1268442 - Failure when generating mipmap (regression)

Status: RESOLVED DUPLICATE of bug 1242347

(Core :: Graphics: CanvasWebGL, defect)

Description

[stephane]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/50.0.2661.87 Safari/537.36

Steps to reproduce:

https://sketchfab.com/models/37e04564b61140f9bb4c89a73b3d84fd
^ WebGL fails at generate mipmap and it ends up with a black texture.

Version 46 fails on both OSX and Windows.
Nightly FF, FF45, chrome, opera, edge don't have issue.


https://www.khronos.org/registry/webgl/sdk/tests/webgl-conformance-tests.html
I don't know if it can help but the test : conformance/textures/misc/texture-npot.html is failing with the error :
gl.texImage2D and gl.generateMipmap with POT texture at level 0 should succeed

Comment 1

[stephane]

Note that this particular test is not failing on nightly and other browsers.
(Also the issue is reproducible on many sketchfab model that have mipmap filtering)

Comment 2

[Loic]

Attachment: Capture d'écran 2016-04-28 à 13.38.17.png

It seems to work for me with FF46 on Win 7 64b.

Could you attach a screenshot of what you see.

In addition, could you test with a fresh profile:
https://support.mozilla.org/en-US/kb/profile-manager-create-and-remove-firefox-profiles

Comment 3

[Loic]

NVM, I see the diff with FF49. :) In fact, there is a regression.

Comment 4

[stephane]

Great!

So far it looks like the mipmap bug occurs only with the textures defined as LUMINANCE.
Not 100% sure about that though, I didn't really dig more than that.

Comment 5

[Loic]

Fix range:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=a1b94785b48437662c696c9143ed92639e9617b2&tochange=ac21d6f878e6fdd69ec4140d1b63a94601524abe

Fixed by:
Morris Tseng — Bug 1242347 - Allow unsized internal format when generate mipmap. r=jgilbert

It's fixed in 47+.

Comment 7

[Loic]

For the record, it has been broken in 46 by bug 1239259 and fixed a few days later by bug 1242347 in 47.

Comment 8

[stephane]

Glad that the issue has been fixed!
Do you know if it will be backported/hotfixed in 46, or if we have to wait for 47?

Comment 9

[Loic]

I have requested tracking for 46 for the fix in bug 1242347. Follow bug 1242347 if you want more details.

Comment 10

[stephane]

Thanks!