#
✕ nick:sven
#asahi-dev · 2022-11-02 19:06
svenI have some old dock that does 3+dp as well
#asahi-dev · 2022-11-02 18:59
svenlol, I wonder if that would actually work :D
#asahi-dev · 2022-11-02 18:46
svennow I just need to understand how to hook up the drm connector to the typec port in the device tree and write some glue code and then hopefully hotplugging displays should also work
#asahi-dev · 2022-11-02 18:45
svencd321x registers are exactly the same as tps6598x for the DP pin assignment
#asahi-dev · 2022-11-02 18:44
svenLinux driver, I just messed up something in my code which confused me
#asahi-dev · 2022-11-02 18:18
svenjannau: great! sounds like a good test for the 4 lane atcphy setup then
#asahi-dev · 2022-11-02 18:17
svenso now just HPD is missing and then I've finally shaved all yaks and can go back to thunderbolt :>
#asahi-dev · 2022-11-02 18:16
svenbecause technically this capture card doesn't have any usb3 connection so switching to 4 lane mode is correct here
#asahi-dev · 2022-11-02 18:15
svenoh, nevermind, my 4 lane code just did something dumb
#asahi-dev · 2022-11-02 17:57
sven(and also to confirm my atcphy setup works for 4 lanes as well)
#asahi-dev · 2022-11-02 17:57
svenprobably, it looks like apple modified some bits inside these cd321x chips and my 2-lane DP pin assignments looks like a 4-lane assignment. would be interesting to see how 4 lane actually looks
#asahi-dev · 2022-11-02 17:37
svenanyone with such a display that connects via display port and not thunderbolt here?
#asahi-dev · 2022-11-02 17:37
svenhrm, what kind of displays would require 4 lane display port? 5k I guess?
#asahi-dev · 2022-11-02 16:03
svenhah, found the HPD signal. it's in one of the "reserved" bits in the tps6598x DATA_STATUS and triggers and irq there
#asahi-dev · 2022-11-02 14:53
svenI think usb4 devices are pretty new
#asahi-dev · 2022-11-02 14:53
svenyeah, thanks.
#asahi-dev · 2022-11-02 14:51
sventhat's the same as my tb3 dock I think :(
#asahi-dev · 2022-11-02 14:50
svenI want to see if tps6598x identifies it differently because the PD commands to enter USB4 are different from those used for pre-USB4 thunderbolt
#asahi-dev · 2022-11-02 14:49
svenI care about the _device_ here
#asahi-dev · 2022-11-02 14:49
svenI e.g. have an older caldigit dock that's advertised as Thunderbolt 3. the newer version of that dock claims to be "Thunderbolt 4 / USB4"
#asahi-dev · 2022-11-02 14:48
svencheck if the marketing materials mention USB4 :D
#asahi-dev · 2022-11-02 14:47
svenyou guys don't happen to have a device that specifically claims to be USB4 and not "just" thunderbolt around, do you?
#asahi-dev · 2022-11-02 13:36
svenmight even prevent people who only know C from contributing
#asahi-dev · 2022-11-02 13:36
svenrust makes a lot of sense in certain areas but converting from C to rust just because "C is obsolete" doesn't really gain anythiing
#asahi-dev · 2022-11-02 13:35
sveni think that's a pretty bad perspective
#asahi-dev · 2022-11-02 13:35
svenbut if the ADT is malicious all bets are off anyway
#asahi-dev · 2022-11-02 13:34
svenrust for e.g. fat32 makes sense because that's a potential attack vector
#asahi-dev · 2022-11-02 13:34
svenwhat would be the benefit of that?
#asahi-dev · 2022-11-02 12:50
svenbut no idea if they already started
#asahi-dev · 2022-11-02 12:50
svendon't think so, I only vaguely recall someone working on linux for x86 macs with that T2 chip wanted to write one
#asahi-dev · 2022-11-02 12:16
svenI can test that one myself later though, but good idea
#asahi-dev · 2022-11-02 12:15
svenno idea, maybe
#asahi-dev · 2022-11-02 11:48
svenanyone with a device that specifically claims to be USB4 (i.e. not just thunderbolt) here who could run this script? https://f.svpe.de/ae4da979c62f4048dfebf0f55568ff27a81bbb66d9248e63d9a25584fd74c57b_tipd.py
#asahi-dev · 2022-11-01 22:04
svennext up is hotplug support then I guess
#asahi-dev · 2022-11-01 22:04
svenlooks like it’s stable now :)
#asahi-dev · 2022-11-01 20:04
svenyeah, and then I get bored with $work or something again and continue :-P
#asahi-dev · 2022-11-01 20:02
svensure, my plan is to trigger the breakpoint on the first atcphy register write that I care about
#asahi-dev · 2022-11-01 20:02
svenhttps://developer.apple.com/documentation/kernel/1575328-iodelay
#asahi-dev · 2022-11-01 20:01
sven*inlined
#asahi-dev · 2022-11-01 20:01
sveniodelay is literally a XNU function that's not inline for some reason iirc
#asahi-dev · 2022-11-01 19:59
svenmight get too spammy, but let's see
#asahi-dev · 2022-11-01 19:59
svenmaybe. I assume xnu uses iodelay all over the place
#asahi-dev · 2022-11-01 19:58
svenhuh, maybe I can just set a breakpoint on iodelay
#asahi-dev · 2022-11-01 19:56
svendo you still feel envy? :P
#asahi-dev · 2022-11-01 19:54
svenI could probably hack some bind/unbind thing but since reboots itself are fast enough it doesn't matter
#asahi-dev · 2022-11-01 19:54
sventhere's no hotplug support yet :D
#asahi-dev · 2022-11-01 19:53
svenand usually 20 retries are enough to catch it
#asahi-dev · 2022-11-01 19:53
svenit works, I dunno, 9/10 times without any issue
#asahi-dev · 2022-11-01 19:53
svenyeah, it's just unstable
#asahi-dev · 2022-11-01 19:52
sventhe recompile really isn't the issue

plus ›