1696819 - Color mapping is off on some cases

Status: RESOLVED FIXED

(Core :: Graphics: Color Management, defect)

Description

[x]

Attachment: index.png

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:87.0) Gecko/20100101 Firefox/87.0

Steps to reproduce:

Open different images on websites or this test image for example
https://commons.wikimedia.org/wiki/File:NTV7_Testpattern.png

Setting:
gfx.color_management.mode is on 2 (standard)

Commit that introduced the issue:
https://hg.mozilla.org/integration/autoland/rev/fa296c4907207b2d7658a290dfef7999a632c13b

Actual results:

Color offset is really noticeable for some colors by eye.

Attached image:
Hex value is #00ffff
In broken version it is #95fcfd (Firefox color picker & system utility)
In broken version with gfx.color_management.mode on 0 it looks good.

Expected results:

Correct colors 5 row 3 col should be #00ffff

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

Since the status are different for nightly and release, what's the status for beta?
For more information, please visit auto_nag documentation.

Comment 2

[Jeff Muizelaar [:jrmuizel]]

Can you attach the CMSOutputProfile section of about:support?

Comment 3

[Jeff Muizelaar [:jrmuizel]]

Can you also attach the output of: xprop -root XFree86_DDC_EDID1_RAWDATA

Comment 4

[Jeff Muizelaar [:jrmuizel]]

For my own reference, here's Chrome's handling of this kind of thing:
https://source.chromium.org/chromium/chromium/src/+/master:ui/display/util/display_util.cc;drc=65b725cc309dff780b21d4f289e86847b97200d6;bpv=1;bpt=1;l=75

x, do colors look correct in Chrome?

Comment 5

[x]

(In reply to Release mgmt bot [:sylvestre / :calixte / :marco for bugbug] from comment #1)

Since the status are different for nightly and release, what's the status for beta?
For more information, please visit auto_nag documentation.

Beta build is broken in latest bin version i tried - added the commit hash/link in the report

Comment 6

[x]

(In reply to Jeff Muizelaar [:jrmuizel] from comment #2)

Can you attach the CMSOutputProfile section of about:support?

CMSOutputProfile: AAABOAAAAAAAAAAAbW50clJHQiBYWVogAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAGclhZWgAAAMwAAAAUZ1hZWgAAAOAAAAAUYlhZWgAAAPQAAAAUclRSQwAAAQgAAAAOZ1RSQwAAARgAAAAOYlRSQwAAASgAAAAOWFlaIAAAAAAAAJ8uAABNoQAAAalYWVogAAAAAAAAMwsAAKXzAAAL0lhZWiAAAAAAAAAkmwAADGsAAMWwY3VydgAAAAAAAAABAjMAAGN1cnYAAAAAAAAAAQIzAABjdXJ2AAAAAAAAAAECMwAA

XFree86_DDC_EDID1_RAWDATA(INTEGER) = 0, -1, -1, -1, -1, -1, -1, 0, 77, 16, -115, 20, 0, 0, 0, 0, 5, 28, 1, 4, -91, 34, 19, 120, 14, -44, 19, -87, 83, 53, -68, 37, 12, 80, 82, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 77, -48, 0, -96, -16, 112, 62, -128, 48, 32, 53, 0, 88, -62, 16, 0, 0, 24, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -2, 0, 70, 78, 86, 68, 82, -128, 76, 81, 49, 53, 54, 68, 49, 0, 0, 0, 0, 0, 2, 65, 3, 40, 1, 18, 0, 0, 11, 1, 10, 32, 32, 0, -90

Chromium and every other program i tried including pre-change had the correct color code with an color picker 00ff00.

Comment 7

[Jeff Muizelaar [:jrmuizel]]

Chrome doesn't seem to use XFree86_DDC_EDID1. Instead it uses XRandR EDID.
https://source.chromium.org/chromium/chromium/src/+/master:ui/base/x/x11_display_util.cc;l=139;drc=01f1c9aca9c897a5ce6af3b7a0ca2517f069fcf8

Comment 8

[Jeff Muizelaar [:jrmuizel]]

x, can you also attach the contents of xrandr --verbose?

Comment 9

[x]

Attachment: xrandr

```

Comment 10

[x]

Comment 11

[x]

I tried it on another PC and the colors are also off in the testpattern, but not that far. In the testpatterns #00ffff is #14ffff for example

Comment 12

[Jeff Muizelaar [:jrmuizel]]

Attachment: Bug 1696819. Disable generating a profile from the EDID data.

This code doesn't seem to be working correctly and was broken
accidentally before. Intentionally break it for now to avoid the
regression.

Comment 13

[Jeff Muizelaar [:jrmuizel]]

Comment on attachment 9212679 [details]
Bug 1696819. Disable generating a profile from the EDID data.

Beta/Release Uplift Approval Request

• User impact if declined: Wrong colors sometimes on Linux
• Is this code covered by automated tests?: No
• Has the fix been verified in Nightly?: No
• 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): This reverts us to the same behaviour we had in 87
• String changes made/needed:

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/88275f615ea5
Disable generating a profile from the EDID data. r=aosmond

Comment 15

[Andreea Pavel [:apavel]]

https://hg.mozilla.org/mozilla-central/rev/88275f615ea5

Comment 16

[Ryan VanderMeulen [:RyanVM]]

Comment on attachment 9212679 [details]
Bug 1696819. Disable generating a profile from the EDID data.

Reverts us back to Fx87 behavior. Approved for 88.0b6.

Comment 17

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-beta/rev/c152c506ef05

Comment 18

[x]

Looks fixed in stable. V87 on Android is unusable. Colors are so far of. Black is more like #888 (not measured) and is horrible on an OLED.

Comment 19

[Jeff Muizelaar [:jrmuizel]]

x, the commit from your initial comment is not in 87. It sounds like the Android problem might be something different. Can you file a separate issue for that?

Comment 20

[x]

OK, but looks really similar and in nightly it is gone. I will test a little bit before i open an issue.