Closed Bug 737367 Opened 12 years ago Closed 12 years ago

Playing Runfield and rotating to landscape yields a black screen on the Samsung Galaxy SII

Categories

(Firefox for Android Graveyard :: General, defect)

Product:
Firefox for Android Graveyard
Component:
General
Version:
14 Branch
Platform:
ARM
Android
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(blocking-fennec1.0 +)

Status:
VERIFIED WORKSFORME
Tracking Flags:
Tracking Status
blocking-fennec1.0 --- +

People

(Reporter: csuciu, Assigned: jrmuizel)

Details

(Keywords: regression, Whiteboard: [gfx])

Attachments

(2 files)

Log file
176.91 KB, text/plain
Details
logcat
6.54 KB, text/plain
Details
Attached file Log file — 
Nightly 14.0a1 (2012-03-19) 20120319031122
Samsung Galaxy SII (Android 2.3.4)

Steps:
1. Load http://paulrouget.com/mwc-demos/maple.html
2. Tap on Runfield

Expected:
Runfield should load

Actual:
Runfield doesn't load. Demos page freezes (cannot pan or zoom and sometimes the entire page starts to flicker continuously).
blocking-fennec1.0: --- → ?
Adding qawanted to see if this happens on anyone else's Galaxy SII.

Runfield works for me in portrait just fine on my SII (03/22 Nightly). On landscape I just get a black-screen.
Keywords: qawanted
Yes black on landscape for my GSII as well. Works in portrait mode.
Keywords: qawanted
Keywords: regression
Seems to be working now on portrait. 
Once started in portrait, if I switch to landscape, the screen turns black. 

The actual result from comment #0 still stands if Runfield is started in landscape mode
Summary: Runfield doesn't load on Samsung Galaxy SII → Playing Runfield and rotating to landscape yields a black screen on the Samsung Galaxy SII
Assignee: nobody → bugmail.mozilla
blocking-fennec1.0: ? → beta+
For QA regression finding : Try taking a look between 19/20
I suspect Jeff is on the hook for this though I'm cc'ing the other two people who checked in during that range.
Group: core-security
Group: core-security
This is more likely to be the scroll-clip stuff than it is double-paint after invalidation, I would think, though I'd have also thought that trying to pan or zoom afterwards would then fix it if that were the cause.
You can't pan/zoom on runfield, it eats all the touch events.
I built the code at 70c7baef22df and it still has the bug. I think the bisection range in comment #5 isn't right. Then again I'm building debug builds so maybe that makes a difference.
I'm confused, because I got both 18/19 builds working fine, not showing the issue.  Do I have to tap while the runfield is loading for the STR?
This is still broken for me with today's nightly I'll check the regression range.
(In reply to Naoki Hirata :nhirata from comment #10)
> I'm confused, because I got both 18/19 builds working fine, not showing the
> issue.  Do I have to tap while the runfield is loading for the STR?


No, just load and rotate. For me loading in portrait and rotating to landscape could repro 100%. Loading while in landscape already sometimes worked and sometimes behaved strangely with some brown background appearing.
The regression range in comment 5 is spot on for my phone.
Looking at the regression range, its not kats, re-assigning.
Assignee: bugmail.mozilla → jmuizelaar
Whiteboard: [gfx]
blocking-fennec1.0: beta+ → +
Please see if we can get this on any other canvas apps/demos with the SII.
Keywords: qawanted
I did not have an issue on Samsung Galaxy SII w/ Android OS 2.3.3; The demo eventually loads.

I retried this issue on the Galaxy Nexus, Nexus S, and Droid Pro and did not have any issues.  Note: My Galaxy S II is rooted.
I'm still getting the black screen in landscape mode on Runfield.

On MineGL, a white blank page is displayed instead of the actual demo. This happens in portrait and landscape.

Also, the demos page (http://paulrouget.com/mwc-demos/maple.html) becomes unresponsive after pressing back to exit from Runfield (in landscape) or MineGL (portait and landscape).

Nightly 14.0a1 (2012-04-17) 20120417030641
Samsung Galaxy SII (Android 2.3.4)

Please see the attached video.
I'm getting more convinced that it's a device + Android OS specific issue.  This does not occur on my 2.3.3 device.
I believe that the back button is a separate issue.  Catalin, can you please report that in a separate bug?  I am finding that I can no longer navigate after hitting the back button from runfield.
qawanted: assigning to kevin to reproduce.
http://kevs3d.co.uk/dev/canvask3d/ultralight.html
http://kevs3d.co.uk/dev/lsystems/ 
http://www.kevs3d.co.uk/dev/eglogo/ 

The above urls reproduce the issue. The other urls at the site some render ok others have problems in a different manner. We need to determine what these have in common.
Other canvas demos at http://kevs3d.co.uk/dev/
I am able to reproduce this issue on the latest Nightly build. Also I get these errors when I tap on MineGL link:

E/GeckoConsole(10354): [JavaScript Warning: "Error: WebGL: Can't get a usable WebGL context" {file: "http://paulrouget.com/mwc-demos/webgl/PhiloGL.cls.js" line: 76}]
D/GeckoFavicons(10354): Loading favicon from DB for URL = http://paulrouget.com/mwc-demos/webgl/minegl.html
E/GeckoConsole(10354): [JavaScript Warning: "Error: WebGL: Can't get a usable WebGL context" {file: "http://paulrouget.com/mwc-demos/webgl/PhiloGL.cls.js" line: 76}]
E/GeckoConsole(10354): [JavaScript Error: "NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIDOMHTMLCanvasElement.getContext]" {file: "http://paulrouget.com/mwc-demos/webgl/PhiloGL.cls.js" line: 76}]

Here is a video regarding this bug: http://youtu.be/AuCDkOVziAs

--
Firefox 14.0a1 (2012-04-17)
Device: Samsung Galaxy S (Captivate)
OS: Android 2.2
(In reply to Naoki Hirata :nhirata from comment #20)
> I believe that the back button is a separate issue.  Catalin, can you please
> report that in a separate bug?  I am finding that I can no longer navigate
> after hitting the back button from runfield.

Bug #746509 was logged for the "back button" issue
Attached file logcat — 
Ok.  I see the issue on the Samsung Galaxy S Captivate with today's build 4/18/2012;

04-18 15:20:43.410: E/libEGL(4546): call to OpenGL ES API with no current context (logged once per thread)
removing qawanted.  last 5 comments have STRs and other sites.
Keywords: qawanted
I can not reproduce this on a Galaxy SII Mali-400MP. However I can reproduce the MineGL bug, so I have split that out into bug 752971
I could reproduce the runfield issue with an old build on a Galaxy S, but current builds no longer show the problem. Can anyone else reproduce the problem with a recent nightly.
QA, can you retest on the SII?  If other issue reproduce, let's split them out like in #28.
Keywords: qawanted
Works for me in the current trunk.
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → WORKSFORME
This is broken in current Aurora going to look into what fixed this.
Runfield started working again in landscape on Nightly 2012-04-30. On today's Aurora this is still broken.
Keywords: qawanted
Do we have an un-regression range? We should uplift whatever fixed this to aurora.
The most likely candidate in that range is bug 743247 but that was already uplifted to aurora. I don't see much else in the range that might have fixed this. We'll probably need to bisect the inbound tinderbox builds.
Can people retest (Catalin, kbrosnan)?  I see it working on both nightly and aurora for 5/15/2012
WFM on both Aurora and Nightly.

Aurora 14.0a2 (2012-05-15)
Nightly 15.0a1 (2012-05-15)

Samsung Galaxy SII (2.3.4)
Status: RESOLVED → VERIFIED
Keywords: regressionwindow-wanted
Yes, I concur. Working in today's Aurora and Nightly.
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: