Bug 62042

Summary: X11 mouse cursor alpha blending on OpenGL application broken
Product: xorg Reporter: Steven Noonan <steven>
Component: Server/GeneralAssignee: Xorg Project Team <xorg-team>
Status: RESOLVED MOVED QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium    
Version: unspecified   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Intel HD 4000 showing the issue
none
Radeon HD 6850 demonstrating the issue
none
Expected rendering
none
Intel HD 4000 showing the issue none

Description Steven Noonan 2013-03-09 00:58:36 UTC
Created attachment 76203 [details]
Intel HD 4000 showing the issue

I'm encountering a rendering issue with a alpha-blended X11 cursor (created with libxcursor and set with XDefineCursor) on top of an OpenGL application. The application in question is TOME4 (Tales of Maj'Eyal). It uses SDL 2.x to interact with the X11 cursor API and set up an OpenGL rendering context.

I've reproduced this on both a ThinkPad X230 (Ivy Bridge, Intel HD 4000 graphics) and on a Westmere desktop system with a Radeon HD 6850. Both are using open source drivers. I have *not* however been able to reproduce it on a system with an NVIDIA card using the proprietary NVIDIA driver. I *think* the common component between the two affected systems is mesa, but I'm not sure what components of the graphics stack could be implicated in this issue.

I have also attempted using mesa from git (728240b64d) as well as the mesa 9.1 release.

I'm attaching photos of the screens on both systems. Screenshots are unable to capture the issue, as the image is rendered correctly there (screenshot also attached for comparison).
Comment 1 Steven Noonan 2013-03-09 00:58:54 UTC
Created attachment 76204 [details]
Radeon HD 6850 demonstrating the issue
Comment 2 Steven Noonan 2013-03-09 00:59:08 UTC
Created attachment 76205 [details]
Expected rendering
Comment 3 Steven Noonan 2013-03-09 01:02:07 UTC
Created attachment 76206 [details]
Intel HD 4000 showing the issue

Re-attaching photo of Intel HD 4000 demonstrating the issue because it was incorrectly added as text/plain. Oops.
Comment 4 Michel Dänzer 2013-03-12 09:30:48 UTC
Mesa doesn't deal with the hardware cursor directly. The photos look like something isn't handling premultiplied vs. non-premultiplied alpha correctly, or some kind of format mismatch. That something could be the app or SDL, but reassigning to X for now.

(In reply to comment #3)
> Re-attaching photo of Intel HD 4000 demonstrating the issue because it was
> incorrectly added as text/plain. Oops.

You could have just fixed the MIME type in the attachment details. :)
Comment 5 Steven Noonan 2013-03-12 14:26:23 UTC
(In reply to comment #4)
> You could have just fixed the MIME type in the attachment details. :)

Well, since it read as text/plain, the page to change the MIME type also provided a text box to "edit" the file. I figured if I saved it, the binary data would be corrupted. safer to re-upload.
Comment 6 GitLab Migration User 2018-12-13 22:27:57 UTC
-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/xorg/xserver/issues/436.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.