mbox series

[0/3] Add mclk0 clock source for SAI

Message ID 20190528132034.3908-1-daniel.baluta@nxp.com
Headers show
Series Add mclk0 clock source for SAI | expand

Message

Daniel Baluta May 28, 2019, 1:20 p.m. UTC 
This patch series brings together patches [1] and [2] which
introduce mclk0 clock source via DT.

mclk0..3 are the four clock sources options
of SAI's clock MUX. 

mclk0 option selects:
 - Bus Clock on i.MX8
 - MCLK1 on i.MX6/7

Finally we also update the DT binding information for SAI clocks.

In [1] and [2] Nicolin had a very good point on the fact that
mclk0 might not be needed in the DT. Anyhow, there are two reasons
for which I think mlck0 should be added to DT:

1) SAI clock source select MUX is really part of the hardware
2) flexibility! We let DT tell us which is the option for MUX
option 0.


[1] lkml.org/lkml/2019/4/20/141
[2] lkml.org/lkml/2019/4/20/56

Daniel Baluta (2):
  dt-bindings: sound: Clarify the usage of clocks in SAI
  ASoC: fsl_sai: Read SAI clock source 0 from DT

Shengjiu Wang (1):
  ARM: dts: imx: Add mclk0 clock for SAI

 Documentation/devicetree/bindings/sound/fsl-sai.txt | 5 +++--
 arch/arm/boot/dts/imx6sx.dtsi                       | 6 ++++--
 arch/arm/boot/dts/imx6ul.dtsi                       | 9 ++++++---
 arch/arm/boot/dts/imx7s.dtsi                        | 9 ++++++---
 sound/soc/fsl/fsl_sai.c                             | 3 +--
 5 files changed, 20 insertions(+), 12 deletions(-)

Comments

Nicolin Chen May 29, 2019, 10:57 p.m. UTC | #1 
On Tue, May 28, 2019 at 01:21:00PM +0000, Daniel Baluta wrote:
> From: Shengjiu Wang <shengjiu.wang@nxp.com>
> 
> Audio MCLK source option is selected with a 4:1 MUX
> controller using MCLK Select bits in SAI xCR2 register.
> 
> On imx6/7 mclk0 and mclk1 always point to the same clock
> source. Anyhow, this is no longer true for imx8.
> 
> For this reason, we need to add mclk0 and handle it
> in a generic way in SAI driver.
> 
> Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
> Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com>
> ---
>  arch/arm/boot/dts/imx6sx.dtsi | 6 ++++--
>  arch/arm/boot/dts/imx6ul.dtsi | 9 ++++++---
>  arch/arm/boot/dts/imx7s.dtsi  | 9 ++++++---

These are dtsi/dts files that have SAI missing mclk0:
arch/arm/boot/dts/imx6ul.dtsi
arch/arm/boot/dts/imx6sx.dtsi
arch/arm/boot/dts/ls1021a.dtsi
arch/arm/boot/dts/imx7s.dtsi
arch/arm/boot/dts/vfxxx.dtsi
arch/arm64/boot/dts/freescale/imx8mq.dtsi
arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi
arch/arm64/boot/dts/freescale/fsl-ls1012a.dtsi
arch/arm64/boot/dts/freescale/fsl-ls1012a.dtsi

Would it be possible for you to update the others also?

Thanks
Nicolin
Nicolin Chen May 29, 2019, 11:03 p.m. UTC | #2 
On Tue, May 28, 2019 at 01:20:46PM +0000, Daniel Baluta wrote:

> 1) SAI clock source select MUX is really part of the hardware
> 2) flexibility! We let DT tell us which is the option for MUX
> option 0.

I think the "MUX" is plausible comparing to your previous version.
As long as DT maintainers ack the DT binding doc, I would be okay
to ack too. Just one comments at the dts/dtsi changes, I know the
driver would just warn old DTs, but it does change the behavior at
the mclk_clk[0] from previously bus_clk to NULL after this series.
Daniel Baluta May 30, 2019, 8:32 a.m. UTC | #3 
On Wed, 2019-05-29 at 16:03 -0700, Nicolin Chen wrote:
> On Tue, May 28, 2019 at 01:20:46PM +0000, Daniel Baluta wrote:
> 
> > 1) SAI clock source select MUX is really part of the hardware
> > 2) flexibility! We let DT tell us which is the option for MUX
> > option 0.
> 
> I think the "MUX" is plausible comparing to your previous version.
> As long as DT maintainers ack the DT binding doc, I would be okay
> to ack too. Just one comments at the dts/dtsi changes, I know the
> driver would just warn old DTs, but it does change the behavior at
> the mclk_clk[0] from previously bus_clk to NULL after this series.

Thanks for review, will fix in v2.