1553828 - Displaying images in combination with transform: translateY(0) and mix-blend-mode: multiply

Status: RESOLVED FIXED

(Core :: Web Painting, defect, P2)

Description

[Andy Bestvater (:paquan)]

Attachment: Firefox67.png

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.14; rv:67.0) Gecko/20100101 Firefox/67.0

Steps to reproduce:

I placed an image with transform: translateY(0) and over the image I placed a div with text and a before-element with a background-color and mix-blend-mode: multiply.
https://codepen.io/paquan/pen/PvQJbX

Actual results:

The image disappears except of that part underneath the div with the mix-blend-mode.

Expected results:

The image should be fully displayed like in Firefox 66 or in Chrome.

Comment 1

[Alice0775 White]

I can reproduce the iss on Windows10 if HWA is disabled.

Regression window:
8a15f91d48c7541337d16c5e3f6f4551e7371e3d

Regressed by: Bug 1526941

Miko Mynttinen,
Your patch seems to cause the regression. Can you please look into this?

Comment 2

[Alice0775 White]

(In reply to Alice0775 White from comment #1)

Regression window:
8a15f91d48c7541337d16c5e3f6f4551e7371e3d

Oops, sorry, paste miss

Regression window:
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=8a15f91d48c7541337d16c5e3f6f4551e7371e3d&tochange=02fe822afa8c6252970f779dcbfd4a7b57fefd13

Comment 3

[Andy Bestvater (:paquan)]

I'm new to this and am pretty sure I did something wrong. Anyways, thank you @Alice0775 White for helping me out and thanks for checking this out.

Comment 4

[Miko Mynttinen]

(In reply to Andy Bestvater (:paquan) from comment #3)

I'm new to this and am pretty sure I did something wrong. Anyways, thank you @Alice0775 White for helping me out and thanks for checking this out.

Thank you for the report Andy. Having a testcase is always very appreciated. I have managed to reproduce this bug and I am working on getting it fixed. :)

Comment 5

[Miko Mynttinen]

Attachment: 1553828.html

Comment 6

[Miko Mynttinen]

Attachment: dl-dump.txt

Comment 7

[Miko Mynttinen]

Attached a further simplified testcase and a DL dump.

The testcase seems to work with WR enabled, so this looks like a FLB regression due to changed bounds calculation. Since the image sometimes loads, the bug might also be related to image decoding.

Comment 8

[Miko Mynttinen]

Attachment: 1553828.html

Simplified testcase without image.

Comment 9

[Miko Mynttinen]

Attachment: Bug 1553828 - Compute intermediate bounds for transform items that do not extend 3D context r=mattwoodrow

Comment 10

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/80e2c9488704
Compute intermediate bounds for transform items that do not extend 3D context r=mattwoodrow

Comment 11

[Stefan Hindli [:stefan_hindli]]

https://hg.mozilla.org/mozilla-central/rev/80e2c9488704

Comment 12

[Ryan VanderMeulen [:RyanVM]]

Is this something we should consider backporting to Beta ahead of the ESR or can it ride the trains to release?

Comment 13

[Miko Mynttinen]

(In reply to Ryan VanderMeulen [:RyanVM] from comment #12)

Is this something we should consider backporting to Beta ahead of the ESR or can it ride the trains to release?

I would like to get this uplifted.

Comment 14

[Miko Mynttinen]

Comment on attachment 9069809 [details]
Bug 1553828 - Compute intermediate bounds for transform items that do not extend 3D context r=mattwoodrow

Beta/Release Uplift Approval Request

• User impact if declined: Blend modes inside 3D transforms might not work correctly.
• Is this code covered by automated tests?: Yes
• Has the fix been verified in Nightly?: Yes
• Needs manual test from QE?: No
• If yes, steps to reproduce:
• List of other uplifts needed: None
• Risk to taking this patch: Low
• Why is the change risky/not risky? (and alternatives if risky): The patch is small and restores the previous behavior of 3D transform bounds calculations, which is well tested. There is also an automated testcase.
• String changes made/needed:

Comment 15

[Julien Cristau [:jcristau]]

Comment on attachment 9069809 [details]
Bug 1553828 - Compute intermediate bounds for transform items that do not extend 3D context r=mattwoodrow

painting fix for 68.0b9

Comment 16

[Razvan Maries]

I tried to uplift this but I got an conflict while merging on layout/painting/nsDisplayList.cpp

Comment 17

[Julien Cristau [:jcristau]]

https://hg.mozilla.org/releases/mozilla-beta/rev/cce82ca49cbd