603040 - Black Screen, No GUI on Maximize w. YouTube (See Comment 3)

Status: RESOLVED WONTFIX

(Core :: Graphics, defect)

Description

[Alex]

User-Agent:       Mozilla/5.0 (Windows NT 6.1; rv:2.0b8pre) Gecko/20101008 Firefox/4.0b8pre
Build Identifier: Mozilla/5.0 (Windows NT 6.1; rv:2.0b8pre) Gecko/20101008 Firefox/4.0b8pre

Black screen for a less than half a second with D3D10 when maximizing Firefox.

Reproducible: Always

Steps to Reproduce:
1. Enabled D3D10
2. Minimize Browser
3. Maximize Browser
Actual Results:  
Black screen for less than half a second.

Expected Results:  
Instant show of currently opened webpage.

Plugins (like Adobe Flash) do NOT go black, everything around them does.
That is, the page will go black for half a second - except for the plugins.

This happens with and without Direct2D.

Comment 1

[(mostly gone) XtC4UaLL [:xtc4uall]]

What's your Graphics Info from about:support (Bottom)?

Comment 2

[Bas Schouten (:bas.schouten)]

(In reply to comment #0)
> User-Agent:       Mozilla/5.0 (Windows NT 6.1; rv:2.0b8pre) Gecko/20101008
> Firefox/4.0b8pre
> Build Identifier: Mozilla/5.0 (Windows NT 6.1; rv:2.0b8pre) Gecko/20101008
> Firefox/4.0b8pre
> 
> Black screen for a less than half a second with D3D10 when maximizing Firefox.
> 
> Reproducible: Always
> 
> Steps to Reproduce:
> 1. Enabled D3D10
> 2. Minimize Browser
> 3. Maximize Browser
> Actual Results:  
> Black screen for less than half a second.
> 
> Expected Results:  
> Instant show of currently opened webpage.
> 
> Plugins (like Adobe Flash) do NOT go black, everything around them does.
> That is, the page will go black for half a second - except for the plugins.
> 
> This happens with and without Direct2D.

It's interesting that this happens without Direct2D. Since disabling Direct2D disables D3D10 as well.

Comment 3

[Alex]

(o_o)'

You are right, my bad! False correlation; this is reproducible with any rendering method.

Is it duplicate of 594821?

1. GUI is not drawn for ~0.4 seconds; it just shows up as a slab of glass. 
--- The maximize, minimize, exit buttons are there.
2. Flash Plugin gets painted ASAP. Rest of the webpage is black ~0.4 seconds.
3. Switching from Tab_A to Tab_B:
--- If Tab_B has adobe flash content (ex. YouTube Video),
---- Tab_B's flash content gets drawn first on top of Tab_A's page.
---- Rest of Tab_B's page gets painted. Tab switch done
--- If Tab_B has no flash content,
---- Tab_B's page gets painted. Tab switch done.

Comment 4

[Bas Schouten (:bas.schouten)]

(In reply to comment #3)
> (o_o)'
> 
> You are right, my bad! False correlation; this is reproducible with any
> rendering method.
> 
> Is it duplicate of 594821?
> 
It's certainly possible!

Comment 5

[Alex]

This still happens.
Open YT video.
Minimize browser.
Maximize browser.
It is longer (the blackness and only glass) when the tab with YT vid is active.
When the tab with the flash vid is in background, it goes away a bit faster.

Switching tabs results in the YT video being drawn first FOLLOWED by the GUI (redrawing to show the YT tab as active) and the YT page itself.

Without the YouTube video (I closed the tab), minimizing then maximizing leads Firefox 8.0a to be fully white for a split second.

The white background (no flash) or black window + glass without any GUI (flash) goes away much faster without hardware acceleration enabled. It is still there thought

Comment 6

[Chris Lord [:cwiiis]]

I'm not sure how it differs on Windows, but non-windowed plugins (such as on YouTube) are generally drawn by the plug-in process, and so there's no guarantee that we can draw synchronously with them.

I also don't know what capabilities you have with regards to synchronous drawing after restoring a window on Windows (Linux/X11, for example, is asynchronous, but there are extensions that allow the window-manager and applications to collude to draw things synchronously).

