#
✕ nick:sven
#asahi-dev · 2023-08-17 11:29
svennot resetting the co-processor between connecting different devices often resulted in a full reset after N replugs
#asahi-dev · 2023-08-17 11:28
svenit used to do that
#asahi-dev · 2023-08-17 11:28
svenI fixed the crash the system part :D
#asahi-dev · 2023-08-17 11:27
svenat least I can restart that processor without a full reboot but suddenly disconnecting everything connected to the type c port isn't very nice :D
#asahi-dev · 2023-08-17 11:26
svenand unreliably even when everything's disabled
#asahi-dev · 2023-08-17 11:26
svenyeah, i can reliably crash the tbt co-processor if I enable debug prints inside thunderbolt otherwise
#asahi-dev · 2023-08-17 11:24
sventhunderbolt is a different mess as usual since it seems that I have to lock some NHI operations against PCIe hotplug *sigh*
#asahi-dev · 2023-08-17 11:24
svenif nothing breaks over a few days maybe we can actually include it for people to test then
#asahi-dev · 2023-08-17 11:23
svenokay, i'll see if I can get at least the usb 3 stuf ready then
#asahi-dev · 2023-08-17 11:23
svenas long as you make sure tipd + dwc3 are from the same branch everything's fine
#asahi-dev · 2023-08-17 11:23
sventhe new tipd thing also doesn't break DT compatibility fwiw\
#asahi-dev · 2023-08-17 11:22
svenmy plan is to prepare something that a few people can test on the weekend or so
#asahi-dev · 2023-08-17 11:22
svenmarcan: any other patches except for the pcie multiple controller thing you want asap? i've rebase and cleaned up my tree a bit but I'm not convinced the new usb 3 stuff is ready for wide testing yet
#asahi-dev · 2023-08-17 10:49
sveneiln: nice!! :)
#asahi-dev · 2023-08-16 07:34
svenstill reading code and haven’t found anything suspicious :/
#asahi-dev · 2023-08-16 07:02
sven maybe some code is reordered which then maybe triggers a missing barrier bug or something like that somewhere :/
#asahi-dev · 2023-08-16 07:01
sven🤨
#asahi-dev · 2023-08-15 20:20
svenweird
#asahi-dev · 2023-08-15 20:14
svenfirst one after boot or first one after a wrap around? and i guess that buffer is mapped to dcpext?
#asahi-dev · 2023-08-13 12:24
svenyeah, I’ll prepare my current branch. The magic bit that now turns on is just some BIST completed successfully for the SS lanes iirc
#asahi-dev · 2023-08-13 12:23
svenit’s not a big change, just a mutex plus a counter for that notifier iirc
#asahi-dev · 2023-08-13 12:23
sventhe PCIe multiple controller thing didn’t change iirc
#asahi-dev · 2023-08-13 12:20
svenI hope the new usb3 stuff is fine now
#asahi-dev · 2023-08-13 12:19
svenbut I can split off the reasonable parts from my current branch
#asahi-dev · 2023-08-13 12:19
sventoo many hacks to work around the old tipd stuff
#asahi-dev · 2023-08-13 12:19
sventhere’s not much stuff that’s reasonable in there :D
#asahi-dev · 2023-08-13 12:18
svenYeah, it still has the old tipd code
#asahi-dev · 2023-08-13 12:16
svenyeah, that’s when I stopped pushing branches because people started to annoy me too much
#asahi-dev · 2023-08-13 12:16
svenhm… maybe? who knows
#asahi-dev · 2023-08-13 12:16
svenoh, right, there’s still an old janky tree that maybe works around
#asahi-dev · 2023-08-13 12:15
svenokay, I’ll try to clean the current mess up in the next days
#asahi-dev · 2023-08-13 12:15
svenno other Linux machine with tb support?
#asahi-dev · 2023-08-13 12:15
svenah, heh
#asahi-dev · 2023-08-13 12:14
svenjannau has at least one nvme controller that didn’t work (BAR read as ff) so testing the 10gbe might still be a good idea
#asahi-dev · 2023-08-13 12:14
svenlspci will tell you what it really is
#asahi-dev · 2023-08-13 12:13
sven10gbe sounds like something useful to test though
#asahi-dev · 2023-08-13 12:13
svenI have the owc tbt hub so that’s probably goshen ridge
#asahi-dev · 2023-08-13 12:12
sven(and another broken titan ridge)
#asahi-dev · 2023-08-13 12:12
svenmarcan: which dock? I have a tbt3 (titan ridge) and a usb4 (goshen ridge) one here
#asahi-dev · 2023-08-13 10:37
svenlets see about DP-altmode and tunneling now
#asahi-dev · 2023-08-13 10:37
svenlooks like finally usb3 and tbt works with the new tipd and atcphy code
#asahi-dev · 2023-08-13 09:45
svenlol, i thought i just broke tbt completely but it turns out that it's just this old caldigit dock that broke completely now :D
#asahi-dev · 2023-08-11 14:00
svenwith a very helpful „assert 4816 failed“ printf because everything thunderbolt seems to be heavily redacted
#asahi-dev · 2023-08-11 13:59
sventhe mess starts when you have chained tbt devices that possibly also tunnel DP. I still can crash the tbt coprocessor in those cases :/
#asahi-dev · 2023-08-07 14:03
svenmarcan: I have a patch in some branch to teach the pcie driver about multiple controllers fwiw, also needed that for thunderbolt
#asahi-dev · 2023-08-06 17:06
svenand now thunderbolt works again but usb3 broke. wtf. there's something really really dumb going on here
#asahi-dev · 2023-08-02 19:46
svenusually by the time we manage to get them upstream (with some minor exceptions)
#asahi-dev · 2023-08-02 19:46
sveneventually the dtbs will be stable enough that you can mix and match then with kernels
#asahi-dev · 2023-08-01 09:52
sventhey also have a (somewhat broken) virtual GIC but that’s only relevant for VMs
#asahi-dev · 2023-08-01 09:51
svenyeah, and just to confirm: the root interrupt controller is the AIC on apple machines

plus ›