05:51fdobridge: <gfxstrand> I really doubt the buffer memory requirements stuff is needed. There's already a common `vkGetBufferMemoryRequirement2()` implementation in `src/vulkan/runtime/vk_buffer.c` which is in terms of `GetDeviceBufferMemoryRequirements()`.
05:53fdobridge: <gfxstrand> If it is actually doing something, I'd like to know what because that strikes me as very odd.
06:06fdobridge: <![NVK Whacker] Echo (she) 🇱🇹> I think that needs a custom kernel (and I gave up on that)
07:07fdobridge: <![NVK Whacker] Echo (she) 🇱🇹> Removing that function hunk makes no difference
07:09fdobridge: <![NVK Whacker] Echo (she) 🇱🇹> The problem is why nouveau is tossing the BO 🐸
11:53fdobridge: <esdrastarsis> Why?
12:02fdobridge: <![NVK Whacker] Echo (she) 🇱🇹> People are combining new uAPI (which needs an entire kernel swap) and GSP into one
14:45fdobridge: <gfxstrand> Oh? What do you mean by "tossing the BO"?
14:46fdobridge: <![NVK Whacker] Echo (she) 🇱🇹> https://cdn.discordapp.com/attachments/1034184951790305330/1132685265757032498/message.txt
15:31fdobridge: <gfxstrand> Well that's annoying...
23:36fdobridge: <airlied> @gfxstrand okay gone from 190us to 70us with an internal flag
23:36fdobridge: <gfxstrand> 🥳
23:39fdobridge: <gfxstrand> Branches?
23:40fdobridge: <airlied> https://gitlab.freedesktop.org/nouvelles/kernel/-/commits/airlied-testing-mctestface/
23:40fdobridge: <airlied> https://gitlab.freedesktop.org/airlied/mesa/-/commits/nvk-new-uapi-wip
23:40fdobridge: <airlied> the mesa commit probably needs some more work
23:46fdobridge: <gfxstrand> Yeah, I'll tweak it. I'll give it all a ton tomorrow and see if it gets my CTS times down.
23:46fdobridge: <gfxstrand> Not that that's the primary concern.
23:52fdobridge: <airlied> fixed mesa to supply the flag to all the internal driver bos
23:54fdobridge: <gfxstrand> Yay