1596513 - [SVG] drop-shadow blur becomes solid when WebRender is on

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, defect, P3)

Description

[Harry Twyford]

Attachment: tab-drag-indicator.svg

I was working with the attached SVG and noticed that its blur becomes solid when WebRender is enabled. The issue only exists when it uses drop-shadow; it renders fine when using other forms of SVG shadows instead. The issue was resolved by disabling WebRender.

I'll attach screenshots of how it rendered when I pasted the markup into SVGOMG.

Comment 1

[Harry Twyford]

Attachment: WebRender_disabled.png

This is the way it should render, and how it does render when WebRender is disabled.

Comment 2

[Harry Twyford]

Attachment: WebRender_enabled.png

Here it is with WR enabled.

Comment 3

[Timothy Nikkel (:tnikkel)]

I still get the drop shadow blur with webrender enabled (tested on macOS). So maybe it's something specific to your setup.

Can you paste your about:support?

Comment 4

[Harry Twyford]

Attachment: about:support

```

Comment 5

[Connor Brewster [:cbrewster]]

WebRender is not taking into account scale factors when determining the standard deviation for the blur when rendering the shadow for drop-shadows.

Blur: https://searchfox.org/mozilla-central/source/gfx/wr/webrender/src/picture.rs#3980

Drop Shadow: https://searchfox.org/mozilla-central/source/gfx/wr/webrender/src/picture.rs#4122

Comment 6

[Brian Birtles (:birtles)]

See also the examples on https://developer.mozilla.org/en-US/docs/Web/SVG/Element/feDropShadow, two of which appear to be affected by this.

Comment 7

[Connor Brewster [:cbrewster]]

Attachment: Bug 1596513: Part 1: Take scale factors into account when rendering drop shadows in WebRender

Comment 8

[Connor Brewster [:cbrewster]]

Attachment: Bug 1596513: Part 2: Allow for floating point dx/dy on feDropShadow attributes

WebRender expects drop shadow dx/dy to be in user space, which is a floating point value.
Before this, a dx/dy such as (0.2, 0.2) was truncated to (0, 0) resulting in no offset of the shadow.

Depends on D63441

Comment 9

[Daniel Varga [:dvarga]]

Connor, this failed to land with the following error: applying /tmp/tmpbJJCpy 2 out of 2 hunks ignored -- saving rejects to file layout/svg/nsFilterInstance.cpp.rej abort: patch command failed: exited with status 256

Comment 10

[Connor Brewster [:cbrewster]]

Not sure what happened there. I have rebased the commits onto the latest central.

Comment 11

[Daniel Varga [:dvarga]]

Same thing: Details: applying /tmp/tmptoYfGg 2 out of 2 hunks ignored -- saving rejects to file layout/svg/nsFilterInstance.cpp.rej abort: patch command failed: exited with status 256

Comment 12

[Connor Brewster [:cbrewster]]

Looks like my commit was conflicting with a reformat commit that was in autoland (https://hg.mozilla.org/mozilla-central/rev/06762c633bf6f4609c31add3b3a6df7c2ee5373e). I've rebased onto central which now has the reformat commit. This should be able to be applied cleanly now.

Comment 13

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/889cee140af8
Part 1: Take scale factors into account when rendering drop shadows in WebRender r=gfx-reviewers,kvark
https://hg.mozilla.org/integration/autoland/rev/49e61b566da1
Part 2: Allow for floating point dx/dy on feDropShadow attributes r=nical

Comment 14

[Andreea Pavel [:apavel]]

Backed out for webrender wrench bustages

Push with failures: https://treeherder.mozilla.org/#/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception&revision=49e61b566da14b8e34e0efa332d9c7ff91625062&selectedJob=290314849

Failure log: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=290314849&repo=autoland&lineNumber=18520

Backout: https://hg.mozilla.org/integration/autoland/rev/e9a85eb47d024d9defa5e0a0eaf17c6403eddf98

Comment 15

[Connor Brewster [:cbrewster]]

Attachment: Bug 1596513: Part 3: Ensure drop shadow blur radius does not exceed MAX_BLUR after scale factors are applied

This was causing one of the large drop-shadow wrench reftests to timeout.
This is only a partial fix, as we should be checking the scale factors earlier on when sanitizing the
filter input. This will ensure we match what the non-WR backend is doing and will prevent overinflation.

Comment 16

[Noemi Erli[:noemi_erli]]

Tried to land latest changes but received:
"We're sorry, Autoland could not rebase your commits for you automatically. Please manually rebase your commits and try again. applying /tmp/tmp_sGwPN gfx/wr/webrender/src/picture.rs Hunk #1 FAILED at 96. 1 out of 2 hunks FAILED -- saving rejects to file gfx/wr/webrender/src/picture.rs.rej abort: patch command failed: exited with status 256 "

Comment 17

[Connor Brewster [:cbrewster]]

Rebased on top of central.

Comment 18

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/a9d8b62a3993
Part 1: Take scale factors into account when rendering drop shadows in WebRender r=gfx-reviewers,kvark
https://hg.mozilla.org/integration/autoland/rev/ce1593bf9d51
Part 2: Allow for floating point dx/dy on feDropShadow attributes r=nical
https://hg.mozilla.org/integration/autoland/rev/b7ab4c947688
Part 3: Ensure drop shadow blur radius does not exceed MAX_BLUR after scale factors are applied r=gfx-reviewers,nical

Comment 19

[Cristian Brindusan [:cbrindusan]]

https://hg.mozilla.org/mozilla-central/rev/a9d8b62a3993
https://hg.mozilla.org/mozilla-central/rev/ce1593bf9d51
https://hg.mozilla.org/mozilla-central/rev/b7ab4c947688