HDAUniversal - AppleHDA-like Audio Kext for macOS Tahoe and Hackintosh Systems

Droga · July 10

Olá MaLd0n,

Passando aqui para agradecer demais pelo seu trabalho incrível com o HDAUniversal. Está funcionando perfeitamente com a placa de som integrada da minha placa-mãe ASUS ROG Crosshair X470, que usa o codec ROG SupremeFX S1220.

O som está se ouvindo muito mais nítido e limpo agora!

Gostaria muito de saber como está o progresso geral desse projeto. Além disso, queria saber se existe alguma previsão ou plano para adicionar a opção de áudio em 24-bit (ou superior) nas próximas atualizações?

Valeu demais pela dedicação com a comunidade! Tamo junto!

Hello MaLd0n,

I just wanted to pass by and thank you so much for your amazing work with HDAUniversal. It is working flawlessly with the built-in sound card on my ASUS ROG Crosshair X470 motherboard, which uses the ROG SupremeFX S1220 codec.

The audio output feels noticeably clearer and much more crisp now!

I would love to know how the overall progress of this project is going. Also, I was wondering if there are any plans to include an option for 24-bit audio output (or higher) in future updates?

Thanks again for your dedication to the community! Enjoy!

MaLd0n · July 10

Fala, @Cristian Abad

Estou trabalhando nisso, já está quase saindo o update completo, mas como é um driver de áudio, de uma complexidade absurda, demora um pouco mesmo.

HackintoshEUA · July 11

Thankssssssss

Edited July 11 by HackintoshEUA

MaLd0n · July 11

Enjoy!!!!

josh anderson · July 14

Hey Maldon, thank you so much for this HDA Universal! It really solved the problem of having to depend on OCLP for audio on Tahoe or having to use Voodoo (which has terrible quality). It's fantastic! I'm incredibly grateful. In fact, it confirms my suspicion. I thought that as soon as they announced Tahoe was the last macOS for Intel, they wouldn't receive macOS 2.7 anymore, and they wouldn't focus on adding support but rather on stabilizing the remaining macOS as much as possible and finding workarounds for current problems. And look, HDA Universal! I wonder if something similar for Wi-Fi will be released soon? Thank you so much! The only problem is that I couldn't get audio through HDMI or DisplayPort, but otherwise everything is excellent, just like Apple's own ALC. If you could check it out, that would be excellent, really, thank you. (Please confirm if what I think is true. Are they going to stabilize macOS Tahoe to the maximum because it's the latest? haha)

MaLd0n · July 14

Big Update is comming and soon can try add digitam/hdmi too

licobar · July 14

Valeu maldon

MaLd0n · July 14

Flw!

mishealgo · July 16

Can I use it when SIP is enabled or not?

MaLd0n · July 16

On installation no, use 03080000, install, reboot then enable

Droga · July 17

i attached my codec info report if that helps.

Codec-Info-Report.zip

MaLd0n · July 19

HDAUniversal v0.1.353 — PCM Test Results

I individually tested all 12 PCM formats currently available on the analog output.

Formats tested

Sample rate	Bit depths
44.1 kHz	16, 20 and 24-bit
48 kHz	16, 20 and 24-bit
96 kHz	16, 20 and 24-bit
192 kHz	16, 20 and 24-bit

32 kHz is not available as an output format on this device, so I did not test it.

32-bit is also not exposed as a native physical bit depth on this output.

Results

All 12 selected formats were accepted by IOAudioFamily and HDAUniversal:

performFormatChange: 12 occurrences
setFormat returned 0x0: 12 occurrences
setFormat failures: 0

The 20-bit formats were correctly transported inside a 32-bit container:

fBitDepth = 20
fBitWidth = 32

The 24-bit formats also correctly used a 32-bit container:

fBitDepth = 24
fBitWidth = 32

Physical readback proof

The final selected format was 192 kHz, 24-bit, stereo.

The physical stream readback confirmed:

rate=192000
depth=24
width=32
fmt=0x1831
fmtok=1
swrun=1
hwrun=1
pos=!0
cblok=1
bdlok=1

The codec converter confirmed the same format:

stream=0x10
fmt=0x1831
path=0x14>0x0c>0x02
pinctl=0x40
eapd=0x02
complete=1

Real non-zero PCM data was also captured during playback:

nz=7288
peak=35750656
rate=192000
depth=24
fmt=0x1831

Stability results

The Playback Blackbox reported:

playback anomaly: 0
repeated-window: 0
stagnant-position: 0
timeout: 0
underrun: 0
overrun: 0
xrun: 0
clip-guard: 0
clip-normalize: 0
select failed: 0

Conclusion

The complete native output matrix of 44.1, 48, 96 and 192 kHz at 16, 20 and 24-bit passed the format programming, physical readback, DMA and playback stability tests.

No driver, IOAudioFamily, DMA, timeout, underrun, overrun or xrun failures were recorded.

