#asahi · 2021-01-16 06:46
marcanthis is why I ported mini first, because it makes figuring out this stuff way easier#asahi · 2021-01-16 06:46
marcannot that I know of, but judging by the length of the corellium driver, it's trivial enough to do#asahi · 2021-01-16 06:43
marcannot worth it#asahi · 2021-01-16 06:43
marcananyway I'm not even going to look at their code any more#asahi · 2021-01-16 06:43
marcan4) wifi, which I guess is just again instantiating the broadcom fullmac stuff and maybe some fixes for that chip?#asahi · 2021-01-16 06:42
marcan3) apparently bluetooth... which I think was just adding missing flow control support to their UART and instantiating bluetooth on top of that? :D#asahi · 2021-01-16 06:42
marcan2) a trivial AIC driver#asahi · 2021-01-16 06:42
marcanthey wrote 1) a uart driver... even though linux already has a driver for this kind of UART#asahi · 2021-01-16 06:41
marcantheir repo is a copyvio just out of that#asahi · 2021-01-16 06:41
marcanthere is just no attribution because it's a single commit#asahi · 2021-01-16 06:41
marcantheir APFS driver is from linux-apfs#asahi · 2021-01-16 06:41
marcanShiz: they didn't even write the APFS driver#asahi · 2021-01-16 06:28
marcanhttps://mrcn.st/t/Screenshot_20210116_152820.png#asahi · 2021-01-16 06:27
marcanhttps://mrcn.st/t/Screenshot_20210116_152736.png#asahi · 2021-01-16 06:26
marcanso now he's turning it into a competition#asahi · 2021-01-16 06:26
marcanbrinly: the corellium CTO is pissed off at me because I said (in a twitter *reply*) that I couldn't use their code as is#asahi · 2021-01-15 18:43
marcanit probably makes more sense to just add caching to qemu or something like that#asahi · 2021-01-15 14:55
marcan*resolve reg properties#asahi · 2021-01-15 14:55
marcanfee free to write a wiki page :)#asahi · 2021-01-15 14:55
marcanit's just a trivial linear smash of everything#asahi · 2021-01-15 14:55
marcanShiz has a parser in Python too#asahi · 2021-01-15 14:54
marcanclayfreeman: it is not, but the format is quite trivial, it should be fairly obvious from the two structures and the source#asahi · 2021-01-15 14:54
marcanyrlf: yeah, I'm thinking a variant of that that returns a node path that can then be crawled to resolver reg properties#asahi · 2021-01-15 14:22
marcanwhat I'm wondering right now is how I want to implement that, because I can't recurse upwards through the tree given the current adt implementation, so I'll need a more self-contained lookup regs function that takes a path :)#asahi · 2021-01-15 14:21
marcanmem_size_actual has nothing whatsoever to do with it, that's just a coincidence#asahi · 2021-01-15 14:21
marcanyou have to recursively go up the range blocks in the devicetree and apply those as translation functions#asahi · 2021-01-15 14:21
marcanat all#asahi · 2021-01-15 14:21
marcanthat's not how it works#asahi · 2021-01-15 13:57
marcanneed to write or find some code to do that (I'm sure I'm not the first one to do this on top of libfdt or a lookalike)#asahi · 2021-01-15 13:56
marcanyeah, I'm not resolving devicetree memory regions properly#asahi · 2021-01-15 13:35
marcanthe weird halo/sharpening is my capture card, that's.... let's call it a 4hr stream to figure out and fix in these chinese capture cards :p#asahi · 2021-01-15 13:35
marcanat 720p, forget about it#asahi · 2021-01-15 13:35
marcanI do assume people are watching in 1080p#asahi · 2021-01-15 13:35
marcanso it's a bit smaller than normal for 1080p, but no unreadably small#asahi · 2021-01-15 13:35
marcanbasically you get 1.5/2 scaling vs what would be 1080p-looking for me#asahi · 2021-01-15 13:34
marcanyeah#asahi · 2021-01-15 13:34
marcanI run at 1.5x hidpi mode and then for the stream mirror-scale down to 1080p into an HDMI capture card#asahi · 2021-01-15 13:34
marcanbut at 4K because 5K needs another 8hr stream from me to figure out ( :-) )#asahi · 2021-01-15 13:34
marcanyorick: iMac retina 5K 27"#asahi · 2021-01-15 05:55
marcan(you just need to look for any of the million chips that include it, and find one with a good TRM)#asahi · 2021-01-15 05:54
marcanit's not hard at all to find documentation for that core#asahi · 2021-01-15 05:54
marcanhttps://rockchip.fr/RK312X%20TRM/chapter-26-usb-otg-2-0.pdf <- same controller I believe#asahi · 2021-01-15 05:43
marcandavidrysk[m]: drivers/usb/dwc2/gadget.c#asahi · 2021-01-15 05:41
marcanthough I will probaby port over some minimal debug print support#asahi · 2021-01-15 05:41
marcanthere should already be a kernel driver for it#asahi · 2021-01-15 05:40
marcanand once I have that developers will be able to hack on mini without any special cables#asahi · 2021-01-15 05:40
marcanNecrosporus: mini's USB driver#asahi · 2021-01-15 05:39
marcandavidrysk[m]: we do not yet, but AIUI it should be handled by a separate controller#asahi · 2021-01-15 05:12
marcanbut I want USB support first because loading linux over serial is going to be painful#asahi · 2021-01-15 05:12
marcanyou could just writemem(somewhere, linux) and call(somewhere, devtree) and that should boot linux, for "well it's missing drivers so it won't actually boot" values of booting linux