#asahi · 2021-09-17 19:56
svenand i'll probably submit https://github.com/AsahiLinux/linux/commits/tipd/v1 sometime this weekend if someone want to take a look already (and find my typos :-P)#asahi · 2021-09-17 19:24
svenhow about this weekend? :p#asahi · 2021-09-17 19:22
svenj_ey: time to submit it then :>#asahi · 2021-09-17 18:19
sven(unless you prefer my kthread that just calls the interrupt handler every second or so 😋)#asahi · 2021-09-17 18:18
svenfor the tipd i need the gpio interrupt#asahi · 2021-09-17 18:18
svenright, but for pcie i can kinda ignore the pinctrl part and it'll still work#asahi · 2021-09-17 18:18
sveni keep forgetting pinctrl/gpio#asahi · 2021-09-17 18:18
svenugh. and pinctrl#asahi · 2021-09-17 18:17
svenfor that i "just" need i2c, the patches to tipd and the quirk to dwc3 (or a better solution if we find one)#asahi · 2021-09-17 18:17
svennope#asahi · 2021-09-17 18:14
svenrtkit.. dunno... i guess i'd first need to submit it together with something that uses it? :P#asahi · 2021-09-17 18:14
svenmbox needs to be reviewed by the mailbox maintainer first#asahi · 2021-09-17 18:11
svensame way the initial bringup did - through the soc tree i think#asahi · 2021-09-17 18:10
svennot impossible to handle all that, but that will require some coordination if it all goes into the same merge window#asahi · 2021-09-17 18:09
svenrtkit itself will depend on the mailbox code. and then nvme and dcp and smc all depend on rtkit#asahi · 2021-09-17 18:09
sventhis is still okay because only the dt change depends on all of them. rtkit will be interesting. #asahi · 2021-09-17 18:08
svenoh, it's already three thanks to pinctrl :D#asahi · 2021-09-17 18:07
sventhis will all be "fun" anyway. those driver changes will go through at least two separate tree. possibly three if I get the PHY driver ready for usb3#asahi · 2021-09-17 18:05
svenyeah#asahi · 2021-09-17 18:04
svenuh, it works fine that way but i'm not sure about how the usb phy will be connected to that yet. it also needs to be split into separate commits#asahi · 2021-09-17 17:59
sven(also hopefully also get the kernel builder bots to pick it up and see if i broke anything)#asahi · 2021-09-17 17:58
svenjust wanted to make sure the pci driver still compiles#asahi · 2021-09-17 17:57
svenyeah#asahi · 2021-09-17 17:55
svenfixed, thanks.#asahi · 2021-09-17 17:54
svenah, whoops.#asahi · 2021-09-17 17:47
svenhttps://github.com/AsahiLinux/linux/tree/sven/i2c-tipd-WIP with some updates to the i2c code. usb hotplugging and automatic switching to device mode should work there#asahi · 2021-09-17 13:26
svenah, true. that would work as well#asahi · 2021-09-17 12:51
svenwe could just pretend pci is always revision = 0 then#asahi · 2021-09-17 12:51
svenhm. i wonder if the first non-pci controller just has revision = 1#asahi · 2021-09-17 12:49
svenyeah#asahi · 2021-09-17 12:49
sventhat's actualy quite useful. the transactions to the PD chip are always less than 65 byte iirc#asahi · 2021-09-17 12:48
svenprobably not required for the initial merge. you can always add it together with interrupt support later :>#asahi · 2021-09-17 12:47
svenah, makes sense#asahi · 2021-09-17 12:46
svenyeah. so for some reason it does that#asahi · 2021-09-17 12:46
svenor, well, after writing to the fifo i think#asahi · 2021-09-17 12:46
svendid you clear it again after the transaction?#asahi · 2021-09-17 12:45
svenno#asahi · 2021-09-17 12:45
svenmacos sets that before it starts a transaction and clears it afterwards iirc#asahi · 2021-09-17 12:45
svenit also doesn't seem to make a difference#asahi · 2021-09-17 12:44
svenregister 0x44, BIT(0)#asahi · 2021-09-17 12:44
svensorry :)#asahi · 2021-09-17 12:44
sven0x44#asahi · 2021-09-17 12:43
sventhere's another bit it sets and clears around transactions#asahi · 2021-09-17 12:41
svenalright, good to know#asahi · 2021-09-17 12:40
svenyeah#asahi · 2021-09-17 12:38
sventhanks#asahi · 2021-09-17 12:38
svenhrm. okay. guess i'll just duplicate the if (hw_rev >= 6) logic then#asahi · 2021-09-17 12:07
svenwhich is REG_REV in the corellium driver#asahi · 2021-09-17 12:07
sven0x28#asahi · 2021-09-17 12:06
svenmacos checks some other register and only sets that bit if that other register is >= 6 or so