#asahi-dev · 2024-04-21 17:25
jannaupushed to https://github.com/jannau/linux/tree/beta-6.8/asahi-wip tests on m2 pro/max laptops or m2 ultra mac studio would be interesting#asahi-dev · 2024-04-21 15:13
jannauno, it doesn't#asahi-dev · 2024-04-21 15:10
jannaumight require a fix in m1n1 though#asahi-dev · 2024-04-21 15:10
jannauDP/HDMI audio working on m2 pro#asahi-dev · 2024-04-21 13:07
jannautpw_rules: I haven't tested#asahi-dev · 2024-04-21 11:40
tpw_rulesjannau: were you able to get usb boot working with it? i assume it's not in the installer yet?#asahi-dev · 2024-04-21 06:51
jannautpw_rules: it's rolled out as fedora package, source are atm only in my github repo https://github.com/jannau/u-boot-1/tree/asahi-releng-v2024.04 or with additional fedora patches in https://github.com/jannau/u-boot-1/tree/fedora-asahi-v2024.04#asahi-dev · 2024-04-20 23:00
tpw_rulesthe problems being that the bootflow stuff invalidates existing usb booting instructions and that my kernel crashed when i tried to boot it from usb#asahi-dev · 2024-04-20 22:55
tpw_rulesbut there was no tag on the asahi repo#asahi-dev · 2024-04-20 22:55
tpw_rulesjannau: is u-boot 2024.04 available to asahi users yet? i saw some activity and ran into some problems#asahi-dev · 2024-04-20 21:13
jannauI didn't think and have already adjusted user space to use i2c-pasemi-platform so if we switch back we have to adjust 2 things#asahi-dev · 2024-04-20 21:11
jannauthat resulting in a initramfs without dart and other important modules#asahi-dev · 2024-04-20 21:10
jannauI think I'll keep it. it matters for initramfs generation with dracut. we specified i2c-apple to be included. That itself seems not to be necessary but dracut has the annoying behaviour that it stop looking for other modules if it doesn't find one. it doesn't handle that as fatal error though and creates a initramfs with missing modules#asahi-dev · 2024-04-20 20:47
arndjannau: only in the sense that it seemed easier than renaming the files again or having a multi-object file with a single object in it. Usually the name doesn't matter much since it just gets autoloaded, but feel free to change the name back if that helps you#asahi-dev · 2024-04-20 20:20
jannauarnd: did you change the module names in f44bff19268517ee98e80e944cad0f04f1db72e3 intentionally? instead of i2c-pasemi and i2c-apple we now have i2c-pasemi-pci and i2c-pasemi-platform#asahi-dev · 2024-04-20 09:41
jannauworking speakers on f40 with https://src.fedoraproject.org/rpms/wireplumber/pull-request/6 and asahi-audio 2.0#asahi-dev · 2024-04-20 09:34
chadmedi was inclined to believe there was an issue given how loose upstream seem to be with these things. the reason i had to change what we match on in the first place is because the naming convention for nodes just... changed one day #asahi-dev · 2024-04-20 09:33
chadmedah ok phew#asahi-dev · 2024-04-20 09:32
jannauI don't think there is an issue with pipewire-1.0.5, I just booted a broken dev kernel (hdmi audio work)#asahi-dev · 2024-04-20 09:30
chadmedim updating my system now and will check what's going on with pw 1.0.5#asahi-dev · 2024-04-20 09:26
jannaucause for the non-working speakers is a dev kernel with broken audio#asahi-dev · 2024-04-20 09:25
jannauI didn't look hard enough and was confused by the find_match fails in the wireplumber logs#asahi-dev · 2024-04-20 09:17
chadmedthey will be hidden in pw-dump if the dsp rule matched#asahi-dev · 2024-04-20 09:11
jannauchadmed: pipewire-1.0.5 on f40 has no nodes with hw:AppleJxxx,1 entries in 'api.alsa.path'. should they exists in pw-dump?#asahi-dev · 2024-04-19 05:38
apalos /q kettenis #asahi-dev · 2024-04-18 06:27
mazhaving recently experienced an unexpected lock-up on a CS while debugging remotely (which I suspect is from the USB stack), I've added a 5s watchdog reset. Patch pushed out.#asahi-dev · 2024-04-16 21:17
ilmofrom a linux machine#asahi-dev · 2024-04-16 21:16
ilmobooting the non-development kernel#asahi-dev · 2024-04-16 21:16
ilmodevice macbook pro m1, linux install on seperate volume, macOSTest hypervisor install on seperate volume#asahi-dev · 2024-04-16 21:15
ilmothis happens only if far into macOS when rebooting, meaning login screen or further#asahi-dev · 2024-04-16 21:15
jannauwithout any action from the connected computer? how is m1n1 installed (linux install + tethered boot setup)? which device?#asahi-dev · 2024-04-16 21:13
ilmofirst it reboots to m1n1, then after around 5s it reboots again#asahi-dev · 2024-04-16 21:07
jannauexperiments work also against bare-metal m1n1 (using /dev/m1n1 as proxy device)#asahi-dev · 2024-04-16 21:06
jannauwhat do you mean by boot twice? it should reboot only once. do you hear the startup chime twice (if not already disabled)?#asahi-dev · 2024-04-16 20:57
ilmoOr make it faster somehow#asahi-dev · 2024-04-16 20:56
ilmoCan I run experiments without using run_guest? Also when using reboot in m1n1 it reboots twice, can I circumvent that somehow?#asahi-dev · 2024-04-16 20:49
ilmoah okay, thank you#asahi-dev · 2024-04-16 20:46
sven*them#asahi-dev · 2024-04-16 20:46
sventhey’re setup by m1n1 in the iommu pagetable and we can chose then however we want #asahi-dev · 2024-04-16 20:45
ilmowhy do those IOVA defines all have easialy recognizable values like 0xdead/0xbeef? are they set by m1n1/the cpu/arbitrary?#asahi-dev · 2024-04-16 07:25
jannauI hope this explains why I never got sio to boot on t6020#asahi-dev · 2024-04-16 07:25
jannauDP/HDMI audio is working on the m1 max mbp 14" and I finally found the issue why sio crashes after the first few buffers in my test setup. https://github.com/AsahiLinux/m1n1/pull/354 is even more broken than I knew#asahi-dev · 2024-04-15 07:36
jannaubah, with hdmi audio it's a kernel bug though. due to the way the dp/hdmi audio device is instantiated there is no automatic power domain handling#asahi-dev · 2024-04-15 06:50
chadmedah yeah that makes more sense#asahi-dev · 2024-04-15 05:43
jannauno, I don;t think so. the main issue is that the adt is an incomplete maybe even misleading documentation of power-domains#asahi-dev · 2024-04-14 23:14
chadmed_is there something fundamentally broken about power domains in the kernel? we seem to have these sorts of issues quite frequently with various hw blocks...#asahi-dev · 2024-04-14 22:23
jannauHDMI audio on j314 almost working as well. breaks with a SError on sio on playback start, most likely a power-domain issue despite pd_ignore_unused and marking some domains as always-on#asahi-dev · 2024-04-14 11:40
jannaufor some reason's dp-audio1's power-domain is critical, marking the node as always-0n in the devicetree avoids the revious "DMA shiom unavailable" error#asahi-dev · 2024-04-14 11:38
jannaufinally some (very little) audio from the m2 mini's hdmi port. sio still crashes almost immediately but at least it clearly processes some data#asahi-dev · 2024-04-12 06:34
jannauno problem, just wanted to make sure they are not forgotten