09:38MrCooper: time to restrict access to the main mesa repo? https://gitlab.freedesktop.org/mesa/mesa/-/tags/libclc-overwritten
10:51daniels: MrCooper: the two wrinkles I can think of is that stable branches aren't gated by CI, and also that tags + tarballs are manually generated and pushed rather than coming out of a pipeline; apart from that we could just restrict pushes to maintainers only, make Marge a maintainer, and trust our maintainers to not screw up :P
10:51bnieuwenhuizen: daniels: we could limit it to just marge + release managers though?
10:51bnieuwenhuizen: doesn't eliminate the risk but reduces it by a lot :)
10:52daniels: bnieuwenhuizen: our only levers are no-one / everyone / developers + maintainers / maintainers
10:52daniels: we can't put specific named people in rules
10:52daniels: (that comes in the pay-for edition)
10:52bnieuwenhuizen: can't you make release managers maintainers?
10:53daniels: we can!
10:53bnieuwenhuizen: also does the maintainer lists include owners? that is a reasonably big list already there?
10:53daniels: right, owners have all maintainer permissions
10:54daniels: so we can make RMs maintainers, but we still allow all the others to screw up, unless we remove them all and then only the RMs can actually add people into the project, etc
10:55bnieuwenhuizen: try to do a push for owners to only have a https remote for the main repo?
10:55bnieuwenhuizen: (yes that won't be perfectly enforced ...)
10:55daniels: no, it's far from perfect, but it's a very good pragmatic solution I think :)
10:58MrCooper: Marge is a GitLab admin, doesn't that imply maintainer status as well?
11:01daniels: wait, why is Marge an admin?
11:02daniels: I do wonder if it implies maintainer access - I know that I can go merge things in projects I'm not part of, but not sure if that would pass the protection check
11:03daniels: in order to delete that tag I had to first make the tag unprotected, then delete it, then reprotect all the tags
11:03daniels: (we should also do some tag/branch gardening tbh)
12:37pcercuei: Is it possible to give my KMS driver a CMA area, where it will allocate dumb buffers from?
16:16pcercuei: drm-misc-fixes isn't periodically merged into drm-misc-next?
18:07DPA: The program linked over here: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3425#note_602968Will it fail with any driver?
20:54DPA: I just found the piglit tests. On the librem5 with etnaviv, when running the sanity tests, they all get sikipped or fail.
20:54DPA: Of the opengl tests, only 7 pass (8 if I set MESA_GL_VERSION_OVERRIDE=2.1) Is this normal, or am I doing something wrong?
21:02DPA: Nevermind, looks like some of those need a running Xserver.
21:10dcbaker[m]: Most of then should work with whatever backend you select. I think most of us use gbm