102.74 KB, text/plain
275.21 KB, image/png
3.70 KB, patch
|Details | Diff | Splinter Review|
User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:12.0) Gecko/20100101 Firefox/12.0 Build ID: 20120420145725 Steps to reproduce: Browse to www.google.com and leave my phone alone Actual results: When I came back Firefox Beta 14 was using 130+M of memory anf 70+% of CPU Settings > Battery indicates that Firefox is responsible for 21% of CPU usage (2nd position) Expected results: Firefox should not use has much CPU and battery
Did you setup Firefox Sync ? Any other webpages loaded? How long were you "away" from Firefox?
Sync is setup. That was the only loaded page at this time. I was away for maybe 15 minutes or so. I did not use the phone during that time.
I don't see this with a Galaxy SII. Sync was not enabled. Memory usage was around 50 MB and CPU usage was always below 5%.
I do not know if this is relevant but I have Galaxy S II with Ice Cream Sandwich. I have seen this battery and CPU usage problem with Firefox 10.0 as well on Ice Cream Sandwich but never seen it on Gingerbread with Firefox 10. I did not try Firefox 14 on Gingerbread. It does not do it all the time though. I do not know if there is a special pattern that triggers it. When it does not happen memory is abut 30MB and CPU usage is 1 or 2 %. Is there some logfiles or special things I could do to give you more information? I really like FF14 but this battery thing is really annoying.
If you are able to reproduce it, getting the logcat output from when the CPU usage is high would be useful. See https://wiki.mozilla.org/Mobile/Fennec/Android#Using_logcat for how to get logcat output - there's an app you can use if you don't have the Android SDK installed. Even better would be if we could reproduce it so that we could attach gdb and see exactly what it's doing, but hopefully the logcat will shed some light on it.
This morning my battery went under 50% in about 2 hours. Reported battery usage was Firefox 73%, screen 16%. Attached is a log of when Firefox had high CPU usage.
CCing gcp, looks like infinite loop of migration errors?
There's nothing from migration in this log.
Oh, sorry. Do you know what all those PlacesDBUtils errors are, or why they would be in the log?
If I would I'd have taken the bug :) The question is what is trying to use Places. We don't have Places in Native Fennec, it's compiled out.
User agent is Mozilla/5.0 (Amdroid; Mobile; rv:14.0) Gecko/14.0 Firefox/14.0 Screenshot of about:home attached I got it from Google Play > Firefox Beta: http://play.google.com/store/apps/details?id=org.mozilla.firefox_beta
Looks like the JS errors are related to Telemetry: http://mxr.mozilla.org/mozilla-aurora/source/toolkit/components/places/PlacesDBUtils.jsm#985
blocking-fennec1.0 was already set to soft, so minusing for tracking. With Firefox 15, I expect we'll move back to tracking Fennec issues using the tracking flags.
Not sure if this is related but Firefox Beta also has a lot of Background Data Usage
I suggested time ago to undef MOZ_PLACES to really stop compiling places... now we are adding a further ifdef around everything... sigh. I'll review shortly, but please file a bug to throw away MOZ_ANDROID_HISTORY and properly use ifdef MOZ_PLACES (that requires doing properly the history implementation instead of the current internal module redirect of History.h).
Filed bug 761588 for fixing building with MOZ_PLACES=
(In reply to Matthieu Pupat from comment #21) > Not sure if this is related but Firefox Beta also has a lot of Background > Data Usage Actually seems very related, see bug 762590
I've created bug 762620 to deal with the idle-daily thing that seems to misfire. Getting rid of the unwanted database stuff can stay in this bug.
I cannot reproduce this issue on the latest Beta 7 build. The power consumption by Fennec is decent now (about 7% in 15 minutes). Closing bug as verified fixed on: Device: Galaxy Nexus OS: Android 4.0.2