Open Bug 1422116 Opened 3 years ago Updated 3 years ago

http://keithclark.co.uk/labs/css-fps/nojs/ is broken because gfx.content.azure.backends begins with direct2d1.1

Categories

(Core :: Graphics, defect, P3)

x86_64
Windows 10
defect

Tracking

()

Tracking Status
firefox-esr52 --- affected
firefox57 --- affected
firefox58 --- affected
firefox59 --- affected

People

(Reporter: jan, Unassigned)

References

()

Details

(Keywords: correctness, nightly-community, Whiteboard: [gfx-noted])

Attachments

(3 files)

Nightly 59 x64 20171130101246 de_DE @ Windows 10 1709 (Radeon RX480)
fresh profile (without any modification/without webrender)

(Seen while testing WebRender bug 1422051 on Windows.)

http://keithclark.co.uk/labs/css-fps/nojs/

This demo is broken as long as gfx.content.azure.backends contains direct2d1.1.

If I set "gfx.content.azure.backends;skia,cairo" it looks like on Linux and is fine.
Attached image screenshot.png
Also affected:
* Beta 58.0b7 x64 20171127135700 de_DE @ Windows 10 1709 (Radeon RX480)
* Firefox 57.0.1 x64 20171128222554 de_DE @ Windows 10 1709 (Radeon RX480)
* ESR 52.5.0 x64 20171107091003 de_DE @ Windows 10 1709 (Radeon RX480)

Fixed by setting "gfx.content.azure.backends;skia,cairo".
Has STR: --- → yes
Interesting, it looks like nojs could be stuck before lighting is calculated? I'm having some difficulty reproducing this. Is this a regression?

If so running mozregression could give some more information. [1]

[1] http://mozilla.github.io/mozregression/
Flags: needinfo?(jan)
Whiteboard: [gfx-noted]
Attached image 2014-10-01_direct2d.png
http://keithclark.co.uk/labs/css-fps/nojs/

cmd @ Windows 10
mozregression --good 2014-02-01 --bad 2017-11-30 --pref "browser.tabs.remote.autostart:false" "browser.tabs.remote.autostart.1:false" "browser.tabs.remote.autostart.2:false"
> [...]
> 5:20.51 INFO: Narrowed nightly regression window from [2014-09-30, 2014-10-02] (2 days) to [2014-10-01, 2014-10-02] (1 days) (~0 steps left)
> 5:20.51 INFO: Got as far as we can go bisecting nightlies...
> 5:20.51 INFO: Last good revision: 14665b1de5ee (2014-10-01)
> 5:20.51 INFO: First bad revision: 5d6ec4dddf14 (2014-10-02)
> 5:20.52 INFO: Pushlog:
> https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=14665b1de5ee&tochange=5d6ec4dddf14
> [...]
> 7:01.50 WARNING: Skipping build 4ff52be673f6: Unable to find build info using the taskcluster route 'buildbot.revisions.4ff52be673f6ce9c95ac3af20dbac02b463cb5be.mozilla-central.win64'
> [...again and again...]

it may be this one:
> d954ed24e795	Bas Schouten — Bug 902952 - Part 2: Use Direct2D 1.1 where available. r=jrmuizel
which did:
> -pref("gfx.canvas.azure.backends", "direct2d,skia,cairo");
> -pref("gfx.content.azure.backends", "direct2d,cairo");
> +pref("gfx.canvas.azure.backends", "direct2d1.1,direct2d,skia,cairo");
> +pref("gfx.content.azure.backends", "direct2d1.1,direct2d,cairo");

Before, elements had textures (images), but they weren't always at the correct position (2014-10-01_direct2d.png).
Now, the raw structure (attachment 8933442 [details]) seems to be correct, but doesn't have textures.
Flags: needinfo?(jan)
Here is an intermediate step. The first day with direct2d1.1 where we only see the raw structure. Elements aren't always at the correct position.
Also that big vertical bar at the right side flickers.
Attachment #8935561 - Attachment description: before.png → 2014-10-02_direct2d1.1.png
You need to log in before you can comment on or make changes to this bug.