Re: [PATCH] usb: dwc2: handle OTG HNP SetFeature requests

[NG, TZE YEE]

On 17/8/2026 4:00 pm, NG, TZE YEE wrote:
> From: Adrian Ng Ho Yin <adrian.ho.yin.ng@xxxxxxxxxx>
> 
> Without handling for b_hnp_enable, a_hnp_support, and a_alt_hnp_support,
> HNP cannot be enabled when two OTG controllers are connected.
> 
> Handle SetFeature for these OTG selectors, gated on otg_caps.hnp_support.
> The OTG specification only defines SetFeature for them, so reject
> ClearFeature with -EINVAL instead of silently accepting it. Use dev_dbg
> for status updates to avoid noisy logs.
> 
> Advertise the core's HNP/SRP capability during gadget init by restoring
> GUSBCFG_HNPCAP/SRPCAP from otg_caps, mirroring dwc2_gusbcfg_init() on the
> host side, instead of clearing them unconditionally.
> 
> Clear the HNP flags and GOTGCTL_DEVHNPEN on disconnect/reset so OTG state
> does not leak across sessions.
> 
> Signed-off-by: Adrian Ng Ho Yin <adrian.ho.yin.ng@xxxxxxxxxx>
> Signed-off-by: Tze Yee Ng <tze.yee.ng@xxxxxxxxxx>
> ---
>   drivers/usb/dwc2/gadget.c | 51 +++++++++++++++++++++++++++++++++++++--
>   1 file changed, 49 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/usb/dwc2/gadget.c b/drivers/usb/dwc2/gadget.c
> index c8b02c27d27d..4d905989c09f 100644
> --- a/drivers/usb/dwc2/gadget.c
> +++ b/drivers/usb/dwc2/gadget.c
> @@ -1780,6 +1780,7 @@ static int dwc2_hsotg_process_req_feature(struct dwc2_hsotg *hsotg,
>   	struct dwc2_hsotg_ep *ep;
>   	int ret;
>   	bool halted;
> +	u32 otgctl;
>   	u32 recip;
>   	u32 wValue;
>   	u32 wIndex;
> @@ -1809,6 +1810,36 @@ static int dwc2_hsotg_process_req_feature(struct dwc2_hsotg *hsotg,
>   
>   			hsotg->test_mode = wIndex >> 8;
>   			break;
> +		case USB_DEVICE_B_HNP_ENABLE:
> +			if (!hsotg->params.otg_caps.hnp_support)
> +				return -ENOENT;
> +			if (!set)
> +				return -EINVAL;
> +
> +			otgctl = dwc2_readl(hsotg, GOTGCTL);
> +			otgctl |= GOTGCTL_DEVHNPEN;
> +			dwc2_writel(hsotg, otgctl, GOTGCTL);
> +			hsotg->gadget.b_hnp_enable = 1;
> +			dev_dbg(hsotg->dev, "HNP enabled\n");
> +			break;
> +		case USB_DEVICE_A_HNP_SUPPORT:
> +			if (!hsotg->params.otg_caps.hnp_support)
> +				return -ENOENT;
> +			if (!set)
> +				return -EINVAL;
> +
> +			hsotg->gadget.a_hnp_support = 1;
> +			dev_dbg(hsotg->dev, "a_hnp_support set\n");
> +			break;
> +		case USB_DEVICE_A_ALT_HNP_SUPPORT:
> +			if (!hsotg->params.otg_caps.hnp_support)
> +				return -ENOENT;
> +			if (!set)
> +				return -EINVAL;
> +
> +			hsotg->gadget.a_alt_hnp_support = 1;
> +			dev_dbg(hsotg->dev, "a_alt_hnp_support set\n");
> +			break;
>   		default:
>   			return -ENOENT;
>   		}
> @@ -3322,6 +3353,13 @@ void dwc2_hsotg_disconnect(struct dwc2_hsotg *hsotg)
>   	hsotg->connected = 0;
>   	hsotg->test_mode = 0;
>   
> +	if (hsotg->params.otg_caps.hnp_support) {
> +		hsotg->gadget.b_hnp_enable = 0;
> +		hsotg->gadget.a_hnp_support = 0;
> +		hsotg->gadget.a_alt_hnp_support = 0;
> +		dwc2_clear_bit(hsotg, GOTGCTL, GOTGCTL_DEVHNPEN);
> +	}
> +
>   	/* all endpoints should be shutdown */
>   	for (ep = 0; ep < hsotg->num_of_eps; ep++) {
>   		if (hsotg->eps_in[ep])
> @@ -3416,9 +3454,18 @@ void dwc2_hsotg_core_init_disconnected(struct dwc2_hsotg *hsotg,
>   	usbcfg &= ~GUSBCFG_TOUTCAL_MASK;
>   	usbcfg |= GUSBCFG_TOUTCAL(7);
>   
> -	/* remove the HNP/SRP and set the PHY */
> +	/*
> +	 * Configure HNP/SRP capability from params (same idea as
> +	 * dwc2_gusbcfg_init() for host). Unconditionally clearing these
> +	 * bits leaves an HNP-capable OTG gadget unable to negotiate.
> +	 */
>   	usbcfg &= ~(GUSBCFG_SRPCAP | GUSBCFG_HNPCAP);
> -        dwc2_writel(hsotg, usbcfg, GUSBCFG);
> +	if (hsotg->params.otg_caps.hnp_support &&
> +	    hsotg->params.otg_caps.srp_support)
> +		usbcfg |= GUSBCFG_HNPCAP;
> +	if (hsotg->params.otg_caps.srp_support)
> +		usbcfg |= GUSBCFG_SRPCAP;
> +	dwc2_writel(hsotg, usbcfg, GUSBCFG);
>   
>   	dwc2_phy_init(hsotg, true);
>

Hi,

Gentle ping on this patch. It has been over a week with no feedback.

Happy to address any review comments or provide more testing details.

Thanks,
Tze Yee