Would you be able to test this on a Nightly build with and without acceleration enabled, and comment on the results?

Comment 7

[Alex]

Chrome and Opera do _not_ slow down or black screen with flash.

Minimizing and Restoring Chrome & Opera (click really fast taskbar icon) results in me seeing the Chrome interface and Opera interface and webpage AT THE SAME time as I see the flash content on the page. No blackness - no slowdown - very very snappy. Flash content does not slow down the browser which you are competing with.

Minimizing and Restoring Firefox 10 Nightly (clicking really fast on the taskbar icon) - I get a black box with empty glass on top (100% black = where the webpage should be rendered, glass without anything = where the GUI is).
It is very easy to see with Hardware Acceleration and harder to see without hardware acceleration. Hardware Acceleration makes it much more painful.
Without flash Firefox does not do the BLACK screen instead of webpage content for a large portion of a second, it does a WHITE screen instead for a very small portion of a second.
Opera and Chrome do not do the WHITE screen either.

I can click with my notebook touchpad fast enough to have Firefox never render its iterface and the webpage.

100% reproducible.

STR,
0. Enable HW, use Windows 7
1. Go to site with Flash
3. Minimize / Maximize

Windows 7 x64 SP1 Pro.
Athlon II P340 2.2 Ghz
ATi Radeon 4250M (256MB)
3 GB DDR3
1336*768 @ 60Hz
Clean & up-to-date

Windows 7 x86 SP1 Pro.
Atom N270 1.6 Ghz
NV ION (256MB)
2GB DDR3
1920*1080 @ 60 hz
Clean & up-to-date


IE9 also doesn't white out for a fraction of a second when making it pop-up really fast.
I refuse to get IE9 version of flash plugin and will not test that.

Comment 8

[Alex]

This also happens when starting Firefox. 
First it is all white, then I see start page and interface.

Comment 9

[Alex]

Videos that show all the problems I described,

http://youtu.be/llc2cLbfk6I
http://youtu.be/Q8HSjUtn1Po
http://youtu.be/6UF4-Qsz9lc

Comment 10

[Alex]

netbook with D2D,

records samples that occured in during periods when we did not service the event loop for more than > 100 ms.
NOTE: about:jank doesn't interact well with the Gecko Profiler Addon
about:jank results (2632 samples)

1751 - c-gfx::DrawThebesLayer
173 - c-Paint::PresShell::Paint
143 - c-JS::CallEventHandler
113 - c-layout::DoReflow
83 - c-image::imgFrame::Draw
83 - c-event::nsViewManager::DispatchEvent
76 - c-layout::FlushPendingNotifications
75 - c-nsEventListenerManager::HandleEventInternal
60 - c-JS::EvaluateString
40 - c-html5::RunFlushLoop
20 - c-CSS::ProcessRestyles
5 - c-GC::CycleCollectNow
3 - c-Input::nsInputStreamPump::OnStateTransfer
2 - c-nsInputStreamPump::OnStateStart
2 - c-JS::EvaluateStringWithValue
1 - c-Timer::Fire
1 - c-network::nsHttpChannel::OnDataAvailable
1 - c-Input::DispatchMouseEvent

netbook with no D2D,

records samples that occured in during periods when we did not service the event loop for more than > 100 ms.
NOTE: about:jank doesn't interact well with the Gecko Profiler Addon
about:jank results (1689 samples)

630 - c-gfx::DrawThebesLayer
370 - c-Paint::PresShell::Paint
136 - c-layout::DoReflow
100 - c-image::imgFrame::Draw
87 - c-JS::EvaluateString
83 - c-event::nsViewManager::DispatchEvent
69 - c-nsEventListenerManager::HandleEventInternal
68 - c-html5::RunFlushLoop
65 - c-layout::FlushPendingNotifications
62 - c-JS::CallEventHandler
13 - c-CSS::ProcessRestyles
2 - c-Input::nsInputStreamPump::OnStateTransfer
2 - c-network::nsHttpChannel::OnDataAvailable
1 - c-network::nsHttpChannel::OnStopRequest
1 - c-layout::nsLayoutUtils::DrawBackgroundImage

Comment 11

[Alex]

I am getting some comments confirming this behavior now that Firefox 12 is released on the videos in Comment 9 now that Firefox 12 is out.

