Re: [PATCH v1 1/6] usb: dwc3: core: only provide legacy gadget IRQ handler when USB_DWC3_GADGET

On Wed, Aug 26, 2026 at 09:15:01AM +0200, Stefan Eichenberger wrote:
> On Tue, Aug 25, 2026 at 01:33:27AM +0200, Marek Vasut wrote:
> > On 8/24/26 8:29 PM, Stefan Eichenberger wrote:
> > > From: Stefan Eichenberger <[email protected]>
> > > 
> > > Building a board with USB_DWC3=y (DWC3 used purely as an XHCI host
> > > controller), USB_DWC3_GADGET unset, and DM_USB_GADGET disabled (e.g.
> > > to use CONFIG_CI_UDC for a separate ChipIdea gadget/OTG controller)
> > > fails to link:
> > > 
> > > drivers/usb/gadget/ci_udc.o: in function `dm_usb_gadget_handle_interrupts':
> > > drivers/usb/gadget/ci_udc.c:983: multiple definition of `dm_usb_gadget_handle_interrupts';
> > > drivers/usb/dwc3/core.o:drivers/usb/dwc3/core.c:1034: first defined here
> > > drivers/usb/dwc3/core.o: in function `dm_usb_gadget_handle_interrupts':
> > > drivers/usb/dwc3/core.c:1044: undefined reference to `dwc3_gadget_uboot_handle_interrupt'
> > > 
> > > Add "&& CONFIG_IS_ENABLED(USB_DWC3_GADGET)" to the guard so this code
> > > is only compiled when the DWC3 gadget driver it depends on is actually
> > > present, matching the Makefile dependency. Boards that combine DWC3
> > > (host) with a separate legacy gadget driver like CI_UDC, with
> > > DM_USB_GADGET disabled, now link correctly. Boards that already use
> > > DWC3 in gadget mode without DM_USB_GADGET are unaffected since
> > > USB_DWC3_GADGET is set for them.
> > > 
> > > Fixes: 4d1589808970 ("usb: udc: dwc3: Fold board dm_usb_gadget_handle_interrupts() into DWC3 gadget")
> > > Signed-off-by: Stefan Eichenberger <[email protected]>
> > > ---
> > >   drivers/usb/dwc3/core.c | 2 +-
> > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
> > > index be198041f08..9bff5149c5f 100644
> > > --- a/drivers/usb/dwc3/core.c
> > > +++ b/drivers/usb/dwc3/core.c
> > > @@ -1016,7 +1016,7 @@ MODULE_AUTHOR("Felipe Balbi <[email protected]>");
> > >   MODULE_LICENSE("GPL v2");
> > >   MODULE_DESCRIPTION("DesignWare USB3 DRD Controller Driver");
> > > -#if !CONFIG_IS_ENABLED(DM_USB_GADGET)
> > > +#if !CONFIG_IS_ENABLED(DM_USB_GADGET) && CONFIG_IS_ENABLED(USB_DWC3_GADGET)
> > >   __weak int dwc3_uboot_interrupt_status(struct udevice *dev)
> > Simply enable DM_USB_GADGET , then the drivers/usb/gadget/udc/udc-uclass.c
> > dm_usb_gadget_handle_interrupts() is used which correctly invokes the
> > per-controller IRQ handler. DWC3_GADGET should likely depend on
> > DM_USB_GADGET.
> 
> Unfortunately, CI_UDC is not a DM driver yet and it depends on
> !DM_USB_GADGET that's why I can't enable DM_USB_GADGET in that specific
> case. I think NXP is working on changing that driver to DM (at least
> they did that in their BSP). So my hope was to have this as intermediate
> solution. Also in our case we use the DWC3 controller in host mode only,
> so we don't need DWC3_GADGET at all.

So the NXP downstream tree has a DM_USB_GADGET driver that needs to be
upstreamed, to solve this? We're quite a ways past where enhancing the
non-DM side things is the normal workflow.

-- 
Tom

-----BEGIN PGP SIGNATURE-----

iHUEABYKAB0WIQTzzqh0PWDgGS+bTHor4qD1Cr/kCgUCao8QDAAKCRAr4qD1Cr/k
CoUFAP9S8+xhUqTYm/mFedP7lCmO2X520eQLmB94bi5oQ7rb0QEA+S7ZnnExqRiU
GGheKVfIv2UzalrzXatku1m48ftcagQ=
=ZQbH
-----END PGP SIGNATURE-----