feat: Add aw87xxx codec for AyaNeo/Ayn/Orange Pi handheld speaker amplifiers

[hijae]

Context

Several handhelds (AyaNeo Flip KB/DS, AIR 1S, AIR Plus, Kun; Ayn Loki MiniPro; Orange Pi NEO-01) use the Awinic AW87XXX Smart K PA as their speaker amplifier, identified over ACPI as AWDZ8830. Mainline Linux has no driver for this chip — the closest match (aw87390.c) doesn't recognize it. Without this driver these devices have no speaker output at all.

This driver was lost when Bazzite's kernel source switched to OGC, unrelated to the driver itself. Ported here referencing ChimeraOS's kernel, which carries the same driver. See ublue-os/bazzite#5374 for the resulting regression report (AyaNeo AIR 1S, no audio since the kernel switch).

Changes

• First commit: the original driver + DMI-based firmware filename selection, ported from hhd-dev/patchwork with attribution preserved.
• Second commit: drop a dead linux/of_gpio.h include (header no longer exists in this tree) so it builds on current kernels.
• Third commit: add an ALC269VB pin fixup for the AIR 1S's internal speaker, ported from hhd-dev/patchwork — needed in addition to the aw87xxx driver above, since the AIR 1S's main HDA codec pin routing isn't otherwise configured for internal speaker output.

Testing

Built out-of-tree against this branch's exact source (kernel-devel 7.2.0-ogc4.1) and verified on real AyaNeo Flip KB hardware — chip probes correctly (chip_id 0x5a, product aw87559), and speaker audio works, including after a fresh reboot.

The AIR 1S pin fixup (third commit) sets the same pin/value bl1nk verified fixes internal speaker output on real AIR 1S hardware (see comments below) — but this commit ports the original, simpler hhd-dev/patchwork implementation (a static pin table) rather than bl1nk's own version (a runtime-DMI-checked function), and I don't have an AIR 1S to test with myself. The underlying fix is verified; this exact commit hasn't been re-tested as-is, so it'd be good to confirm it works before merging.

[hijae]

Companion PR enabling this in the build config: OpenGamingCollective/kernel-packages#36 — CONFIG_SND_SOC_AW87XXX isn't on by default, so without that PR this driver is present but never actually built.

[pastaq]

Are we planning to upstream this, or rework it before hand? Authorship makes it somewhat of a challenge.
@BoukeHaarsma23 I would love your input here.

[bl1nk]

For the AYANEO Air 1S another quirk in the ALC269 codec is necessary to fix audio output. I commented on the initial bug report in ublue-os/bazzite: ublue-os/bazzite#5374 (comment)

@hijae Do you want to include a fix for that in this PR as well or should this be done via a separate PR?

Working patch to fix audio output on the AYANEO Air 1S

I went back and forth with an LLM to write this, but I am neither a kernel contributor nor am I experienced in C. I understand the code, but I do not understand why 0x90170110 results in the correct pinout. It seems to be some conventional value that is used in other fixes for the realtek codec across the tree to indicate what an internal speaker is.

This fix is should even be safer than the upstream one as we are only targeting the Air 1S specifically.

diff --git a/sound/hda/codecs/realtek/alc269.c b/sound/hda/codecs/realtek/alc269.c
index cae327e1d9..390bf19966 100644
--- a/sound/hda/codecs/realtek/alc269.c
+++ b/sound/hda/codecs/realtek/alc269.c
@@ -3998,6 +3998,7 @@
 	ALC269VB_FIXUP_ASUS_ZENBOOK,
 	ALC269VB_FIXUP_ASUS_ZENBOOK_UX31A,
 	ALC269VB_FIXUP_ASUS_MIC_NO_PRESENCE,
+	ALC269_FIXUP_AYANEO_AIR_1S,
 	ALC269_FIXUP_LIMIT_INT_MIC_BOOST_MUTE_LED,
 	ALC269VB_FIXUP_ORDISSIMO_EVE2,
 	ALC283_FIXUP_CHROME_BOOK,
@@ -4299,6 +4300,19 @@
 	ALC285_LENOVO_DAC_RENAME,
 };
 
