#
#asahi-dev · 2026-03-14 08:26
JamesCalligeros[m]We have Some Ideas on getting around that
#asahi-dev · 2026-03-14 08:25
xarkesmy understanding is that tracing was possible on m1-m3 thanks to loading m1n1 as an hypervisor, which is no more doable on m4 because of SPTM, correct?
#asahi-dev · 2026-03-14 08:24
xarkesright, thanks for the reminder :)
#asahi-dev · 2026-03-14 08:22
jannauI wouldn't start looking at iboot for cpu frequency handling but before you start looking at any apple binaries please read and understand the copyright and reverse engineering policy https://asahilinux.org/copyright/
#asahi-dev · 2026-03-14 08:20
jannauxarkes: I think the offsets for m1-m3 were discovered by tracing macos
#asahi-dev · 2026-03-14 08:20
jannauxarkes: I guess it's not really necessary anymore now that chainload writes the new image to same location. I'm not sure whether the secondaries can wake spuriously but in that case it would be preferable to use the current RVBAR
#asahi-dev · 2026-03-14 08:12
xarkesalso, I was checking to implement cpufreq for M4, where would I find information like the features offsets (e.g. ppt-thrtl), is it part of iBoot2 (or maybe iBoot1?)
#asahi-dev · 2026-03-14 08:02
xarkeswhy is it necessary for chainload.py to set secondary CPU RVBARs? wouldn't it be the responsability of whoever plan to call smp_start_secondaries?
#asahi-dev · 2026-03-14 04:38
fmzakariThere were only 3 tasks: https://asahilinux.org/docs/project/help-wanted/ I was looking to see what was small to start with
#asahi-dev · 2026-03-14 04:38
fmzakariis https://github.com/AsahiLinux/linux/issues a valid issue list to look through to contribute?
#asahi-dev · 2026-03-14 04:34
nicolas17almost all code is Swift and it's a pain to decompile
#asahi-dev · 2026-03-14 04:32
nicolas17I have not reverse engineered the part that matters so idk :p
#asahi-dev · 2026-03-14 04:19
chadmedis there an interlock? does SIL need to report that it's active for the mic/webcam to come up?
#asahi-dev · 2026-03-14 04:08
nicolas17https://support.apple.com/en-us/guide/security/sec75a2d237d/1/web/1
#asahi-dev · 2026-03-14 04:07
nicolas17then again I'm not sure yet if M4/M5 were already using SIL for the microphone
#asahi-dev · 2026-03-14 04:06
nicolas17this could add extra fun to supporting the webcam on the Neo
#asahi-dev · 2026-03-14 04:05
nicolas17Secure Indicator Light (SIL)
#asahi-dev · 2026-03-14 04:04
nicolas17(this was already being used on some iOS devices)
#asahi-dev · 2026-03-14 04:03
nicolas17MacBook Neo does not have a webcam LED, instead an Exclave applet draws an on-screen indicator in such a way that even the XNU kernel can't bypass
#asahi-dev · 2026-03-14 03:39
UnknownArtistBotHi all! check this out, An Unknown Artist - https://www.anunknownartist.com/
#asahi-dev · 2026-03-13 21:54
oliverYea I assumed that. I was hoping that we decide what the connection should speak and that we can still select dp-alt if the device supports both.
#asahi-dev · 2026-03-13 21:42
jannauusb-c to displayport input should work
#asahi-dev · 2026-03-13 21:41
jannauif it is a thunderbolt/usb4 input apple's type-c port controller will use thunderbolt and we have to tunnel displayport over it. This is something completely different than dp-altmode
#asahi-dev · 2026-03-13 21:37
oliverleio: yea, it works with another small portable screen I have laying around. dell uses the thunderbolt wording a lot, not sure if the screen actually does dp-alt or only supports dp via tb? I'll check with a usb-c to display port cable tomorrow in hopes that that is more explicit in what it actually does. And when I have some more time I'll try to understand where it actually breaks.
#asahi-dev · 2026-03-13 21:31
jannauhelpful to define debug in pmgr-pwrstate and see what actually happens
#asahi-dev · 2026-03-13 21:30
chaos_princessvery confused about how it all is supposed to work now
#asahi-dev · 2026-03-13 21:29
jannauno, pmp crashes without "apple,always-on" in ps_pms_sram
#asahi-dev · 2026-03-13 21:25
chaos_princessi guess that too, but also the pmp_reporting node has a reference to the sram pmgr as one of the power domains, and it shouldn't shut down, as it does not declare any pm functions, right?
#asahi-dev · 2026-03-13 21:23
jannauI think the sram shouldn't be powered down due to "apple,always-on" in one of its children
#asahi-dev · 2026-03-13 21:22
jannauor am I mixing it up with runtime-pm?
#asahi-dev · 2026-03-13 21:20
chaos_princessand not rtkit core itself?
#asahi-dev · 2026-03-13 21:20
chaos_princessso it is probably powering off the sram
#asahi-dev · 2026-03-13 21:19
chaos_princessthat explains some things
#asahi-dev · 2026-03-13 21:18
jannaunot automatically, would need to be implemented explicitly in each driver using rtkit
#asahi-dev · 2026-03-13 21:15
chaos_princessi think i just screwed up the power dependencies most likely and it works on t6000 by luck
#asahi-dev · 2026-03-13 21:15
chaos_princessdo we shut down rtkit coprocessors in sleep?
#asahi-dev · 2026-03-13 21:14
jannaucrash is gone with always-on ps_pmp{,_sram}
#asahi-dev · 2026-03-13 21:11
jannauif my recollection is correct we probably broke it during smc upstreaming
#asahi-dev · 2026-03-13 21:09
jannauI think smc log_power was previously logging during suspend no_console_suspend
#asahi-dev · 2026-03-13 21:07
chaos_princessand "active" mostly by getting it as idle as possible and then eyeballing it
#asahi-dev · 2026-03-13 21:07
chaos_princesscan you try always-on too?
#asahi-dev · 2026-03-13 21:07
chaos_princessmy measurements were also via reading from smc in a tight loop and then entering/exiting suspend
#asahi-dev · 2026-03-13 21:06
jannauno, without always-on
#asahi-dev · 2026-03-13 21:06
jannauhard to tell how much power it saves while active and 5% dispaly brightness but for directly after suspend it's from 6W down to 4W
#asahi-dev · 2026-03-13 21:05
chaos_princessis that with ps_pmp and ps_pmp_sram as always-on?
#asahi-dev · 2026-03-13 21:04
jannaupmp still crashes after waking from suspend
#asahi-dev · 2026-03-13 21:00
jannaubut it looks like the pmp reporting reduced power use
#asahi-dev · 2026-03-13 20:59
jannauargh. I forgot to enable pmp
#asahi-dev · 2026-03-13 20:58
chaos_princesst6020->t6021 should also in theory differ in 0x60000000
#asahi-dev · 2026-03-13 20:56
chaos_princessoh well

plus ›