Repository navigation
Description
Mac and installed build
- Model / board / SoC: MacBook Pro 14-inch M2 Max (2023) /
apple,j414c/apple,t6021 - Installed Aurora tag: not determinable. Package
linux-aurora 7.2.11.aurora12-1from the omarchy repo is what we know. The repo has no matching release tag. uname -r:7.2.11-12-1-ARCH- Bootloader:
m1n1-aurora 1.6.1.aurora22-1,uboot-asahi 2026.07.asahi2-4.asahi,m1n1-stage2-versionreadsbc79acc4-dirty. - Audio:
asahi-audio 4.1-1,alsa-ucm-conf-asahi 10-1,pipewire 1:1.6.10-1,wireplumber 0.5.18-1 - Distribution: Omarchy on Arch Linux ARM with the omarchy package repo.
omarchy-mac 0.2.3-1andomarchy-mac-boot 20261027-1are installed. We use Asahi's audio packages, notaurora-audio. appledrm.hdmi_audioreadsYwith no option set, so drm/apple: turn DP/HDMI audio on by default #300's default is active.
Both firmware versions and provisioning
asahi,os-fw-version: 13.5
asahi,system-fw-version: 26.6.2
Provisioning: not relevant.
Reproduction, expected result, and actual result
The output is the first DP entry, alsa_output.platform-39b540000.audio-controller.stereo-fallback, linked to dcp@289c00000 (dcpext0).
- Play sound on the first DP entry.
- Lose the display: swap HDMI to USB-C, or unplug USB-C.
Expected: sound moves or recovers when the display returns.
Actual: /proc/asound/card0/pcm0p/sub0/status shows state DISCONNECTED, PipeWire logs about 30 hw:0p errors a second (92 and 93 lines in 3 s samples), and there's no sound. YouTube in Brave stalled while this was happening. The link to the stall is our inference.
Error runs from the journal, EDT:
| Start | What happened | End |
|---|---|---|
| 09:26:28 | HDMI to USB-C, at the dcpext0 audio timeout | 09:26:34 |
| 09:41:47 | USB-C unplugged | 09:42:37 |
| 09:43:32 | HDMI to USB-C, at the dcpext0 audio timeout | 09:45:16 |
| 09:46:13 | USB-C unplugged | 09:47:40 |
| 09:49:04 | HDMI to USB-C, at the dcpext0 audio timeout | 09:49:11 |
We also saw brief HDMI link drops with no cable swap (HPD connected:0 then connected:1) at 09:41:58 and 09:46:21. We can't attribute them to the Mac or the monitor. Both fell inside an error run that had already started, so they didn't start one.
Recovery, not consistent:
pactl suspend-sink ... 1then0at 09:45:15 was followed by the errors ending and the card goingRUNNING. The default also moved to MacBook Speakers within that second, so we can't separate the two.- Switching to MacBook Speakers and back restored sound once.
- Another time, switching back went silent, with the
playing into the voidline at 09:47:32 on39b580000.pactl set-default-sinkto39b540000fixed it at 09:48:19. A misclick on the second entry can't be ruled out. - The 09:42, 09:45 and 09:47 runs each ended within about 5 s of the default moving to MacBook Speakers.
Last known good: unknown.
Graphics or session details
Not a graphics problem. Players were Brave (brave-origin-bin 1:1.98.63-1) and cliamp.
Relevant error lines
Date, hostname and kernel: trimmed.
# PipeWire, repeating
09:41:47.883845 pipewire[1280]: spa.alsa: hw:0p: snd_pcm_status error: File descriptor in bad state
09:41:47.883852 pipewire[1280]: spa.alsa: hw:0p: snd_pcm_drop: File descriptor in bad state
09:41:47.883862 pipewire[1280]: spa.alsa: hw:0p: (0 suppressed) snd_pcm_avail after recover: File descriptor in bad state
# HDMI link drop with no cable swap
09:46:21.161449 apple-dcp 289c00000.dcp: DP2HDMI HPD irq, connected:0
09:46:21.754441 apple-dcp 289c00000.dcp: DP2HDMI HPD irq, connected:1
# Second entry selected with the monitor attached, during recovery
09:47:32.400455 dcp-dp-audio 39b580000.audio-controller: no sink, playing into the void
Related issues, pull requests, or code examples
- drm/apple: turn DP/HDMI audio on by default #300: turned DP/HDMI audio on by default.
- Display audio: qualify HDMI/DP audio before changing its default #48: display audio qualification, which includes reconnect. The J314S report there used the
apple_dcpprofile from apple_dcp: add HDMI/DisplayPort output AsahiLinux/alsa-ucm-conf-asahi#9 and saw HDMI unplug and replug recover by itself. We don't have that profile, and we see the stuck state instead. That difference may matter, but we haven't tested it. - aurora12 with asahi-audio: DCP outputs take the default from MacBook Speakers, which go silent #314: DCP outputs taking the default, on the same Mac. USB-C hot-swap from HDMI: every mode rejected while routed to dcpext1, Hyprland aborts #313: Hyprland crash on the same swap.
What was not tested
- The
aurora-audiostack - Other M1/M2 Macs
- With drm/apple: turn DP/HDMI audio on by default #300 reverted
- Suspend
Activity