#
✕ nick:marcan
#asahi · 2021-01-07 17:35
marcanit's an MSR or something, I'll implement it as a prctl()
#asahi · 2021-01-07 12:11
marcanI'll be on #asahi-dev for this stream.
#asahi · 2021-01-07 11:41
marcanalright, let me get set up
#asahi · 2021-01-07 11:03
marcan"I'll just reply to emails and get on the stream" *2 hours later still replying to emails*
#asahi · 2021-01-07 10:48
marcanit is
#asahi · 2021-01-07 10:40
marcan(I have learned the lesson many times that this pays off *big time* for this kind of project)
#asahi · 2021-01-07 10:40
marcangoing to work on scaffolding for speeding up the debug/test cycle
#asahi · 2021-01-07 10:40
marcanyup, was just readying to do a little livestream shortly actually :)
#asahi · 2021-01-07 05:07
marcanIntel: "hey, ARM had a terrible idea, why don't we copy it?" :-)
#asahi · 2021-01-07 05:06
marcanyes, SGX is the same nonsense
#asahi · 2021-01-07 05:06
marcanand modern SoCs are so complex it's *impossible* to get this right and plug all the holes
#asahi · 2021-01-07 05:06
marcansame with anything related to the memory subsystem
#asahi · 2021-01-07 05:06
marcane.g. if the EL1 kernel can do CPU voltage/frequency scaling, EL3 is just one clock glitch away from getting pwned
#asahi · 2021-01-07 05:05
marcanand even if you do, there's the entire massive field of speculation/etc microarchitectural vulnerabilities
#asahi · 2021-01-07 05:05
marcanbecause unless you put half of your OS kernel in TrustZone, EL1 has immense power over EL3
#asahi · 2021-01-06 20:06
marcanok, ok
#asahi · 2021-01-06 20:06
marcannope
#asahi · 2021-01-06 20:05
marcan:P
#asahi · 2021-01-06 20:05
marcanthat was my point
#asahi · 2021-01-06 20:05
marcanit isn't
#asahi · 2021-01-06 19:37
marcan(e.g. I could've written kexec.kext way back too but...)
#asahi · 2021-01-06 19:37
marcandon't want to waste my time with alternate boot methods if the real one is coming soon
#asahi · 2021-01-06 19:36
marcananyway, whatever is missing should drop like next week or something
#asahi · 2021-01-06 19:36
marcanthe manpages are still ahead of the actual code
#asahi · 2021-01-06 19:35
marcanbut I guess the customkc thing is in there in the previous beta?
#asahi · 2021-01-06 19:35
marcanI thought we still needed the configure-boot mode
#asahi · 2021-01-06 18:55
marcanyes
#asahi · 2021-01-06 18:55
marcanthe kmutil changes to set up the secureboot policy are not released yet
#asahi · 2021-01-06 18:55
marcanCDFH: citation needed?
#asahi · 2021-01-06 18:40
marcanand get some proper work done
#asahi · 2021-01-06 18:40
marcanspeaking of not taking long, I'm going to sleep; let's see if I can actually wake up fresh tomorrow
#asahi · 2021-01-06 18:39
marcanbut once I figure out the dwc usb controller you'll have an interface you can use
#asahi · 2021-01-06 18:39
marcanif you don't have a serial / VDM cable you won't be able to debug that anyway
#asahi · 2021-01-06 18:39
marcanbut that won't take long
#asahi · 2021-01-06 18:39
marcanthe low level bringup is a choke point, and there isn't much for anyone else to do there until I get mini up at least
#asahi · 2021-01-06 18:39
marcanright now, the most useful thing is to poke around macos and document
#asahi · 2021-01-06 18:38
marcanrjeli: check out https://github.com/AsahiLinux/docs/wiki/Tasks
#asahi · 2021-01-06 18:38
marcanso that driver might well be totally clean based on those docs
#asahi · 2021-01-06 18:38
marcanalso APFS is documented IIRC
#asahi · 2021-01-06 18:37
marcanthe good thing is a FS driver is stand-alone and unrelated to everything else, so I don't care if we use it for now, I just won't be upstreaming it myself
#asahi · 2021-01-06 18:37
marcanbut I intend to try to avoid APFS unless I really have to
#asahi · 2021-01-06 18:37
marcanthat's actually probably the most useful part of that kernel
#asahi · 2021-01-06 18:36
marcanhttps://github.com/a2en/linux-sandcastle/tree/sandcastle-5.4/fs/apfs
#asahi · 2021-01-06 18:36
marcanthey also have APFS
#asahi · 2021-01-06 18:36
marcanI'm going to consider it ~the same as APSL code unless they speak up about it and sign off on parts
#asahi · 2021-01-06 18:36
marcanbut we can at least use it as a reference
#asahi · 2021-01-06 18:36
marcanI do not trust them
#asahi · 2021-01-06 18:36
marcanI'm wary of the sandcastle stuff because it seems they aren't interacting with the community about this, and they haven't signed off anything, so I am extremely wary of taking any code from there even though it's ostensibly GPL
#asahi · 2021-01-06 18:35
marcanthe sandcastle kernel already has support for some subset of AIC: https://github.com/a2en/linux-sandcastle/blob/sandcastle-5.4/drivers/irqchip/irq-hx-aic.c
#asahi · 2021-01-06 18:32
marcandefinitely not pci

plus ›