DMIC support on AMD RPL platform produces I/O error after quirk addition

DMIC support on AMD RPL platform produces I/O error after quirk addition

[Jason Mo]

Approximately one year ago, I raised a concern regarding the lack of DMIC support for AMD Raphael (RPL) platforms in the following thread:
https://lore.kernel.org/linux-sound/BY5PR07MB66126978095912E5D4A26E50A1802@BY5PR07MB6612.namprd07.prod.outlook.com/T/#u

Recently, I noticed the patch series submitted at:
https://lore.kernel.org/linux-sound/20260213055904.110256-1-qby140326@gmail.com/T/#u
which adds initial DMIC support for these platforms. After applying the relevant changes and adding my device to the quirk list, the DMIC capture device is now correctly enumerated by the kernel.

However, when attempting to record audio, the operation fails with an I/O error. Below are the reproduction steps and corresponding logs from a Gentoo live environment:

gentoo@livecd ~ $ arecord -l
**** List of CAPTURE Hardware Devices ****
card 2: Generic_1 [HD-Audio Generic], device 0: SN6140 Analog [SN6140 Analog]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 3: acp6x [acp6x], device 0: DMIC capture dmic-hifi-0 []
  Subdevices: 1/1
  Subdevice #0: subdevice #0
gentoo@livecd ~ $ arecord -D hw:3,0 -f S32_LE -r 48000 -c 2 -d 10 /tmp/test.wav
Recording WAVE '/tmp/test.wav' : Signed 32 bit Little Endian, Rate 48000 Hz, Stereo
arecord: pcm_read:2272: read error: Input/output error
gentoo@livecd ~ $

The device node appears functional at the ALSA layer, but data transfer fails during capture. No obvious dmesg errors appear during device initialization beyond the I/O failure.

Could you kindly advise whether further logs (e.g., ftrace, ACP debug registers) would be helpful for diagnosis?
I am happy to provide additional system information, test patches, or perform targeted debugging as needed.

thanks,
Jason Mo

Re: DMIC support on AMD RPL platform produces I/O error after quirk addition

[Mukunda,Vijendar]

On 19/02/26 19:50, Jason Mo wrote:
> Approximately one year ago, I raised a concern regarding the lack of DMIC support for AMD Raphael (RPL) platforms in the following thread:
> https://lore.kernel.org/linux-sound/BY5PR07MB66126978095912E5D4A26E50A1802@BY5PR07MB6612.namprd07.prod.outlook.com/T/#u
>
> Recently, I noticed the patch series submitted at:
> https://lore.kernel.org/linux-sound/20260213055904.110256-1-qby140326@gmail.com/T/#u
> which adds initial DMIC support for these platforms. After applying the relevant changes and adding my device to the quirk list, the DMIC capture device is now correctly enumerated by the kernel.
>
> However, when attempting to record audio, the operation fails with an I/O error. Below are the reproduction steps and corresponding logs from a Gentoo live environment:
>
> gentoo@livecd ~ $ arecord -l
> **** List of CAPTURE Hardware Devices ****
> card 2: Generic_1 [HD-Audio Generic], device 0: SN6140 Analog [SN6140 Analog]
>   Subdevices: 1/1
>   Subdevice #0: subdevice #0
> card 3: acp6x [acp6x], device 0: DMIC capture dmic-hifi-0 []
>   Subdevices: 1/1
>   Subdevice #0: subdevice #0
> gentoo@livecd ~ $ arecord -D hw:3,0 -f S32_LE -r 48000 -c 2 -d 10 /tmp/test.wav
> Recording WAVE '/tmp/test.wav' : Signed 32 bit Little Endian, Rate 48000 Hz, Stereo
> arecord: pcm_read:2272: read error: Input/output error
Typically, I/O errors will be reported when DMA is not progressing.
RPL based platforms are intended for desktop variants, and DMIC is not
expected to be present on these platforms.  Before proceeding with further
debugging, we would like to double-check whether DMIC actually exists on this
platform.
Additionally, please log a kernel Bugzilla ticket and include detailed
information such as
Alsa info dump, lspci -vvv output, ACPI dump and dmidecode output. 

-
Vijendar

> gentoo@livecd ~ $
>
> The device node appears functional at the ALSA layer, but data transfer fails during capture. No obvious dmesg errors appear during device initialization beyond the I/O failure.
>
> Could you kindly advise whether further logs (e.g., ftrace, ACP debug registers) would be helpful for diagnosis?
> I am happy to provide additional system information, test patches, or perform targeted debugging as needed.
>
> thanks,
> Jason Mo

Re: DMIC support on AMD RPL platform produces I/O error after quirk addition

[Jason Mo]

On 24/02/26 17:42, Mukunda,Vijendar <vijendar.mukunda@amd.com> wrote:
> Typically, I/O errors will be reported when DMA is not progressing.
> RPL based platforms are intended for desktop variants, and DMIC is not
> expected to be present on these platforms.  Before proceeding with further
> debugging, we would like to double-check whether DMIC actually exists on this
> platform.
> Additionally, please log a kernel Bugzilla ticket and include detailed
> information such as
> Alsa info dump, lspci -vvv output, ACPI dump and dmidecode output. 

Hi Vijendar,

Thanks for your reply. I have created a Bugzilla ticket for this issue 
and attached all the requested logs (ALSA info, lspci, ACPI dump, and 
dmidecode).

You can find the ticket here: 
https://bugzilla.kernel.org/show_bug.cgi?id=221132

Please let me know if any further information is needed.

Thanks,
Jason