Droga · July 19

Alpha22 · July 20

HDAUniversal.kext test, version 0.1.323, no audio; layout-id <01000000> used in Device-Properties

Codec-Info-Report.txt.zip

MaLd0n · July 20

This version is just for test. There are many unfinished things and bugs.

MaLd0n · August 5

HDAUniversal 1.0.0 — Major Release Update

For a clean and proper uninstallation, use this then reboot then install new version HERE

Problems? Back to previous version! The driver is currently being built, and we will certainly encounter issues with some hardware.

Approximately one month after HDAUniversal 0.1.190, released on July 2, HDAUniversal reaches its first production milestone with version 1.0.0.

The list below includes only changes that remain in the final source. Reverted experiments, temporary diagnostics and superseded implementations have been excluded. Version 1.0.0 freezes the physically tested 0.1.414 / Round178 runtime; the final version promotion itself did not modify the established audio paths.

What Changed Since 0.1.190

Expanded hardware detection

Added support for both PCI multimedia audio classes 0x040300 and 0x040100.
Improved matching for standard analog HDA controllers and GPU digital-audio controllers.
Separate analog and digital controller roles prevent HDMI/DisplayPort devices from interfering with motherboard or laptop audio.
Detection remains dynamic, without motherboard, SMBIOS, fixed PCI-path or codec-address hardcoding.

More complete analog audio support

Dynamic analog codec and widget-topology discovery.
Dynamic DAC-to-pin and ADC-to-input path construction.
Multiple analog outputs supported on the same controller.
Multi-DAC routing and controller-wide route transactions.
Dedicated playback and capture engines.
Support for internal speakers, line out, headphones, internal microphone, external microphone, headset microphone and line input when exposed by the codec.
Reliable input and output selector handling.
Safer route rollback when a hardware transaction cannot be completed.
EAPD, pin control, converter stream and amplifier state are validated through hardware readback.

Playback and PCM corrections

Correct IOAudioFamily mix-ring indexing for 16-, 20-, 24- and 32-bit PCM.
Fixed repeated, reordered and robotic audio caused by incorrect source-ring positioning.
Restored the proven 16,384-frame analog playback ring.
Improved BDL construction, descriptor boundaries and ring-end interrupt handling.
Improved playback reopen behavior after short CoreAudio client sessions.
More reliable DMA start, stop and position tracking.
Better handling of long playback sessions, notification sounds and rapid client attach/detach cycles.
No EQ, DSP, compressor, limiter, resampler or artificial output boost was added.

Volume, mute and control handling

Codec amplifier capabilities are used to generate native volume ranges.
Improved startup volume restoration and cached CoreAudio control replay.
Fixed stale volume commands that could override the current endpoint.
Improved first-unmute and first-volume-change behavior.
Volume and mute commands are coalesced during rapid CoreAudio control bursts.
Warm-idle playback can retain the route without repeatedly shutting down and restarting the complete codec path.
Invalid or unsupported amplifier operations now fail safely instead of leaving a partially programmed state.

Headphones, microphones and combo-jack

Runtime headphone insertion and removal detection.
Unsolicited codec events are used when supported.
Automatic polling fallback is available when unsolicited events are unavailable or unreliable.
Dynamic combo-jack classification and input retasking.
Headset microphone, external microphone and line-input routing.
Pin sense, impedance, VREF and pin capabilities are used to determine the physical connector role.
Debounce and rollback protection prevent unstable jack events from leaving the wrong route active.
Combo-jack and headphone behavior remains topology-driven rather than model-specific.

HDMI, DisplayPort, MST and S/PDIF

Added capability-gated HDMI and DisplayPort audio support.
Separate GPU digital-audio controller discovery.
Dynamic digital codec and converter detection.
ELD and presence-detection validation before publishing a digital output.
Disconnected ports do not receive unnecessary CoreAudio engines or DMA resources.
Support for multiple physical HDMI/DisplayPort endpoints and DP-MST topology.
Late-appearing digital codecs and displays can be discovered without disturbing analog audio.
Digital engines are removed safely when their physical endpoint disappears.
Optional native HDMI/DisplayPort handoff remains available through -nohdmi.
S/PDIF infrastructure is published only when the codec reports the required capabilities.

The current runtime exposes analog audio together with capability-gated HDMI, DisplayPort, DP-MST and S/PDIF paths.

Sleep, wake and anti-pop improvements

Added an analog system-sleep power barrier.
Every active analog path is hard-muted before the controller loses power.
Playback DMA must report both software-stopped state and physical RUN=0.
EAPD, pin output and converter stream assignment are disabled before codec D3.
The complete analog silence barrier is applied again during wake before delayed endpoint activation.
Physical buffers and playback state are sanitized across power transitions.
Added controlled post-wake cold-prime handling before gain and EAPD are released.
Improved wake route restoration, stream rearming and fallback recovery.
Added terminal anti-pop handling for restart and shutdown.
Command transport shutdown now uses bounded CORB/RIRB stop attempts and preserves exact failure telemetry when a controller requires reset recovery.

These changes significantly harden HDA power transitions. Platform-level USB, ACPI and scheduled DarkWake events remain independent of the audio driver and can still wake a computer separately.

Kernel lifecycle and stability

Stronger bounds checking for widgets, pins, paths and fixed-size kernel arrays.
Malformed layout or topology data is rejected before hardware programming.
Improved IORegistry property precedence, including valid zero and false values.
Safer engine, stream, control and port ownership.
Improved interrupt-generation and teardown serialization.
Removed duplicate engine publication and stale endpoint ownership cases.
Safer asynchronous IOAudioPort retirement.
Better cleanup when an engine, route, format change or capture start fails.
Improved controller-wide recovery without abandoning active registry objects.
More reliable unload, restart, format-change and client-close paths.
Failed verb programming is reported instead of being silently ignored.

The previous 0.1.190 release already introduced substantial bounds and malformed-data protection; 1.0.0 extends that work across the complete runtime, power, digital-audio and service-lifecycle paths.

Diagnostics and verification

Exact runtime version and source-fingerprint verification.
Detailed playback DMA, PCM, BCIS and clip-gap telemetry.
Persistent first-start and worst-start latency snapshots.
Absolute worst delay until the first non-zero PCM sample.
Route-authority and endpoint-qualification diagnostics.
Jack polling and unsolicited-event diagnostics.
Optional-codec read isolation and retry telemetry.
Detailed sleep command-transport state, including CORB/RIRB registers and failure reasons.
Dedicated playback blackbox for underrun, overrun, xrun, stall, timeout and deadlock analysis.
Better filtering of unrelated GPU, webcam and normal CoreAudio error 0 messages.

The playback blackbox remains compiled and available, but it is disabled by default in all production personalities.

Codec and layout database

The current embedded database contains:

667 layouts
109 codec definitions
0 duplicate keys
0 validation errors

These entries provide broader automatic coverage, but physical compatibility still depends on the exact codec, board wiring and selected layout.

Release Validation

HDAUniversal 1.0.0 completed:

75/75 configured historical and current validators
123/123 Python syntax checks
23/23 shell command checks
7/7 C++ translation units compiled with warnings treated as errors
7/7 optimized Mach-O x86_64 objects generated
0 compiler warnings
0 compiler errors
Valid Info.plist and Xcode project structure
Clean production configuration without HIL fault injection

Important Notes

HDAUniversal remains a hardware-driven audio implementation. Features are published only when the controller, codec, widgets, pins and connected digital sink provide sufficient physical evidence.

No fixed motherboard rules, forced DeviceProperties, ACPI patches, artificial volume boosts or sound-processing effects are included.

®Daniel Maldonado — Olarila.com

nosborsousa · August 5

Conseguiu ajustar para não mostrar duplicados os dispositivos no painel de controle do som ? Pois no meu S145 8Th vc tinha ajustado layout 171 pra eu usar, mas tava com esse problema de mostrar duplicado, apesar de está funcionando perfeitamente e com extrema qualidade de som.

Alpha22 · August 5

Hello MaLd0n

HDAUniversal.kext test, version 1.0.0 It doesn't work when using DisplayPort

Codec-Info-Report.txt.zip

nosborsousa · August 5

3 hours ago, nosborsousa said:

Conseguiu ajustar para não mostrar duplicados os dispositivos no painel de controle do som ? Pois no meu S145 8Th vc tinha ajustado layout 171 pra eu usar, mas tava com esse problema de mostrar duplicado, apesar de está funcionando perfeitamente e com extrema qualidade de som.

Testei aqui a nova versão e ainda estão duplicados. Audio continua funcionando com extrema qualidade.

Droga · August 5

Hello MaLd0n,

I'm testing the latest update of HDAUniversal on my ASUS ROG Crosshair VII Hero (X470 / Realtek ALC1220 - SupremeFX S1220) on macOS Tahoe (26.6).

The previous version of HDAUniversal worked great on my board, but after installing the latest update into /Library/Extensions/ and rebuilding the AuxKC (kmutil install), macOS fails to load HDAUniversal at boot and falls back to AppleHDAController.

I checked the system boot logs (log show --last boot), and kernelmanagerd / kernelmanager_helper throws the following static linking error during the AuxKC build phase:

building auxiliary kext collection failed: Error occurred while building a collection: Superclass vtable '__ZTV8OSObject' is not exported from 'com.apple.driver.AppleHDA' or its dependencies

Because of this C++ vtable symbol export error, kernelmanagerd cancels the AuxKC compilation for HDAUniversal.kext on macOS Tahoe when installed in /Library/Extensions/.

Also, I noticed in IORegistry that the onboard HD Audio controller on my AMD X470 motherboard is named AZAL (/_SB/PCI0/GP13/AZAL), while Info.plist matches HDEF, B0D3, and HDAU.

Could you please check this vtable dependency issue and consider adding AZAL to IONameMatch in the next build?

Thank you so much for your awesome work and dedication!

Sign In