Laptop speakers, headphone jack or internal mic never work despite the card being detected
Checked on this Omarchy 4 workstation (omarchy 4.0.2-1, kernel 7.1.9-arch1-2, mkinitcpio 41.1-1, limine-mkinitcpio-hook 1.37.1-1) and against the cited pages, all four of which I retrieved and all four of which still say what the record claims. docs.kernel.org/sound/hd-audio/notes.html returned 200 and carries the nofixup wording, the generic wording, the SSID alias introduced in 5.15 and the request_firmware and /lib/firmware sentence verbatim. models.html returned 200. The Arch ALSA troubleshooting page still carries the three chip-identification commands and the dell-headset-multi tip including the not-a-Dell caveat and both codec families, and the power_save line verbatim. alsa-tools is extra 1.2.15-3. The two identification commands work here: both report ALC1220 on card 0. Two real defects and two gaps. The worst is in the danger field: the record states that nothing here touches the initramfs, and on Omarchy 4 that is false. /etc/mkinitcpio.conf.d/omarchy_hooks.conf assigns HOOKS containing modconf, and /usr/lib/initcpio/install/modconf line 5 is add_full_dir /etc/modprobe.d '*.conf', so anything written to /etc/modprobe.d is bundled into the UKI at the next rebuild. The conclusion the record draws is still right, because snd_hda_intel is not in MODULES and autodetect does not pull sound modules, so the option takes effect from the real root with no rebuild needed. The reasoning was wrong, and the corpus has already been bitten by records assuming the wrong initramfs model. I added the correct rebuild command for the case where someone does need one: /etc/mkinitcpio.d/ is empty on this machine, so mkinitcpio -P finds no presets, and the UKI comes from limine-mkinitcpio. Second, step 3's service stop omitted the sockets. pipewire.socket and pipewire-pulse.socket socket-activate the daemon on the next client connection, so the fuser gate can pass and the card be retaken a second later. The record's own audit note spotted this in 2026 and never put it in the fix, so it is now in the fix. Third, the record offered model=XXXX:YYYY without saying where the subsystem ID comes from. It is in the same file as the codec name: grep -i 'Subsystem Id' /proc/asound/card0/codec#0 returns 0x1458a0c3 on this machine, which is model=1458:a0c3. Fourth, alsa-info.sh is what both the kernel docs and the Arch wiki ask for in a bug report and it is already in alsa-utils 1.2.16-1, so that is now step 8. I also named omarchy-restart-audio, which is the supported way to bring audio back here. NOT exercised, and this is the important limit on the confidence. This machine is a desktop: card 0 is an Intel PCH with an ALC1220, card 1 is NVidia HDMI, and the mic is a USB Blue Yeti. There are no laptop speakers, no combined TRRS jack and no internal mic, so the symptom cannot be reproduced and no model string could be tested. I did not run modprobe, did not write anything under /etc/modprobe.d and did not rebuild the initramfs, all of which the audit forbids on the operator's daily workstation. alsa-tools is not installed, so hdajackretask and hda-verb were confirmed from the package file list and the kernel docs rather than by running them, and the exact hda-verb argument form is reproduced from the record and the hwdep section of notes.html rather than from its own --help. The /etc/modprobe.d finding was read from the hook files and the modconf install script, not proven by inspecting a built UKI, because /boot is a vfat ESP mounted dmask=0077 and unreadable unprivileged.
A wrong model= string can leave you with less working audio than you started with, including no sound at all, until you delete the file from /etc/modprobe.d and reboot. Keep a note of what the file contained. hda-verb writes raw verbs straight to codec registers with no driver state tracking, and on some machines it can unmute an amplifier past its safe level, so lower the volume before experimenting and do not run verbs you cannot explain.
Nothing here can stop the machine booting, because snd_hda_intel is not loaded in early userspace. It does reach the initramfs on Omarchy 4, though, and the record used to say otherwise. Omarchy assigns HOOKS wholesale in /etc/mkinitcpio.conf.d/omarchy_hooks.conf and that list contains modconf, whose install script does add_full_dir /etc/modprobe.d '*.conf'. So every file you write there is copied into the UKI at the next rebuild. That is harmless for a sound quirk, but it means /etc/modprobe.d is not an initramfs-free zone here, and a stale copy of the file survives inside the image until the next rebuild.
If you ever do need that rebuild, **mkinitcpio -P is the wrong command on Omarchy 4**: /etc/mkinitcpio.d/ is empty, so there are no presets and it does nothing. The image is a UKI built by limine-mkinitcpio-hook:
sudo limine-mkinitcpio
Symptom
aplay -l lists the HDA card and wpctl status shows a sink, but the built-in speakers are permanently silent, or the headphone jack outputs nothing, or plugging a headset into the combined 4-pole (TRRS) jack gives you audio out but the headset microphone is never offered as an input. Nothing in alsamixer is muted and there are no obvious errors in dmesg. Installing sof-firmware and reinstalling PipeWire made no difference.
