LLM provided analysis, using GLM-5.3 model (note: I'm not familiar with
this area of the distro, so I couldn't vet it completely, although it
looks sensible to me).

-------- 8< -----
WirePlumber probes every ALSA card via factory api.alsa.acp.device
(scripts/monitors/alsa.lua sets api.alsa.use-acp=true by default). Inside
ACP (spa/plugins/alsa), acp_card_new() defaults use_ucm=true and queries UCM
via pa_alsa_ucm_query_profiles() using the SplitPCM-prefixed card
identifier "<<<SplitPCM=1>>>hw:0". When that query fails, ACP falls back to
its built-in profile set, which has no analog profile definitions for
SOF HDA (sof-hda-dsp) cards — leaving only "off" and "pro-audio".

In the broken state the card is therefore probed without UCM profiles.
WirePlumber's saved default-profile for this card
("HiFi (HDMI1, HDMI2, HDMI3, Headphones, Mic1, Mic2)" in
~/.local/state/wireplumber/default-profile — a UCM split-verb name that
worked on 7.0.0-42) no longer matches any offered profile, and since
api.acp.auto-profile=false, the card stays at profile "off" → no nodes
created → auto_null only.

The UCM query succeeds when the system is quiescent (see evidence), so the
failure appears to be a boot-time race in the ACP/UCM probe. Notably, at
login the GDM greeter's WirePlumber (uid 60578) and the user session's
WirePlumber probe the same card nearly simultaneously (both stacks start
within ~10 ms in the journal). The failure is silent at wireplumber's
default log level (ACP's "UCM not available for card …" is logged at info
level and does not appear in the journal by default).

Notable non-causes (to preempt mis-triage):
- "ALC257: picked fixup  for PCI SSID 17aa:0000" (empty fixup) appears in
  dmesg on every boot including working ones back to 2026-10-26 — not the
  trigger (also noted for this hardware in bug 2121736).
- "Topology: ABI 3:29:1 Kernel ABI 3:23:1" appears in dmesg but the card
  registers fine — not bug 2142979.
- The card IS present at the ALSA level, so this is NOT the 7.0.0-43
  SoundWire "CODEC DAI cs35l56-ot25 not registered" regression (bug 2173618)
  — please do not mark as duplicate.

Related bugs:

- bug 2170445 — same class: boot-time SOF/ACP probe race, WirePlumber
  persists the broken profile, workaround = restart wireplumber until a
  profile appears (HP ZBook, 7.0.0-38)
- bug 2172716 — WirePlumber misses ALSA card at boot on 26.04 /
  wireplumber 0.5.13-1ubuntu1.1; reload works, reboot loses it
- bug 2173618 — the 7.0.0-43 "no sound" regression wave (different
  mechanism: SoundWire DAI, no card at ALSA level)
- bug 2170869 — boot-time audio init failure → Dummy Output on 26.04
  (kernel-level codec probe failure)
- bug 2121736 comment (Hugoj, 2026-07-10) — same hardware (P14s Gen 6,
  ALC257, SSID 17aa:0000), different failure (SOF DSP boot -110)
- upstream alsa-ucm-conf #594 — same "only off/pro-audio on sof-hda-dsp"
  profile signature (Fedora, stalled)

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2174755

Title:
  ThinkPad P14s Gen 6 Intel (Arrow Lake, ALC257) No audio sink after
  boot on 7.0.0-43: WirePlumber ACP probe leaves sof-hda-dsp card at
  profile "off"

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/wireplumber/+bug/2174755/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to