+static void alc269_fixup_ayaneo_air_1s(struct hda_codec *codec,
+				       const struct hda_fixup *fix, int action)
+{
+	if (action != HDA_FIXUP_ACT_PRE_PROBE)
+		return;
+
+	if (!dmi_match(DMI_SYS_VENDOR, "AYANEO") ||
+	    !dmi_match(DMI_PRODUCT_NAME, "AIR 1S"))
+		return;
+
+	snd_hda_codec_set_pincfg(codec, 0x1a, 0x90170110);
+}
+
 /* A special fixup for Lenovo C940 and Yoga Duet 7;
  * both have the very same PCI SSID, and we need to apply different fixups
  * depending on the codec ID
@@ -4530,6 +4544,10 @@
 			{ }
 		},
 	},
+	[ALC269_FIXUP_AYANEO_AIR_1S] = {
+		.type = HDA_FIXUP_FUNC,
+		.v.func = alc269_fixup_ayaneo_air_1s,
+	},
 	[ALC269_FIXUP_AMIC] = {
 		.type = HDA_FIXUP_PINS,
 		.v.pins = (const struct hda_pintbl[]) {
@@ -8213,6 +8231,7 @@
 	SND_PCI_QUIRK(0x1f4c, 0xb020, "Minisforum AI X1 Pro",
 		      ALC245_FIXUP_MINISFORUM_JACK_DETECT),
 	SND_PCI_QUIRK(0x1f4c, 0xe001, "Minisforum V3 (SE)", ALC245_FIXUP_BASS_HP_DAC),
+	SND_PCI_QUIRK(0x1f66, 0x0103, "AYANEO AIR 1S", ALC269_FIXUP_AYANEO_AIR_1S),
 	SND_PCI_QUIRK(0x1f66, 0x0105, "Ayaneo Portable Game Player", ALC287_FIXUP_CS35L41_I2C_2),
 	SND_PCI_QUIRK(0x2014, 0x800a, "Positivo ARN50", ALC269_FIXUP_LIMIT_INT_MIC_BOOST),
 	SND_PCI_QUIRK(0x2039, 0x0001, "Inspur S14-G1", ALC295_FIXUP_CHROME_BOOK),

[hijae]

@bl1nk
I can include it if you'd like. However, I don't have an Air 1s device and can't test it, so I recommend that you create a separate PR yourself.
Also, I'm not sure if my PR will be accepted because of the authorship issue.

[pastaq]

We'll most likely carry this patch as it is needed for many devices. The authorship issue is administrative. I asked bl1nk to see if you could cherry-pick that patch as well. It will depend on these being merged so it makes it simpler to include.

[hijae]

Thanks for tracking this down and for the fix! I found the original source for this — it's already in hhd-dev/patchwork (89eb9630d9, authored by fewtarius, co-developed by linh1987, carried by Antheas) — so I ported that version instead of yours: it's simpler (a static pin table, no runtime DMI check needed) and keeps clean original authorship. Pushed as a third commit on this PR, same pin/value you verified (0x1a → 0x90170110).

I don't have an AIR 1S myself, so if you get a chance to confirm this exact commit works on your hardware, that'd be great before it merges.

[pastaq]

IIRC the mini pro and the air 1s have inverted pin configurations or something similar that necessitated the DMI match but I'm not certain what it exactly was. Hopefully we can find a mini pro user to test as well

[bl1nk]

Thank you both, again.

I built both modules and loaded them against the current -testing kernel and everything worked just fine (after restarting steam, so it could pick up the new pipewire sink, I think).

If anyone wants to try it out on their device(s) to see if it breaks on something that is not the AYANEO Air 1S

Clone the kernel repo:

git clone https://github.com/OpenGamingCollective/linux.git
cd linux

git fetch origin pull/102/head:pr-102
git switch pr-102

Set some variables:

KDIR=/lib/modules/$(uname -r)/build
REALTEK="$PWD/sound/hda/codecs/realtek"
AW87="$PWD/sound/soc/codecs/aw87xxx"

Then build the patched modules.

Realtek codec modules:

make -C "$KDIR" \
    M="$REALTEK" \
    modules

AW87xxx amplifier driver:

make -C "$KDIR" \
    M="$AW87" \
    CONFIG_SND_SOC_AW87XXX=m \
    modules

The relevant resulting modules are:

sound/hda/codecs/realtek/snd-hda-codec-realtek-lib.ko
sound/hda/codecs/realtek/snd-hda-codec-alc269.ko
sound/soc/codecs/aw87xxx/snd-soc-aw87xxx.ko

Optional check:

strings "$REALTEK/snd-hda-codec-alc269.ko" | grep -i ayaneo

should return something like:

ayaneo-speaker-pin-fix
ALC269VB_FIXUP_AYANEO_SPKR_PIN_FIX

to verify that the compiled module contains the fix.

Temporarily disable PipeWire/WirePlumber:

systemctl --user mask --runtime \
    pipewire.service pipewire.socket \
    pipewire-pulse.service pipewire-pulse.socket \
    wireplumber.service

systemctl --user stop \
    pipewire.service pipewire.socket \
    pipewire-pulse.service pipewire-pulse.socket \
    wireplumber.service

Check that nothing still has the ALSA devices open:

sudo fuser -v /dev/snd/*

(it should return nothing)

Unload the existing HDA modules:

sudo modprobe -r snd_hda_intel
sudo modprobe -r snd_hda_codec_alc269
sudo modprobe -r snd_hda_codec_realtek_lib

Load the patched modules:

sudo modprobe snd_hda_scodec_component
sudo modprobe snd_hda_codec_generic

sudo insmod "$REALTEK/snd-hda-codec-realtek-lib.ko"
sudo insmod "$REALTEK/snd-hda-codec-alc269.ko"
sudo insmod "$AW87/snd-soc-aw87xxx.ko"

sudo modprobe snd_hda_intel

You should be able to play a sound now:

speaker-test -D hw:1,0 -c 2 -r 48000 -t sine

Reboot to reset the device to its previous state.