#asahi-dev · 2022-08-16 04:30
marcanheh, CONFIG_TRACE_MMIO_ACCESS showed up on 6.0, cute#asahi-dev · 2022-08-15 19:18
marcanaaaand the WQ fix already got applied to the wq fixes branch. nice.#asahi-dev · 2022-08-15 18:26
marcanyes#asahi-dev · 2022-08-15 17:42
marcanpackages are on -dev#asahi-dev · 2022-08-15 17:42
marcanplease do test :)#asahi-dev · 2022-08-15 17:42
marcanalso just pushed asahi-5.19-4 with just the TTY/WQ fix on top of -3, if nobody reports any regressions for the other u-boot/m1n1/etc stuff I'll push to production tomorrow#asahi-dev · 2022-08-15 17:37
marcanfor reference: https://mrcn.st/t/DeviceTree.j413ap.adt.txt https://mrcn.st/t/DeviceTree.j493ap.adt.txt (those are pre-iBoot-munging)#asahi-dev · 2022-08-15 17:36
marcanpovik: when you get a chance, can you review the t8112 audio stuff? both to see whether I got it right, and to update it to the newer bindings#asahi-dev · 2022-08-15 17:35
marcanpushed the rebase, but not even build tested yet#asahi-dev · 2022-08-15 17:25
marcanI think povik/asahi-sound-wip has the right commits, let me see if I can pull them from there#asahi-dev · 2022-08-15 17:24
marcanit still has the filter_controls thing instead of the fixup_controls one that went in#asahi-dev · 2022-08-15 17:23
marcanah, but unfortunately, no#asahi-dev · 2022-08-15 17:19
marcanthanks!#asahi-dev · 2022-08-15 17:19
marcan:)#asahi-dev · 2022-08-15 17:19
marcanwhat would I do without you reminding me of everything I forget people told me already#asahi-dev · 2022-08-15 17:14
marcanpovik: what's the right branch to get the macaudio stuff on top of what was already upstreamed? I'm rebasing on 6.0-rc1 and I think that's the only thing which was lagging in the asahi tree.#asahi-dev · 2022-08-12 05:45
marcanjannau: as in the IOVA space as mapped to DART, not the DVA space? (DVA has always been at >0)#asahi-dev · 2022-08-10 03:31
marcanheh :)#asahi-dev · 2022-08-10 03:24
marcanhttps://marcan.st/2011/01/safe-hacking/ etc etc#asahi-dev · 2022-08-10 03:23
marcan(I will fight AppleCare 5 times if need be, if it means *nobody* blows up their speakers without hacking DTs/kernel drivers)#asahi-dev · 2022-08-10 03:23
marcan(and I'm serious about not minding blowing them if it's for science and it means nobody else will)#asahi-dev · 2022-08-10 03:22
marcaneither way I am definitely putting on I Won The Loudness War on loop for a few hours once we settle on good max settings ;)#asahi-dev · 2022-08-10 03:21
marcanyeah, I'm getting the feeling that with that HPF + max volume settings, we're probably doing pretty good, especially if as you say the speakers are resilient enough not to blow on brief excursions#asahi-dev · 2022-08-10 03:21
marcanthat should give us a good idea for where we stand re speaker safety#asahi-dev · 2022-08-10 03:20
marcanonce we have the right knobs implemented in the driver I should probably do some tests in macOS, see how loud it gets (and what response I get), and see if I can safely go that loud in Linux, and then optimize the HPF/gain ceilings, and then feed evil stuff to the raw speakers at those settings and see if I can make them blow or start to rattle badly#asahi-dev · 2022-08-10 03:19
marcanbut for speaker protection you'd probably want even slower attack, tbh, since it would be more of a running average thing than a limiter, at least for power reasons#asahi-dev · 2022-08-10 03:18
marcanand it can attenuate up to 15dB#asahi-dev · 2022-08-10 03:17
marcanthe slowest configurable rate is 82ms/dB attack, 8.2s/dB release (!), in powers of 2#asahi-dev · 2022-08-10 03:15
marcanso it should work#asahi-dev · 2022-08-10 03:15
marcanyeah, I think they have the ICC pins all linked#asahi-dev · 2022-08-10 03:14
marcanusing either a dedicated pin or the TDM bus, let me take a quick look at the schematic...#asahi-dev · 2022-08-10 03:14
marcanthe chips can also align the limiter values between themselves for stereo/multichannel linking#asahi-dev · 2022-08-10 03:11
marcanit's your typical fully configurable limiter (attack, release, hold, threshold, max attn)#asahi-dev · 2022-08-10 03:11
marcan(you can configure it statically)#asahi-dev · 2022-08-10 03:11
marcanit's intended to deal with low-battery conditions, but worth noting in case it's useful for some speaker protection use case#asahi-dev · 2022-08-10 03:10
marcanalso, not sure if this is useful, but the speaker amp also has a limiter feature#asahi-dev · 2022-08-10 03:06
marcanheh :)#asahi-dev · 2022-08-10 03:06
marcananyway, not critical initially#asahi-dev · 2022-08-10 03:05
marcanI assume it's smart enough to not run the DSPs if there are no inputs#asahi-dev · 2022-08-10 03:05
marcaninteresting#asahi-dev · 2022-08-10 03:05
marcanyeah, that makes sense#asahi-dev · 2022-08-10 03:04
marcans/flat/balanced/ (because marketing)#asahi-dev · 2022-08-10 03:04
marcanand then we could even have a "null" profile that just exposes the raw device for people who want to play around#asahi-dev · 2022-08-10 03:03
marcansounds like it should fit in with the general device profile stuff somehow?#asahi-dev · 2022-08-10 03:03
marcanyou know, "flat", "macOS style", etc#asahi-dev · 2022-08-10 03:03
marcanspeaking of this, maybe we should try to make sure whatever PW solution we end up with supports selectable profiles#asahi-dev · 2022-08-10 03:03
marcanyup...#asahi-dev · 2022-08-10 02:59
marcanI should really try your FIRs, I like good DSP but right now macOS on these things is the best audio I've heard from a laptop, so I don't have a point of reference for how properly DSPed decent laptop speakers *should* sound#asahi-dev · 2022-08-10 02:50
marcanyeah, that sounds like something Apple would do#asahi-dev · 2022-08-10 02:50
marcanFletcher–Munson curves, eh...