#
✕ nick:povik
#asahi-dev · 2021-10-11 20:42
povikhttps://github.com/povik/linux/commits/asahi/audio
#asahi-dev · 2021-10-11 20:42
povikbehold!
#asahi-dev · 2021-10-11 19:22
povikso that's that
#asahi-dev · 2021-10-11 19:22
povikmca1 (as device node) does not depend on sio_adma directly
#asahi-dev · 2021-10-11 19:21
povikoh sorry mca1 is assigned to PD called mca1 which in turn depends on sio_adma
#asahi-dev · 2021-10-11 19:18
povikanyway, this should be testable
#asahi-dev · 2021-10-11 19:18
povikthat's what i meant
#asahi-dev · 2021-10-11 19:18
povikso the question is, if the dependence of mca1 on sio_adma is an actual dependence, or something someone threw in there
#asahi-dev · 2021-10-11 19:18
povikmca1 also depends on PD called mca1
#asahi-dev · 2021-10-11 19:17
povikthe dma controller also depends on sio_adma
#asahi-dev · 2021-10-11 19:17
povikand in the ADT, mca1 depends on sio_adma, which is presumably the PD of the dma controller
#asahi-dev · 2021-10-11 19:17
povikwell you have distinct nodes for the i2s port (mca1) and the dma controller (admac) in both ADT and the linux DT i am putting together
#asahi-dev · 2021-10-11 15:54
povikafk for ~2h
#asahi-dev · 2021-10-11 15:40
povikthat dependence is bit fishy
#asahi-dev · 2021-10-11 15:40
povikmaybe sio_adma need not be a parent for mca1, being the power domain of the dma controller
#asahi-dev · 2021-10-11 15:38
poviktell me about it
#asahi-dev · 2021-10-11 15:38
poviki know
#asahi-dev · 2021-10-11 15:38
povikmarcan: only sio_adma is off, audio_p is on
#asahi-dev · 2021-10-11 15:36
poviki believe so. let me check
#asahi-dev · 2021-10-11 15:34
povikdidn't remember what chime means for a moment
#asahi-dev · 2021-10-11 15:33
povikright, probably, mca1 is the speaker
#asahi-dev · 2021-10-11 15:33
povikoh, as in the boot sound
#asahi-dev · 2021-10-11 15:33
povikboot chime?
#asahi-dev · 2021-10-11 15:31
povikbeing turned on at boot
#asahi-dev · 2021-10-11 15:31
povikps_mca1 is offending
#asahi-dev · 2021-10-11 15:31
povikmarcan: https://paste.debian.net/1215064/
#asahi-dev · 2021-10-11 15:29
povikwell i can post it if you want
#asahi-dev · 2021-10-11 15:29
povikoh right
#asahi-dev · 2021-10-11 15:29
povikmarcan: 5.15-rc3 with pmgr, i2c, pinctrl and maz's pcie patches
#asahi-dev · 2021-10-11 15:22
povikwill patch my copy of the linux pwrstate driver for the time being
#asahi-dev · 2021-10-11 15:22
povikyeah but that runs against my current goal of having a free-standing audio driver with no special m1n1 preinit :(
#asahi-dev · 2021-10-11 15:05
povikmarcan: yep. disabling the domain before linux init makes the error go away
#asahi-dev · 2021-10-11 14:58
povikAIUI it happens due to iboot leaving some domain on where the parent is off
#asahi-dev · 2021-10-11 14:58
povikhere it errors out
#asahi-dev · 2021-10-11 14:58
povikhttps://github.com/torvalds/linux/blob/master/drivers/base/power/domain.c#L1830
#asahi-dev · 2021-10-11 14:58
povikand i ran into this
#asahi-dev · 2021-10-11 14:57
povikmarcan: so i started adding additional child-parent relations to the power domains as we discussed earlier
#asahi-dev · 2021-10-09 07:52
povikreluctantly, but okay
#asahi-dev · 2021-10-09 07:52
poviki remeber there were voltage translaters/buffers of some sort on the connection between the speaker amp and the soc
#asahi-dev · 2021-10-09 07:47
povikAIUI you have one mini disassembled... :D
#asahi-dev · 2021-10-09 07:45
povikwhat about that power rail conjecture
#asahi-dev · 2021-10-09 06:57
poviki have been picturing them as bringing up some voltage rail on the board
#asahi-dev · 2021-10-09 06:54
povik*do those writes
#asahi-dev · 2021-10-09 06:54
povikit's pmgr[23]+0xc000 from the point of view of ADT
#asahi-dev · 2021-10-09 06:54
povikmarcan: ^ does those writes look like anything to you?
#asahi-dev · 2021-10-09 06:54
povikhttps://github.com/AsahiLinux/m1n1/blob/main/proxyclient/experiments/speaker_amp.py#L132
#asahi-dev · 2021-10-08 18:12
povikthought it is underflowing periodically
#asahi-dev · 2021-10-08 18:12
povikthat beats at the beginning, clipped, confused me at first
#asahi-dev · 2021-10-08 18:11
povik(that's what it was)
#asahi-dev · 2021-10-08 18:11
povikhttps://www.youtube.com/watch?v=KzZz9Aak2kE

plus ›