#asahi-alt · 2026-11-17 09:18
jannaugit rebase should notice that and just drop the commit#asahi-alt · 2026-11-17 09:17
jannauyes, that would work but is pointless as the only commit between asahi-7.3.4-1 and asahi-7.3.4-2 is already in fairydust. I did the fix while I was rebasing fairydust#asahi-alt · 2026-11-17 09:05
jannauyep, since fairydust are just a few patches you can determine $OLD_TAG with git log. Do this all while the fairydust branch is chacked out#asahi-alt · 2026-11-16 20:50
jannauI've also rebased fairydust. I use `git rebase --onto $NEW_TAG $OLD_TAG` for that. usually that works without issues#asahi-alt · 2026-11-16 20:44
jannaumps: fixes in asahi-7.3.4-2#asahi-alt · 2026-11-16 19:57
jannaulikely the change allowing more formats for fbdev emulation and simpledrm + xrgb2101010 not working#asahi-alt · 2026-11-10 12:53
jannauI wouldn't want to guarantee that clock issue always results in speakersafetyd getting bogus data that result it to fail but I think it's very unlikely that something happened to your (or anyones else's) speakers#asahi-alt · 2026-11-10 10:33
jannaudo not use speakers on asahi-7.3.3-1. It likely fails into the safe state but I wouldn't want to guarantee that#asahi-alt · 2026-11-10 10:31
jannaus/bot/not/#asahi-alt · 2026-11-10 10:30
jannauI've tagged asahi-7.3.3-2 please do not use asahi-7.3.3-1 or upgrade from it immediately#asahi-alt · 2026-11-09 23:13
jannauI mostly noticed it because `BIT(ffs(port_mask))` looks wrong. ffs() returns 1 + 0-based index of the first set bit (or 0 for ffs(0))#asahi-alt · 2026-11-09 23:09
jannauyou changed it after Joshua asked about "6 + 1" in https://lore.kernel.org/all/sCELYiNQSvCxqjQFP0uhkQ@gmail.com/#asahi-alt · 2026-11-09 23:07
jannauin mca#asahi-alt · 2026-11-09 23:07
jannaudiff between 7.2.14-1 and 7.3.3-1 is very small. the most suspicious thing I saw was that REG_SYNCGEN_MCLK_SEL gets now BIT(1) or-ed instead of writing 7#asahi-alt · 2026-11-09 23:03
jannauit's unfortunately not reproducible just by playing the same audio#asahi-alt · 2026-11-09 22:21
jannaubut it seems to have that already#asahi-alt · 2026-11-09 22:20
jannauI was thinking if we might need the same for tas2764#asahi-alt · 2026-11-09 22:19
jannaumyria: I picked that patch up but it is the wrong speaker amp. tas2770 is only use on devices with base m1#asahi-alt · 2026-11-09 18:11
jannauthe issue is that speakersafetyd either gets bogus ivsense data or it processes it in strange way#asahi-alt · 2026-11-09 18:09
jannaureproduce now with a different symptom, speakersafetyd panicked with "Gain reduction exceeded threshold", audio wasn't even load#asahi-alt · 2026-11-09 17:42
jannaunot sure what triggers it#asahi-alt · 2026-11-09 17:35
jannauback on gentoo with kernel 7.3.3 and it hasn't reproduced yet at 95% volume#asahi-alt · 2026-11-09 17:08
jannauno issues after upgrading#asahi-alt · 2026-11-09 17:02
jannaufedora system is still on pipewire 1.6.9 / wireplumber 0.5.18, updating now#asahi-alt · 2026-11-09 16:35
jannaufedora is on 4.1 as well#asahi-alt · 2026-11-09 16:34
jannauasahi-audio is the least likely source of the issue#asahi-alt · 2026-11-09 16:32
jannauelisaado: I can't reproduce with fedora on the same device (j314c) booted via HV vs self-compiled kernel gentoo. Fedora lags a little on speakersafetyd / asahi-audio but nothing looks like it could explain this#asahi-alt · 2026-11-09 14:22
jannauhuh, I thought I switched to the upstream macaudio but did not#asahi-alt · 2026-11-09 14:15
jannauI have no issues at 50% but the point where issues start certainly depends on the audio signal#asahi-alt · 2026-11-09 14:09
jannauI can reproduce issues at higher (not near 100% though) volumes#asahi-alt · 2026-11-09 14:01
jannauis speakersafetyd restarting?#asahi-alt · 2026-11-05 13:28
jannaumps: I pushed asahi-7.2.15-2 which should fix the issue#asahi-alt · 2026-11-04 14:56
jannau...#asahi-alt · 2026-11-04 14:55
jannaumeston setup --prefix /#asahi-alt · 2026-11-04 13:59
jannauerr, yes I meant avd-fw-v0.2.tar.xz#asahi-alt · 2026-11-04 13:29
jannauavd-fw-all.tar.xz https://github.com/AsahiLinux/avd-fw/releases/download/v0.2/avd-fw-all.tar.xz#asahi-alt · 2026-11-04 13:02
jannaumps: get the tarball from the release and not the from the tag#asahi-alt · 2026-11-04 11:54
jannauso I supect the gentoo folks are hitting something else (unless they roll their only kernel)#asahi-alt · 2026-11-04 11:53
jannauthe kernel issue is something I introduced in asahi-wip and backported to asahi-7.2.15-1#asahi-alt · 2026-11-04 11:33
jannauI'll fix the root problem later today#asahi-alt · 2026-11-04 11:33
jannaucan you try to add mux_apple_display_crossbar to the initramfs#asahi-alt · 2026-11-04 10:45
jannauM1 dcpext via fairydust on asahi-7.2.15-1 wors for me#asahi-alt · 2026-11-04 10:44
jannauI wouldn't be surprised if the inverted check for dcpext initialization messed things up for M1#asahi-alt · 2026-11-04 09:51
jannauusing m1n1 main or a tag? if it's main it might be fixed since this morning#asahi-alt · 2026-11-04 09:43
jannausounds like dcp did not probe#asahi-alt · 2026-11-01 09:16
jannauor rpm fusion could just ship the all codecs firmware with firmware/updates as path without conflicting with linux-firmware#asahi-alt · 2026-11-01 09:15
jannaublocking in firmware has the advantage that the workaround on user size is curl -L https://github.com/AsahiLinux/avd-fw/releases/download/v0.2/avd-fw-all.tar.xz | tar xv -C /usr/lib/firmware/updates/`#asahi-alt · 2026-10-31 18:00
jannauno HDMI or fairydust support#asahi-alt · 2026-10-31 17:59
jannaudisplay is hot off the press and not yet at the same state as M1 / M2#asahi-alt · 2026-10-31 17:21
jannauCONFIG_RESET_APPLE_CIO=m is important as well