#asahi-dev · 2023-09-28 05:11
marcaneiln: so as far as I'm concerned we're in a good spot to merge webcam and push out some asahi-dev builds and see how it goes on all the other platforms/machines#asahi-dev · 2023-09-27 05:31
marcanthank you for finding that!#asahi-dev · 2023-09-26 14:27
marcanmostly was thinking if I'm going to be traveling early september, I can work on speakersafetyd on the go, but not this :p#asahi-dev · 2023-09-26 14:27
marcanI can make time this weekend any time I think#asahi-dev · 2023-09-26 14:01
marcanchadmed: btw, do you have any time in the next week and a half ish? it might be useful to be on a call or something to start making DSP profiles for the other laptops, to make sure I'm not doing anything dumb :)#asahi-dev · 2023-09-23 12:44
marcanjanneg: done#asahi-dev · 2023-09-23 03:46
marcanare there actually per-CPU optimizations in Xcode LLVM? if Apple are doing that there too I can't imagine there's much benefit#asahi-dev · 2023-09-22 02:09
marcansince my 6.5.0 merge works, but the Fedora 6.5.4 rebase is all kinds of broken with PCIe IRQs#asahi-dev · 2023-09-22 02:09
marcanand now it's starting to look like something in 6.5.4 stable (or some other prior version) regressed PCIe for us#asahi-dev · 2023-09-21 11:53
marcanmaz: well the kernel without that patch landed in Fedora for a while last night and... yeah, everything is on fire.#asahi-dev · 2023-09-21 09:06
marcanIIRC the "first one to ever unlock an iphone" thing was also mostly based on work from others#asahi-dev · 2023-09-21 09:05
marcanchadmed: he's always coasted through everything, from the beginning#asahi-dev · 2023-09-21 08:57
marcanamusing at least :)#asahi-dev · 2023-09-21 08:54
marcaneiln: not sure what you want to reply :)#asahi-dev · 2023-09-21 08:52
marcan"improving my docs"#asahi-dev · 2023-09-21 08:52
marcanLOL geohot#asahi-dev · 2023-09-21 08:43
marcan:(#asahi-dev · 2023-09-21 08:00
marcanso there's no supporting the entire range without conditionals, which I didn't want to bother with#asahi-dev · 2023-09-21 08:00
marcanbasically upstream have dragged their feet so much on bindgen that they've actually blown through the deprecation period for certain commandline options...#asahi-dev · 2023-09-21 07:59
marcanI've been using 0.62 recently but now let's just jump forward#asahi-dev · 2023-09-21 07:59
marcanaxboe: yeah I just unconditionally upped the bindgen compat, upstream is doing it for 6.6 anyway and the old bindgen is all kinds of unusable#asahi-dev · 2023-09-21 07:58
marcanany news on the pipewire stuff btw?#asahi-dev · 2023-09-21 07:58
marcan:)#asahi-dev · 2023-09-21 07:50
marcanchadmed: I did already have that#asahi-dev · 2023-09-21 06:35
marcanthanks#asahi-dev · 2023-09-21 06:34
marcangot it, will be in the next merge#asahi-dev · 2023-09-20 21:27
marcancy8aer: next kernel will require a new bindgen anyway, none of us have been using that old bindgen for a long time now.#asahi-dev · 2023-09-20 14:11
marcankernel rebased on 6.5, kernel&m1n1 updates pushed to asahi-dev.#asahi-dev · 2023-09-19 21:37
marcanit's SPMI, not I2C#asahi-dev · 2023-09-14 02:08
marcanI'm pretty sure it doesn't matter#asahi-dev · 2023-09-13 17:21
marcan(also fine to implement in the C side and call from python, since nvram writes *could* be a useful thing to have in m1n1 proper)#asahi-dev · 2023-09-13 17:21
marcan(also if anyone's bored: one thing we need to be able to do for CI is switch boot volumes to select different firmware, and ideally we'd do that from m1n1, so a m1n1/python implementation of nvram writes would be useful)#asahi-dev · 2023-09-13 17:20
marcanawesome :D#asahi-dev · 2023-09-13 17:19
marcanyup, will do! :)#asahi-dev · 2023-09-13 17:17
marcanstill need to buy my CI rack machines but that's pretty much the next project after the Fedora GA release :)#asahi-dev · 2023-09-13 17:17
marcanyes, CI would definitely all run in the HV, among other things because we want to use that to validate GPU initialization and stuff like that#asahi-dev · 2023-09-13 08:32
marcaneiln: t6000 worked for me without any of those pointers. I suspect the dsid thing is for t8xxx only, since those addresses make no sense for t6xxx. But I think that whole thing is a memory controller optimization and we can just skip it (as long as there is some valid way to do that, e.g. all 0).#asahi-dev · 2023-09-11 02:20
marcanjannau: done#asahi-dev · 2023-09-11 02:01
marcansince there are many more variables (e.g. GPU usage)#asahi-dev · 2023-09-11 02:00
marcanchadmed: per-soc should be fine I think. only the fanless models should throttle and I don't think that behavior is going to be very usefully encoded in the EAS data anyway#asahi-dev · 2023-09-10 18:56
marcananyway, I really need to sleep :p#asahi-dev · 2023-09-10 18:55
marcan(that can do this for you)#asahi-dev · 2023-09-10 18:55
marcanI would hope edk2 already has some semblance of PCIe support?#asahi-dev · 2023-09-10 18:54
marcanamarioguy: yes (after doing a whole bunch of other pcie init)#asahi-dev · 2023-09-10 18:54
marcanj`ey: same codebase#asahi-dev · 2023-09-10 18:52
marcanI mean if you're not running on u-boot of course#asahi-dev · 2023-09-10 18:52
marcan(via ECAM writes)#asahi-dev · 2023-09-10 18:51
marcanthere's no firmware, you are the firmware so you get to *assign* a BAR.#asahi-dev · 2023-09-10 17:34
marcanprobably easier to hack on the code separately at first#asahi-dev · 2023-09-10 17:34
marcanbut that refactoring can come later too, up to you