#asahi · 2022-04-04 12:19
svenmost of them probably won't like to ship downstream kernel hacks to their users#asahi · 2022-04-04 12:18
sveni think it'll be a few more months before there will be an official way for any major distribution#asahi · 2022-04-03 18:06
svenkinda like how I can never reproduce the nvme bug without the command submission spin locks #asahi · 2022-04-03 18:06
sventhat patch fixes a timing related bug. those are very hard to reproduce #asahi · 2022-04-03 12:56
sventhat patch more or less fixes a race condition#asahi · 2022-04-03 11:08
sven*event#asahi · 2022-04-03 11:08
svenbecause we need to restart dwc3 on every plug even and that’s the easiest hack to do that#asahi · 2022-04-02 17:23
sven:)#asahi · 2022-04-02 17:21
sven:D#asahi · 2022-04-02 17:09
sven*usb-c ports#asahi · 2022-04-02 17:09
svennot on the usb3 ports unless you use my WIP branch#asahi · 2022-04-02 16:54
svenit’s more likely the kernel just isn’t compiled with exfat support#asahi · 2022-04-02 16:53
svenhm, no, they should be able to fall back to usb2#asahi · 2022-04-02 16:47
sventhen you need usbstorage ;)#asahi · 2022-04-02 16:46
svenonly if you want your machine to pretend to be a usb flash device#asahi · 2022-04-02 16:45
sventhey essentially allow your machine to pretend to be some usb device (storage, serial, etc.)#asahi · 2022-04-02 16:44
svenusb gadget drivers are something like the CDC ACM device m1n1 emulates #asahi · 2022-04-02 16:43
sventhe first step is to be in a constant state of confusion, which seems to be close enough to whatever the usb people were in when they made all this up#asahi · 2022-04-02 16:43
svensounds like you’re starting to understand usb!#asahi · 2022-04-01 17:41
svenso maybe that custom apple chip would be easier because it makes sense :D#asahi · 2022-04-01 17:40
svenApple’s designs are mostly sane. Broadcom not so much…#asahi · 2022-04-01 17:40
svenmaybe it’s similar to something, no idea. I just remember that someone concluded that there’s no driver that would work with small modifications #asahi · 2022-04-01 17:39
svenits Broadcom but not standard #asahi · 2022-04-01 17:39
svenapparently not#asahi · 2022-04-01 17:39
svenand they all work #asahi · 2022-04-01 17:39
svenwifi and the usb a ports and Ethernet are on there as well#asahi · 2022-04-01 17:39
svenbut yeah, pcie itself works#asahi · 2022-04-01 17:39
sveniirc someone at least briefly looked into and it didn’t seem to be a standard component that already has a driver#asahi · 2022-04-01 17:38
svenyes#asahi · 2022-03-29 19:09
svenstill takes time to track down bugs like this even when the problem is technically known in another codebase :)#asahi · 2022-03-29 19:08
svennice work!#asahi · 2022-03-29 13:12
sven“I don’t, but let me explain how I actually do”#asahi · 2022-03-29 12:57
svendo the problems always appear without that patch? do they always disappear with that patch?#asahi · 2022-03-29 12:45
svenotherwise we end up with N slightly different kernels and won’t know what caused or fixed any bugs#asahi · 2022-03-29 12:44
svenIt’s generally a bad idea to distribute any random patch to users when we’re still discussing it#asahi · 2022-03-29 12:42
sven*sigh*#asahi · 2022-03-29 12:40
svenit’s just big enough to fit any commands for the tps6598x#asahi · 2022-03-29 12:40
sveni checked back when I wrote the m1n1 driver#asahi · 2022-03-29 12:39
sventhe fifo is 64 bytes iirc#asahi · 2022-03-29 12:39
svenbut the register names are all documented somewhere in m1n1#asahi · 2022-03-29 12:39
svenor, well, Marcan has some but isn’t allowed to share#asahi · 2022-03-29 12:39
svenno#asahi · 2022-03-29 12:32
svenif it waits at least once we probably need that additional wait#asahi · 2022-03-29 12:32
sven:D#asahi · 2022-03-29 12:31
svenok, and you confirms that it waits at least once?#asahi · 2022-03-29 12:30
svenwould it exit immediately if that thing was already set before you call wait ready?#asahi · 2022-03-29 12:29
svenid first check if it’s actually some ready bit that we need to wait for or if it’s just an additional delay that fixes something by accident #asahi · 2022-03-29 12:28
svenbut i2c is slow anyway#asahi · 2022-03-29 12:28
svensomeone really should add interrupt support ;)#asahi · 2022-03-29 12:28
sven*absolutely