[PATCH] configs: am62x_evm_*: Fix USB DFU configuration

[Siddharth Vadapalli]

On Fri, Dec 06, 2024 at 11:17:28AM +0200, Roger Quadros wrote:
> Hello Siddharth,
> 
> On 06/12/2024 09:19, Siddharth Vadapalli wrote:

[...]

> > While this fixes the issue, I am wondering if the issue lies elsewhere.
> > In U-Boot, the compatible "snps,dwc3" is associated with:
> > drivers/usb/host/xhci-dwc3.c
> > while in Linux, the compatible "snps,dwc3" is associated with:
> > drivers/usb/dwc3/core.c
> > 
> > So there seem to be two alternatives that I could think of:
> > 1. Modify U-Boot to match Linux in the sense that we associate
> > "snps,dwc3" with drivers/usb/dwc3/core.c in U-Boot.
> 
> Many platforms (grep for CONFIG_USB_XHCI_DWC3
> in configs/) use xhci-dwc3 to get Host mode working
> without the need for core.c. Maybe it was more simpler that way?
> Also see USB_XHCI_DWC3_OF_SIMPLE.
> 
> So if we drop "snps,dwc3" from xhci-dwc3, we will have to ensure each
> and every platform works.

Yes, that will require significant effort and testing.

> 
> > 2. With the understanding that "dr_mode" doesn't have to be host/otg for
> > the compatible "snps,dwc3" which is tied to drivers/usb/host/xhci-dwc3.c,
> > do the following to exit probe when "dr_mode" is "peripheral":
> > -------------------------------------------------------------------------
> > diff --git a/drivers/usb/host/xhci-dwc3.c b/drivers/usb/host/xhci-dwc3.c
> > index e3e0ceff43e..edfd7b97a73 100644
> > --- a/drivers/usb/host/xhci-dwc3.c
> > +++ b/drivers/usb/host/xhci-dwc3.c
> > @@ -208,6 +208,8 @@ static int xhci_dwc3_probe(struct udevice *dev)
> >         writel(reg, &dwc3_reg->g_usb2phycfg[0]);
> > 
> >         dr_mode = usb_get_dr_mode(dev_ofnode(dev));
> > +       if (dr_mode == USB_DR_MODE_PERIPHERAL)
> > +               return -ENODEV;
> >         if (dr_mode == USB_DR_MODE_OTG &&
> >             dev_read_bool(dev, "usb-role-switch")) {
> >                 dr_mode = usb_get_role_switch_default_mode(dev_ofnode(dev));
> > -------------------------------------------------------------------------
> > which will still show "xhci-dwc3" in the output of "dm tree" after "usb
> > start", but the driver won't be probed (absence of "+" in the "Probed"
> > column of "dm tree" output).
> 
> I agree with your second proposal.
> But it is a separate bug (in xhci-dwc3 driver) which also
> fixes the issue this patch is fixing.
> 
> Regardless, I think both fixes should go in.
> 
> Could you please send a patch to fix xhci-dwc3? Thanks!

Sure. Thank you for reviewing the suggestions and sharing your feedback.

Regards,
Siddharth.