#asahi-dev · 2023-07-15 00:44
marcandesktops and machines that are just base model test devices, no#asahi-dev · 2023-07-15 00:44
marcanTramtrist: I get applecare for machines I actually intend to use daily/travel with#asahi-dev · 2023-07-14 14:44
marcanand I figure you'd have more uses for it than me, I have too many machines already :)#asahi-dev · 2023-07-14 14:44
marcanI don't need that thing for anything except working out the speaker profile, since it's a unique chassis (so far)#asahi-dev · 2023-07-14 14:44
marcanchadmed: I was wondering... any chance you're interested in the MBA 15"?#asahi-dev · 2023-07-14 03:59
marcan6.5-rc1 merge is trivial except for Rust stuff#asahi-dev · 2023-07-14 03:48
marcangot some Rust DRM issues, I'll let Lina deal with that. other than that 6.4 build succeeded, pushing that for now.#asahi-dev · 2023-07-14 03:31
marcanlooks like a good 50 or so patches dropped off the GPU branch in 6.4 :)#asahi-dev · 2023-07-14 03:11
marcanI'm going to rebase on 6.4 now and probably go all the way to 6.5-rc1#asahi-dev · 2023-07-14 03:11
marcanI pushed -13 with more of lina's fixes yesterday to -dev and pushing to prod now, I think it's fine#asahi-dev · 2023-07-13 14:53
marcannow we actually check the csrutil status (which we get in the DT) which should be safe#asahi-dev · 2023-07-13 14:53
marcansven: older iboot filtered -v, newer does not, that's why I changed it#asahi-dev · 2023-07-11 07:50
marcanjannau: thanks, pushing now#asahi-dev · 2023-07-11 06:30
marcanjannau: if you still have ALARM on your Ultra, can you confirm it fixes it? (I just wiped mine and put Fedora on it, haven't tried a Fedora kernel build yet)#asahi-dev · 2023-07-11 06:30
marcanpushed kernel -12 with a fixed missing ethernet alias which broke the MAC address on the Mac Studio#asahi-dev · 2023-07-11 03:34
marcanit bypasses the coprocessor and talks directly to the hardware#asahi-dev · 2023-07-09 10:44
marcan(this is for eventual HDMI support in the Asahi CI cluster)#asahi-dev · 2023-07-09 10:44
marcanI ordered a MuxLab 4x4 which sounds like it'll do what I want, worst case I have other use cases for it#asahi-dev · 2023-07-09 10:40
marcanwhich might happen every frame with aggressive power management#asahi-dev · 2023-07-09 10:40
marcanydalton: forgetting the firmware when powered down wouldn't surprise me, the DCP M3 coprocessors get reloaded every time too.#asahi-dev · 2023-07-09 10:39
marcansven: nice :)#asahi-dev · 2023-07-03 22:02
marcankettenis: ohh that, yeah, I should fix that. IIRC the split is just off by 12 bits, so APPLE_RTKIT_OSLOG_SIZE should start at 36 and IOVA end at 35, and SIZE is in bytes, not 4k pages.#asahi-dev · 2023-07-03 09:28
marcanu-boot needs to just answer the buffer request with a chunk stolen from the end of the MTP SRAM to make it work (that's the easiest way I can think of)#asahi-dev · 2023-07-03 09:27
marcankettenis: we just ignore it other than the buffer request, I think I mentioned this earlier?#asahi-dev · 2023-06-29 04:14
marcanpushed a kernel to asahi-dev with lina's fix from yesterday for the Firefox hangs/lockups, please smoke test a bit :)#asahi-dev · 2023-06-26 13:49
marcanah, that explains why it went away#asahi-dev · 2023-06-26 13:48
marcanah but ^^ this applies to M2 and newer, T600x should still have the old trap bit#asahi-dev · 2023-06-26 13:48
marcanif I had to guess it'd move to HACR?#asahi-dev · 2023-06-26 13:47
marcanamarioguy: also yeah, I think the SMC trap bit went away, but I don't know if it moved somewhere else or it's now always enabled or what#asahi-dev · 2023-06-26 13:46
marcanevery tier tries to handle it in sequence, if nothing does you get dumped into an exception shell (though for sysregs I think the default handler is passthrough for everything)#asahi-dev · 2023-06-26 13:45
marcanthere are actually 3 tiers for sysregs: fast (does not take the BHL), normal (takes the BHL), and python#asahi-dev · 2023-06-26 13:44
marcanamarioguy: whether it needs to be fast or not :p#asahi-dev · 2023-06-18 02:57
marcancool#asahi-dev · 2023-06-18 02:53
marcanchadmed: is that because it sees one end only? how does it break?#asahi-dev · 2023-06-17 13:45
marcanah right, you'd want that#asahi-dev · 2023-06-17 13:45
marcan*backlight#asahi-dev · 2023-06-17 13:45
marcanwonder if there's a way to get notifications out of a brightness device (uevent?)#asahi-dev · 2023-06-17 13:45
marcancould we try to tie touch bar brightness to screen brightness instead?#asahi-dev · 2023-06-17 13:33
marcanlibinput quirk?#asahi-dev · 2023-06-17 05:59
marcanmodesetting should probably just be fixed to ignore those devices#asahi-dev · 2023-06-17 05:57
marcanworks for me :)#asahi-dev · 2023-06-17 05:14
marcanwhich are already kinda broken :p#asahi-dev · 2023-06-17 05:14
marcanand $newdistro has the wayland snapshot sddm I believe, so it would only break xorg sessions#asahi-dev · 2023-06-17 05:12
marcanyes, but on $newdistro we *can* get stuff fixed easily#asahi-dev · 2023-06-17 05:12
marcanand there's no way to work around it in the kernel?#asahi-dev · 2023-06-17 05:00
marcanwouldn't be a first for DRM :p#asahi-dev · 2023-06-17 05:00
marcanalternatively we could do another stupid kernel patch where if(xorg) no touchbar#asahi-dev · 2023-06-17 04:58
marcanwe already screwed up with libinput once#asahi-dev · 2023-06-17 04:58
marcanso I'm not really comfortable with either of those solutions :/#asahi-dev · 2023-06-17 04:56
marcanand sddm/xorg are system critical, if they break you can't (easily) log in to fix it