#asahi-dev · 2026-02-24 19:20
svenstill only compile tested :D#asahi-dev · 2026-02-24 19:20
svendropped#asahi-dev · 2026-02-24 19:20
svenyou probably also want to drop that enter KIS commit, that's just there because I started from the wrong base when adding the other code#asahi-dev · 2026-02-24 19:18
svennickchan: i think https://github.com/AsahiLinux/m1n1/tree/dockchannel-hacks was my latest version, only compile tested though#asahi-dev · 2026-02-24 08:25
svenugh, sorry, forgot about it yesterday. will do this evening #asahi-dev · 2026-02-22 15:06
svensure, probably tomorrow evening #asahi-dev · 2026-02-22 13:33
sventhere's no need for all that weird stuff you're doing, there's a register that gives you the number of bytes available iirc#asahi-dev · 2026-02-22 13:32
svenotherwise you're just duplicating working for no reason#asahi-dev · 2026-02-22 13:32
svenuh, i mean, you could've just asked and i can push it once i'm back at a computer#asahi-dev · 2026-02-22 13:17
sveni already have an implementaiton fwiw#asahi-dev · 2026-02-14 09:29
sventhe MMIO address is different from the SoC's dockchannel so this might just be the other end of the FIFO and it might have no access to any other blocks#asahi-dev · 2026-02-14 09:28
sveni wonder what else that thing has access to#asahi-dev · 2026-02-14 09:28
sven> PAR seq=4352, portal=PAM, offset=0, addr = 0x23d13402c, len = 0x00000004 and that's checking RX_FREE#asahi-dev · 2026-02-14 09:28
sven> PAR seq=4432, portal=PAM, offset=0, addr = 0x23d134014, len = 0x00000004 but there it's clearly checking TX_FREE just before that#asahi-dev · 2026-02-14 09:26
sventhough i would've expected 0x4004 / TX8 there instead of 0x4010 / TX32#asahi-dev · 2026-02-14 09:25
sven> PAW seq=4448, portal=PAM, offset=0, addr = 0x23d134010, len = 0x0000000c, data = bytearray(b'Hello, test\n')#asahi-dev · 2026-02-14 09:25
svenalso, lol, no idea how I didn't realize that debugusb literally is just MMIO access to some dockchannel block#asahi-dev · 2026-02-14 09:19
svenoh.. but i guess that wouldn't reset the PHY inside that debugusb co-processor #asahi-dev · 2026-02-14 09:18
svenit's connected to the tipd chip via i2c after all#asahi-dev · 2026-02-14 09:18
svenmaybe there's some way to reset it with another VDM message#asahi-dev · 2026-02-14 09:18
svenso i guess it makes sense that it's in the way as well#asahi-dev · 2026-02-14 09:17
sven*repeater#asahi-dev · 2026-02-14 09:17
sventhat repeat is a eusb level shifter, yeah. i found its datasheet back then#asahi-dev · 2026-02-14 09:17
svenhttps://f.svpe.de/b0e64c56f05b94e9b64adeb55ef71083c9c532e9196fc865a892400bab03a181_0001-HACK-enter-debugusb.patch#asahi-dev · 2026-02-14 09:13
svener wait.. that's outdated i think#asahi-dev · 2026-02-14 09:12
svenhttps://f.svpe.de/19cda7de66f180f06aa60471f0b989d5b6d45e130e666927ea25b9ebe1331e3b_0001-enter-KIS.patch is my hack to enter that debugusb/kis mode fwiw#asahi-dev · 2026-02-14 09:10
sveni guess the eusb repeater is also in the debugusb -> soc path#asahi-dev · 2026-02-14 09:10
svenyeah, i've tried switching to usb and then back to debugusb and that didn't help#asahi-dev · 2026-02-14 08:59
svensecond time around nothing shows up#asahi-dev · 2026-02-14 08:59
svenat least on m1 i can only plug the cable in once and have the debugusb device show up#asahi-dev · 2026-02-14 08:59
svenlol, just realized that debugusb / kis mode has the same hotplug issue like the main dwc controller :D#asahi-dev · 2026-02-12 18:22
svenETA 10 days for m3 gpu support? awesome!#asahi-dev · 2026-02-12 18:18
svenShiz: so how about that m3 gpu? :>#asahi-dev · 2026-02-12 16:01
sventhere's no simple fix for this#asahi-dev · 2026-02-12 16:01
svenand again, there is no simple solution. multiple people who understand how this platform works have put thought into this and we came up with a few complex options to look into#asahi-dev · 2026-02-12 15:59
svenyeah#asahi-dev · 2026-02-12 15:59
svenif there was a simple fix for this we would've already done it#asahi-dev · 2026-02-12 15:58
svenplease read again what i just wrote#asahi-dev · 2026-02-12 15:50
svenpre-SPTM there was PPL which does similar things and that one just gets virtualized inside GL1 under m1n1 on m1-m3#asahi-dev · 2026-02-12 15:49
svenon m4 stuff like GXF gets disabled in raw boot mode which makes this much harder#asahi-dev · 2026-02-12 15:49
svenwhat matters is that on m1-m3 all hardware extensions are still available in raw boot mode so we could even virtualize SPTM there if we wanted to#asahi-dev · 2026-02-12 11:02
svenyou can just look at the devicetrees to figure out which variant is used by which device#asahi-dev · 2026-02-12 11:01
svenyeah, it's hardware#asahi-dev · 2026-02-12 06:09
svenwe know that its SWD lines are hooked up to some SoC gpio#asahi-dev · 2026-02-12 06:09
sventhat kext is probably for the pd controller #asahi-dev · 2026-02-11 19:37
sventhe real debug interface was ofc disabled#asahi-dev · 2026-02-11 19:36
svenhe talked to JTAG or SWD but all that worked was the "get device id" or whatever it's called command#asahi-dev · 2026-02-11 19:36
sventhe title is clickbait and misleading