Re: [PATCH v2 11/13] mmc: sdhci-esdhc-imx: Add sdhc support for i.MXRT series

Hello Rob, Jesse, All,

Il giorno 4 nov 2021, alle ore 02:05, Rob Herring [off-list ref] ha scritto:

On Wed, Nov 03, 2021 at 12:30:17AM +0100, Giulio Benetti wrote:

quoted

Hi Fabio, Jesse, All,

quoted

On 11/3/21 12:25 AM, Jesse Taube wrote:


On 11/2/21 19:17, Fabio Estevam wrote:

quoted

On Tue, Nov 2, 2021 at 7:57 PM Jesse Taube [off-list ref] wrote:

quoted

  static struct esdhc_soc_data usdhc_imx8qxp_data = {
         .flags = ESDHC_FLAG_USDHC | ESDHC_FLAG_STD_TUNING

@@ -357,6 +363,7 @@ static const struct of_device_id imx_esdhc_dt_ids[] = {
         { .compatible = "fsl,imx7ulp-usdhc", .data = &usdhc_imx7ulp_data, },
         { .compatible = "fsl,imx8qxp-usdhc", .data = &usdhc_imx8qxp_data, },
         { .compatible = "fsl,imx8mm-usdhc", .data = &usdhc_imx8mm_data, },
+       { .compatible = "fsl,imxrt-usdhc", .data = &usdhc_imxrt_data, },

I thought Rob suggested to use the SoC name, so this would be:

Uh i think that may have been for the UART.

quoted

{ .compatible = "fsl,imxrt1050-usdhc", .data = &usdhc_imxrt1050_data, },

The same applies to the other bindings in the series.

This way it would be possible to differentiate between future
supported i.MX RT devices.

This makes sense will do in V3.

If we add every SoC we will end up having a long list for every device
driver. At the moment it would be 7 parts:
1) imxrt1020
2) imxrt1024
.
.
.
7) imxrt1170

You don't need a driver update if you use a fallback. When you add 
the 2nd chip, if you think it is 'the same', then you do:

compatible = "fsl,imxrt1024-usdhc", "fsl,imxrt1050-usdhc";

That requires no driver update until the driver needs to handle some 
difference. And when there is a difference, you don't need a DT update.

This solution is pretty fine, we’re going with that then, for this and every driver involved.

Thank you for pointing us.

Best regards
Giulio Benetti
Benetti Engineering sas

You could make "fsl,imxrt-usdhc" the fallback from the start if you are 
adverse to the first way.

Rob