#asahi-dev · 2022-03-19 10:58
svenalso please boot with tp_printk trace_event=appletypecphy:*#asahi-dev · 2022-03-19 10:58
svenstill early WIP, make sure to also merge https://github.com/AsahiLinux/m1n1/pull/170 into m1n1 if you want to try it#asahi-dev · 2022-03-19 10:57
svenjust pushed https://github.com/AsahiLinux/linux/commits/atc-WIP#asahi-dev · 2022-03-19 10:25
sventhe right approach is to figure out what macOS does, then understand how the hardware works and then write a Linux driver#asahi-dev · 2022-03-18 21:26
svenI guess. I just stole whatever Linux itself used in some kexec path#asahi-dev · 2022-03-18 21:18
sven:-)#asahi-dev · 2022-03-18 20:37
svendon’t jinx it ;)#asahi-dev · 2022-03-18 16:15
svenand that ROM is already running and waiting for commands anyway. no need to even set a "start plz" bit#asahi-dev · 2022-03-18 16:14
sventhat was the thing I vaguely remembered this morning#asahi-dev · 2022-03-18 16:13
svenyup!#asahi-dev · 2022-03-18 16:01
svenyeah, i think there's even a WARN_ON in there if multiple namespaces are present#asahi-dev · 2022-03-18 15:53
svenah, yes, that's part of axboe's flush workaround. it probably also needs to support multiple namespaces#asahi-dev · 2022-03-18 15:52
svenuh, sure, i must've missed that#asahi-dev · 2022-03-17 18:12
svenI’ll grab dinner and at least collect the patches for that somewhere #asahi-dev · 2022-03-17 18:10
svenyeah, that too :)#asahi-dev · 2022-03-17 18:08
svenbut I guess I should clean up atcphy first anyway and get more people to test it :D#asahi-dev · 2022-03-17 18:06
svenacio needs some reversing as well ofc#asahi-dev · 2022-03-17 18:05
svenhuh, so thunderbolt looks less painful than I thought. The biggest challenge will be that typec subsystem and how to pass the “usb4 cable connected” along and that most thunderbolt crap can’t be mapped before this weird “clock/power domain” is up and that acio coprocessor is running #asahi-dev · 2022-03-17 14:48
svencould be the same for pcie#asahi-dev · 2022-03-17 14:48
sveniirc when atcphy is powered down all tunables are lost#asahi-dev · 2022-03-17 14:19
svenyeah, single RFC. sart+nvme can go together anyway because it's not used anywhere else. and for rtkit we can probably create a frozen branch after review if we actually end up with multiple drivers in different subsystems that need it#asahi-dev · 2022-03-17 14:17
svensure!#asahi-dev · 2022-03-17 14:17
svenj`ey: i actually already wrote a cover letter for the nvme patchset. just need to collect the individual patches now and submit it. maybe this weekend :)#asahi-dev · 2022-03-17 14:12
svenj`ey: sure :D#asahi-dev · 2022-03-17 14:12
sven+get#asahi-dev · 2022-03-17 14:12
svenbut i should at least nvme upstream soon-ish#asahi-dev · 2022-03-17 14:11
svenand implementing new hardware has been more fun that sending patches recently. at least for me.#asahi-dev · 2022-03-17 13:09
svenah :D#asahi-dev · 2022-03-17 13:08
sveni only have a m1 air and a m1 mini#asahi-dev · 2022-03-17 12:16
svenhrm, then i'd also have to bring up atcphy inside m1n1. oh well.#asahi-dev · 2022-03-17 12:11
svenjannau: have you ever tried to bring up dcpext? (either inside m1n1 or linux)#asahi-dev · 2022-03-16 18:19
sven*required for usb2 when you don’t want to bring up the usb3 phy to be exact #asahi-dev · 2022-03-16 18:19
svenbut the kext knows a third mode. It’s just never used#asahi-dev · 2022-03-16 18:18
svenhuh, so there are actually only two mux modes that macOS actually uses: usb3 and that usb4-tunnel mode that also seems to be required for usb2 #asahi-dev · 2022-03-16 10:10
svensounds like a plan!#asahi-dev · 2022-03-16 10:06
svenuh. i, er, need to bring up displayport!#asahi-dev · 2022-03-16 10:06
sven... just one more thing ;)#asahi-dev · 2022-03-16 08:29
sveni can do some proof reading this evening#asahi-dev · 2022-03-15 20:36
svenbedtime for me as well then!#asahi-dev · 2022-03-15 20:35
sventhe other phys all don't need the cio3pll_dll_start_capcode_workaround#asahi-dev · 2022-03-15 20:35
sventhey are in AppleT8103TypeCPhy::performEFuseProgramming. just guessing them from HV+serial output is going to be hard since they are just bits inside a u32. here's how the one phy I have looks like https://f.svpe.de/178c18c473b504077492d9bdc7acfa70224467c1dd13c3698945ee0484cde7f5_t8103-atcphy.txt#asahi-dev · 2022-03-15 20:34
svenalright, the atcphy stuff still needs some more cleanup but i'll push it tomorrow. if someone is bored and wants to help until then: I still need to track down the fuse locations for the other t8103 and the nodes+fuses for all t6000 phys.#asahi-dev · 2022-03-15 18:15
svenyou’ll probably also need a blog post or something to go along with it!#asahi-dev · 2022-03-15 18:14
svenThat’s still plenty of time :P#asahi-dev · 2022-03-15 18:14
svenGoodnight!#asahi-dev · 2022-03-15 18:13
svenyeah#asahi-dev · 2022-03-15 18:13
svenpart of that is tunables, and then there are two or three registers