1057716 - [OMTC] Popping shadows

Status: VERIFIED FIXED

(Core :: Graphics: Layers, defect)

Description

[Elbart]

Attachment: screencap.mp4

Moving the mouse over the titles under "Recent Threads" on the right side opens an animated preview-balloon with a shadow-effect. In about every other try, the shadow has some sort of popping-effect shortly before the animation is finished.

Doesn't happen with OMTC disabled.

Application Basics
------------------

Name: Firefox
Version: 34.0a1
User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:34.0) Gecko/20100101 Firefox/34.0
Multiprocess Windows: 0/1

Graphics
--------

Adapter Description: NVIDIA GeForce GT 640
Adapter Drivers: nvd3dumx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um
Adapter RAM: 2048
ClearType Parameters: Gamma: 2200 Pixel Structure: R ClearType Level: 100 Enhanced Contrast: 200
Device ID: 0x0fc1
Direct2D Enabled: true
DirectWrite Enabled: true (6.2.9200.16571)
Driver Date: 7-2-2014
Driver Version: 9.18.13.4052
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 11 (OMTC)
Subsys ID: 125819da
Vendor ID: 0x10de
WebGL Renderer: Google Inc. -- ANGLE (NVIDIA GeForce GT 640 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote: true
AzureCanvasBackend: direct2d
AzureContentBackend: direct2d
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

Important Modified Preferences
------------------------------

gfx.direct3d.last_used_feature_level_idx: 0

Comment 1

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

Jeff, can you take this on to speed up OMTC blockers resolution?

Comment 2

[Jeff Muizelaar [:jrmuizel]]

I can't seem to reproduce this on Intel hardware.

Comment 3

[Jeff Muizelaar [:jrmuizel]]

Attachment: before.png

Comment 4

[Jeff Muizelaar [:jrmuizel]]

Attachment: during.png

Comment 5

[Jeff Muizelaar [:jrmuizel]]

Attachment: after.png

Comment 6

[Jeff Muizelaar [:jrmuizel]]

I just confirmed that running the same apitrace that reproduces the problem on an nvidia driver does not reproduce the problem on WARP

Comment 7

[Jeff Muizelaar [:jrmuizel]]

The difference seems to be the results of a RenderTarget after a CopySubresourceRegion call.

Comment 8

[Jeff Muizelaar [:jrmuizel]]

Interestingly enough the CopySubresourceRegion call doesn't seem to be actually copying the background into the resource as we would expect it to.

Comment 9

[Jeff Muizelaar [:jrmuizel]]

Attachment: This might help

Comment 10

[Bas Schouten (:bas.schouten)]

(In reply to Jeff Muizelaar [:jrmuizel] from comment #9)
> Created attachment 8478559 [details] [diff] [review]
> This might help

I had to make some adjustments to make it build and not crash. But it at least doesn't seem to be solving the problem for me. Interestingly the bug I'm looking at is also NVidia only, and -also- CopySubresourceRegion related. But in the case of what I'm looking at it didn't feel that CopySubResourceRegion wasn't working. But maybe that idea is wrong.

Comment 11

[Jeff Muizelaar [:jrmuizel]]

Attachment: Don't use a zero sized box to copy the background

This should do the trick

Comment 12

[Bas Schouten (:bas.schouten)]

Comment on attachment 8479031 [details] [diff] [review]
Don't use a zero sized box to copy the background

Review of attachment 8479031 [details] [diff] [review]:
-----------------------------------------------------------------

::: gfx/layers/d3d11/CompositorD3D11.cpp
@@ +450,5 @@
>      srcBox.top = aSourcePoint.y;
>      srcBox.front = 0;
>      srcBox.right = aSourcePoint.x + aRect.width;
>      srcBox.bottom = aSourcePoint.y + aRect.height;
> +    srcBox.back = 1;

I bow my head in shame for r+'ing this.

Comment 13

[Jeff Muizelaar [:jrmuizel]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/b5ccef6b01ec

Comment 14

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

https://hg.mozilla.org/mozilla-central/rev/b5ccef6b01ec

Comment 15

[Lawrence Mandel [:lmandel] (use needinfo)]

This just landed on m-c. Assuming the fix is good, please request approval to uplift to 33.

Comment 16

[Sylvestre Ledru [:Sylvestre]]

Jeff, could you fill the uplift request for 33? Thanks

Comment 17

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

Comment on attachment 8479031 [details] [diff] [review]
Don't use a zero sized box to copy the background

This is for the uplift to 33, which by now is Beta?  Either way, this only needs to make it to 33.

Approval Request Comment
[Feature/regressing bug #]: 899785
[User impact if declined]: Intermittent correctness problems on Nvidia/Windows. 
[Describe test coverage new/current, TBPL]:
[Risks and why]: Low.
[String/UUID change made/needed]: n/a

Comment 18

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-beta/rev/9eb4dff42df0

Comment 19

[Ada [:adalucinet]]

Couldn't reproduce this issue on Nightly 2014-08-23 nor 2014-08-25 with STR from comment 0 on 4 different Windows 7 64-bit machines - 1 AMD and 3 nVidia (NVIDIA GeForce 210 and NVIDIA GeForce GT 620). I see no difference between those Nightly builds and latest Nightly 35.0a1 (Build ID: 20140910030204) or Firefox 33 beta 2 (Build ID: 20140908190852). Am I missing something here?
Elbart, could you please look into this with Firefox 33 beta 2 [1]? Is this issue still reproducible for you? Thanks in advance! 

[1] - http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/33.0b2-candidates/build1/win32/en-US/

Comment 20

[Elbart]

(In reply to Alexandra Lucinet, QA Mentor [:adalucinet] from comment #19)

> Is this
> issue still reproducible for you?

No.

Comment 21

[Ada [:adalucinet]]

Elbart, I need your help again. Could you please verify this fix on latest Aurora 34.0a2 [1] too? Many thanks!

[1] http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-mozilla-aurora/

Comment 22

[Elbart]

(In reply to Alexandra Lucinet, QA Mentor [:adalucinet] from comment #21)
> Could you please verify this fix on latest
> Aurora 34.0a2 [1] too?

Yes, fixed.