01:24imirkin: skeggsb: what do you think about adding an ioctl for changing the config of an existing bo? thinking about ways to get started on a vk driver without the full thing "being there"
17:05imirkin: looks like NVDEC docs got posted: https://nvidia.github.io/open-gpu-doc/classes/video/
17:05imirkin: and also NVENC
17:05imirkin: (and, i'm sure most exciting of all, NVJPG)
17:06ccr: :P
17:07imirkin: from what i understand, NVDEC is quite similar to MSDEC as well, so probably worth going over our existing impl and seeing what we missed
17:07imirkin: maybe we can finally put that h264 bug to bed, although i'm unconvinced - it felt like a communication issue between VLD and VP. but who knows.
18:13RSpliet: imirkin: mJPEG streamers will be overjoyed
18:14RSpliet: but more seriously, yes, if this helps track down the x264 bug I'd be pretty stoked,although I might be EOLing my Atom 330 "media centre"
18:15RSpliet: C5B0 has context switching registers... somehow this shouldn't surprise me, but it's also not something I've ever thought about
18:16RSpliet: is it supposed to switch along with FECS/GPCCS?
19:12dviola: hey skeggsb, RSpliet, did you see my comment about moving the reset code before line 105? https://github.com/torvalds/linux/blob/master/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/base.c#L112-L114
19:12dviola: that helps my issue, no timeout after that