https://bugzilla.redhat.com/show_bug.cgi?id=2212099
Matthew Woehlke mwoehlke.floss@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |CLOSED Flags|needinfo?(mwoehlke.floss@gm | |ail.com) | Resolution|--- |WORKSFORME Last Closed| |2023-06-05 04:36:15
--- Comment #3 from Matthew Woehlke mwoehlke.floss@gmail.com --- That's... interesting. It seems to be behaving itself now.
I've rebooted, but it was doing it from a relatively fresh boot previously.
I'm going to chalk it up to either: - A *VERY* strange fluke. - The previous installation was corrupt, somehow. - Somehow caused by having run 3.3.fc36 previously followed by 3.5.fc38, and somehow fixed after running 3.4.fc38?
I'd been using a somewhat-old version of the fc36 package for some time (circa August 2022 at a guess). A few days ago, Blender wedged my machine (thank you Intel GPU drivers), which I used as an excuse to upgrade to fc38. I installed the latest fc36 version (while the older version was running, I think). I don't recall if I launched the latest fc36 version. I then did a distro-upgrade. After my first boot into fc38, it was broken. I subsequently downgraded to 3.4.fc38, which worked. After upgrading back to 3.5.1-3.fc38, it seems okay. I have no clue if something peculiar to that chain of actions caused it, or what.
As a professional software developer that's sometimes worked with 3D code, I'm confident what I saw was completely bogus and often non-finite vertex positions being generated, probably with invalid normals (i.e. values not in [-1, +1]) as well (explaining why, when I could see *anything*, it was solid black). What would cause *that specifically* (i.e. subdivision) to go haywire without Blender just plain crashing, I have not a clue. Very, very strange.
I guess I'll close this; if it happens again, I'll re-open with a screen shot. Meanwhile, this will be 'on file' in case anyone else runs into the same issue.