#asahi-dev · 2023-02-06 08:55
sventhat's kinda what i expected#asahi-dev · 2023-02-06 08:02
sveninteresting#asahi-dev · 2023-02-05 13:19
svendoesn't get any modes yet, either it needs additional DPTX calls or I'm missing something inside NHI#asahi-dev · 2023-02-05 13:18
sven*almost*#asahi-dev · 2023-02-05 13:18
svenIOMFBStatus VideoInterfaceIOAV::plug_gated(IOAVVideoInterface *): display HPD asserted#asahi-dev · 2023-02-03 22:50
svenhrm, i get the hpd interrupt in dpin, the nhi part claims the tunnel is up, I set up the crossbar but dcpext doesn’t see the hpd :/#asahi-dev · 2023-02-03 20:12
svenI don't even remember how that all started but I do remember that I just wanted to figure out how $something worked#asahi-dev · 2023-02-03 20:11
svendistraction is how I ended up with this usb4 mess :P#asahi-dev · 2023-02-03 20:11
sventsk tsk ;)#asahi-dev · 2023-02-03 20:05
svenmostly just want to have a good understanding how that part interacts with the rest before I start cleaning up the usb4 and pcie changes#asahi-dev · 2023-02-03 20:04
svenonce is good enough for now ;)#asahi-dev · 2023-02-03 20:04
svensounds good, thanks#asahi-dev · 2023-02-03 19:56
svenisn’t review on mailing lists fun? :p#asahi-dev · 2023-02-03 19:54
svenjannau: oh, and what’s your latest DCP branch? I wanna try to bring the DP tunnel up this weekend #asahi-dev · 2023-02-03 19:52
svenah, then bt probably also needs these memory offers to be mapped into the BAR extracted as well#asahi-dev · 2023-02-03 19:51
svenI’d only send each individual patch to the relevant maintainers fwiw#asahi-dev · 2023-02-03 19:50
svendid you already check if Bluetooth just works btw.?#asahi-dev · 2023-02-03 19:49
svenmaybe just send patch 1 to the corresponding maintainers and don’t include them for the rest?#asahi-dev · 2023-02-03 19:48
svenOh… maintainers for N different subsystems?#asahi-dev · 2023-02-03 19:48
svenuh. that sounds a bit too many#asahi-dev · 2023-02-03 18:40
svenlooks like there are even more magic pokes inside atcphy required to make DP tunneling work which is weird #asahi-dev · 2023-02-03 18:39
svenalright, figured out how dpin0/1 work. they have a single “dptx enable” bit and a slightly cumbersome way to get HPD changes#asahi-dev · 2023-02-03 08:08
svenah.. and a few sentences later the spec talks about an "implementation-specific mechanism that is more appropriate for the architecture of the Host"#asahi-dev · 2023-02-03 08:05
svenhrm, so all thunderbolt devices end up in the same iommu group because not even the root ports seem to support PCIe ACS. if I'm reading the usb4 spec correctly they should support that though :/#asahi-dev · 2023-02-02 13:22
sventrue. i was mostly joking fwiw ;)#asahi-dev · 2023-02-02 13:18
svenjannau: there's also the kext and m1n1 traces, how could you forget these :p#asahi-dev · 2023-02-02 10:57
svenjannau: good point. I'm still a bit sad that the USB4 variant's datasheets are hidden behind an NDA :(#asahi-dev · 2023-02-02 10:56
svenand then finally probably create some debugfs file or something like that to trigger the commands from userspace#asahi-dev · 2023-02-02 10:54
svenyes#asahi-dev · 2023-02-02 10:53
sventhen you can probably implement a small proof of concept inside the tipd driver to unconditionally send some of those commands using tps6598x_exec_cmd to make sure they work#asahi-dev · 2023-02-02 10:52
svenprobably a good start is to look at that macvdmtool and see what those commands are#asahi-dev · 2023-02-02 10:52
svenhttps://github.com/AsahiLinux/macvdmtool uses that i2c interface to send apple-specific commands to e.g. enable serial or reboot the target machine #asahi-dev · 2023-02-02 10:51
svent: tipd sits on a i2c bus and there's a public datasheet that's close enough (https://www.ti.com/lit/ug/slvuan1a/slvuan1a.pdf). #asahi-dev · 2023-02-02 10:50
svent: sure!#asahi-dev · 2023-02-01 19:58
svenI guess I can also see if I can convince my Mac mini running Linux and my air running macOS to establish a connection #asahi-dev · 2023-02-01 19:57
svenohhh… looks like Linux also has support for interdomain connections #asahi-dev · 2023-02-01 19:52
svensure, I’ll ping you once I cleaned this up a bit more. that enclosure sounds like a good test to see if we get a decent speed on the bus#asahi-dev · 2023-02-01 17:23
svenit’s usually ~20% and then the shipping company will try to essentially scam me in addition to that for “automated import and customs processing” so it’s probably not worth the hassle. but thanks for the offer!#asahi-dev · 2023-02-01 17:06
sventb4 is usb4 with some required features and certification #asahi-dev · 2023-02-01 17:05
svenI think usb4 works but I only have tb1 and half-broken tb3 devices here#asahi-dev · 2023-02-01 17:01
svenis that a “true” usb4 hub? you’d also have to ship it to .de and I’d probably have to pay import taxes on it so that might just not be worth the effort #asahi-dev · 2023-02-01 15:47
svenurgh.. looks like my usb4 changes for atcphy broke usb3 :(#asahi-dev · 2023-01-31 15:47
sven(that connection also contains load-bearing duct tape on the caldigit connector since that thing only works as long as there's some pressure on it :>)#asahi-dev · 2023-01-31 15:43
svenhotplug on the downstream port of the caldigit dock also works so I assume that there's just something wrong with the apple thunderbolt display in general where hotplug doesn't work#asahi-dev · 2023-01-31 15:42
svenfor some reason everything ends up in the same iommu group though#asahi-dev · 2023-01-31 15:42
svenhttps://f.svpe.de/ccf815cbb0af10eaa716af0cace6d5b289224f58acaf13def93813b11e0c77f3_lspci.txt old thunderbolt display via 3->2 adapter via half-broken caldigit dock and it seems to work :D#asahi-dev · 2023-01-31 13:11
sventhis is going to be 'fun' to debug#asahi-dev · 2023-01-31 13:11
svenurgh.. the rescan->unplug leaves stale buses around in e.g. /sys/class/pci_bus/#asahi-dev · 2023-01-31 12:46
svenbut only with the more complex topology created by the thunderbolt display. just the ethernet adapter seems to be fine#asahi-dev · 2023-01-31 12:46
svenugh, looks like plug tbt device -> manual pci bus rescan -> unplug device confuses something in the pci subsystem and crashes the kernel eventually :/