01:09 fdobridge: <n​orablackcat> @karolherbst but what about perfectly safe bindgen functions 😭
01:10 fdobridge: <k​arolherbst🐧🦀> they are safe by definition and therefore need no documentation
01:17 fdobridge: <n​orablackcat> If someone buys me what I need then you can consider it 4 more people :p
01:17 fdobridge: <n​orablackcat> If someone buys/sends me what I need then you can consider it 4 more people :p (edited)
02:16 fdobridge: <a​irlied> okay haven't reproduced but just force focused my brain into reading codepaths and I think I see a hole
02:38 fdobridge: <a​irlied> @karolherbst https://paste.centos.org/view/raw/84b52971
02:44 fdobridge: <a​irlied> attached to the gitlab MR
05:18 fdobridge: <m​arysaka> In those case there is generic comment I saw on rust stdlib for that being like "Always safe to call"
05:18 fdobridge: <m​arysaka> In those cases there is generic comment I saw on rust stdlib for that being like "Always safe to call" (edited)
05:20 fdobridge: <!​[NVK Whacker] Echo (she) 🇱🇹> I don't think Triang3l knows that much about NVIDIA GPUs
08:23 OftenTimeConsuming: Hmm, for some reason the nouveau on the 780 loves hanging from lots of context switches. I can easily replicate the hang by starting up i3 and holding the create virtual terminal key. Any suggestions for debugging?
08:24 OftenTimeConsuming: Should I change power modes and what to? Mode 0f seems to hang more readily.
08:38 fdobridge: <k​arolherbst🐧🦀> yeah I mean...
08:38 fdobridge: <k​arolherbst🐧🦀> cool, I'll give it a try
08:38 fdobridge: <m​arysaka> yeah it's not that great still :painpeko:
09:31 fdobridge: <k​arolherbst🐧🦀> @airlied sooo.. I think you were not hitting it because you ran the VK CTS. And my patches to convert the GL driver to the `bo_wait` model NVK uses seems to also mitigate the issue reliable enough so I don't hit it: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19543
09:32 fdobridge: <k​arolherbst🐧🦀> kinda funky
19:59 fdobridge: <a​irlied> No I've been running gl CTS and even threw some piglit at it
20:03 fdobridge: <k​arolherbst🐧🦀> interesting.. tu102 as well I assume?
20:03 fdobridge: <k​arolherbst🐧🦀> mhh
20:04 fdobridge: <k​arolherbst🐧🦀> maybe you need my ultra script 🙃
20:05 fdobridge: <k​arolherbst🐧🦀> https://gist.github.com/karolherbst/323e91cb7d77be060c02169a535bdb70
20:05 fdobridge: <k​arolherbst🐧🦀> (+ force enabling 460 for nvc0)
20:05 fdobridge: <k​arolherbst🐧🦀> anyway
20:05 fdobridge: <k​arolherbst🐧🦀> your patch clearly fixes buggy code and it does seem to help
20:52 fdobridge: <a​irlied> I just think my cpu wasn't fast enough to get those two paths to collide, or I added too many debug options to my kernel to make it slow down 😛
20:52 fdobridge: <k​arolherbst🐧🦀> ahhh
20:52 fdobridge: <k​arolherbst🐧🦀> fair
20:53 fdobridge: <k​arolherbst🐧🦀> I only enabled KASAN and my CPU is plenty fast though
20:53 fdobridge: <k​arolherbst🐧🦀> i7-12700 here