Open Bug 1705916 Opened 3 years ago Updated 2 years ago

foreignobject overflow

Categories

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

Product:
Core
Component:
Graphics: WebRender
Version:
Firefox 88
Platform:
x86_64
Windows 10
Type:
defect

Tracking

()

People

(Reporter: naraku7686, Unassigned)

References

(Blocks 1 open bug)

Details

(Keywords: regression)

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:88.0) Gecko/20100101 Firefox/88.0

Steps to reproduce:

https://jsfiddle.net/Naraku/4m3v0gb9/

inserted an overflowing html element at runtime inside a <foreignObject>

Actual results:

the dynamical element was not overflowing correcly and not triggering :hover selector properly

Expected results:

an identical "static" (pre-existent) element is rendering and hovering.
If there are "static" element pre-existent, the dynamic element is working too, up to the size of the overflow boundary defined by the pre-existent "static" elements and broken if it overflows past that point.
more details and live example inside the fiddle.

OS: Unspecified → Windows 10
Hardware: Unspecified → x86_64
Status: UNCONFIRMED → NEW
Component: Untriaged → Graphics: WebRender
Ever confirmed: true
Keywords: regression
Product: Firefox → Core
Blocks: wr-correctness
Severity: -- → S3
Flags: needinfo?(gwatson)
Priority: -- → P3

I was able to reproduce this locally and reduce a WR capture.

From what I can see, it looks like the clip_rect supplied in the display item for those text runs is incorrect (extending the clip rect in the WR capture scene.ron file makes it draw correctly).

So it seems like a bug in the display list creation or conversion code that gets sent to WR.

Miko, is this something you would be able to investigate? Or is there someone else who might be better to look at it?

Flags: needinfo?(gwatson) → needinfo?(mikokm)
Flags: needinfo?(mikokm)
You need to log in before you can comment on or make changes to this bug.