#asahi-dev · 2022-08-10 20:15
poviki did that for mac mini and j314, which i have access to#asahi-dev · 2022-08-10 20:15
povikhttps://github.com/povik/linux/blob/macaudio-rfc-v2/sound/soc/apple/macaudio.c#L924#asahi-dev · 2022-08-10 20:15
povikbtw, earlier i collected by tracing the volume maximums macos sets (they don't seem to dynamically adjust volume at codec level), and wrote them into the wip macaudio driver#asahi-dev · 2022-08-10 20:15
povikthe macaudio driver will later lock the controls to the desired values...#asahi-dev · 2022-08-10 20:15
povikapplying it to a tree with the j314 speakers brought up, you should have 6 new controls in alsamixer#asahi-dev · 2022-08-10 20:14
povikin the meantime i wrote this (untested) patch to control the tas2764 highpass: https://tpaste.us/8bZO#asahi-dev · 2022-08-10 20:14
povikfyi i will be away for a week, then i should be at hand for any testing we might want to do#asahi-dev · 2022-08-09 13:32
povikyeah#asahi-dev · 2022-08-09 13:28
povikit's -3 dB at 800 hz#asahi-dev · 2022-08-09 13:28
povikthough i can't find a shape for the filter in the datasheet#asahi-dev · 2022-08-09 13:27
povikmarcan: how did i miss that! that's not too low#asahi-dev · 2022-08-09 11:45
povikthat's ok (although bug in spewing false error messages)#asahi-dev · 2022-08-09 11:05
povik(kernel_neon_begin still has zero hits in sound/*)#asahi-dev · 2022-08-09 11:03
povikj`ey: well i was wondering if someone already did the unspeakable of signal processing in kernel#asahi-dev · 2022-08-09 10:46
povikthen no wonder#asahi-dev · 2022-08-09 10:46
povikah#asahi-dev · 2022-08-09 10:41
povik0 matches#asahi-dev · 2022-08-09 10:41
povikSearching 2545 files for "kernel_fpu_begin"#asahi-dev · 2022-08-09 10:41
povikso, looking into sound/* of the kernel:#asahi-dev · 2022-08-09 10:25
poviki guess no. if you want airtight protection from userspace#asahi-dev · 2022-08-09 10:22
povikChaosPri1: ah, then the disadvantage may be that the frequency check in kernel isn't all that different from doing the full processing, and is additional computational work#asahi-dev · 2022-08-09 09:58
povikthe crossover/filtering is natural to do as a single step#asahi-dev · 2022-08-09 09:57
poviki am not sure what you mean by that split#asahi-dev · 2022-08-09 09:52
povikthe filters are the safety feature#asahi-dev · 2022-08-09 09:48
povikmaybe we will end up with the handshake to userspace#asahi-dev · 2022-08-09 09:47
poviki will bring up we may have unsafe PCMs even with volume caps next time i push the macaudio driver, we will see what the kernel folk say#asahi-dev · 2022-08-09 09:47
povikchadmed: it is way more work#asahi-dev · 2022-08-09 09:31
povikor compute the frequency image of the FIRs on load and check them against a safe envelope :p#asahi-dev · 2022-08-09 09:30
povikanyway i will be happy if the issue goes either way#asahi-dev · 2022-08-09 09:30
povikthe solution is authenticating the FIRs, of course!#asahi-dev · 2022-08-09 09:29
povikha, found the loophole :)#asahi-dev · 2022-08-09 09:27
povikbut then you can open the device for 6 channels if you are an outright evil userspace#asahi-dev · 2022-08-09 09:27
povikwell that covers some cases of accidental damage#asahi-dev · 2022-08-09 09:25
povikyeah, but the argument against is we have to trust userspace#asahi-dev · 2022-08-09 09:23
povikfranky the original idea is a sane suggestion#asahi-dev · 2022-08-09 09:23
povik(the impulse responses)#asahi-dev · 2022-08-09 09:23
povikonly the data would have to be loaded#asahi-dev · 2022-08-09 09:23
povikthe suggestion is having the filter chain baked into kernel code#asahi-dev · 2022-08-09 09:22
povikthat's not the case though#asahi-dev · 2022-08-09 09:22
poviknow sven got me interested#asahi-dev · 2022-08-09 09:21
povik:)#asahi-dev · 2022-08-09 09:21
povikwell that's no different from other firmware kernel loads#asahi-dev · 2022-08-09 09:13
povikthat didn't print anything new, but maybe i should try on a machine that has nonempty aop audio (mac mini has no mic, but the driver still probes)#asahi-dev · 2022-08-09 08:46
povikbut aop-audio-debug?#asahi-dev · 2022-08-09 08:46
povikno, that's not it, that debugs the ADMAC instance of course#asahi-dev · 2022-08-09 08:41
povikoh hello there 'admac-aop-audio.debug'#asahi-dev · 2022-08-09 08:21
poviki tried looking around for plausible bootargs, but that didn't do it#asahi-dev · 2022-08-09 08:21
povikthere's a level of verbosity the AOP driver can do i am not unlocking#asahi-dev · 2022-08-09 08:21
povikwhat typical ways are there to enable verbose logging from kexts?#asahi-dev · 2022-08-05 15:01
povikyeah