#asahi · 2021-04-20 19:42
svenbut well.. if i was them i'd rather spend my time improving their hypervisor especially since there are enough other people interested in doing that upstreaming work#asahi · 2021-04-20 19:41
sven their proof of concept was good work that showed off their skills and probably even helped them validate their hardware models.#asahi · 2021-04-20 19:38
svenchoozy: i don't think that makes much sense from a business perspective from them. their core knowledge is virtualization and not maintaining linux drivers ;)#asahi · 2021-04-20 19:38
svenspeaking of which... i should get back to that iommu patch#asahi · 2021-04-20 19:37
svenbut since then there's been no interaction from them on the mailing lists#asahi · 2021-04-20 19:37
sventhere was one upstreaming attempt by a third-party afaik#asahi · 2021-04-19 20:06
sveni don't see any problem with posting such a list either. especially if it's easily obtained by asking the smc nicely#asahi · 2021-04-18 13:45
svenor you could just shave another yak and finish that usb-pd cable you've been meaning to build :P#asahi · 2021-04-18 13:44
sventhat sounds like a neat hack :D#asahi · 2021-04-18 13:40
sveni.e. no debug port mode#asahi · 2021-04-18 13:40
sventhat's all the xhci register space claims to support#asahi · 2021-04-18 13:40
sven0x900 0x3100002 Supported Protocol#asahi · 2021-04-18 13:40
sven0x8f0 0x2000402 Supported Protocol#asahi · 2021-04-18 13:40
sven0x8e0 0x401 USB Legacy Support#asahi · 2021-04-18 13:32
svennot really. you'll probably also need to bring up the iommu#asahi · 2021-04-18 13:31
sveni might've messed up, but afaict the dwc3 xhci doesn't support this debug port#asahi · 2021-04-18 13:30
svenbecause the kernel won't be able to find the xhci controller otherwise#asahi · 2021-04-18 13:30
svenyou'll probably still need the dwc3 node though#asahi · 2021-04-18 13:30
svensure :)#asahi · 2021-04-18 13:21
svenit comes up quite some time before userspace afaik, but i haven't played much with it because macos doesn't seem to attach their cdc kext to it#asahi · 2021-04-18 13:20
sven"The xHCI debug capability (DbC) is an optional but standalone functionality provided by the xHCI host controller. " let's see#asahi · 2021-04-18 13:20
svenit might also just not be supported on apple hw. they don't seem to care much about standards#asahi · 2021-04-18 13:19
svenprobably#asahi · 2021-04-18 13:16
svenyeah, but i'm not so sure if the mac mini even has a usb3 debug port#asahi · 2021-04-18 13:14
svenor alternatively: enable full bypass mode in the usb iommu, add dma-ranges to the device tree, patch the dwc3 driver to use 64bit dma addresses and add a node for it to the device tree as well#asahi · 2021-04-18 13:13
svenmake sure to enable g_serial as well ofc ;)#asahi · 2021-04-18 13:13
svenit's somewhat hacked together, but tl;dr: grab my iommu patches, make sure the usb PHY has been brought up from m1n1 and add a dwc3 node attached to the iommu to the device tree#asahi · 2021-04-18 13:12
svennah, but i have usb gadget mode working in linux as well#asahi · 2021-04-18 08:30
sven*custom#asahi · 2021-04-18 08:30
sventpw_rules: the reason i used a common device class is such that we don't need customer kernel drivers or require libusb in proxyclient/*#asahi · 2021-04-17 12:13
svenlgtm :)#asahi · 2021-04-17 11:15
svenjust not sure what that it yet#asahi · 2021-04-17 11:15
svenyup#asahi · 2021-04-17 11:15
svenit's weird#asahi · 2021-04-17 11:15
svenyeah#asahi · 2021-04-17 11:14
svenmarcan: i mean.. it works if i just manually force the dwc3 part to reinit the device even without touching the phy#asahi · 2021-04-17 11:07
svenbut it just stays there then if i connect it again#asahi · 2021-04-17 11:07
sven0x502200024 starts at 0xa2000042, then goes to 0xa1000042 after i connect the cable and then finally to 0xa0000042 when i disconnect the cable again#asahi · 2021-04-17 11:04
svenit's not part of the known dwc3 range though i think#asahi · 2021-04-17 11:03
svenhrm. wait. maybe i just found a bit that's flipped#asahi · 2021-04-17 10:55
sveni didn't look at the atc-phy space though#asahi · 2021-04-17 10:55
sveni've looked at the whole dwc register space using RegMonitor and I didn't see any interesting bits flip when connecting a usb cable#asahi · 2021-04-17 09:54
svenwhich we could poll as well but ugh...#asahi · 2021-04-17 09:53
svenmy best bet still is that it somehow relies on interrupts from the TI chip :/#asahi · 2021-04-17 09:53
svenit does enable the PD negotiation stuff though#asahi · 2021-04-17 09:52
sveneven bringing up the TI chip with that special apple command doesn't help#asahi · 2021-04-17 09:52
svenyeah, it's something strange#asahi · 2021-04-17 09:22
svenbut nothing afterwards#asahi · 2021-04-17 09:22
sventhe first connection to a port works#asahi · 2021-04-17 09:22
svenand then use linux's xhci driver