1420754 - regression: Animation rendered all over the place with webrender

Status: RESOLVED WORKSFORME

(Core :: Graphics: WebRender, defect, P1)

Description

[noszalyaron4]

Attachment: withwebrender.png

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:59.0) Gecko/20100101 Firefox/59.0
Build ID: 20171126100425

Steps to reproduce:

Visited https://www.reddit.com/r/dankmemes/ with webrender enabled on latest nightly (standard way: gfx.webrender.enabled, gfx.webrender.blob-images and layers.acceleration.force-enabled enabled on a new profile).  My GPU is integrated intel. 


Actual results:

The animation of the road goes all over the place (there're pieces on chrome, web content and even in dev tools, see my attachments)

Comment 1

[noszalyaron4]

Attachment: devtools.png

Comment 2

[noszalyaron4]

Attachment: withoutwebrender.png

Comment 3

[Darkspirit]

Thank you! Confirmed in
Nightly 59 x64 20171126100425 de_DE 3e14872b31a7b1b207605d09b78fbaaf21f1bba7 @ Debian Testing (KDE, Radeon RX480).
fresh profile: layers.acceleration.force-enabled, gfx.webrender.enabled, gfx.webrender.blob-images

The forward button, the animation when hovering a tab and the animation of the history arrow when hovering the locationbar are showing anomalies.

-----

https://www.reddit.com/r/dankmemes/

The surface the car is driving on wasn't shown in the past (bug), but these visual anomalies are new (regression).

mozregression --good 2017-09-15 --bad 2017-11-26 --pref "layers.acceleration.force-enabled:true" "gfx.webrender.enabled:true" "gfx.webrendest.enabled:true" "gfx.webrender.layers-free:true" "gfx.webrender.blob-images:true"
> 20:12.99 INFO: Last good revision: f7b7bdd7b05c992b4fb435a26a6278829063748b
> 20:12.99 INFO: First bad revision: b7bc562d77f1f987b623c11cd6e6713f7be2c846
> 20:12.99 INFO: Pushlog:
> https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=f7b7bdd7b05c992b4fb435a26a6278829063748b&tochange=b7bc562d77f1f987b623c11cd6e6713f7be2c846

> b7bc562d77f1	Kartikaya Gupta — Bug 1419440 - Update for API change in WR PR 2082. r=mstange
> 85a4275ed544	Dzmitry Malyshau — Bug 1419440 - Update for API changes in WR PR 2050. r=kats
> 4bd54453fd2e	Kartikaya Gupta — Bug 1419440 - Update webrender to commit b7b07562fda338fcb2faff66ce01aafb6235fbcf. r=mstange

Comment 4

[Kartikaya Gupta (email:[e-mail])]

This is a regression from the same WR range as bug 1420748:

*   02d9b6d9 Auto merge of #2078 - kvark:rtree, r=glennw
|\
| * 08a32a9e RenderTaskTree refactor
* |   0cc47eb8 Auto merge of #2072 - nical:dbg-flags, r=glennw
|\ \
| * | b32c990d Make servo-tidy happy.
| * | 72161329 Add toggle_debug_flags.
| * | 1365e063 Properly handle debug flags when creating the renderer.
| * | 0d934778 Use DebugFlags for time and sample queries.
|  /
* |   e83612df Auto merge of #2069 - glennw:xf, r=kvark
|\ \
| * | 0aebe2b7 Address review comments
| * | 8253bbb3 Remove manual interpolation of local space in transform shaders.
|  /
* | 00e70b3c Auto merge of #2077 - kvark:pass-kind, r=glennw
|/
* 40349b38 Strong typing for the main framebuffer pass

Comment 5

[Kartikaya Gupta (email:[e-mail])]

I narrowed down the regression range to servo/webrender#2069

Comment 6

[Kartikaya Gupta (email:[e-mail])]

As with bug 1420748 this has a corresponding issue in GH that is assigned to Glenn.

Comment 7

[Andrew Overholt [:overholt]]

Given we're not going to let WR ride in 59, I'm going to mark it "disabled" so we don't have to track this regression elsewhere.

Comment 8

[Kartikaya Gupta (email:[e-mail])]

This should be fixed by the next WR update

Comment 9

[Kartikaya Gupta (email:[e-mail])]

This seems to still happen in the latest nightly.

Comment 10

[Darkspirit]

Not fixed. Win10 (GeForce GTX 1060 with/without ANGLE), Debian Testing (Radeon RX 480), 2560x1440 (Dell U2515H)

Comment 11

[Kartikaya Gupta (email:[e-mail])]

(In reply to Jan Andre Ikenmeyer [:darkspirit] from comment #10)
> Not fixed. Win10 (GeForce GTX 1060 with/without ANGLE), Debian Testing
> (Radeon RX 480), 2560x1440 (Dell U2515H)

Do you still see the anomalies showing up in browser chrome areas? The only issue I see on the page now (on Windows, anyway) is the car's driving surface is missing, which was a pre-existing bug and less severe.

Comment 12

[Kartikaya Gupta (email:[e-mail])]

Same question (comment 11) to the original reporter.

Comment 13

[Darkspirit]

(In reply to Kartikaya Gupta (email:[e-mail]) from comment #11)
> Do you still see the anomalies showing up in browser chrome areas? The only issue I see on the page now (on Windows, anyway) is the car's driving surface is missing, which was a pre-existing bug and less severe.

On Win10 with GTX 1060 it's as described by you.

Comment 14

[Kartikaya Gupta (email:[e-mail])]

Thanks. I filed bug 1470905 for the driving surface issue, and will close this bug since it doesn't seem to be happening any more. Would still like the original reporter to verify if possible.

Comment 15

[noszalyaron4]

I can verify too that the issue is gone and only the driving surface is missing with intel igpu on linux.