#asahi-dev · 2021-03-08 07:42
marcanneed to run some experiments...#asahi-dev · 2021-03-08 07:42
marcanI kind of want to make this dwc thing work for the hypervisor, because that will provide much better bandwidth for mmiotrace. though I'm also considering doing it a different way, where I keep a ring buffer in RAM and fetch it from darwin itself.#asahi-dev · 2021-03-08 07:40
marcanso like m1n1 can be logically divided into two stages, what it does for itself (which is along the lines of how iBoot would work), and what it does to boot a devicetree-land OS (which in this case involves all that kboot stuff, and also tunables)#asahi-dev · 2021-03-08 07:40
marcan(among other things, because I intend to boot darwin from m1n1, and I don't want to do anything unnecessary - same reason we don't bring up SMP by default until needed)#asahi-dev · 2021-03-08 07:39
marcanI'm thinking it's worth treating tunable application as a kboot thing, and keep m1n1 usage of dwc from doing that#asahi-dev · 2021-03-08 07:39
marcanbtw, we shouldn't need the tunables for this, right?#asahi-dev · 2021-03-08 07:38
marcanvery nice :D#asahi-dev · 2021-03-07 16:45
marcanoff to sleep, let me know how it goes :)#asahi-dev · 2021-03-07 16:43
marcanseems to work here#asahi-dev · 2021-03-07 16:42
marcanlet me do a quick test#asahi-dev · 2021-03-07 16:42
marcanbut that would have to be a pretty big snafu to have broken in that way :)#asahi-dev · 2021-03-07 16:41
marcanyeah I haven't even tested it either#asahi-dev · 2021-03-07 16:41
marcanyay!#asahi-dev · 2021-03-07 16:35
marcankettenis: pushed a pile of fixes, let me know if you run into anything else#asahi-dev · 2021-03-07 16:15
marcan:)#asahi-dev · 2021-03-07 16:10
marcanyeah, let me do a pass#asahi-dev · 2021-03-07 16:04
marcanah, found it#asahi-dev · 2021-03-07 16:03
marcan(I think linux does this...)#asahi-dev · 2021-03-07 16:03
marcanlet me see if I can figure out a canonical way of getting rid of the libc include paths while keeping the gcc ones#asahi-dev · 2021-03-07 16:02
marcanI know dlmalloc gratuitously pulls in some stuff which it shouldn't; I tried to remove most of it but probably missed some#asahi-dev · 2021-03-07 16:02
marcanlet me fix those#asahi-dev · 2021-03-07 16:02
marcanoh tinflate, heh#asahi-dev · 2021-03-07 16:01
marcanassert I just realized is not in gcc#asahi-dev · 2021-03-07 16:01
marcanstdlib... what on earth is pulling that#asahi-dev · 2021-03-07 16:01
marcanstring.h should be "string.h"#asahi-dev · 2021-03-07 15:59
marcanwhat headers are you missing?#asahi-dev · 2021-03-07 15:59
marcanI do default to the linux triple (aarch64-linux-gnu-) because I expect most people to have that and it should be equivalent to the native compiler once we're all on M1s as our daily drivers (;)), but m1n1 is intended to work with any sane AArch64 gcc or gcc-like compiler (clang support is in there, though less tested)#asahi-dev · 2021-03-07 15:57
marcanI don't even have any normal-use aarch64 machines lying around, other than a few dedicated rPis#asahi-dev · 2021-03-07 15:57
marcanI do all my builds on an x86#asahi-dev · 2021-03-07 15:57
marcanI absolutely am#asahi-dev · 2021-03-07 15:57
marcanso am I :)#asahi-dev · 2021-03-07 15:56
marcanthey should be *cross-compiler* headers#asahi-dev · 2021-03-07 15:56
marcanbut either way none of these should be *host* headers#asahi-dev · 2021-03-07 15:56
marcanit's possible the malloc and other 3rd party stuff in there are spuriously pulling in proper libc headers; if so that's a bug#asahi-dev · 2021-03-07 15:56
marcanah, stdbool is used in 3rdparty stuff, yeah#asahi-dev · 2021-03-07 15:55
marcanah, stddef and assert too#asahi-dev · 2021-03-07 15:54
marcanI don't think I use anything other than stuff that should come with any reasonable compiler (stdint, stdarg, can't think of anything else?)#asahi-dev · 2021-03-07 15:54
marcanyou mean host?#asahi-dev · 2021-03-07 15:53
marcankettenis: ... hos?#asahi-dev · 2021-03-06 16:23
marcan*tomorrow#asahi-dev · 2021-03-06 16:23
marcanI'm going to spend today writing the first progress report, let's see how that goes :)#asahi-dev · 2021-03-06 16:21
marcanthe only thing that would be somewhat annoying is if we can't power down peripherals without re-applying tunables either, i.e. power gating kills all those settings too, but we'll worry about that when we get there and figure out how much power it saves us#asahi-dev · 2021-03-06 16:21
marcanbut we can just ignore all that stuff for now#asahi-dev · 2021-03-06 16:20
marcanso we can then do the tunables there, however we decide to handle that#asahi-dev · 2021-03-06 16:20
marcan(and to apply chicken bits etc)#asahi-dev · 2021-03-06 16:20
marcanbecause of the RVBAR issue#asahi-dev · 2021-03-06 16:20
marcanfor full system sleep, AIUI we're going to need some kind of extended hook anyway#asahi-dev · 2021-03-06 16:20
marcansven: as long as the tunables don't need re-applying except for full system sleep, it makes sense to do the tunables entirely in m1n1#asahi-dev · 2021-03-04 13:54
marcan(this yak shave brought to you by maz asking me to add _ELx suffixes to the registers... which means testing at EL1 :p)#asahi-dev · 2021-03-04 13:53
marcancool!