#asahi-dev · 2026-01-04 18:29
svenbut i found no evidence of either in the firmware I dumped from the apple-variant of those chips#asahi-dev · 2026-01-04 18:28
sventhe TI chips both have a mode where we control the raw PD messages (i.e. they effectively act like a dumb PD PHY) and another mode where we can tell them to enter other altmodes#asahi-dev · 2026-01-04 18:28
rosalyntg_interesting, ok makes sense#asahi-dev · 2026-01-04 18:28
svenyup#asahi-dev · 2026-01-04 18:27
rosalyntg_yep, hitting that on my TB dock :/#asahi-dev · 2026-01-04 18:27
rosalyntg_yeah saw that part, just wasn't familiar enough to know exactly what functionality the altmodes/dp.c implements. so that logic is...directly in the PD chip?#asahi-dev · 2026-01-04 18:27
svenwhich is a bit unfortunate because that means we can't use the usb3 fallback that most tbt docks have right now :(#asahi-dev · 2026-01-04 18:27
svenwe just get told by that pd controller what mode it negotiated and have to deal with it#asahi-dev · 2026-01-04 18:27
rosalyntg_I see, that makes sense then#asahi-dev · 2026-01-04 18:26
svenwe don't have any control over altmode negotiation or the PD VDO/PDO messages#asahi-dev · 2026-01-04 18:26
svenwe deliberately bypass it even #asahi-dev · 2026-01-04 18:26
svenwe can't use that :D#asahi-dev · 2026-01-04 18:26
rosalyntg_sven: thanks, interesting. drm_bridge makes sense for the DP->HDMI probably (need to read more about it, the DRM drive I've written doesn't use it). Not sure that it makes sense to me for the type-c ports--esp because the generic dp altmode driver (typec/altmodes/displayport.c) already has a way to send the hotplug events to DRM...but maybe there's some reason we can't use that#asahi-dev · 2026-01-04 18:25
svenokay, that should have a slightly higher chance of working#asahi-dev · 2026-01-04 18:25
yuka(proxyclient/tools/linux.py)#asahi-dev · 2026-01-04 18:25
yukabare metal#asahi-dev · 2026-01-04 18:25
svenare you running bare metal or under the HV?#asahi-dev · 2026-01-04 18:24
yuka(I did add earlycon, but also no output)#asahi-dev · 2026-01-04 18:24
svenearlycon is also very helpful for early bringup#asahi-dev · 2026-01-04 18:24
yukayes. so probably still guessed some values wrong ;)#asahi-dev · 2026-01-04 18:24
svenas long as you have very basic stuff like aic and the serial in there it should work#asahi-dev · 2026-01-04 18:23
yukaif the device tree doesn't have any pmgr nodes would it still try to bring it up at all?#asahi-dev · 2026-01-04 18:23
yukaI just now tried making a minimal t8132 device tree inspired by the m3 max minimal device tree (posted by integral_pilot as a gist a week ago), and loading it baremetal, but so far getting only "Vectoring to next stage..." (at least improvements from previous tries where I forgot again that on t8132 the m1n1 exception handlers throw exceptions again...)#asahi-dev · 2026-01-04 18:23
svenpd_ignore_unused#asahi-dev · 2026-01-04 18:22
sventhere's a kernel bootarg, something something unused_pd#asahi-dev · 2026-01-04 18:22
yukaas expected, thx#asahi-dev · 2026-01-04 18:22
svenyuka: you can boot without the pmgr driver, but if you instantiate the pmgr driver you have to make sure it doesn't disable the serial power domain#asahi-dev · 2026-01-04 18:22
yuka(kernel bringup I should say)#asahi-dev · 2026-01-04 18:21
svenrosalyntg_: https://social.treehouse.systems/@janne/115808166298942165 might be interesting for you#asahi-dev · 2026-01-04 18:21
yukais pmgr strictly needed for serial or can I leave power-domain unset during bringup?#asahi-dev · 2026-01-04 18:18
svenmaybe, didn’t look at the your dmesg tbh. I stumbled upon this while starting with tbt bringup #asahi-dev · 2026-01-04 18:17
GlanzmannIf I understand correctly, https://gist.github.com/svenpeter42/a2b3602ca1f6e27e5cadd912a0ae7dec would have caught my outdated m1n1 and report an error instead of dereferencing a null pointer when trying to set tunables which were not there.#asahi-dev · 2026-01-04 18:16
svenhuh?#asahi-dev · 2026-01-04 18:16
Glanzmannsven: Thank you for making atc Glanzmann proof.#asahi-dev · 2026-01-04 18:15
rosalyntg_jannau: from my looking at what my x86 laptop does, I think we should probably have 2 drm_connector's for the DCPext. That way, the oob hotplug event can be fed to the right connector in DRM so that it gets wired correctly. Am I correct in thinking that its the dptx_phy argument of dptxport_connect that tells the DCP which ATC to send the signal to? Or do I have some wires crossed....#asahi-dev · 2026-01-04 18:13
svenreasons I dislike C: https://gist.github.com/svenpeter42/a2b3602ca1f6e27e5cadd912a0ae7dec#asahi-dev · 2026-01-04 18:08
svenwe'll have to build this in a way that eventually also supports the dp-via-tbt tunnels and supports more complicated routings on max/pro machines that have N dcpext and M ports with 3 different DP stream inputs (dp-altmode, 2x tbt tunnel)#asahi-dev · 2026-01-04 18:06
rosalyntg_Sounds good, thx. I've taken a first pass (not functioning yet) and ideally would run that by someone to make sure I'm understanding all the interactions properly#asahi-dev · 2026-01-04 18:06
svennah, just keep it in this channel. he's probably already noticed now that i mentioned him#asahi-dev · 2026-01-04 18:06
rosalyntg_yep, I understand#asahi-dev · 2026-01-04 18:05
svenm1 air will also only ever support one external display concurrently#asahi-dev · 2026-01-04 18:05
rosalyntg_ok, thanks. what's best IRC ediquette here, send a DM?#asahi-dev · 2026-01-04 18:05
svenjannau is the right person to talk to#asahi-dev · 2026-01-04 18:05
svenso that's playing the dcpext mix-and-match game then which is complicated#asahi-dev · 2026-01-04 18:04
rosalyntg_no, talking about, for example M1 Macbook Air, so one dcpext shared between two USB-C ports#asahi-dev · 2026-01-04 18:03
svenrosalyntg_: i assume you're talking about a machine that has multiple dcpext?#asahi-dev · 2026-01-04 18:03
chaos_princessaditya: i would suggest running checkpatch before sending it to other people#asahi-dev · 2026-01-04 18:02
rosalyntg_sven: curious if you sway my messages above about enabling both USB-C ports for DP alt mode (or maybe you can point me to someone who's better to talk to?)#asahi-dev · 2026-01-04 18:00
svensame!#asahi-dev · 2026-01-04 18:00
adityaAnd happy new year!