On 27/05/2022 21:05, Nícolas F. R. A. Prado wrote: > On Thu, May 26, 2022 at 08:49:39AM +0200, Krzysztof Kozlowski wrote: >> On 09/05/2022 22:58, Nícolas F. R. A. Prado wrote: >>> The Mediatek AFE PCM controller for MT8192 allows two I2S interfaces to >>> share the same clock and act as a single interface with both input and >>> output. Add patterns for these properties in the dt-binding. The >>> property is split into two patterns in order to allow all valid >>> interface pairings. >>> >>> Signed-off-by: Nícolas F. R. A. Prado <nfraprado@xxxxxxxxxxxxx> >>> >>> --- >>> The series from v1 of this patch was merged although some changes were >>> still needed in this patch, so the v1 of this patch was reverted [1] and >>> this standalone commit addresses the feedback from v1 and readds the >>> property. >>> >>> [1] https://lore.kernel.org/all/20220509185625.580811-1-nfraprado@xxxxxxxxxxxxx >>> >>> v1: https://lore.kernel.org/all/20220429203039.2207848-2-nfraprado@xxxxxxxxxxxxx/ >>> >>> Changes in v2: >>> - Added "mediatek," prefix to property >>> - Rewrote and added more information to property description >>> - Split into two patterns to validate that output-input pairings are >>> done >>> >>> .../bindings/sound/mt8192-afe-pcm.yaml | 16 ++++++++++++++++ >>> 1 file changed, 16 insertions(+) >>> >>> diff --git a/Documentation/devicetree/bindings/sound/mt8192-afe-pcm.yaml b/Documentation/devicetree/bindings/sound/mt8192-afe-pcm.yaml >>> index 7a25bc9b8060..2abf43c6c2c3 100644 >>> --- a/Documentation/devicetree/bindings/sound/mt8192-afe-pcm.yaml >>> +++ b/Documentation/devicetree/bindings/sound/mt8192-afe-pcm.yaml >>> @@ -54,6 +54,22 @@ properties: >>> - const: aud_infra_clk >>> - const: aud_infra_26m_clk >>> >>> +patternProperties: >>> + "^mediatek,i2s[13579]-share$": >>> + description: >>> + Each I2S interface has a single data line, input if its index is even or >>> + output if the index is odd. An input and an output I2S interface can be >>> + used together as if they were a single I2S interface with both input and >>> + output data lines by sharing the same clock. This property represents this >>> + pairing. The value should be the name of the interface whose clock is >>> + used, and the property name the other interface that depends on this >>> + clock. >>> + pattern: "^I2S[0268]$" >>> + >>> + "^mediatek,i2s[0268]-share$": >>> + description: Same as above. >>> + pattern: "^I2S[13579]$" >> >> Rob's question is still valid - why these are not phandles? > > So, instead of having > > i2s9-share = "I2S8"; > > on the DT, you want us to have something like this: > > afe_i2s8: mediatek,i2s8 { }; > > mediatek,i2s9 { > mediatek,share-clock = <&afe_i2s8>; > }; > > Or do you mean something else? Not entirely. This is a binding for mediatek,mt8192-audio, not for I2S9, so the property should be in this binding referencing your existing I2S devices. > > It seems like a lot more syntax to express the same thing (and the empty node > seems awkward), but if that's the DT way, I can change it no problem. Hm, then the follow up question: why you do not have i2s8 defined in DTS? Best regards, Krzysztof
