usb: wrong compatible string for MX23?

usb: wrong compatible string for MX23?

[Stefan Wahren]

Hi Peter,

i was inspired by the discussion about the USB runtime suspend issue on MX28. I
think i've found another issue.
In the dtsi for the MX23 the usb node has the following compatible:

usb0: usb at 80080000 {
        compatible = "fsl,imx23-usb", "fsl,imx27-usb";
...

But in ci_hdrc_imx.c there is only "fsl,imx27-usb" and "fsl,imx28-usb" defined.

So in case of a MX23 board it would fallback to "fsl,imx27-usb", which is IMHO
not correct.

Which way to fix this issue would you suggest:

a) extend ci_hdrc_imx.c and the binding by copying the imx28 platform and name
it imx23

b) rename the compatible string from "fsl,imx23-usb" to "fsl-imx28-usb"

Best regards
Stefan

usb: wrong compatible string for MX23?

[Peter Chen]

On Tue, Dec 29, 2015 at 12:45:51AM +0100, Stefan Wahren wrote:
> Hi Peter,
> 
> i was inspired by the discussion about the USB runtime suspend issue on MX28. I
> think i've found another issue.
> In the dtsi for the MX23 the usb node has the following compatible:
> 
> usb0: usb at 80080000 {
>         compatible = "fsl,imx23-usb", "fsl,imx27-usb";
> ...
> 
> But in ci_hdrc_imx.c there is only "fsl,imx27-usb" and "fsl,imx28-usb" defined.
> 
> So in case of a MX23 board it would fallback to "fsl,imx27-usb", which is IMHO
> not correct.

No, it is correct. All i.mx SoC series use the same controller core, and
the first user is imx27, the reason why we have a compatible string for
imx28 is that imx28 needs special handling when write register.

> 
> Which way to fix this issue would you suggest:
> 
> a) extend ci_hdrc_imx.c and the binding by copying the imx28 platform and name
> it imx23
> 
> b) rename the compatible string from "fsl,imx23-usb" to "fsl-imx28-usb"
> 

But one thing we can improve that is imx23 needs the flag
CI_HDRC_TURN_VBUS_EARLY_ON, the platform flag would like below:

static const struct ci_hdrc_imx_platform_flag imx23_usb_data = {
	.flags = CI_HDRC_TURN_VBUS_EARLY_ON |
		CI_HDRC_DISABLE_STREAMING,

I prefer the way a), thanks.

-- 

Best Regards,
Peter Chen