diff mbox series

arm: dts: imx7d-sdb: Fix enet by setting soft spi mode to 3 for extended io

Message ID 20210219072358.20157-1-yocto@bsorensen.net
State Changes Requested
Delegated to: Stefano Babic
Headers show
Series arm: dts: imx7d-sdb: Fix enet by setting soft spi mode to 3 for extended io | expand

Commit Message

Christian Sørensen Feb. 19, 2021, 7:23 a.m. UTC
This patch sets SPI MODE 3 for the fairchild 74HC595 device that is used
to reset the ethernet device, which fixes ethernet.

Reasoning:
The imx7dsabresd board uses SPI for setting the extended IO.
One of the purposes of this extended IO, is having the ability to set
ENET_RST i.e. reset the ethernet PHY. Without proper reset of the PHY,
ethernet does not work.

commit 0e146993bb3d ("spi: add support for all spi modes with soft spi")
broke network support for boards using imx7d-sdb, since it actually
corrected default SPI mode setting for the software emulated soft_spi to 0,
i.e. CPHA to 0 and CPOL to 0.
Previously, the driver defaulted to CPHA 1 and CPOL 1 (i.e. spi mode 3).
After the default SPI mode is now 0 (which is the correct default value)
we now need to actively set CPHA and CPOL to 1 for the specific devices
that actually needs this.

Signed-off-by: Christian Bräuner Sørensen <yocto@bsorensen.net>
---

 arch/arm/dts/imx7d-sdb.dts | 2 ++
 1 file changed, 2 insertions(+)

Comments

Fabio Estevam Feb. 19, 2021, 1:44 p.m. UTC | #1
Hi Christian,

On Fri, Feb 19, 2021 at 10:32 AM Christian Bräuner Sørensen
<yocto@bsorensen.net> wrote:
>
> This patch sets SPI MODE 3 for the fairchild 74HC595 device that is used
> to reset the ethernet device, which fixes ethernet.
>
> Reasoning:
> The imx7dsabresd board uses SPI for setting the extended IO.
> One of the purposes of this extended IO, is having the ability to set
> ENET_RST i.e. reset the ethernet PHY. Without proper reset of the PHY,
> ethernet does not work.
>
> commit 0e146993bb3d ("spi: add support for all spi modes with soft spi")
> broke network support for boards using imx7d-sdb, since it actually

Shouldn't this commit be reverted then?

I suppose it introduced regressions on other boards too.

> corrected default SPI mode setting for the software emulated soft_spi to 0,
> i.e. CPHA to 0 and CPOL to 0.
> Previously, the driver defaulted to CPHA 1 and CPOL 1 (i.e. spi mode 3).
> After the default SPI mode is now 0 (which is the correct default value)
> we now need to actively set CPHA and CPOL to 1 for the specific devices
> that actually needs this.
>
> Signed-off-by: Christian Bräuner Sørensen <yocto@bsorensen.net>
> ---
>
>  arch/arm/dts/imx7d-sdb.dts | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/arch/arm/dts/imx7d-sdb.dts b/arch/arm/dts/imx7d-sdb.dts
> index 8191ac7c33..6ed21ce8b0 100644
> --- a/arch/arm/dts/imx7d-sdb.dts
> +++ b/arch/arm/dts/imx7d-sdb.dts
> @@ -58,6 +58,8 @@
>                         reg = <0>;
>                         registers-number = <1>;
>                         spi-max-frequency = <100000>;
> +                       spi-cpha
> +                       spi-cpol;

It is not good that we deviate from the devicetree used in Linux.

Johannes/Jagan,

Any suggestions as to how to properly fix the regression?

Thanks
diff mbox series

Patch

diff --git a/arch/arm/dts/imx7d-sdb.dts b/arch/arm/dts/imx7d-sdb.dts
index 8191ac7c33..6ed21ce8b0 100644
--- a/arch/arm/dts/imx7d-sdb.dts
+++ b/arch/arm/dts/imx7d-sdb.dts
@@ -58,6 +58,8 @@ 
 			reg = <0>;
 			registers-number = <1>;
 			spi-max-frequency = <100000>;
+			spi-cpha;
+			spi-cpol;
 		};
 	};