1196539 - [e10s] After switching the tab, flash plug-in would not repaint until mouse hover the plug-in or scrollbar

Status: RESOLVED FIXED

(Core Graveyard :: Plug-ins, defect)

Description

[Alice0775 White]

Steps to reproduce:
1. Open 2 tabs
   about:home and http://edition.cnn.com/2014/11/23/showbiz/music/katy-perry-super-bowl/index.html?hpt=en_bn1
2. Wait for the video autoplay
3. Switch between tabs

Actual Results:
Flush plug-in would not repaint until mouse hover the plug-in or scrollbar

Expected results:
The flash plugin should be repainted immediately after switching tab

Regression window:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=f00b7bf7b9f83d74ca8413d04da4942c8bcc84db&tochange=0f5421c28b14

Suspect:  Bug 1163570

Comment 1

[Alice0775 White]

Attachment: screencap zipped ogg

Comment 2

[Alice0775 White]

Attachment: dance.swf

Comment 3

[Alice0775 White]

Attachment: testcase html

Steps to reproduce:
1. Open 2 tabs
   about:home and the attached testcase
2. Wait for the Flash autoplay
3. Switch between tabs

Actual Results:
Flush plug-in would not repaint until mouse hover the scrollbar or resize window

Expected results:
The flash plugin should be repainted immediately after switching tab

Comment 4

[Alice0775 White]

This problem was also reproduced on ubuntu14.04
https://hg.mozilla.org/mozilla-central/rev/095988abdc560bf8ba07a94a425c6922a3e9bfd6
Mozilla/5.0 (X11; Linux i686; rv:43.0) Gecko/20100101 Firefox/43.0 ID:20150821030204

Comment 6

[Jim Mathies [:jimm]]

You don't need the html test case really, just load the swf directly.

Comment 7

[Jim Mathies [:jimm]]

I can fix this, but it's going to relay on some code in bug 1137944.

Comment 8

[Liz Henry (:lizzard)]

[Tracking Requested - why for this release]:

[Tracking Requested - why for this release]:

This seems likely to affect Firefox 40 and onwards if the culprit is bug 1163570.

Comment 9

[Jim Mathies [:jimm]]

bug 1163570 is only in 43 at this point.

Comment 10

[Liz Henry (:lizzard)]

Thanks Jim, I misread version reported vs. target milestone!

Comment 11

[Chris Peterson [:cpeterson]]

I also see this problem playing Silverlight video on Netflix. The plugin tab does not repaint, but I can still hear Silverlight audio playing. If you switch back and forth between the tabs, the plugin tab will sometimes repaint.

I can reproduce this in a Windows 8.1 VM and Windows 10 VM, but not OS X. I don't have a real Windows machine to test. Maybe this is a VM painting issue?

I can only reproduce the problem when e10s is enabled.

Comment 12

[Jim Mathies [:jimm]]

(In reply to Chris Peterson [:cpeterson] from comment #11)
> I also see this problem playing Silverlight video on Netflix. The plugin tab
> does not repaint, but I can still hear Silverlight audio playing. If you
> switch back and forth between the tabs, the plugin tab will sometimes
> repaint.
> 
> I can reproduce this in a Windows 8.1 VM and Windows 10 VM, but not OS X. I
> don't have a real Windows machine to test. Maybe this is a VM painting issue?
> 
> I can only reproduce the problem when e10s is enabled.

Thanks for the details report. This is related to a problem with compositor code that manages plugin windows.

Comment 13

[Jim Mathies [:jimm]]

Attachment: use AutoResolveRefLayers logic to update plugins

More reliable fix than the checks I was using for roots and such via composition events.

Comment 14

[Jim Mathies [:jimm]]

Attachment: tests

Comment 15

[Jim Mathies [:jimm]]

Comment on attachment 8658437 [details] [diff] [review]
use AutoResolveRefLayers logic to update plugins

Review of attachment 8658437 [details] [diff] [review]:
-----------------------------------------------------------------

Matt, mind looking this over?

Comment 16

[Jim Mathies [:jimm]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=f1117158b7ef

Comment 17

[Jim Mathies [:jimm]]

Attachment: use AutoResolveRefLayers logic to update plugins

fixes a build failure due to an unused variable.

Comment 18

[Matt Woodrow (:mattwoodrow)]

Comment on attachment 8658668 [details] [diff] [review]
use AutoResolveRefLayers logic to update plugins

Review of attachment 8658668 [details] [diff] [review]:
-----------------------------------------------------------------

::: gfx/layers/ipc/CompositorParent.cpp
@@ +1987,5 @@
>  }
>  
>  #if defined(XP_WIN) || defined(MOZ_WIDGET_GTK)
> +// static, sends plugin window state changes to the main thread
> +void 

Trailing whitespace

Comment 19

[Matt Woodrow (:mattwoodrow)]

Comment on attachment 8658439 [details] [diff] [review]
tests

Review of attachment 8658439 [details] [diff] [review]:
-----------------------------------------------------------------

I'm not really the best person to review JS test additions.

::: dom/plugins/test/testplugin/nptest_windows.cpp
@@ +50,5 @@
>    HWND childWindow;
>    ID3D10Device1 *device;
>    ID3D10Texture2D *frontBuffer;
>    ID3D10Texture2D *backBuffer;
> +  uint32_t showWindowCount;

What is this added for? Doesn't seem to be used anywhere.

Comment 20

[Jim Mathies [:jimm]]

(In reply to Matt Woodrow (:mattwoodrow) from comment #19)
> Comment on attachment 8658439 [details] [diff] [review]
> tests
> 
> Review of attachment 8658439 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> I'm not really the best person to review JS test additions.
> 
> ::: dom/plugins/test/testplugin/nptest_windows.cpp
> @@ +50,5 @@
> >    HWND childWindow;
> >    ID3D10Device1 *device;
> >    ID3D10Texture2D *frontBuffer;
> >    ID3D10Texture2D *backBuffer;
> > +  uint32_t showWindowCount;
> 
> What is this added for? Doesn't seem to be used anywhere.

sorry, that was bleed through from another test patch. I'll bounce this off roc, he already reviewed a couple of these for me.

Comment 21

[Jim Mathies [:jimm]]

Attachment: added tests

Comment 22

[Jim Mathies [:jimm]]

Comment on attachment 8658750 [details] [diff] [review]
added tests

more plugin + compositor tests

Comment 23

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/7d64aea33c44
https://hg.mozilla.org/integration/mozilla-inbound/rev/34bd669481c2

Comment 24

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/7d64aea33c44
https://hg.mozilla.org/mozilla-central/rev/34bd669481c2