1373335 - Top border of Find Toolbar & Web Console blinking/flashing/flickering on black on scrolling

Status: VERIFIED FIXED

(Core :: Graphics: Layers, defect)

Description

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

STR:
1. Open some long website page to get scrollbar
2. Scroll with it or use mouse scroll button
3. See that top border of Find Toolbar blinking/flashing/flickering on black on scrolling

Comment 1

[Alice0775 White]

Regression window:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=7c4ca88d519f1f4a1ea4938cbba047e89d3ff3ec&tochange=45fde181a497a187d01d5412f5b72897c7520517
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=958d2a5d10091401fd5e900e8e063d21940c137e&tochange=7f894f791cdf170d788507d0eff30024ce699523

Regressed by: Bug 1359709 , Bug 1361970

Comment 2

[Alice0775 White]

I can also see the border blinking when scroll web content if Web Console(Ctrl+Shift+K) is opened

Comment 3

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

Thank you very much Alice0775 White for finding specific regression range!
For Web Console, the issue is most visible when it's in "Light" or "Firebug" theme.

Comment 5

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

[Tracking Requested - why for this release]: Regression

Comment 6

[Markus Stange [:mstange]]

The video on https://www.wired.com/story/flight-rail-vectorr-atmospheric-railway-train overlaps the chrome as a black box during scrolling. I'm going to assume that that's also this bug.

Comment 7

[Matt Woodrow (:mattwoodrow)]

Attachment: Bug 1373335 - Make sure we clip the opaque region in render target coordinate space, since complex transforms mean that we can't always have a rectangular representation of clips in layer space.

Review commit: https://reviewboard.mozilla.org/r/151084/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/151084/

Comment 8

[Matt Woodrow (:mattwoodrow)]

Try push: https://treeherder.mozilla.org/#/jobs?repo=try&revision=bd8be33db1e07d8fcc2ebcb25b027bcc802fa6f0

Comment 9

[Alice0775 White]

(In reply to Matt Woodrow (:mattwoodrow) from comment #8)
> Try push:
> https://treeherder.mozilla.org/#/
> jobs?repo=try&revision=bd8be33db1e07d8fcc2ebcb25b027bcc802fa6f0

FYI,
The try build fixes Bug 1373830 and Bug 1373528.
However, The try build does not fix Bug 1373479.

Comment 11

[Markus Stange [:mstange]]

Comment on attachment 8879741 [details]
Bug 1373335 - Make sure we clip the opaque region in render target coordinate space, since complex transforms mean that we can't always have a rectangular representation of clips in layer space.

https://reviewboard.mozilla.org/r/151084/#review156432

This is all so complicated... your patch looks plausible but I might have missed bugs.

::: gfx/layers/composite/LayerManagerComposite.cpp:386
(Diff revision 1)
>    for (Layer* child = aLayer->GetLastChild(); child; child = child->GetPrevSibling()) {
> -    PostProcessLayers(child, localOpaque, descendantsVisibleRegion, ancestorClipForChildren);
> +    MOZ_ASSERT(aLayer->AsContainerLayer()->UseIntermediateSurface());
> +    LayerComposite* childComposite = static_cast<LayerComposite*>(child->AsHostLayer());
> +    PostProcessLayers(child, localOpaque, descendantsVisibleRegion,
> +                      ViewAs<RenderTargetPixel>(childComposite->GetShadowClipRect(),
> +                                                PixelCastJustification::RenderTargetIsParentLayerForRoot),

not sure this PixelCastJustification completely matches what you're doing

Comment 12

[Matt Woodrow (:mattwoodrow)]

(In reply to Markus Stange [:mstange] from comment #11)
 
> not sure this PixelCastJustification completely matches what you're doing

It doesn't really, the true reason is RenderTargetIsParentLayerIfParentLayerHasIntermediateSurface, but that doesn't exist, so I went for the closest :)

Comment 13

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/e9c2da54cfed
Make sure we clip the opaque region in render target coordinate space, since complex transforms mean that we can't always have a rectangular representation of clips in layer space. r=mstange

Comment 14

[Markus Stange [:mstange]]

(In reply to Matt Woodrow (:mattwoodrow) from comment #12)
> (In reply to Markus Stange [:mstange] from comment #11)
>  
> > not sure this PixelCastJustification completely matches what you're doing
> 
> It doesn't really, the true reason is
> RenderTargetIsParentLayerIfParentLayerHasIntermediateSurface

Seems like it's worth adding :)

Comment 16

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/e9c2da54cfed

Comment 17

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

I'm marking this bug as VERIFIED, as issue was fixed, starting with Mozilla Firefox Nightly 56.0a1 (2017-06-23). Thanks.