New issue
Advanced search Search tips

Issue 922899 link

Starred by 1 user

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Mac
Pri: 2
Type: Bug-Regression



Sign in to add a comment

Regression:Dark black colour is seen on Background in print preview overlay.

Project Member Reported by shru...@virtusa.com, Jan 17 (5 days ago)

Issue description

Chrome Version:73.0.3674.0 (Official Build) Revision	69010a60db037ed7b973da72c8e4171434473102-refs/branch-heads/3674@{#1}(64-bit)
OS : Mac(10.13.6, 10.13.1, 10.14.3)

What steps will reproduce the problem?
1. Launch chrome and Give print command.
2. Click on 'Change' button and Observe background.

Actual Result   : Dark black colour is seen on Background in print preview overlay.
Expected Result : Dark black colour should not be  seen on Background in print preview overlay. 

This is a regression issue broken in M-73 and will soon update bisect information:
Good Build :73.0.3673.0
Bad Build : 73.0.3674.0


Kindly refer the attached screen-cast.
Thank You.
 
 
Actual_result.mov
5.6 MB View Download
Expected_result.mov
6.3 MB View Download

Comment 1 by shru...@virtusa.com, Jan 17 (5 days ago)

Labels: hasbisect
Owner: dbeam@chromium.org
Status: Assigned (was: Unconfirmed)
Update :
Bisect-info:
https://chromium.googlesource.com/chromium/src/+log/f01172ea04d6dc7a4d50975a10e442fe16749d4a..fda5fb7a4c9a5a312eec6744c9442481fc8b9ba5?pretty=fuller&n=10000

Suspect:r623047 ?

@dbeam: Could you please check whether this is caused with respect to your change, if not please help us in assigning it to the right owner.

Note: 
1. Unable to provide 'per-revision' bisect as it shows "We don't have enough builds to bisect" error message for above range. (tried on other machines but still getting the same error again)
2. Hence providing suspect via perform Chromium bisect .
3. Issue is not seen on OS: Windows(7,8,10) and Linux(14.04 LTS)
4. Issue is also seen for other backgrounds Eg..Chrome export password overlay.

Thank You..

Comment 2 by dbeam@chromium.org, Jan 18 (5 days ago)

Cc: rbpotter@chromium.org
not seeing this problem on a local Chromium build (73.0.3675.0, r623833) nor a local Google Chrome 73.0.3673.0 dev on Linux

Comment 3 by dbeam@chromium.org, Jan 18 (5 days ago)

Labels: -Pri-1 Pri-2

Comment 4 by dbeam@chromium.org, Jan 18 (5 days ago)

Cc: andrescj@chromium.org dbeam@chromium.org
Owner: ----
yeah, don't think it was my change.  i also don't see any functional difference in the CSS that styles the dialog::backdrop in use here (just 60% black in both stable and canary).  I think they're just RENDERING differently.

+andrescj@ for https://crrev.com/c/1387385

note that there are some skia rolls in that range as well (nothing in the content of the rolls looked particularly offending, though).

Comment 6 by dbeam@chromium.org, Jan 18 (5 days ago)

> Issue is not seen on OS: Windows(7,8,10) and Linux(14.04 LTS)

btw, I eventually saw this (sorry I didn't earlier)

Comment 7 by shru...@virtusa.com, Jan 18 (4 days ago)

Cc: pkasting@chromium.org
Owner: andrescj@chromium.org
Update :
Re-bisected the issue on different machines and got the same CL using Chromium Bisect.

Change Log:
https://chromium.googlesource.com/chromium/src/+log/f01172ea04d6dc7a4d50975a10e442fe16749d4a..fda5fb7a4c9a5a312eec6744c9442481fc8b9ba5?pretty=fuller&n=10000

Suspect: r622972 ? 

@Andres Calderon Jaramillo: Could you please check whether this is caused with respect to your change, if not please help us in assigning it to the right owner.

Thank You..

Comment 8 by andrescj@chromium.org, Jan 18 (4 days ago)

Status: Started (was: Assigned)

Comment 9 by andrescj@chromium.org, Jan 18 (4 days ago)

Cc: dcasta...@chromium.org
Owner: ccameron@chromium.org
Status: Assigned (was: Started)
ccameron@: It looks like my CL [1] was the one that introduced the regression, but there seems to be a different underlying issue. dcastagna@ and I explored this further and we found that the issue does not come up when running chrome with --disable-mac-overlays. Here's a minimal example that should reproduce the issue:

<html>
<head>
<title>Alpha Test</title>
<style type="text/css">
#d1 {
  background-color: rgba(0, 0, 0, 0.6);
  width: 100px;
  height: 100px;
  position: absolute;
  top: 0;
  transform: translateZ(0);
}
</style>
</head>
<body>
</body>
ABC DEF
<div id="d1"></div>
</html>

We are hitting the workaround in [2]. At that point, the |background_color| was as expected (#99000000). We're not sure what's happening after that point that is drawing the color incorrectly. Perhaps some color space correction? IIRC, forcing |allow_solid_color_layers_| to be true in the constructor also fixed the issue. Can you look into this?

[1] https://crrev.com/c/1387385
[2] https://cs.chromium.org/chromium/src/ui/accelerated_widget_mac/ca_renderer_layer_tree.mm?l=384-389&rcl=6b8849c12913ae09934a4f4cea658cd3c7088bed

Sign in to add a comment