#asahi-dev · 2026-05-27 17:08
jskAny hints on where I would start with this? I will read up a bit on spmi in general but if there are any resources on this, I would appreciate it. Is the M3 air the first that uses spmi for usb?#asahi-dev · 2026-05-27 16:59
jskyuka: yess I tried wip and also saw the file you linked. just thought someone probably worked on this before.#asahi-dev · 2026-05-27 16:17
yukaSo, it needs more than device tree changes#asahi-dev · 2026-05-27 16:17
yukaApparently the M3 Air has spmi#asahi-dev · 2026-05-27 16:16
yukahttps://github.com/AsahiLinux/linux/blob/af7dd7b48f235c040682e23808c4e13ab866a190/arch/arm64/boot/dts/apple/t8122-usbpd-i2c.dtsi#asahi-dev · 2026-05-27 16:16
chaos_princessis the air the one with i2c pd controller or the spmi one?#asahi-dev · 2026-05-27 16:15
yukajsk: did you try the DT from asahi-wip?#asahi-dev · 2026-05-27 15:54
jskare there any usb DT patches for M3 (the air) flying around? I couldnt find them in asahi-wip nor upstream.#asahi-dev · 2026-05-26 22:50
chaos_princesswhich one were you poking at?#asahi-dev · 2026-05-26 22:50
fl0_dbut hope will resume asap#asahi-dev · 2026-05-26 22:49
fl0_dchaos_princess, yes it's quite different. I think I got quite close, but then fell into a rabbit whole of restoring my work stuff from backup / fixing a broken dmg...#asahi-dev · 2026-05-26 19:48
StanfordAppliedCyber[m]ie, i had to fix a lot of timer stuff in the hypervisor to support 26.5, not to mention a bunch of sysregs#asahi-dev · 2026-05-26 19:48
StanfordAppliedCyber[m]unfortunately i'm not sure if it's sptm emulation related or other stuff related#asahi-dev · 2026-05-26 19:47
StanfordAppliedCyber[m]i suppose i'll check again#asahi-dev · 2026-05-26 19:46
StanfordAppliedCyber[m]yeah, trust me that's the first thing i did#asahi-dev · 2026-05-26 19:42
sveni haven't been able to get tbt working since march or so due some weird bug as well#asahi-dev · 2026-05-26 19:41
svenother than that... getting stuck at $weird-crap is unfortunately part of the game here#asahi-dev · 2026-05-26 19:40
svenor if there's some timer inaccuracies going on (e.g. your sptm emulation takes too long or is too fast and then xnu runs into an unexpected timeout)#asahi-dev · 2026-05-26 19:40
sveni'd triple check if you're really doing all the things SPTM does in your emulation#asahi-dev · 2026-05-26 19:30
StanfordAppliedCyber[m]i'm willing to put in the work but it's clear that on this one i'm out of my depth#asahi-dev · 2026-05-26 19:30
StanfordAppliedCyber[m]magical opaque hardware fabric between the reported error and the thing causing the error#asahi-dev · 2026-05-26 19:29
StanfordAppliedCyber[m]i've made essentially 0 progress in a week because i'm not sure how to debug this#asahi-dev · 2026-05-26 19:29
StanfordAppliedCyber[m]this is the same panic i posted two days ago#asahi-dev · 2026-05-26 19:29
StanfordAppliedCyber[m]it is power related#asahi-dev · 2026-05-26 19:28
chaos_princessdpe sounds power-related#asahi-dev · 2026-05-26 19:27
StanfordAppliedCyber[m]is there anyone i can talk to who knows about the piogw and/or soc-dpe fabric?#asahi-dev · 2026-05-26 12:32
chaos_princesssure, why not#asahi-dev · 2026-05-26 12:32
yukaon some private channel?#asahi-dev · 2026-05-26 12:31
yukachaos_princess: can I just send you some of the adts with tunables?#asahi-dev · 2026-05-26 12:04
yukahave to check later, not on a laptop rn#asahi-dev · 2026-05-26 12:04
chaos_princessalso pcie-pcieclkgen-tunables#asahi-dev · 2026-05-26 12:04
chaos_princessdoes m4 have all of 'apcie-gp-axi2af-tunables', 'apcie-ge-axi2af-tunables', 'apcie-cio3pllcore-tunables' ?#asahi-dev · 2026-05-26 12:03
yukaand I think m3 max is similar to m4#asahi-dev · 2026-05-26 12:02
chaos_princesspcie on m3 pro and m3 max seem materially different, pro has something closer to vanilla m3 (completely wild guess - just take https://github.com/AsahiLinux/m1n1/blob/dc1a56521c2eaf9775a83456adc2ebdd000c8002/src/pcie.c#L333-L334 drop second line, and change offset to 0x4000 in first)#asahi-dev · 2026-05-26 05:19
nickchan...and of course things decided to magically to start working again after rebooting to linux again#asahi-dev · 2026-05-26 05:16
nickchanand this must be software issue same cable same machines but host on macos with macvdmtool and it's perfectly fine#asahi-dev · 2026-05-26 04:56
nickchanFrom what I can tell it's almost as if the dfu command is acting like the reboot command#asahi-dev · 2026-05-26 04:53
nickchanlike the j680ap screen would still instantly go black just nothing gets detected over usb#asahi-dev · 2026-05-26 04:52
nickchanon j314sap host#asahi-dev · 2026-05-26 04:52
nickchanfor some reason I have not been able to put a j680ap target into dfu mode with tuxvdmtool then detect the dfu usb device reliably in the new tuxvdmtool version that does not use cd321x_vdm#asahi-dev · 2026-05-26 01:01
yukaI hope there is a better way#asahi-dev · 2026-05-26 01:00
yukaSo essentially I would need to keep my own set of mappings and then decide based on that if I go through the IOMapped functions or treat the address in the kernel memory?#asahi-dev · 2026-05-26 00:58
yukaAnd I can not just go and map it where I need it, because I'm in the shared kernel address space#asahi-dev · 2026-05-26 00:58
yukaBut IOMemoryDescriptor just maps it somewhere and gives me a special function to map my physical address back to a virtual one#asahi-dev · 2026-05-26 00:56
yukaIf I understand it correctly, m1n1 enables the mmu and then maps those ranges with virtual address = physical address#asahi-dev · 2026-05-26 00:49
yukaFor the kextproxy I can do mmio reads/writes fine using IOMapped{Read,Write} but it's a bit of a pain to detect whether something should go through that path or be read directly#asahi-dev · 2026-05-25 22:42
berdankaWill try it out now#asahi-dev · 2026-05-25 22:42
berdankaThanks, got it#asahi-dev · 2026-05-25 22:42
chaos_princesssth like that#asahi-dev · 2026-05-25 22:41
berdankaarch-chroot "${MNTDIR}" update-grub