#asahi-dev · 2025-03-02 11:11
chaos_princessthere are pcie devices that read from random memory addresses because, idk, they feel like it. it is "fine" on x86#asahi-dev · 2025-03-02 11:10
chaos_princessthis will need to be controlled by a module flag or something when we will get thunderbolt though#asahi-dev · 2025-03-01 22:49
chaos_princesskexts are not there since quite a while ago, they are now in dyld shared cache#asahi-dev · 2025-03-01 18:43
chaos_princessfair :P#asahi-dev · 2025-03-01 18:41
chaos_princesshttps://github.com/AsahiLinux/linux/commit/90a611c90ed0a60ac6fef340a05f55ff45ae1d23 - this one can just become copy_from_slice, i will fix it later if nobody else gets to it first#asahi-dev · 2025-03-01 15:25
chaos_princessthanks, will take a look#asahi-dev · 2025-02-27 18:34
chaos_princesssven: idk, i just know they exist#asahi-dev · 2025-02-27 18:23
chaos_princessthere is at least 1 PR to m1n1#asahi-dev · 2025-02-25 06:47
chaos_princessnickchan: it is in fact my understanding that cpu voltages are locked and can't be controlled by the os. But, for obvious reasons, i can be wrong#asahi-dev · 2025-02-23 08:50
chaos_princessbut actual code for dma-range is in a "different" series#asahi-dev · 2025-02-23 08:50
chaos_princessso i will need to update the bindings to include the dma-range thing.#asahi-dev · 2025-02-23 08:49
chaos_princessi am submitting the dart nodes with the isp series#asahi-dev · 2025-02-23 08:49
chaos_princessthat is what makes it annoying#asahi-dev · 2025-02-23 08:46
chaos_princessi guess dma-range is a small enough patch so i can rebase it on top of torvalds/master and throw it in with my series. And then when locked darts land i will just drop that and reuse the original patch#asahi-dev · 2025-02-23 08:37
chaos_princesssure, i can figure something out then#asahi-dev · 2025-02-22 23:11
chaos_princessalyssa jannau: wdyt?#asahi-dev · 2025-02-22 23:10
chaos_princessim kinda not sure what to do w/ isp series v2. i rewrote the thing to use iommu-dma instead of drm mm, but that needs the apple,dma-range support, and it does not apply cleanly w/o the locked dart series that is currently in review#asahi-dev · 2025-02-22 16:53
chaos_princesssven: in my case it is for the sep tracer#asahi-dev · 2025-02-22 16:14
chaos_princesshttps://github.com/AsahiLinux/docs/wiki/HW:SEP, ctrl+f "gigalocker/xart"#asahi-dev · 2025-02-22 16:11
chaos_princessit is for the SEP gl file#asahi-dev · 2025-02-19 09:55
chaos_princesscow is in core, hashmap would need to be replaced#asahi-dev · 2025-02-19 09:43
chaos_princessyes, but all of it is in the `pread` function, and that can be swapped out without much trouble#asahi-dev · 2025-02-19 09:33
chaos_princessi already wrote "just enough apfs to get a list of containers" for another thing (asahi-bless)#asahi-dev · 2025-02-19 09:32
chaos_princessno, just check if the role bit is set and go#asahi-dev · 2025-02-19 09:22
chaos_princessand yes, it all will be vendored into m1n1, and probably also kernel#asahi-dev · 2025-02-19 09:22
chaos_princessprobably can even skip doing real CoW and do it in-place#asahi-dev · 2025-02-19 09:22
chaos_princessspecifically we need just enough apfs to find a single partition with a specific role bit, open a single file on it, and do r/w to it#asahi-dev · 2025-02-19 09:21
chaos_princesstechnically, we do not need a "real" apfs driver#asahi-dev · 2025-02-16 17:52
chaos_princessall relevant ones are on internet archive#asahi-dev · 2025-02-14 13:46
chaos_princesswow, thats a bad link, just ctrl+f "24 mhz" to get to relevant part#asahi-dev · 2025-02-14 13:45
chaos_princessyes#asahi-dev · 2025-02-14 13:45
chaos_princesshttps://developer.apple.com/documentation/macos-release-notes/macos-15-release-notes#New-Features#asahi-dev · 2025-02-14 13:43
chaos_princessit is just a scale factor#asahi-dev · 2025-02-14 13:41
chaos_princessyou can run diff frequency between el2 and el1 from what i understand#asahi-dev · 2025-02-14 07:10
chaos_princessHmm, i wonder if there is a system that enables one to upload new patch versions w/o creating a ton of spam and annoyance to the receivers, maybe something closely integrated with git :P#asahi-dev · 2025-02-13 19:25
chaos_princessThere are per-soc differences in agx for sure, for the rest of the ones we support - iirc no. #asahi-dev · 2025-02-13 19:20
chaos_princesshttps://github.com/yu-re-ka/m1n1, https://github.com/AsahiLinux/m1n1/pull/426, https://github.com/mildsunrise/m1n1/ - the ones i know of. for the last one, i'd avoid anything involving SPMI, might be a product of binary RE#asahi-dev · 2025-02-13 19:18
chaos_princesskinda?#asahi-dev · 2025-02-13 19:16
chaos_princessidk, i personally do not want people to run random dev branches unless they are determined enough to find them on their own#asahi-dev · 2025-02-13 19:16
chaos_princessthe _major_ issue is that the gpu got heavily redesigned, and the ISA has to be re-reversed again#asahi-dev · 2025-02-13 19:15
chaos_princessrandom dev branches that can boot to somewhere between single-thread m1n1 and kernel. #asahi-dev · 2025-02-09 21:20
chaos_princessty#asahi-dev · 2025-02-09 18:39
chaos_princesshttps://pastebin.com/syMdQJn6 - not seeing anything that suspicious tbh#asahi-dev · 2025-02-09 18:36
chaos_princesssec#asahi-dev · 2025-02-09 18:28
chaos_princessyep, it was that#asahi-dev · 2025-02-09 18:26
chaos_princesswonder what set it to n, iirc i had it as y before#asahi-dev · 2025-02-09 18:26
chaos_princesssimpledrm was n#asahi-dev · 2025-02-09 18:24
chaos_princessif yes - it is not there#asahi-dev · 2025-02-09 18:24
chaos_princessshould that be in dmesg?#asahi-dev · 2025-02-09 18:23
chaos_princessdo we have a way to bisect w/o a ton of pain?