Skip to contentSkip to conversation

7.1.6: No sound on MacBook12,1 (Cirrus CS4208) - DAC receives data but speaker silent

REGRESSION Takashi Iwai 1 participant 1 message Aug 10 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw
Quoted 40 lines · On Mon, 10 Aug 2026 10:13:11 +0200, nqse0035 wrote:

Hi,

Regression introduced in: v7.1.5 -> v7.1.6 (tested both; config unchanged)

Hardware: MacBook9,1 (2016 MacBook 12"), HDA Intel PCH, Cirrus Logic CS4208 codec (SSID 106b:0000)
Kernel: 7.1.5.arch1-2 (works) / 7.1.6.arch1-1 (broken) on Arch Linux Audio stack: PipeWire 1.6.8, WirePlumber 0.5.15

Symptom:
After upgrading to 7.1.6 and rebooting, the internal speaker is completely silent. Volume UI responds, playback apps run, but no sound comes out. Reboot does not help (this is NOT the pre-existing occasional no-sound-after-boot issue - it is persistent).

What I verified (all identical between 7.1.5 and 7.1.6):
- Codec enumerates fine; snd_hda_intel + snd_hda_codec_cs420x load OK
- PipeWire/WirePlumber up; default sink alsa_output.pci-0000_00_1f.3.analog-stereo present and active
- Mixer controls unmuted (Master 72%, PCM 100%, Line Out 100%)
- Routing correct: Active Port = analog-output-speaker, headphones not available
- DAC Node 0x02 shows "Converter: stream=1" -> data IS reaching the codec, but the speaker amplifier never produces sound
- dmesg/journalctl: no ALSA/HDA errors at all
- Codec fixup/autoconfig logs identical between kernels:
"CS4208: picked fixup for codec SSID 106b:0000", line_outs=1 (0x1d) type:speaker, hp_outs=1 (0x10)
- Verified power_save=0 (our setting) is NOT the cause: reloaded module with power_save=10, still silent; and 7.1.5 works with the same configuration

Conclusion:
A regression between 7.1.5 and 7.1.6 breaks the CS4208 amplifier output path. Suspect: the HD-audio pending-IRQ handling rework from the 7.1-rc5 cycle (stated to affect "odd machines"), which likely changed hda-intel/hda-core behaviour on this platform.

Quoted 4 lines

Workaround: staying on 7.1.5 (pinned). Happy to run tests or provide any further debug info.

#regzbot introduced: v7.1.5..v7.1.6

Through a quick glance, there is no relevant change between v7.1.5 and v7.1.6. At best, please do bisect.

thanks,

Takashi