#asahi-dev · 2026-09-18 11:40
chaos_princessnp, but you may consider pruning your kernel config, mine is less than 10 mins from cold :P#asahi-dev · 2026-09-18 11:26
chaos_princesshttps://github.com/AsahiLinux/linux/pull/608 https://github.com/AsahiLinux/linux/pull/607#asahi-dev · 2026-09-18 11:24
chaos_princessthere are two prs for dcp support, mind testing them on t8122?#asahi-dev · 2026-09-18 11:23
chaos_princessremind me, are you on t6031 or t6030#asahi-dev · 2026-09-18 11:14
chaos_princesssolves even that difficulty faster than that page takes to reload#asahi-dev · 2026-09-18 11:13
chaos_princesshttps://github.com/kasper93/anubis_webgpu/blob/master/anubis_webgpu.user.js#asahi-dev · 2026-09-18 10:56
chaos_princesscan try it out on the mac mini, but it will take a bit as it's linux was wiped due to #experiments. #asahi-dev · 2026-09-18 10:33
chaos_princesswhich are those?#asahi-dev · 2026-09-17 07:52
chaos_princessas for optimizations - feel free to read through https://oftc.catirclogs.org/asahi/2026-09-12#35689608; and https://oftc.catirclogs.org/asahi/2026-09-14#asahi-dev · 2026-09-17 07:51
chaos_princessi mean, sure, why not. There is a bit of a downside in that the j314s like you have is "mostly done" except for a few things#asahi-dev · 2026-09-17 07:35
chaos_princessnomad_of_tatooine[m]: i am not sure that an "official" internship is something that can be arranged here#asahi-dev · 2026-09-16 19:46
chaos_princessisn't "run on A14 linux" the ~same as "run on m1 linux"?#asahi-dev · 2026-09-16 14:33
chaos_princessproxyclient/tools/smccli.py#asahi-dev · 2026-09-16 07:56
chaos_princessmacos actually starts dcpep before system, but flipping the order around does not help#asahi-dev · 2026-09-16 07:55
chaos_princess^ same thing happens here#asahi-dev · 2026-09-16 06:42
chaos_princesschadmed: it fails earlier, during systemep_init#asahi-dev · 2026-09-16 06:39
chaos_princesss/status/startup/#asahi-dev · 2026-09-16 06:39
chaos_princesslike, during endpoint startus#asahi-dev · 2026-09-16 06:39
chaos_princessi am having the same issue on t6031, it fails to start super early#asahi-dev · 2026-09-16 06:38
chaos_princessnah#asahi-dev · 2026-09-15 20:38
chaos_princesstimes out during systemep_init(#asahi-dev · 2026-09-15 20:38
chaos_princessi am not even getting to the point where it matters#asahi-dev · 2026-09-15 15:57
chaos_princessyou should have examples in code on how to do it#asahi-dev · 2026-09-15 15:56
chaos_princesssure, go ahead then#asahi-dev · 2026-09-15 15:54
chaos_princessdoes it do fd passing, or do you only need data?#asahi-dev · 2026-09-15 10:15
chaos_princesskinda shameful to admit, but i realized that dt-bindings is just json-schema-but-yaml like less than a month ago, and all my previous attempts at writing them was me figuring out the format from examples :P#asahi-dev · 2026-09-14 08:29
chaos_princessyes. Feel free to not bother for stuff that is just in the asahilinux git, and only do it right before sending to lkml#asahi-dev · 2026-09-11 10:20
chaos_princesssame#asahi-dev · 2026-09-11 10:03
chaos_princesseverything else is in carveouts, including vram, which is region 14#asahi-dev · 2026-09-11 10:03
chaos_princessyes#asahi-dev · 2026-09-11 09:54
chaos_princess`struct.unpack('QQQI4xQQQI4xQQQI4xQQQI4xQQQI4xQQQI4xQQQI4x', u.adt['/arm-io/dcp/iop-dcp-nub'].segment_ranges)`#asahi-dev · 2026-09-11 09:29
chaos_princessthat one is not present in carveouts at all#asahi-dev · 2026-09-11 09:28
chaos_princessif we go by segment-ranges, they do match, except for '0x100045c4000', '0x901000', '0x100045c4000', '0x1f000'#asahi-dev · 2026-09-11 09:10
chaos_princessyou are on a t6030, no?#asahi-dev · 2026-09-11 09:09
chaos_princessyes#asahi-dev · 2026-09-11 09:03
chaos_princessyes, they don't seem to#asahi-dev · 2026-09-11 08:57
chaos_princesswell, mbox must be correct since rtkit works at all#asahi-dev · 2026-09-11 08:57
chaos_princessmbox is correct, no?#asahi-dev · 2026-09-11 08:52
chaos_princessjust check if the reserved regions are mapped in those darts?#asahi-dev · 2026-09-11 08:51
chaos_princesshow do i do that. As in, how do i verify that they are correct.#asahi-dev · 2026-09-11 08:49
chaos_princessdo y'all do any other m1n1 changes than adding t6030 to the t602x branches to get dcp working? Trying it on t6031, it runs through the rtkit boot, even logs various things to syslog, but leaves the kernel on read once it tries to start endpoint 0x20, and fails to load#asahi-dev · 2026-09-09 10:00
chaos_princesssomewhat weird, but not w/o precedent, usb nodes on all-things-m3 are a direct copy of t602x#asahi-dev · 2026-09-09 09:49
chaos_princessno, at least on m1pro dcp+llvmpipe was worse than fb+llvmpipe#asahi-dev · 2026-09-09 09:48
chaos_princesssince we didn't have that back in the day#asahi-dev · 2026-09-09 09:47
chaos_princessis it the cursor plane#asahi-dev · 2026-09-08 20:56
chaos_princessall t6031 irqs should be the other way around, first one is zero, not second#asahi-dev · 2026-09-08 19:46
chaos_princessthey are in a triangle pattern#asahi-dev · 2026-09-08 11:26
chaos_princessi've added an attempt to decode it there#asahi-dev · 2026-09-08 11:26
chaos_princesssofus: https://github.com/asahilinux/m1n1/blob/main/proxyclient/tools/pmp_assist.py#asahi-dev · 2026-09-08 10:46
chaos_princesslooks like it