Re: [RFC PATCH 7/8] dt-bindings: sound: qcom: add Tambora WCD9378 SDCA codec

[Jorijn van der Graaf]

On 7/29/26 4:04 PM, Krzysztof Kozlowski wrote:
> All this should be part of TX.

I have prototyped this with my mobile-mode wcd9378 driver for the
FP6. For the paths my driver implements (TX/capture), the following
works:

- the TX slave node owns reset-gpios, the supplies and the mic-bias
  voltages; the RX slave is referenced from the TX node by phandle;
- like the wsa883x/wsa8840 speaker amps (powerdown GPIO and supplies
  on the slave node), the TX slave driver does the supply/reset
  bring-up itself from its probe, and both slaves then enumerate
  normally;
- the ASoC component and DAIs register on the TX slave device; the
  aggregation is TX-over-RX instead of a third device over both.

Verified on the FP6 across two boots: both slaves enumerate, the card
registers, and handset-mic capture passes the same kind of 1 kHz
speaker-to-mic round-trip check used to validate [1]. However RX playback and
MBHC are not implemented in my driver in either shape, so I cannot
speak to those paths.

Since, as agreed with Srinivas, [1] already carries this binding:
whatever shape is decided here, I would like to respin [1] to match
rather than have the two threads resolve differently.

[1] https://lore.kernel.org/all/20260729000649.263426-1-jorijnvdgraaf@xxxxxxxxxxxxx/

Prototype: commit 7b720141754b, the top commit of
https://forgejo.catcrafts.net/Catcrafts/milos-linux/src/branch/wcd9378-parentless-proto
(milos v7.1.2 integration tree; the parentless conversion is that one
commit on top of the posted v2 driver).

Jorijn