1401665 - Thin line visible around a transformed primitive

Status: NEW

(Core :: Graphics: WebRender, defect, P2)

Description

[Darkspirit]

Attachment: 2017-09-20_20-28-53.mp4

Nightly 57 x64 20170920100426 de_DE @ Debian Testing (KDE / Radeon RX 480)
hw accel + webrender + webrendest + layersfree + blob-mimages + gpu-process

See attached video.

It's not a "bad rounded corner on a rotating element" like bug 1401631,
but a "bad corner on a rotating element".

Comment 1

[Darkspirit]

Attachment: wr-testcase.html

Comment 2

[Markus Stange [:mstange]]

Attachment: wr-testcase.html

I've tweaked the testcase a bit to make the issue easier to understand.

It looks like the image is clipped to the axis-aligned bounds of the element instead of the actual rotated shape of the element.

Comment 3

[Markus Stange [:mstange]]

https://github.com/servo/webrender/issues/1732

Comment 4

[Darkspirit]

Nightly 58 x64 20170929220356 de_DE @ Debian Testing (KDE/Radeon RX480)
gpu process + layers force accel + webrender + webrendest + blob-images

This got mostly fixed now, but:

(In reply to Markus Stange [:mstange] from comment #3)
> https://github.com/servo/webrender/issues/1732

> This wasn't completely fixed.
> The thin grey dotted line inside the blue square is still there although it shouldn't. Or could this be a different bug?

Comment 5

[Darkspirit]

Nightly 58 x64 20171005100211 de_DE @ Debian Testing (KDE / Radeon RX480)
gpu process + layers force accel + webrender + webrendest

attachment 8910430 [details] It's now looking as it should, but...

https://github.com/servo/webrender/issues/1797#issuecomment-334459144
> This wasn't completely fixed:
> The thin grey line can come back if
> * I zoom in https://bug1401665.bmoattachments.org/attachment.cgi?id=8910430
> * it's rotating https://bug1401665.bmoattachments.org/attachment.cgi?id=8910415

Comment 6

[Darkspirit]

Attachment: 2017-10-05_15-12-59.mp4

See comment 5

Comment 7

[Markus Stange [:mstange]]

Attachment: black line testcase

I can reproduce the remaining black line with this testcase at default zoom on a 2x display, and with 2x zoom on a 1x display.

Comment 8

[Darkspirit]

Attachment: 2017-11-11_03-49-38.mp4

Nightly 58 x64 20171110100139 de_DE 864174ac0707207f7fc698ed6c3c47c043e99395 @ Debian Testing (KDE, Radeon RX480)
fresh profile: layers.acceleration.force-enabled, gfx.webrender.enabled, gfx.webrender.highlight-painted-layers (with and without blob-images is the same)

This video shows the current state of this bug:
* known: The loading circle has an unwanted thin line.
* NEW IN THIS VIDEO: The loading circle appears for a moment on the top left corner when reloading the website. firefox-esr and chromium don't show this behavior. https://dev2.terrax.net/en/terrax (IPv6-only)
(I think I saw this in the past but thought that this could be a different bug.)

> https://github.com/servo/webrender/issues/1732
> Transformed images with transformed rectangle clips are not clipped properly
had been closed after the following patch:
> https://github.com/servo/webrender/pull/1760
> Fix local clip rect being applied in transform shader.

The (open) follow-up issue is:
> https://github.com/servo/webrender/issues/1797
> Thin line visible around a transformed primitive

Comment 9

[Darkspirit]

Attachment: 2017-12-21_03-56-47.mp4

Testcases from comment 1 + comment 7

Comment 10

[Darkspirit]

Attachment: 2018-07-20 22-18-55.mp4

In reply to https://github.com/servo/webrender/issues/1797#issuecomment-406556570:

Win 10, Nvidia GTX 1060 3GB, 2560x1440, (Right click on Desktop > Display Settings > Scaling: 100%)

try build from bug 1476636 comment 7:
mozregression --repo try --launch b3b87524864a797f75a5b4ce78532a82affc248e --pref gfx.webrender.all:true startup.homepage_welcome_url:"https://bug1401665.bmoattachments.org/attachment.cgi?id=8910415|https://bug1401665.bmoattachments.org/attachment.cgi?id=8910430|https://bug1401665.bmoattachments.org/attachment.cgi?id=8916296"

Comment 11

[Maire Reavy [:mreavy]]

We can't release this to the field, but we can let this ride to beta

Comment 12

[Darkspirit]

Debian Testing, KDE, Xorg, GTX 1060

mozregression --find-fix --bad 2018-09-01 --good 2018-09-08 --pref gfx.webrender.all:true -a https://bug1401665.bmoattachments.org/attachment.cgi?id=8910430 -a https://bug1401665.bmoattachments.org/attachment.cgi?id=8916296 -a https://bug1401665.bmoattachments.org/attachment.cgi?id=8910415
> 10:36.65 INFO: First good revision: c1583f2f382b23fa742faa4d50d6081ede64f309
> 10:36.65 INFO: Last bad revision: b04a830d12cdaa298b97a8135c10b64255d88ffe
> 10:36.65 INFO: Pushlog:
> https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=b04a830d12cdaa298b97a8135c10b64255d88ffe&tochange=c1583f2f382b23fa742faa4d50d6081ede64f309

> c1583f2f382b	Jeff Muizelaar — Bug 1486377. Avoid repeat sampling when not wanted. r=mstange

Markus' both testcases look like fixed.
The original testcase now only has a dotted line on one side (it was two before): https://bug1401665.bmoattachments.org/attachment.cgi?id=8910415

Comment 13

[Darkspirit]

> The original testcase now only has a dotted line on one side (it was two before)

It looks more like solid and light-grey.

Comment 14

[Chris Pearce [:cpearce (Not reading bugmail)]]

darkspirit: Is this fixed? I can't repro on Markus' test cases.

Comment 15

[Darkspirit]

(Jan Andre Ikenmeyer [:darkspirit] from comment #12)
> Markus' both testcases look like fixed.
> The original testcase now only has a dotted line on one side (it was two before): https://bug1401665.bmoattachments.org/attachment.cgi?id=8910415

Still the same.

Comment 16

[Darkspirit]

Attachment: 2018-10-01_22-27-56.mp4

https://bug1401665.bmoattachments.org/attachment.cgi?id=8910415

Comment 17

[Markus Stange [:mstange]]

The remaining issue might be bug 1452337 / bug 1447093.

Comment 18

[dalley]

I can't reproduce the black line issue. It's been a long time, this might have been fixed incidentally.

Comment 19

[Darkspirit]

I can still reproduce comment 16. (Even with SW-WR, but not with Basic Layers and Chromium.)

Comment 20

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

Bugbug thinks this bug is a regression, but please revert this change in case of error.

Comment 21

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

The severity field for this bug is relatively low, S3. However, the bug has 6 See Also bugs.
:gw, could you consider increasing the bug severity?

For more information, please visit auto_nag documentation.

Comment 22

[BugBot (nomail) [:suhaib / :marco/ :calixte]]

The last needinfo from me was triggered in error by recent activity on the bug. I'm clearing the needinfo since this is a very old bug and I don't know if it's still relevant.