#asahi-dev · 2026-11-09 14:00
flokliBut if there's some consensus on not building this by default, it should probably not be in the distro kernel#asahi-dev · 2026-11-09 13:59
flokliI have some rk3588, I could look which kernel driver gets used#asahi-dev · 2026-11-09 13:58
floklijannau: I think we should just explicitly disable tyr in the common config for now then#asahi-dev · 2026-11-09 12:13
floklijannau: myria opened https://github.com/AsahiLinux/linux/issues/673 about the build regression with tyr I wrote about earlier.#asahi-dev · 2026-11-08 22:41
floklijannau: https://github.com/AsahiLinux/linux/commit/fd637905b0d62de21b94ce58524849a45ab742dd breaks other kernel drivers (Tyr for mali). The nixos kernel build configures most things as modules, so breaks for asahi-7.3.3-1. We could configure it out, just wanted to ask first if it's a mistake #asahi-dev · 2026-11-02 19:35
flokli(on our side)#asahi-dev · 2026-11-02 19:33
flokliI might just wait for that to get picked, none of this is currently exposed yet.#asahi-dev · 2026-11-02 19:32
flokliAh thanks, so not in a tagged kernel yet. I was just very confused#asahi-dev · 2026-11-02 19:27
floklijannau: https://github.com/AsahiLinux/linux/commit/2e6ef8294919b4bd7b23756d836b447b0b4ffb5f doesn't seem to read from the 0x400 offset that https://github.com/AsahiLinux/avd-fw/commit/79854f31bd4809d78fef49167b8eccfe676b6010 claims, or am I holding this wrong?#asahi-dev · 2026-09-29 18:45
flokliyuka: I should get remote hands in 2-3 days#asahi-dev · 2026-08-18 18:43
floklisven: https://github.com/AsahiLinux/linux/blob/53a142828b08cbcacde04538d49e4166901ff254/drivers/thunderbolt/apple.c#L753 s/happens/happen/#asahi-dev · 2026-08-12 17:05
floklinot in the last 6 months#asahi-dev · 2026-07-31 12:30
flokliYup#asahi-dev · 2026-07-31 12:11
flokliAh, as a mode. But then userspace would still know how to interpret these resolutions (one with, one without notch)#asahi-dev · 2026-07-31 12:09
floklihttps://gitlab.gnome.org/World/Phosh/gmobile/-/work_items/17#note_2823911#asahi-dev · 2026-07-31 12:09
flokliBut let's see what they say about that idea.#asahi-dev · 2026-07-31 12:08
flokliI opened another issue about format stabilization, maybe we can carry this info via a device tree property. Would mean we don't need to populate that repository with some static set of cutout data for all models, but can do this in m1n1 possibly #asahi-dev · 2026-07-31 12:07
floklijannau: I'm not sure what you're proposing. But the current cutouts might not get accidentally applied in our show_notch=0 case, as the "resolution" doesn't match.#asahi-dev · 2026-07-31 11:41
flokliright now it just always assumes the entire screen is visible.#asahi-dev · 2026-07-31 11:40
flokliMaybe this tooling could not just look at the compatible string but actually read some properties that m1n1 could add#asahi-dev · 2026-07-31 11:40
flokliJamesCalligeros[m]: jannau: it might make sense to talk with Phosh/gmobile folks on a way to pass this down from the ADT, where we have this info#asahi-dev · 2026-07-31 11:19
floklielisaado: yeah but this is bugs to be squashed.#asahi-dev · 2026-07-31 11:19
flokliI'm happy with the format phosh came up with. and definitely better than just not moving at all. If people ever intend to carry this along in device trees rather than another catalogue repository they can still do that, and then there's already a format to use, and the bikeshed alredy happened in userspace.#asahi-dev · 2026-07-31 11:10
flokliyeah#asahi-dev · 2026-07-31 11:08
floklithat would be sweet! then I assume we'd at some point default apple_dcp.show_notch=1, so compositors have the whole screen to juggle with?#asahi-dev · 2026-07-31 11:06
floklii recall there were also some proposals to add this to the kernel, but it got rejected as too complicated.#asahi-dev · 2026-07-31 11:05
flokliapparently phosh now has a semi-standardized format to describe cutouts: https://phosh.mobi/posts/notch-support/ . opened an issue to maybe move these parts to a separate repo, so this data can be consumed by other compositors (?): https://gitlab.gnome.org/World/Phosh/gmobile/-/work_items/16#asahi-dev · 2026-07-31 10:27
flokliI can reply, yes. I'd probably not want to rely on nul bytes accidentially working as a separator.#asahi-dev · 2026-07-31 10:27
flokliWe have some mechanism to merge dirs at runtime, and maybe can use that mechanism, but other distros might not have the same option.#asahi-dev · 2026-07-31 10:25
flokliThis is a nixos thingie for us, but it might also become an issue for some image-based distros.#asahi-dev · 2026-07-31 10:20
floklimmpf, we already use the first knob to point to an immutable store path.#asahi-dev · 2026-07-31 10:17
flokliI also spoke with yuka a bit, and after reading on https://docs.kernel.org/driver-api/firmware/fw_search_path.html as well we thought maybe just making that take a list would be another option.#asahi-dev · 2026-07-31 10:16
flokliSon_Goku: following up on https://oftc.catirclogs.org/asahi-dev/2026-07-06#1783361776-1783363807; , do you have that firmware patch somewhere?#asahi-dev · 2026-07-28 21:47
floklit6040 has a working `msr((3, 5, 15, 1, 3), mrs((3, 5, 15, 1, 3)))` on 26.6!#asahi-dev · 2026-07-26 16:17
flokliyuka: day after tomorrow, Tuesday #asahi-dev · 2026-07-16 16:35
flokli(and modprobe again)#asahi-dev · 2026-07-16 16:34
floklithat's with iwd and NetworkMamager, had to rmmod brcmfmac_wcc brcmfmac#asahi-dev · 2026-07-16 16:34
floklihttps://paste.debian.net/hidden/070bc0a8#asahi-dev · 2026-07-16 16:32
flokliconnecting to the 8021x one works fine, OWE oopses#asahi-dev · 2026-07-16 16:32
flokliyes#asahi-dev · 2026-07-16 16:31
floklii'm at bornhack and the wifi driver runs into probably the same crashes as during 39c3... it works fine with macos, but linux is unhappy about it#asahi-dev · 2026-07-07 20:29
flokliWe might need to think about some signature mechanism, and just appending to initrd won't fly in any case. The hard part is that the installer runs in the MacOS context, not in the OS we're preparing to install for, or worse, regenerate the firmware files for.#asahi-dev · 2026-07-07 20:27
flokliOn the topic of 'we dump a .cpio file onto the ESP and ask initrd to load it without verifying anything', this might not fly with upstream, at least if we think about going further with secure / measured boot topics.#asahi-dev · 2026-07-07 15:34
floklit6040 with 27.0_26A5378j still has the locked register. DFU restore had to be done on MacOS (see https://github.com/libimobiledevice/idevicerestore/issues/786 for details)#asahi-dev · 2026-07-06 18:26
flokliOr well, initrd would only need to bind-mount / copy it to the rootfs before switch_root. So initializing hardware requiring vendorfw with kernel modules only in the real rootfs stays possible.#asahi-dev · 2026-07-06 18:19
flokliWould it make sense to change the layout inside the cpios though? Right now it contains a vendorfw/ directory, but maybe it should be lib/firmware/vendor, so we don't need to do any silly copying around where the bootloader appends it as an additional initrd.#asahi-dev · 2026-07-06 18:17
flokli<sven> probably better in #asahi-dev, but neal has an updated version and wanted to submit that patch#asahi-dev · 2026-07-04 12:50
flokli(trying to make this remotely DFU'able, with a KVM and some remote button presser. cs-fw loops with https://gist.github.com/flokli/e3209cd9a7611cbf4122b104382885a2 )#asahi-dev · 2026-07-04 12:35
flokliMmh, did someone try to do a DFU restore /through/ the CS (its passthrough port)? it does briefly probe, but then the CS seems to make it do a hard reset and reboot again (into DFU mode), causing it to get to loop in there...#asahi-dev · 2026-07-02 12:55
floklimischa85: nice on t6041! can you try https://cyberchaos.dev/yuka/linux/-/commit/3bf49bf384d90f91503369de26a09380bd88bf77 is also sufficient?