Closed Bug 1278390 Opened 8 years ago Closed 8 years ago

Fennec doesn't display Gitter.im content, after scrolling up/down

Categories

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

48 Branch
ARM
Android
defect

Tracking

(firefox48 wontfix, firefox49 fixed, firefox50 fixed, firefox51 verified)

RESOLVED FIXED
Tracking Status
firefox48 --- wontfix
firefox49 --- fixed
firefox50 --- fixed
firefox51 --- verified

People

(Reporter: speciesx, Unassigned)

References

()

Details

Attachments

(3 files)

Attached image Screenshot
Load a gitter page eg. https://gitter.im/RigsOfRods/rigs-of-rods, scroll up/down and the Gitter.im contend disappear.

Name Fennec
Version 49.0a1
Build-ID 20160606030219
Update-Kanal nightly
User-Agent Mozilla/5.0 (Android 5.0.1; Mobile; rv:49.0) Gecko/49.0 Firefox/49.0
Betriebssystem Linux 21
Aktivierte Plugins about:plugins
Build-Konfiguration about:buildconfig
Speicherverwendung about:memory
Performance about:performance
Angemeldete Service-Worker about:serviceworkers
Fenster mit mehreren Prozessen 0/1 (deaktiviert)
Abgesicherter Modus false
Profile about:profiles

Self-Destructing Cookies 0.4.10 true jid0-9XfBwUWnvPx4wWsfBWMCm4Jj69E@jetpack
uBlock Origin 1.7.5rc2 true uBlock0@raymondhill.net
Adblock Plus 2.7.3.4175-beta false {d10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d}
Ghostery 6.2.0 false firefox@ghostery.com
Grafik
Features
Compositing OpenGL
Asynchrones Wischen und Zoomen Berührungs-Eingabe aktiviert
WebGL-Renderer Qualcomm -- Adreno (TM) 320
Hardware H264 Decoding No
GPU #1
Active Yes
Description Model: GT-I9505, Product: jfltexx, Manufacturer: samsung, Hardware: qcom, OpenGL: Qualcomm -- Adreno (TM) 320 -- OpenGL ES 3.0 V@84.0 AU@ (CL@)
Vendor ID Qualcomm
Device ID Adreno (TM) 320
Driver Version OpenGL ES 3.0 V@84.0 AU@ (CL@)
Diagnostics
AzureCanvasAccelerated 1
AzureCanvasBackend skia
AzureContentBackend cairo
AzureFallbackCanvasBackend none
Decision Log
HW_COMPOSITING
force_enabled by user: Force-enabled by pref
Wichtige modifizierte Einstellungen
Name Wert browser.cache.disk.capacity	204800
browser.cache.disk.filesystem_reported 1
browser.cache.disk.smart_size.first_run false
browser.download.importedFromSqlite	true
browser.startup.homepage_override.mstone	49.0a1
dom.presentation.device.name	GT-I9505
extensions.lastAppVersion	49.0a1
general.useragent.updates.lastupdated	1464941706328
media.gmp-gmpopenh264.abi	arm-eabi-gcc3
media.gmp-gmpopenh264.lastUpdate	1461490628
media.gmp-gmpopenh264.version	1.5.3
media.gmp-manager.buildID	20160606030219
media.gmp-manager.lastCheck	1465246665
media.gmp.storage.version.observed	1
network.cookie.cookieBehavior	1
network.cookie.prefsMigrated	true
network.http.speculative-parallel-limit 0
network.predictor.cleaned-up true
network.prefetch-next false
privacy.trackingprotection.pbmode.enabled	false
storage.vacuum.last.index	0
user.js-Einstellungen


JavaScript
Inkrementelle GC true
Barrierefreiheit
Aktiviert false
Barrierefreiheit verhindern 0
Bibliotheken-Versionen
Minimal vorausgesetzte Version Verwendete Version
NSPR 4.12 4.12
NSS 3.25 Beta 3.25 Beta
NSSSMIME 3.25 Beta 3.25 Beta
NSSSSL 3.25 Beta 3.25 Beta
NSSUTIL 3.25 Beta 3.25 Beta
I was able to reproduce this issue on latest Nightly 50.0a1 (2016-06-26) using ONE A2001 (Android 5.1.1)
I can reproduce this on a Galaxy S3 GT-I9300. Though switching tabs seems to fix the drawing issues. If it does not reproduce in the first scroll up/down repeat this a couple times.

NI for someone in snorp's team to look at this.
Status: UNCONFIRMED → NEW
Component: General → Graphics, Panning and Zooming
Ever confirmed: true
Flags: needinfo?(snorp)
With the minimap on, it looks like when you scroll to either the top or the bottom, the main thread gets blocked, presumably as the page loads more content. During this time it's relatively easy to checkerboard by scrolling fast back the other way.

Another weird thing I noticed is that once you outrun the low-res displayport it looks like additional panning has no effect, like the APZ hit-test isn't finding anything. In particular this makes it so that once you hit the sea of checkerboard, you can't scroll back to the island of painted content that you left behind. I'm not sure if that's specific to this page or a general issue but we should investigate and file another bug to track that if needed.

Finally it looks like some of the time, when you hit the sea of checkerboard, either the repaint requests are not getting through or something else happens but it gets stuck in that state rather than clearing itself. That is what results in the behavior that this bug was filed about.
Dropping stale needinfo.
Flags: needinfo?(snorp)
Version: Trunk → 48 Branch
I think the patch from bug 1292781 helped here. I can no longer reproduce the issue where it gets stuck in an unpainted state, I only see temporary checkerboarding while scrolling. If you reach the top or the bottom of the content then the page seems to be doing something (loading more content?) on the main thread that takes a while and it's easy to scroll fast and checkerboard during that time. However it paints in once the main thread is freed up again.

SpeciesX, are you still able to reproduce this bug on a recent Nightly?
Flags: needinfo?(speciesx)
It would also be good to know if it reproduces on a recent Aurora or Beta build. Both of those channels have the fix from bug 1292781 uplifted so I would expect the issue to be fixed there as well.
(In reply to Kartikaya Gupta (email:kats@mozilla.com) from comment #5)

> SpeciesX, are you still able to reproduce this bug on a recent Nightly?

no.
Flags: needinfo?(speciesx)
Thanks!
Status: NEW → RESOLVED
Closed: 8 years ago
Depends on: 1292781
Resolution: --- → WORKSFORME
Resolution: WORKSFORME → FIXED
Tested again myself and based on comment 7 I will set the flag verified.
Device: LG G4 (Android 5.1).
Build: Nightly (2016-09-06).
sometimes I am  able to reproduce this issue on Firefox for Android 50 beta 1 using ONE A2001 (Android 5.1.1) while scrolling up/down very fast
(In reply to Teodora Vermesan (:TeoVermesan) from comment #10)
> Created attachment 8793319 [details]
> Screenshot_20160921-155203.png
> 
> sometimes I am  able to reproduce this issue on Firefox for Android 50 beta
> 1 using ONE A2001 (Android 5.1.1) while scrolling up/down very fast

When you see this does it stay blank like this forever? Or does it repaint after a few seconds? If it stays blank forever then please file a new bug. I tried to reproduce that but was unable to, but I was using 49b11 (since 50b1 isn't on the play store yet).
Flags: needinfo?(teodora.vermesan)
Please see the following video: https://www.youtube.com/watch?v=txeUc2OtGJk&feature=youtu.be
Flags: needinfo?(teodora.vermesan)
Thanks, that helps a lot! It looks like a different issue from this one, so please file a new bug.
Product: Firefox for Android → Firefox for Android Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: