Hello,
I would like to add that I do not have the dwc3 documentation myself,
given it is not publically available. I deduced the patch from active
debugging on my system.
Please let me know if DWC3_LINK_STATE_U1 and DWC3_LINK_STATE_U2 should
either be removed from the 'needs_wakeup =' line instead. Or if
something else is off with the proposed patch.
Kind regards,
Marcel
On Fri, Apr 23, 2021 at 03:23:14PM +0200, Marcel Hamer wrote:
> In dwc3_send_gadget_ep_cmd() a check is performed if the gadget needs
> waking up. The following condition is checked before calling the
> __dwc3_gadget_wakeup() function:
>
> needs_wakeup = (dwc->link_state == DWC3_LINK_STATE_U1 ||
> dwc->link_state == DWC3_LINK_STATE_U2 ||
> dwc->link_state == DWC3_LINK_STATE_U3);
>
> Inside __dwc3_gadget_wakeup() only link_state DWC3_LINK_STATE_U3 is checked
> and link state DWC3_LINK_STATE_U1 and DWC3_LINK_STATE_U2 are considered
> invalid and generate a warning becaus of the call to:
>
> dev_WARN_ONCE(dwc->dev, ret, "wakeup failed --> %d\n", ret);
>
> because of ret being -EINVAL.
>
> Signed-off-by: Marcel Hamer <marcel@xxxxxxxxxx>
> ---
> drivers/usb/dwc3/gadget.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
> index c7ef218e7a8c..d05a7d630410 100644
> --- a/drivers/usb/dwc3/gadget.c
> +++ b/drivers/usb/dwc3/gadget.c
> @@ -1972,6 +1972,8 @@ static int __dwc3_gadget_wakeup(struct dwc3 *dwc)
> switch (link_state) {
> case DWC3_LINK_STATE_RESET:
> case DWC3_LINK_STATE_RX_DET: /* in HS, means Early Suspend */
> + case DWC3_LINK_STATE_U1:
> + case DWC3_LINK_STATE_U2:
> case DWC3_LINK_STATE_U3: /* in HS, means SUSPEND */
> case DWC3_LINK_STATE_RESUME:
> break;
> --
> 2.25.1
>
