00:43 airlied[d]: _lyude[d]: the sb fix breaks on blackwell because you try to setup fwsec on machines which don't have it
03:46 kode54: matt_schwartz[d]: it fixes my problem with booting into it
03:46 kode54: I should file the other OOPSes as separate issues
03:46 kode54: be sure to check your dmesg if it might happen to you as well
03:47 kode54: it involved dragging windows between outputs on Plasma
03:47 kode54: maybe Ghostty specifically
04:24 airlied[d]: _lyude[d]: https://lore.kernel.org/dri-devel/20260102041829.2748009-1-airlied@gmail.com/T/#u
04:24 airlied[d]: anyone experience broken blackwell, please try the above
04:24 airlied[d]: matt_schwartz[d]
04:27 matt_schwartz[d]: will try a build with that in a bit :frog_thumbs_up:
04:55 kode54: building now
05:01 kode54: hey, airlied[d], would this be the prepare_fb warnings you mentioned in the latest drm pull? https://gist.github.com/kode54/7334c0fd4c933e6e1846873b0b02e1c8
05:02 airlied[d]: Yes
05:02 kode54: ah, great, that's my only other issue for now
05:17 matt_schwartz[d]: airlied[d]: patch seems to work fine here, ty
05:56 kode54: matt_schwartz[d]: you'll want that latest nouveau commit from drm-fixes-2026-01-02 as well
05:56 kode54: if you aren't already running some sort of drm fixes build
05:58 matt_schwartz[d]: I don’t really use nouveau :blobcatnotlikethis: but thanks
06:03 kode54: I don't really use it either
06:03 kode54: but I do test it occasionally
06:04 kode54: strangely it also makes my machine use 40 watts less power
06:04 kode54: could also be because it's using 8bpc video and no HDR output
06:08 kode54: airlied[d]: also verified, thanks
06:08 kode54: and the warnings patch verified as well
06:11 kode54: should we add our T-b to replies on the list?
06:14 kode54: matt_schwartz[d]: should I add a Closes for https://lore.kernel.org/all/176682185563.8256.115798774340102079@copycat/ or is that not a valid bug report since I didn't bisect it
06:14 kode54: I can at least add a T-b
06:19 kode54: dang it, Dodo keeps picking my gmail account as the default sending address
06:30 kode54: f.ex. nouveau doesn't even appear to be supporting DSC here
06:30 kode54: it's limited to 120Hz on my primary panel, and possibly only 6bpc
06:45 DodoGTA: kode54: Dodo?
08:53 kode54: DodoGTA: Dodo Mail
08:53 kode54: https://github.com/akissinger/dodo
08:54 kode54: sorry for the confusion :D
18:18 gfxstrand[d]: _lyude[d]: Maybe? I think we maybe had some flickers but the corruption issue was something to do with not properly handling certain cursor sizes or maybe not rejecting bad cursor sizes.
20:41 mhenning[d]: ooh workgroup size lowering landed
21:28 snowycoder[d]: mhenning[d]: Time for mesh shaders??
21:33 mhenning[d]: yeah, maybe mary will pick mesh shaders back up 🙂
21:34 marysaka[d]: yeah I am planning to do that right after the openrm backend 😄
21:38 mangodev[d]: can't wait for nouveau compression to land in stable kernel sometime soon-ish
21:40 mangodev[d]: combined with mesh shaders and some other things, nvk may actually be a serious looking driver for once
21:40 mangodev[d]: not just a weird foss alternative
21:41 mhenning[d]: hey, who are you calling a "weird foss alternative"? 😛
21:41 mangodev[d]: ideally, nvk could get to a point to where programs don't have to dodge around weird nvidia-isms, and can actually fit in with the rest of mesa :D
21:41 mangodev[d]: (cough cough *dxvk*)
21:43 mangodev[d]: i mean
21:43 mangodev[d]: isn't nvk already not bad for datacenter purposes, given the compute is about the same as the open driver module in most cases?
21:43 mangodev[d]: maybe coopmatrix2 support could pull in more ai bros :P
21:44 mhenning[d]: I think we still have a bit farther to go on the performance front
21:44 mangodev[d]: yeah, though you can't ignore the major strides recently
21:44 mhenning[d]: yep we're working on it
21:45 mangodev[d]: what would you say is next after mesh shaders, zcull, and prepass?
21:46 mangodev[d]: aside from bug and regression fixes
21:47 mhenning[d]: oh I have a very long TODO list
21:47 mhenning[d]: right now I'm trying to figure out the ROOT_TABLE stuff karol mentioned a few days ago
22:51 sueintelabdredhat: intel and red hat get sued, we discussed it through, for billions of dollars.