#
✕ nick:sven
#asahi-dev · 2023-06-03 09:33
svenalright, let’s see if I can get away with just applying the atc fuses inside m1n1 so that we can skip the entire nvmem mess
#asahi-dev · 2023-05-29 11:28
sventhe worst that can happen on these apple Socs is that your device isn’t correctly recognized fwiw
#asahi-dev · 2023-05-29 11:27
svenType-c *everywhere* is a total mess afaict :D
#asahi-dev · 2023-05-29 10:41
svenalright, completely bypassed most of the altmode typec stuff just like the chromeos ec guys and this seems to be pretty stable so far :>
#asahi-dev · 2023-05-28 13:46
svenIirc atcphy is just setting up the lane mode and clocks and stuff like that
#asahi-dev · 2023-05-28 13:45
sveneither through the actual PHY or through one of the two DPIN tbt tunnels
#asahi-dev · 2023-05-28 13:44
svenI think atc just passes through whatever dcpext spits out
#asahi-dev · 2023-05-28 13:44
svenyeah, then I guess they could implement it all inside dcpext
#asahi-dev · 2023-05-28 12:57
svenAt which layer is MST done in DP? Atcphy has some very low level PHY setting for DP but I think most of the protocol is handled inside DCP
#asahi-dev · 2023-05-28 12:56
sven*multi stream tiling displays
#asahi-dev · 2023-05-28 12:55
sventhat string could also “just” be a more generic multi stream display handling across multiple dcpext. That would be very weird though
#asahi-dev · 2023-05-28 12:54
svenif you enter tb altmode or usb4 mode you have to tunnel everything else
#asahi-dev · 2023-05-28 12:54
sventb altmode with two lanes tb and two lanes DP isn’t a thing afaik
#asahi-dev · 2023-05-09 15:27
svenlol
#asahi-dev · 2023-05-09 15:27
svenif you really want some pain you can always pick up fixing the pcie hotplug mess as well :>
#asahi-dev · 2023-05-09 15:27
sventhough you ofc access registers of all the crap on the path to the device via those ringer buffer commands
#asahi-dev · 2023-05-09 15:26
svenat least it’s mostly ring buffers there and not much mixing between registers and ring buffers
#asahi-dev · 2023-05-09 15:26
svennhi/thunderbolt does that dma ring buffer thing as well but it doesn’t seem to be as bad xhci
#asahi-dev · 2023-05-09 15:24
sventhat dwc3 mode switch race was pretty bad as well ;)
#asahi-dev · 2023-05-09 15:23
svenmarcan: at least someone else gets to share my usb pain :p
#asahi-dev · 2023-05-09 13:57
svenAnd then you realize that 2016 is already 6-7 years ago now…
#asahi-dev · 2023-05-08 08:26
svenlol
#asahi-dev · 2023-05-08 08:26
svenso maybe 10 or so different variants?
#asahi-dev · 2023-05-08 08:26
svenwell, there’s also the iPhone/iPad SoCs
#asahi-dev · 2023-05-08 08:26
svenah, right, same kernel for all t8103 models, etc.
#asahi-dev · 2023-05-08 08:25
svenit helps that they run xnu on like 20 different configuration and just compile it separately for those
#asahi-dev · 2023-05-08 08:24
svenor how that gpu mmu is a separate adt node with essentially a stub kext that just calls back into the main kext iirc
#asahi-dev · 2023-05-08 08:24
svenyeah :D
#asahi-dev · 2023-05-08 08:23
sventhat’s all automatically done for x86 but here we have to do it manually
#asahi-dev · 2023-05-08 08:23
svento essentially notify pcie, dwc3 and dcpext when a tunnel state changes
#asahi-dev · 2023-05-08 08:23
svenwe just need something like typec mux/switch for thunderbolt as well but that’s easy enough
#asahi-dev · 2023-05-08 08:22
svennot more than what typec mux/switch already does, yeah
#asahi-dev · 2023-05-08 08:21
svenand the typec maintainer even reviewed all of that so I don’t even see many issues getting it upstream
#asahi-dev · 2023-05-08 08:20
sventhat will also work for us
#asahi-dev · 2023-05-08 08:20
svenyeah, and chrome just skips all that and directly uses the typec mux/switch stuff
#asahi-dev · 2023-05-08 08:20
svenbecause that subsystem assumes that the cpu does all the negotiation with these usb pd messages
#asahi-dev · 2023-05-08 08:19
sventypec itself is fine, typec-altmode isn’t a great fit
#asahi-dev · 2023-05-08 08:19
svenand google does something similar hidden away in platform for some chromebook(?) thing
#asahi-dev · 2023-05-08 08:18
sventhat entire thing is not a great fit for these SoCs
#asahi-dev · 2023-05-08 08:18
svenyeah, my plan is to just skip typec-altmode
#asahi-dev · 2023-05-08 08:16
svenwhich probably also explains that commented out part in atcphy that’s waiting for some bit that never clears with my code
#asahi-dev · 2023-05-08 08:15
svenyeah, I think the atcphy vs dwc3 setup is in the wrong order in asahi-dev right now
#asahi-dev · 2023-05-08 07:20
sven(and any further attempt as well)
#asahi-dev · 2023-05-08 07:20
sven*the second time works though
#asahi-dev · 2023-05-08 07:20
sventhe second time i
#asahi-dev · 2023-05-08 07:20
svendamnit.. rebased the thunderbolt mess to 6.4 and broke something and now I’m back to the first plug just not working
#asahi-dev · 2023-05-07 10:27
svenI am. please complain to the maintainer of the kernel and not here.
#asahi-dev · 2023-05-07 10:27
sventhat still doesn’t make sense. you had a usb3 device from what I can tell and some unsupported kernel broke it 🤷🏻‍♂️
#asahi-dev · 2023-05-07 10:24
svensounds like you’re confused, thunderbolt is not related to lsusb
#asahi-dev · 2023-05-07 10:20
svenI seriously hope no one shipped a kernel with my half broken thunderbolt support to end end users

plus ›