03:31 fdobridge: <a​irlied> @marysaka does the envyhooks handle multiple FIFO mappings in a single USERD page? and dump all the fifos?
03:32 fdobridge: <a​irlied> I'm working through it now, so I can probably figure it out, but just wanted to see
03:42 fdobridge: <a​irlied> ah it appears to try, will keep digging
04:29 fdobridge: <a​irlied> I wonder if they have some way of hiding pushbufs from user submits, those 350 byte ones are very wierd, and I really cannot see how the video stuff gets sent
08:14 fdobridge: <a​irlied> oh now I think I found it
08:14 fdobridge: <m​arysaka> I ignore the "no prefetch" rule btw
08:14 fdobridge: <m​arysaka> so some buffers are empty because they are filled after running others command buffers
08:14 fdobridge: <a​irlied> just not sure how to decode it properly
08:15 fdobridge: <a​irlied> since the method the sets the video object is called in a previous pushbuf for this subch
08:15 fdobridge: <m​arysaka> I have some very rudimentary command parser going on for MME dumping
08:16 fdobridge: <m​arysaka> maybe you could use that to flag it as such somehow
08:16 fdobridge: <a​irlied> I can just hardcode it for now
08:16 fdobridge: <m​arysaka> that work too 😄
08:16 fdobridge: <a​irlied> https://paste.centos.org/view/raw/d17636f9
08:16 fdobridge: <a​irlied> looks believeable
08:17 fdobridge: <a​irlied> guess I'll be starting to write a video decode attempt tomorrow 😛
08:43 fdobridge: <m​arysaka> nice!
08:51 soyoungface: Does it even matter if Nvidia publishes signed firmware images if imirkins script can extract it anyways?
08:54 Ermine: Legal stuff is definitely involved
09:03 soyoungface: That boils down to whether Nvidia license has reverse anti reverse engineering clauses, cause imirkin distributes the tool to extract the firmware, not the firmware itself.
09:07 soyoungface: Ok, true indeed, they have it in 1.6 no reverse engineering, decompilation and blabla in eula, ok thanks
09:20 soyoungface: I decide to quit to the world of cryptography and Bitcoin then, there's some era where my algorithms could be needed and useful definitely. But Intel and AMD don't restrict me doing better drivers, only Nvidia does it, and I have no Nvidia card anymore. I talked with Ermine, that I forward you the superexecution engine, and compression engine, which I soon offer in some form to him, there has been no issues at my side in code though.
09:20 soyoungface: It's very simple.
09:20 soyoungface: Cheers
09:24 fdobridge: <!​DodoNVK (she) 🇱🇹> And the person disconnected 🚪🏃
10:10 fdobridge: <a​irlied> That is mart bot
10:28 fdobridge: <!​DodoNVK (she) 🇱🇹> Mart?
10:30 HdkR: 👍
10:42 fdobridge: <a​irlied> Our regular Estonian correspondant
11:11 Ermine: I forgot to tell him to stay away from #nouveau
15:25 fdobridge: <i​llwieckz> Is that a real person or a markov chain bot?
15:26 fdobridge: <i​llwieckz> (looks more markov chain than real to me)
15:26 fdobridge: <p​ixelcluster> i think they have been shown capable of holding actual conversations
15:40 fdobridge: <!​DodoNVK (she) 🇱🇹> How much code do the `tegra` and `nouveau` KMDs share? :nouveau:
20:17 fdobridge: <a​irlied> the nvidia ones?
20:22 fdobridge: <b​utterflies> none
20:23 fdobridge: <b​utterflies> Tegra Xavier and earlier had a display engine handled via a separate tegra KMD (also upstream)
20:23 fdobridge: <b​utterflies> The tegra KMD also handles legacy pre-Tegra K1 iGPUs but that can be pretty much ignored nowadays
21:07 kwizart: well, I wish the grate driver for pre-tk1 driver would be upstream, but seems like there are some discutions between some in the community and the tegra maintainer
21:08 kwizart: it can be ignore in the sense that it's not relevant for aarch64
21:08 kwizart: (so anything barely modern)
21:24 karolherbst: it's a different GPU anyway
21:25 karolherbst: I doubt it ever gets upstreamed, because it would be dead on arrival