#
✕ nick:chadmed
#asahi-re · 2026-08-14 10:30
chadmedthe DUT needs to be an apple silicon mac, the device doing the tracing can be any linux machine
#asahi-re · 2026-07-22 08:36
chadmedyes extremely, it is only useful for hypervising xnu
#asahi-re · 2026-07-08 13:00
chadmedthere are a LOT more strings/logging in here by the looks of things
#asahi-re · 2026-07-08 12:49
chadmeddcp fw is now 11 mib :)
#asahi-re · 2026-07-05 11:54
chadmedplease refrain from editing messages on the matrix side. this is primarily an IRC room, and editing just spams extra messages
#asahi-re · 2026-07-02 12:55
chadmedtracing anything that does heaps of dumps to stdout will choke the machine. you have to have a lot of patience
#asahi-re · 2026-07-02 12:55
chadmedthe tracer is really really slow. see if async makes it any better
#asahi-re · 2026-07-02 12:22
chadmedyeah i remember needing a little bit more API surface to abstract away some things the same way macos does e.g. force required to register a click and the "type" of haptic provided but again this was aaaages ago
#asahi-re · 2026-07-02 12:16
chadmedidm the default haptics the stm initialises the pad with but macos has some nice settings that it seems libinput intended to support at some point but never got around to
#asahi-re · 2026-07-02 12:16
chadmedlmfao awesome
#asahi-re · 2026-07-02 12:14
chadmedsofus: did you ever find a way to wire this up into libinput's existing forcepad stuff? when i looked at it (c. 2022?) there was basically nothing libinput could do with this sort of touchpad other than go "hm yes this is indeed a forcepad"
#asahi-re · 2026-07-02 11:52
chadmeda m1n1 experiment might be easier than trying to wrangle macos apis into sending custom HID packets across dockchannel, or just hack up a linux driver based on the existing hid dockchannel driver
#asahi-re · 2026-07-01 08:59
chadmedM3 progress, Apple bugs, and more! https://asahilinux.org/2026/06/progress-report-7-1/
#asahi-re · 2026-06-30 08:10
chadmedim sick to death of getting slop-augmented requirements documents that make absolutely no sense because the person we got to gather the requirements did not understand the task at hand and had claude do it for them, and then having to schedule more 4 hour meetings with the customer to actually get the requirements properly
#asahi-re · 2026-06-30 08:09
chadmedwe have people who use them at $work and all they seem to do is create more annoying rework and fixing for those of us who refuse to use them :P
#asahi-re · 2026-06-16 11:25
chadmeddue to age verification laws or some such
#asahi-re · 2026-06-16 11:25
chadmedimgur doesnt work in the UK
#asahi-re · 2026-06-16 11:24
chadmedthe "Contribute" button on github
#asahi-re · 2026-06-16 11:23
chadmedyou can then select which branch you want your changes to be merged into in the original repo (called the upstream)
#asahi-re · 2026-06-16 11:23
chadmedyou fork the repo, clone your fork, make a branch, do your stuff in that branch, push that branch back up to your forked remote, then if you go to your repo github will prompt you to make a pull request
#asahi-re · 2026-06-16 11:22
chadmedhave you used git before or is this your first time
#asahi-re · 2026-06-16 11:21
chadmedfork it and make merge requests?
#asahi-re · 2026-06-16 11:08
chadmedlinux, all our tooling is developed primarily with a linux dev machine in mind
#asahi-re · 2026-06-16 10:47
chadmedthe usual choice for binary re is ghidra, but we do very little binary re anyway
#asahi-re · 2026-05-13 09:58
chadmedthe only useful thing i got out of putting the dcp fw in ghidra was confirmation on the meaning of some enums, because there are a couple of debug print functions that just switch on the enum value and then print what it means :P
#asahi-re · 2026-05-13 09:56
chadmedprobably wouldnt have figured that one out by trying to disassemble 9 mib of nonsense obj-c/c++ code with vtables that go nowhere and 3 (?) different abstraction frameworks
#asahi-re · 2026-05-13 09:55
chadmedanother example, macos never ever uses more than two display controller planes because old versions of dcp (m1 and previous) only support two. in linux we can simply detect which dcp{,ext} version we have and expose the real number of surfaces to kms which lets compositors do more cool things that macos cant
#asahi-re · 2026-04-30 13:26
chadmedwe have enough sacrificial machines between us to build a fort out of macs i think
#asahi-re · 2026-04-30 13:23
chadmedchaos_princess: thats my line
#asahi-re · 2026-04-30 13:22
chadmedvideo decoder or encoder still doesnt work and a lot of the hard work is done for some codecs if you wanted to take a look at those instead
#asahi-re · 2026-04-30 13:19
chadmedif you fuck up xart/keybag everything stops working. again, these machines are effectively unbrickable save for cosmic rays hitting the ROM with the first stage bootloader so you can always recover, but do _not_ keep any important data on the device under test
#asahi-re · 2026-04-30 13:18
chadmedyeah that
#asahi-re · 2026-04-30 13:18
chadmedwe already load it and boot it
#asahi-re · 2026-04-30 13:17
chadmedSEPOS is loaded by the time you need to talk to SEP
#asahi-re · 2026-04-30 13:16
chadmedyou should not need sep's fw, trace macos with m1n1 in the first instance
#asahi-re · 2026-04-30 13:14
chadmedfrom what i remember i dont think xart is in the hot path for mesa but that was *checks notes* almost 5 years ago now
#asahi-re · 2026-04-30 13:12
chadmedimo its not _likely_ if youre very very careful but pobody's nerfect and we havent really played with sep enough to know what the safety limit is, beyond "there is a risk"
#asahi-re · 2026-04-30 13:11
chadmedif you corrupt any of SEP's data your machine will be unbootable
#asahi-re · 2026-04-26 09:22
chadmedNew progress report: https://asahilinux.org/2026/04/progress-report-7-0/
#asahi-re · 2026-03-28 20:43
chadmedbut ill just borrow an adapter from work
#asahi-re · 2026-03-28 20:42
chadmedthis is why i asked for traces from people with suitable machines a few days ago
#asahi-re · 2026-03-28 20:42
chadmedoliver: i havent been able to trace this out yet. the one machine i have still on macos 13.5 has no hdmi port, and i have no usb-c to hdmi adapters
#asahi-re · 2026-03-28 07:01
chadmedi am worried that quartz is just doing it in software and feeding dcp a ycc bt2020 framebuffer
#asahi-re · 2026-03-28 07:01
chadmedyep, confirmed its using the same display colour mode
#asahi-re · 2026-03-28 06:59
chadmedeverything looks washed out and sickly
#asahi-re · 2026-03-28 06:59
chadmedin normal cases dcp is supposed to do the colourspace conversion for each iosurface based on the display's native colourspace and the colourspace of the surface, but thats not working properly when the display is ycc bt2020
#asahi-re · 2026-03-28 06:58
chadmedthe interface is quite unstable so im not entirely sure. all im really missing here is understanding how macos deals with displays that are ycc bt2020 as opposed to rgb bt2020
#asahi-re · 2026-03-05 02:13
chadmedthe scheduler's gotten pretty good :P
#asahi-re · 2026-03-05 02:13
chadmedhopefully we dont have to do much beyond benching and normalising the performance numbers
#asahi-re · 2026-02-18 09:40
chadmedProgress report for 6.19 is up: https://asahilinux.org/2026/02/progress-report-6-19/

plus ›