#asahi · 2022-07-01 20:00
marcanfrom what I saw of the broadcom thing, Apple uses Yet Another Different Thing instead of what brcmfmac normally does for suspend#asahi · 2022-07-01 19:59
marcanit's not just broadcom that's the problem, pcie itself goes bananas if you do actually power gate the hardware. might need a full reinit, which is currently done in m1n1#asahi · 2022-07-01 19:59
marcanI looked into that but it's WIP#asahi · 2022-07-01 19:58
marcanbut it will put some hardware into suspend, though as you found out, pcie is a problem#asahi · 2022-07-01 19:58
marcanaxboe: it's s2idle, we don't actually suspend the CPUs yet (that needs PSCI) nor do proper full system sleep#asahi · 2022-07-01 13:54
marcangoing to stream some more M2 bring-up for a few hours (#asahi-stream as usual)#asahi · 2022-06-28 12:35
marcanso maybe we should go that way#asahi · 2022-06-28 12:35
marcanthough there is an argument for doing it in the kernel: it can be made to keep working even in s2idle, while a userspace daemon wouldn't#asahi · 2022-06-28 12:27
marcanwell, time to pester that upower MR again then...#asahi · 2022-06-28 12:26
marcanugh#asahi · 2022-06-28 12:26
marcanand KDE ended up hitting sysfs directly#asahi · 2022-06-28 12:26
marcanah, nevermind, apparently upower still doesn't do this?#asahi · 2022-06-28 12:25
marcanno need for more hacky single-purpose daemons, let's do things right ;)#asahi · 2022-06-28 12:25
marcanif that is implemented, it will magically work in KDE/GNOME with UI and everything#asahi · 2022-06-28 12:25
marcanclearly this needs to be a upower patch that implements the charging-threshold emulation for devices with on/off control like this one#asahi · 2022-06-28 12:25
marcanmps: upower already has charging-threshold support for drivers that implement that, and logic to handle different battery attribute support levels#asahi · 2022-06-28 12:21
marcanwhat we're missing is the userspace policy, which should probably go in upower#asahi · 2022-06-28 12:21
marcanand it's implemented and shipping :p#asahi · 2022-06-28 12:21
marcancome on give me some credit, I already figured all this out :p#asahi · 2022-06-28 12:21
marcanchadmed: ^#asahi · 2022-06-28 12:20
marcanto turn on charging: echo auto > /sys/class/power_supply/macsmc-battery/charge_behaviour#asahi · 2022-06-28 12:19
marcanto turn off charging: echo inhibit-charge > /sys/class/power_supply/macsmc-battery/charge_behaviour#asahi · 2022-06-28 12:18
marcanmacOS does it itself#asahi · 2022-06-28 12:18
marcanthere is no magic hardware management#asahi · 2022-06-28 12:18
marcanyou need userspace to handle the mode switching depending on state of charge#asahi · 2022-06-28 12:18
marcan/sys/class/power_supply/macsmc-battery/charge_behaviour#asahi · 2022-06-28 10:11
marcanthe iBoot boot protocol was formerly mach-o, but now uses raw binaries. the m1n1 chainload boot protocol is identical. m1n1 can also load ARM64 linux kernel style images; that is the format used to run non-m1n1 payloads like u-boot and linux directly#asahi · 2022-06-28 10:10
marcanradens: I'm not sure what you mean by decrypted, nothing is encrypted#asahi · 2022-06-28 10:09
marcanit seems they moved one of them to the M2#asahi · 2022-06-28 10:09
marcanit has two firmwares#asahi · 2022-06-28 10:09
marcanit was just... running on something in the trackpad itself#asahi · 2022-06-28 10:09
marcanheh, so it turns out trackpad firmware on the old M1s was *already* RTKit#asahi · 2022-06-27 13:03
marcanit has NV, we knew that#asahi · 2022-06-27 12:58
marcanmaz: https://mrcn.st/p/CifGtE2T#asahi · 2022-06-27 04:52
marcanstreaming M2 bring-up in a few minutes (see #asahi-stream)#asahi · 2022-06-25 07:07
marcan(and put it in the right folder)#asahi · 2022-06-25 07:07
marcansounds like you didn't replace boot.bin properly. did you rename the file you downloaded to that?#asahi · 2022-06-25 07:05
marcanthe good news, of course, is that we can probably catch these early in a beta pretty often. it's less likely that Apple makes major changes that could tickle bugs in point releases.#asahi · 2022-06-25 07:01
marcanit's just, you know, bugs :)#asahi · 2022-06-25 07:01
marcan(so older macOS versions also work)#asahi · 2022-06-25 07:01
marcanyes, but we shouldn't have to, since Apple keeps them backwards-compatible#asahi · 2022-06-25 07:01
marcanthe really messy ones with unstable ABIs are DCP, GFX, etc... those are not part of SFR and will remain stable within a given OS install/stub#asahi · 2022-06-25 07:00
marcanin particular NVMe and SMC firmware is global, those are part of SFR#asahi · 2022-06-25 07:00
marcannothing we can do about that, that's just software life. but it should become less and less likely over time, as we fix bugs.#asahi · 2022-06-25 06:59
marcanbut Apple isn't going to test that their perfectly legitimate and backwards-compatible changes don't trigger our own bugs, obviously :)#asahi · 2022-06-25 06:59
marcanSFR updates are supposed to be backwards-compatible, and this one was#asahi · 2022-06-25 06:58
marcanthose happened to break with 13.0#asahi · 2022-06-25 06:58
marcannothing actually changed with 13.0 as far as how the firmware is *supposed* to work. U-Boot was just taking excessive shortcuts, and Linux had an outright one-line bug.#asahi · 2022-06-25 06:57
marcansystem firmware and recovery#asahi · 2022-06-25 06:57
marcanCorrect. There were two bugs, one in u-boot and one in Linux, that broke with 13.0 SFR