Cause
The HD-Audio driver builds its pin configuration from the BIOS pin defaults, and on a lot of laptops (and most cheap or very new ones) those defaults are wrong or incomplete. The driver then wires the wrong DAC to the wrong pin, or does not know the jack is a shared headset jack, and the affected output or input simply never appears. The kernel keeps a whitelist of per-machine 'fixups' selected by PCI SSID, but a machine that is not yet in that list gets generic handling. This is the escalation step past the SOF-firmware and mute-state records: the firmware is loaded and the card is present, it is the pin/model mapping that is wrong.
Fix
**1. Identify the codec chip.**
grep 'Codec:' /proc/asound/card*/codec#*
grep . /sys/class/sound/hwC?D?/chip_name
alsamixer # 'Chip:' in the top-left corner
**2. Note the subsystem ID at the same time.** You need it for the SSID form in step 4, and it is in the same file:
grep -i 'Subsystem Id' /proc/asound/card0/codec#0
# Subsystem Id: 0x1458a0c3 -> model=1458:a0c3
**3. Look your chip up in the kernel's model list** at https://docs.kernel.org/sound/hd-audio/models.html and pick a candidate model string. For a headset mic that is not detected on a combined 4-pole jack, try dell-headset-multi first even if the machine is not a Dell. It is the standard answer for ALC22x/23x/25x/269/27x/28x/29x and ALC66x/67x/892 codecs.
**4. Test it without rebooting.** Nothing may hold /dev/snd while you reload, and stopping the services is not enough on its own: pipewire.socket and pipewire-pulse.socket will socket-activate the daemon again the moment any client connects, and it takes the card back mid-procedure. Stop the sockets too.
systemctl --user stop pipewire.socket pipewire-pulse.socket
systemctl --user stop wireplumber.service pipewire.service pipewire-pulse.service
fuser --all --verbose /dev/snd/* # must be empty, kill whatever is listed
sudo modprobe -r snd_hda_intel
sudo modprobe snd_hda_intel model=dell-headset-multi
Bring audio back. On Omarchy use the shipped wrapper, which starts the same units and additionally recovers a USB audio device left stuck by the reload:
omarchy restart audio # or: omarchy-restart-audio
On plain Arch:
systemctl --user start pipewire.socket pipewire-pulse.socket
systemctl --user start pipewire.service pipewire-pulse.service wireplumber.service
wpctl status
**5. Make it permanent** once a model string works:
sudo tee /etc/modprobe.d/alsa-hda.conf >/dev/null <<'EOF'
options snd_hda_intel model=dell-headset-multi
EOF
No initramfs rebuild is needed for this to take effect. snd_hda_intel is loaded by udev from the real root after switch_root, and modprobe reads /etc/modprobe.d there. See the danger note for the one Omarchy detail that goes with it.
Since kernel 5.15 the option also accepts an SSID alias, which applies another machine's whole quirk set. That is what to reach for when your laptop is a rebadge of a supported one, using the subsystem ID from step 2:
options snd_hda_intel model=103c:8862
Two special values worth knowing: model=nofixup skips the device-specific fixups in the codec parser (use when a *wrong* quirk is being applied and broke things), and model=generic skips the codec-specific parser and uses only the generic one.
**6. If no model string fits, remap the pins yourself.**
sudo pacman -S alsa-tools
hdajackretask # GUI: override a pin's function, then 'Install boot override'
hdajackretask writes a firmware patch file and the matching /etc/modprobe.d entry for you. For hand-written patches, the driver reads the file with request_firmware(), so it must sit on the firmware path, typically /lib/firmware:
options snd_hda_intel patch=my-laptop-patch.fw
For one-off register pokes while experimenting, hda-verb (also in alsa-tools) writes directly to a codec node. It needs CONFIG_SND_HDA_HWDEP, which is what creates the /dev/snd/hwC0D0 device it talks to:
sudo hda-verb /dev/snd/hwC0D0 0x0 PARAMETERS VENDOR_ID
**7. While you are in /etc/modprobe.d**, a second HDA option is worth setting on laptops that pop or clip the first half-second of every sound:
options snd_hda_intel power_save=0 power_save_controller=N
**8. If nothing works, report it with the data the maintainers ask for.** alsa-info.sh is already present in alsa-utils, so there is nothing to install:
sudo alsa-info.sh --no-upload
Send that output, plus the vendor, product and model names and uname -r, with the bug report.
Verify
After the reload, wpctl status lists the previously-missing sink or source. speaker-test -c 2 -D hw:0,0 produces sound from the built-in speakers. Plug in a headset and confirm wpctl status gains a 'Headset Microphone' source and arecord -f cd -d 5 /tmp/t.wav && aplay /tmp/t.wav captures your voice. alsamixer should show new controls that were absent before.
Sources
- https://wiki.archlinux.org/title/Advanced_Linux_Sound_Architecture/Troubleshooting
- https://wiki.archlinux.org/title/Advanced_Linux_Sound_Architecture
- https://docs.kernel.org/sound/hd-audio/notes.html
- https://docs.kernel.org/sound/hd-audio/models.html
- https://archlinux.org/packages/extra/x86_64/alsa-tools/