#asahi-dev · 2023-01-08 01:56
marcanhampus: I think you want to join #asahi-gpu for that#asahi-dev · 2023-01-06 16:53
marcanI think I saw it work (sometimes) with pro/max machines, never on M1 or something like that?#asahi-dev · 2023-01-06 16:52
marcanbut I do want to figure out what causes the death on suspend#asahi-dev · 2023-01-06 16:52
marcandefinitely never resumed out of that though, that would need specific handling#asahi-dev · 2023-01-06 16:52
marcanI've seen it survive suspend before, as in the machine looked quiesced but the HV console was still alive#asahi-dev · 2023-01-06 16:44
marcanI'm pretty sure I could trace shutdown/reboot up to the action that actually triggers them#asahi-dev · 2023-01-06 16:44
marcanare you sure it's incomplete and it doesn't just reboot the machine without cleaning up dcp?#asahi-dev · 2023-01-06 16:10
marcanyou can just watch it until the end, unless you want to do something after the shutdown?#asahi-dev · 2023-01-06 15:26
marcanok, makes sense#asahi-dev · 2023-01-06 15:26
marcanah right, that's the one I used for suspend mode?#asahi-dev · 2023-01-06 15:24
marcanjannau: wait so then when does A500 need to get called?#asahi-dev · 2023-01-06 06:15
marcanthe whole thing with not being able to actually reset firmware state is pretty cursed#asahi-dev · 2023-01-06 06:15
marcanjannau: remember reset doesn't actually reset any memory, so if they have like a static global with the shmem info/state for that, it would survive#asahi-dev · 2023-01-05 08:22
marcansure#asahi-dev · 2023-01-05 08:21
marcanTellowKrinkle: yes (sorry, missed the question earlier)#asahi-dev · 2023-01-05 08:21
marcandon't think so, just that github issue#asahi-dev · 2023-01-05 06:06
marcanI think it was more common in 11.x where the SFR recoveryOS *was* the primary recoveryOS. Now that we have paired per-OS recoveries I think they stopped doing fallbacks for the system wide one, since there is enough redundancy already.#asahi-dev · 2023-01-05 06:05
marcanthe latter is rarely seen these days#asahi-dev · 2023-01-05 06:04
marcanSystem recoveryOS and System Fallback recoveryOS are both part of SFR#asahi-dev · 2023-01-05 06:00
marcanthere, fixed :)#asahi-dev · 2023-01-05 05:56
marcanlet me fix that#asahi-dev · 2023-01-05 05:56
marcanthough that's actually... a bit wrong!#asahi-dev · 2023-01-05 05:56
marcanhttps://github.com/AsahiLinux/docs/wiki/Glossary#s ;)#asahi-dev · 2023-01-05 05:49
marcanso they probably just do SFR-relevant development in the 13.0 branch only, and 12.6.x just requires whatever minimum version from that branch it needs#asahi-dev · 2023-01-05 05:48
marcanif they didn't do that, a newer 12.6 branch could require a newer SFR labeled 12.6 but it would be impossible to install it when SFR is already 13.0#asahi-dev · 2023-01-05 05:48
marcanso SFR versions go 12.5, 13.0 and there are no and never will be any 12.6+ SFRs#asahi-dev · 2023-01-05 05:47
marcanSFR only ever gets rolled forward, so they can't fork off development like they fork off macOS development#asahi-dev · 2023-01-05 05:47
marcanI just realized that the reason why Apple are shipping 13.0 SFR with 12.6 is because they have to#asahi-dev · 2023-01-05 03:40
marcanhttps://github.com/AsahiLinux/linux/issues/101#asahi-dev · 2023-01-05 03:40
marcanmaz: in case you're interested, looks like the kernel oopses if you try to read /sys/kernel/debug/psci. that file presumably shouldn't exist at all on non-psci systems.#asahi-dev · 2023-01-04 17:51
marcanpushed the kernel to asahi-dev, please test :)#asahi-dev · 2023-01-04 16:40
marcanI first tried a rebase but gave up on that real quick, since there are like 900 patches in that tree including tons of junk and early development history that just needs to be thrown away#asahi-dev · 2023-01-04 16:39
marcanmaz: the rust bit was not *super* painful to merge as far as big messy merges go, but I'm more worried that the RfL tree hasn't been rebased ever as far as I know, because although they can certainly keep merging mainline in, picking things out for upstreaming is going to become an exponential mess that way#asahi-dev · 2023-01-04 16:38
marcan:D#asahi-dev · 2023-01-04 12:42
marcanmaz: 6.2-rc2 rebase pushed (that was more painful than usual)#asahi-dev · 2023-01-04 09:56
marcansven: can't really test on one of those right now, but yeah I can just send it blind and see what happens#asahi-dev · 2023-01-04 01:36
marcanyeah no, I'll give up on the project before turning into one of them thanks#asahi-dev · 2023-01-04 01:32
marcanthis time he's part of the problem#asahi-dev · 2023-01-04 01:32
marcanand *he* gave up on *that* one#asahi-dev · 2023-01-04 01:32
marcanyes, he was#asahi-dev · 2023-01-04 01:23
marcanthis is beyond stupid#asahi-dev · 2023-01-04 01:23
marcanseriously, what the actual fsck is up with some kernel developers/maintainers#asahi-dev · 2023-01-04 01:22
marcanguess he couldn't handle me pointing out that his patch series commits all the same sins my patch did, after he lectured me about it?#asahi-dev · 2023-01-04 01:20
marcanlol he actually said good bye#asahi-dev · 2023-01-03 13:40
marcancould be, yeah#asahi-dev · 2023-01-03 13:23
marcankilling the PCIe power domain breaks the init that m1n1 does#asahi-dev · 2023-01-03 13:23
marcanmy hack with DCP was to set the domain to always on and unset that in the DCP driver#asahi-dev · 2023-01-03 13:22
marcanwhich meant that the DART attaching to the DCP power domain before the FB or DCP itself was killing it#asahi-dev · 2023-01-03 13:22
marcanjannau: the problem is probably the same thing I ran into with DCP, that the genpd framework shuts down the PD as soon as *one* device attaches that claims to be RPM aware#asahi-dev · 2023-01-03 11:03
marcanit's not surprising that is broken