way.

Comment 11

[Michael Vines [:m1] [:evilmachines]]

Why do you think "our current platform" (what does that mean BTW) is not affected?

Comment 12

[Michael Vines [:m1] [:evilmachines]]

Alive, feel free to open up a follow-on bug if you think we can do this in a cleaner way on master.  But for v1.0.1 we need your minimal patch landed asap to get voice power numbers back to their expected values.

Comment 13

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

(In reply to Michael Vines [:m1] [:evilmachines] from comment #12)
> Alive, feel free to open up a follow-on bug if you think we can do this in a
> cleaner way on master.  But for v1.0.1 we need your minimal patch landed
> asap to get voice power numbers back to their expected values.

Yap, per discussion it sounds like there's no other way now, so I'll do what you suggest tomorrow. Thanks.

Comment 14

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

Attachment: https://github.com/mozilla-b2g/gaia/pull/9061

Patch v1:
Don't keep CPU awaken when telephony channel is active.

Comment 15

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

Comment on attachment 737135 [details]
https://github.com/mozilla-b2g/gaia/pull/9061

Looks good for v1.0.1. Need a better commit message though.

Comment 16

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

https://github.com/mozilla-b2g/gaia/commit/42767cfa7dc601f910d4de16388ab7ee1b653e7c
master

Comment 17

[Ryan VanderMeulen [:RyanVM]]

James, can you help with the uplift please?

Comment 18

[James Lal [:lightsofapollo]]

v1-train: f95de9db1131ffbe71b83cffd4bcd3168bb0d30e
v1.0.1: ba0538727eabc7f49fbd402a823cf62d923e3a4c

Comment 19

[Sarah Parsons]

This has been verified fixed on Unagi Build ID: 20130429070204

Environmental  Variables:
Kernel Date: Dec 5
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18/rev/49a908f052ec
Gaia: e26eadbc598bd09c80192016c1024a4a999a5361

This has also been verified fixed on Inari Build ID: 20130429070204

Kernel Date: Feb 21
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18_v1_0_1/rev/45aa5ba0ed53
Gaia: cf2d4136f0ebc66039637fdbeb72ed184dfbc0f2

CPU sleep functions properly while using voicemail.