Open Bug 1736569 Opened 3 years ago Updated 8 months ago

CSS transform rotate fails

Categories

(Core :: Graphics: WebRender, defect)

Firefox 92
Unspecified
macOS
defect

Tracking

()

UNCONFIRMED

People

(Reporter: webmaster, Assigned: nical)

References

(Blocks 1 open bug, Regression)

Details

(Keywords: regression)

Attachments

(5 files)

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.13; rv:93.0) Gecko/20100101 Firefox/93.0
Firefox for Android

Steps to reproduce:

Applying the "transform: rotate (45deg);" style on an element.

Actual results:

The element disappears or is represented in a totally abnormal way.

This has been happening since Firefox update 92, only on my computer running OSX 10.13 (High Sierra), and update 93 didn't change anything.
For example, this is very visible on all leaflet cards: the small arrow under the popup is no longer visible.
https://leafletjs.com/examples/quick-start/example.html
By the way : the screenshot tool shows the element as it should be.

Expected results:

The element should be shown tilted 45 °.

The Bugbug bot thinks this bug should belong to the 'Core::CSS Parsing and Computation' component, and is moving the bug to that component. Please revert this change in case you think the bot is wrong.

Component: Untriaged → CSS Parsing and Computation
Product: Firefox → Core
Component: CSS Parsing and Computation → Graphics: WebRender
OS: Unspecified → macOS
Severity: -- → S3

(In reply to webmaster from comment #0)

https://leafletjs.com/examples/quick-start/example.html

I see the arrow on this page on my machine, both in Firefox Nightly and in Firefox 93 release.

It could be a driver specific issue.

When I saw your comment, I tried to disable the hardware acceleration, and it works ! Thank's a lot ! This bug caused me some trouble with a lot of websites.
In case you could do something with my config (witch is a standard iMac config) here are some details :
GPU : AMD Radeon HD 6770M 512 Mo
AMD Radeon HD 6770M :
Jeu de composants : AMD Radeon HD 6770M
Type : Processeur graphique (GPU)
Bus : PCIe
Longueur de la voie PCIe : x16
VRAM (dynamique, max.) : 512 Mo
Fournisseur : AMD (0x1002)
Identifiant du périphérique : 0x6740
Identifiant de révision : 0x0000
Révision de la ROM : 113-C0170F-175
Version VBIOS : 113-C29501-103
Version du gestionnaire EFI : 01.00.560
Moniteurs :
iMac :
Type de moniteur : LCD
Résolution : 2560 x 1440 (QHD/WQHD - Wide Quad High Definition)
L’interface graphique ressemble à ceci : 2560 x 1440
Profondeur de mémoire d’image : Couleurs 24 bits (ARGB8888)
Moniteur principal : Oui
Miroir : Désactivé
Connecté : Oui
Rotation : Géré
Régler la luminosité automatiquement : Non
Type de connexion : DisplayPort

Thank's again

Attached image Webkit, Gecko, Blink

This seems to be fixed.

Status: UNCONFIRMED → RESOLVED
Closed: 3 years ago
Resolution: --- → WORKSFORME

Hi !
I'm sorry but nothing changed for my configuration : when the material acceleration is enabled, the "transform: rotate (45deg)" css style is always represented in an abnormal way !
Regards

Firefox version : 98.0.2

Status: RESOLVED → UNCONFIRMED
Resolution: WORKSFORME → ---

Ha! so this is very specific indeed to the machine you are using probably. I'm on macos too and the test I just did was on Firefox Nightly 100.

I used this page for testing.
https://leafletjs.com/SlavaUkraini/examples/quick-start/

with from about:support

Graphics
Features
Compositing WebRender
Asynchronous Pan/Zoom wheel input enabled; scrollbar drag enabled; keyboard enabled; autoscroll enabled; smooth pinch-zoom enabled
WebGL 1 Driver WSI Info CGL
IsWebglOutOfProcessEnabled: 1
WebGL 1 Driver Renderer Intel Inc. -- Intel(R) Iris(TM) Plus Graphics OpenGL Engine
WebGL 1 Driver Version 4.1 INTEL-18.5.6

Hi ! Yes, this is very specific to my setup. My colleagues who have Macs with another OS and a different hardware configuration do not have this concern. It's really due to a problem between my GPU driver and Firefox when hardware acceleration is used.
The frustrating thing is that the 91.x version of Firefox worked perfectly, and the problem appeared with the 92.x version.

Would you be able to run mozregression?
I would do it but I can't reproduce with my configuration.
https://mozilla.github.io/mozregression/install.html
https://mozilla.github.io/mozregression/quickstart.html

The range to run mozregression would be
mozregression --good 91 --bad 93
if you run it on the command line.

That would take around 20/30 minutes and that would identify what commit broke your configuration.

Flags: needinfo?(webmaster)
Attached image mozregression screen

Thanks for trying to fix this bug. As I don't really know what you need, and I had never used mozregression, I am sending you a screenshot followed by the contents of the "log" and "build" windows.

Flags: needinfo?(webmaster)
Attached file build_infos.txt
Attached file log.txt

I hope it will help you to find the problem. Feel free to ask me other infos if you need.

That's super useful. Thanks a lot.
So potentially Bug 1711648

Regressed by: 1711648

:nical, since you are the author of the regressor, bug 1711648, could you take a look?
For more information, please visit auto_nag documentation.

Flags: needinfo?(nical.bugzilla)
Assignee: nobody → nical.bugzilla
Flags: needinfo?(nical.bugzilla)
Keywords: regression
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: