718364 - Excess background CPU use

Status: RESOLVED WORKSFORME

(Firefox for Android Graveyard :: General, defect, P2)

Description

[Randell Jesup [:jesup] (needinfo me)]

Checked battery summary after battery ran down overnight. (2-3 nights ago)

~7 hours since unplug.  48% fennec.  6:xx minutes foreground CPU, 1:3x:xx background CPU(!!)

Since unplug, I'd looked at G+, looked at my RSS feeds in NewsRoom, opened a link or two (can't remember), and I *think* I left it on a blog page (see URL) and went to sleep.

Droid Charge, running nightly I think (might have been Aurora, but mostly I run nightly).
Android 2.3.x (stock verizon).

Overall, CPU use seems noticeably higher than stock browser (perhaps because of OOM crashes and frequent restarts because of being forced out of memory?)  But it's hard to be sure.

Comment 1

[Cameron McCormack (:heycam)]

After starting up the browser with a single about:blank tab, I find it fluctuates between using 5% and 15% CPU (according to the Task Manager).  I'm on Android 2.3.3 on a Galaxy S 2.

Comment 2

[Aaron Train [:aaronmt]]

Log? Was a sync going on perhaps?

Comment 3

[Doug Turner (:dougt)]

Cameron - what Task Manager?

Comment 4

[Cameron McCormack (:heycam)]

(In reply to Doug Turner (:dougt) from comment #3)
> Cameron - what Task Manager?

The one that pops up when tap the "Task Manager" button after I long press the home button on my phone.  Unfortunately I don't have experience when other Android phones to know whether this is a Samsung or Android thing.

I'm happy to obtain a log.  Do you have a pointer on how I can do this?  (Phone is not rooted, don't have Android SDK installed atm.  I can do these if needed.)

Comment 5

[Brad Lassey [:blassey] (use needinfo?)]

To get a log you can either install the android sdk and run "adb logcat" from the command line (assuming adb is in you path) or install aLogCat from the market.

If you go the sdk route, it might also be useful to run top (adb shell top)

Comment 6

[Sander Lepik]

I have similar problem and i think it's Sync. One night after installing Aurora i forgot to connect cable for loading battery. Phone (first Desire, running CM7.1) was dead on the next morning. I just booted it up and left idle. Didn't even start Aurora, but Sync was configured.
Now today after disconnecting phone from cable, 4 hours later battery is at 65%, Aurora is reported to take 42% of the blame and i had to remove it.

Comment 7

[Randell Jesup [:jesup] (needinfo me)]

Perhaps your problem was Sync; mine wasn't as I hadn't set up Sync ever on the phone.

Comment 8

[Randell Jesup [:jesup] (needinfo me)]

Ok, it happened again.

3h 16m on battery (from 100%); down to ~20%
Aurora used 56%, detailed info from the battery use screen was 33m 25s of CPU time, 25s of foreground CPU time.  

URL was http://www.huffingtonpost.com/mobileweb/2012/02/28/map-michigan-primary-2012-results_n_1307981.html

I don't have sync configured.

Aurora 12.0a2 2012-02-08 (odd, I know tried to update it since then I think).

Comment 9

[Doug Turner (:dougt)]

we do not disable device motion events when we go into the background on android.

Fennec XUL did something like:
   https://mxr.mozilla.org/mozilla-central/source/mobile/xul/chrome/content/browser.js#364

Comment 10

[Mark Finkle (:mfinkle) (use needinfo?)]

I'd like to get some feedback on the situation now that bug 736311 has landed (although not in a Nightly yet).

Comment 11

[Mark Finkle (:mfinkle) (use needinfo?)]

Randell - Can you try this out again in a new Nightly?

Comment 12

[Mark Finkle (:mfinkle) (use needinfo?)]

Will mark WFM if we don't get some feedback soon

Comment 13

[Doug Turner (:dougt)]

Comment #9 still is reproducible.

Comment 14

[Randell Jesup [:jesup] (needinfo me)]

Tried the URL; after an hour all the time (1:19) is foreground CPU (6%, which is as high as cell standby - but that was probably mostly browsing to the site) (though 1 min cpu seems high - but it is a huffpost page).

HOWEVER, that was a "live" map page when this happened (primary); that has long gone static and therefore it's not a good test.

I should note that my phone (with BatterySpy) said it was losing 18% per hour while sitting idle 2 feet from wifi router (but slightly iffy 4g cell coverage, and it turns off wifi when idle).

Comment 15

[Mark Finkle (:mfinkle) (use needinfo?)]

Reopen or file a new bug if this problem re-appears

Comment 16

[Doug Turner (:dougt)]

Mark, this bug is still reproducible.  See comment #13

Comment 17

[Mark Finkle (:mfinkle) (use needinfo?)]

(In reply to Doug Turner (:dougt) from comment #16)
> Mark, this bug is still reproducible.  See comment #13

Sorry. I forgot to say that comment 13 and comment 9 are different than this bug and should be addressed in other bugs by now.

Comment 18

[Doug Turner (:dougt)]

this is dumb.