09:37 <macc24> icecream95: hello

09:37 <macc24> remember gallium nine patch for panfrost?

09:38 <icecream95> macc24: yes

09:38 <macc24> icecream95: got a link to it?

09:38 <icecream95> https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7001

09:39 <macc24> thanks

09:41 <macc24> are there any box86/wine patches needed?

09:42 <icecream95> macc24: box86 patches are already upstream

09:49 <icecream95> Don't expect much to work yet. The closest I got for an actual game was with DarkPlaces, but with that there were a lot of GPU faults because of uninitialised samplers

09:50 <macc24> i got nfs carbon to not crash on my laptop, so it may work with nine

10:01 <HdkR> looks like initial nine integration wasn't that bad

10:01 <HdkR> For nine+panfrost anyway

11:47 <macc24> can panfrost work under kernel with mali blob?

11:52 <jschwart> macc24: I don't think so the interface exposed by that kernel is different

11:54 <jschwart> I have this situation with my TV, it's rooted, but I can't feasibly replace its kernel in any way and it works with the mali blob

12:27 <alyssa> "Use real name for attribute's unknown field"

12:27 <alyssa> See, this one's exra fun

12:27 <alyssa> ...since on midgard it's "reserved - always set to 1" :p

12:28 <macc24> alyssa: "alwayssettoone"

12:30 <alyssa> Yeah I don't know either :p

13:30 <jschwart> maybe "breaks midgard when not 1" is more clear?

13:32 <bbrezillon> alyssa: maybe because the global index offset is always applied for the primary attribute array, and never applied for the secondary one (AKA varyings), while on bifrost, it's also applied for the secondary array if you don't disable it at the attrib desc level

13:39 <bbrezillon> so I think this field also encodes the "enable global index offset" info on midgard, it's just that you can't disable it there

13:56 <alyssa> bbrezillon: Yeah, that sounds likely.