Comment 12

[Alex]

DLBI seemed to make this problem a lot less significant.

Comment 13

[Wayne Mery (:wsmwk)]

Alex, do you still see this?
If yes, surely it must relate to some other already filed bug

Comment 14

[[e-mail]]

black screen in firefox 39 !!!

Comment 15

[Kanchan Kumari QA]

This is to confirm that problem can still be seen in the latest nightly. I could see the issue on my machine as well. Double checked on Chrome browser and things worked fine there.

Below is the version/other information:
Name     Firefox
Version     48.0a1
Build ID     20160411030231
User Agent     Mozilla/5.0 (Windows NT 10.0; WOW64; rv:48.0) Gecko/20100101 Firefox/48.

Adapter Description    Intel(R) HD Graphics
Adapter Drivers    igdumdim64 igd10iumd64 igd10iumd64 igd12umd64 igdumdim32 igd10iumd32 igd10iumd32 igd12umd32
Asynchronous Pan/Zoom    wheel input enabled; touch input enabled
Direct2D Enabled    true
DirectWrite Enabled    true (10.0.10240.16430)
Driver Version    10.18.15.4248
GPU #2 Active    false
GPU Accelerated Windows    1/1 Direct3D 11 (OMTC)
Subsys ID    106d1025
Supports Hardware H264 Decoding    Yes; Using D3D11 API
Vendor ID    0x8086
WebGL Renderer    Google Inc. -- ANGLE (Intel(R) HD Graphics Direct3D11 vs_5_0 ps_5_0)
windowLayerManagerRemote    true
AzureCanvasAccelerated    0

Comment 16

[Emil Pasca [:emilpasca]]

Attachment: youtube_issue.gif

User Agent:  Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:51.0) Gecko/20100101 Firefox/51.0

I have observed a similar behavior following the STR described below:

 1. Navigate to a video on yutube.com (eg: https://www.youtube.com/watch?v=Io0fBr1XBUA).
 2. Click on "Play all" from the side-bar.
 3. Skip the current playing video almost at the end of it.
 4. Minimize the browser before the next video starts to play then wait a few seconds (until you can hear the audio of the next video playing).
 5. Maximize the browser.
 6. Observe how the website content is rendered.

[Expected result]:
The video should be rendered correctly.

[Actual result]:
A black screen is rendered.

[Notes]:
 - Bug occurs when e10s is Enabled.
 - Attached a screen recording of the issue.
 
Should I log a new issue for this bug?

Comment 17

[Wayne Mery (:wsmwk)]

> Should I log a new issue for this bug?
I do not have a strong opinion. I have no special expertise in this area

Comment 18

[Milan Sreckovic [:milan] (needinfo for best results)]

(In reply to Emil Pasca [:emilpasca] from comment #16)
> Created attachment 8778169 [details]
> youtube_issue.gif
> 
> User Agent:  Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:51.0)
> Gecko/20100101 Firefox/51.0
> 
> I have observed a similar behavior following the STR described below:
> 
>  1. Navigate to a video on yutube.com (eg:
> https://www.youtube.com/watch?v=Io0fBr1XBUA).
>  2. Click on "Play all" from the side-bar.
>  3. Skip the current playing video almost at the end of it.
>  4. Minimize the browser before the next video starts to play then wait a
> few seconds (until you can hear the audio of the next video playing).
>  5. Maximize the browser.
>  6. Observe how the website content is rendered.
> 
> [Expected result]:
> The video should be rendered correctly.
> 
> [Actual result]:
> A black screen is rendered.
> 
> [Notes]:
>  - Bug occurs when e10s is Enabled.
>  - Attached a screen recording of the issue.
>  
> Should I log a new issue for this bug?

Yes please.  This bug is so out of date, that I'm just going to close it (we're talking about enabling D3D10, for instance.)

In the new bug, adding the about:support for your system would be helpful.