#asahi · 2026-08-27 16:01
chaos_princessok, so, do you have backups#asahi · 2026-08-27 16:01
chaos_princessits dead#asahi · 2026-08-27 16:00
niko[m]<niko[m]> "This is so much slower than I..." <- 14 hours in and the shrinking is at 69.8%#asahi · 2026-08-27 14:30
fl0_id(if you search logs, in the past there were similar questions and I think it worked for ppl)#asahi · 2026-08-27 14:29
fl0_idBotkilla afaik it should work. but I haven't tried it.#asahi · 2026-08-27 14:26
BotkillaGuys. I'd love to install Asagi on my headless Macbook Pro M1Pro (laptop display removed but it's connected to an ext. monitor via HDMI). Is not having the laptop display a dealbreaker?#asahi · 2026-08-27 13:41
jannauthat will remain separate and probably driven a kms bridge, same for the DP part of atc#asahi · 2026-08-27 13:39
chaos_princesswhat about this thingy? https://github.com/AsahiLinux/linux/blob/asahi-wip/drivers/phy/apple/dptx.c#asahi · 2026-08-27 13:39
jannaudcp's are essentially the crtc devices. unfortunately more than that because all other things KMS abstracts away also have to go through the DCP firmware#asahi · 2026-08-27 13:37
jannauwe have that already with the meta device display-subsystem#asahi · 2026-08-27 13:36
chaos_princess-s#asahi · 2026-08-27 13:36
chaos_princessmaybe do all the muxes as one single devices#asahi · 2026-08-27 13:36
chaos_princessand idk how to wire up all of that#asahi · 2026-08-27 13:36
jannaukms already allows N:M crtcs to connector mappings and we have (and can) to support that#asahi · 2026-08-27 13:36
chaos_princessso, they need to somehow ask the dcp driver for a free crtc, and possibly get a refusal#asahi · 2026-08-27 13:35
chaos_princesschadmed: the dp muxes are part of the typec block, so, there are a bunch of them, and they are semi-independent.#asahi · 2026-08-27 13:34
runxiyuuhh typo, didnt mean to say anything#asahi · 2026-08-27 13:34
runxiyuwhy #asahi · 2026-08-27 13:34
chadmedwe can just refuse a modeset that wants to attach another crtc if we've run out of dcps i think?#asahi · 2026-08-27 13:32
chaos_princessyes, but how do you do M:N with those?#asahi · 2026-08-27 13:30
chadmedi think we will need to do something with connectors or drm_bridge maybe #asahi · 2026-08-27 12:51
chaos_princessj`ey: that is actually a good question, since i am not sure how you are even supposed to represent a M:N crtcs to outputs mapping#asahi · 2026-08-27 12:49
chadmedthere are a few moving parts but my understanding is that the ideal state would be for the phy to get an idle dcp routed to it by the xbar#asahi · 2026-08-27 12:48
chadmedi had a terribly cursed idea a few years ago to get one of those "fell off a truck" tbt dev boards with a pcie slot on it, plug a HBA into it and turn a mac mini into a raid controller#asahi · 2026-08-27 12:47
j`eyhow does the tbt stuff hook into the dcp stuff? like kernel API wise#asahi · 2026-08-27 12:47
chadmedegpu is cool for certain market segments#asahi · 2026-08-27 12:47
chadmedreally fasst external disk cages without the need for some overkill enterprise fibre channel/san type solution are cool#asahi · 2026-08-27 12:47
j`ey:)#asahi · 2026-08-27 12:46
chadmedscience has been trying to answer this for 15 years#asahi · 2026-08-27 12:44
j`eywhat is tbt actually for if not alt-dp?#asahi · 2026-08-27 12:44
fl0_idmakes sense, maybe i wasn't awake enough yet ;)#asahi · 2026-08-27 12:43
fl0_idchadmed ok cool#asahi · 2026-08-27 12:35
chadmedpeople know we would have made a big song and dance about tbt#asahi · 2026-08-27 12:35
chadmedthe phrasing is a little ambiguous but i think in the context of the PHY it's fine#asahi · 2026-08-27 12:33
fl0_idchaos_princess for sure, I just thought it might give ppl ideas about all of tb working. but I'm sure it's fine#asahi · 2026-08-27 12:30
chadmednute gunray was very wise in this regard, no need to report on that until we have something to report #asahi · 2026-08-27 12:29
chadmedthat is exactly what happened lol. i had the draft sitting there for weeks because every morning i'd wake up and the thunderbolt situation would be crazy again#asahi · 2026-08-27 12:27
yukafl0_id: things are moving so fast in this regard, the status probably changed a dozen times while the blog post was being written ;)#asahi · 2026-08-27 12:26
chaos_princesseven if only xdomain (the thing that usb4ip and usb4stream use) worked, that is still a tb link, just the protocol adapters are missing#asahi · 2026-08-27 12:26
fl0_idthx for clarifying#asahi · 2026-08-27 12:26
fl0_idok good to know. some I thought only ip based tb worked#asahi · 2026-08-27 12:25
chaos_princesstb works on m3, the link comes up, you can use xdomain and usb3 over usb4 tunneling#asahi · 2026-08-27 12:24
chaos_princesswdym wrong#asahi · 2026-08-27 12:24
fl0_idso is that completely wrong or just the tb part?#asahi · 2026-08-27 12:23
fl0_idok, but specifically "Both the SPMI interface and ACE3 itself are now working in Asahi Linux, bringing USB 3.0 and Thunderbolt support to all M3 series devices."#asahi · 2026-08-27 12:18
chaos_princessand usb tunneling works too#asahi · 2026-08-27 12:18
chaos_princessthe phy part is working. #asahi · 2026-08-27 12:17
fl0_idbtw, was it intentional in the progress report that tb was almost working on m3? apart from device to device, does it work anywhere yet? might give people wrong ideas#asahi · 2026-08-27 07:43
yukayep#asahi · 2026-08-27 07:43
chaos_princessturned out that base m4 has a very similar usb phy as m3 series, so things pretty much just worked(tm) with very minor changes