Re: [PATCH] usb: dwc2: gadget: Restart the gadget when a new session follows a session end

[Thinh Nguyen <Thinh.Nguyen@synopsys.com>]

On Fri, Oct 23, 2026, Olivier Louvignes wrote:
> When the B-session becomes invalid (the cable is pulled),
> dwc2_hsotg_disconnect() disconnects the gadget and leaves it in L3.
> Nothing brings it back, unless something outside the controller reports
> VBUS through usb_gadget_vbus_connect(): the next USB reset does not
> reinitialise a gadget that is not connected, and dwc2_hsotg_ep_queue()
> then refuses EP0's setup request because lx_state is not L0. Every
> enumeration after a replug fails with
> 
>   dwc2_hsotg_enqueue_setup: failed queue (-11)
> 
> On the session request interrupt that a new session raises, start the
> gadget again when it is in L3, as dwc2_hsotg_vbus_session() does when a
> session becomes active.
> 
> Seen on an Ingenic X1600 (HiBy R1) in peripheral mode with g_serial and
> no VBUS notifier. After a replug the host enumerates the gadget again
> and the serial shell comes back, without a reboot. Tested on v7.3.
> 
> Signed-off-by: Olivier Louvignes <olivier@mgcrea.io>
> ---
> Tested on one board only, in peripheral mode. The alternative is for the
> platform glue to report VBUS with usb_gadget_vbus_connect() (a
> usb_role_switch or extcon notifier), which this board does not have: its
> VBUS is seen by a PMIC with no such driver. If that is the preferred
> place for this, I will look at doing it there instead. The patch is
> build-tested on 734bef548a96 (v7.3-7363) with and without gadget support.
> 
>  drivers/usb/dwc2/core.h      |  2 ++
>  drivers/usb/dwc2/core_intr.c |  3 +++
>  drivers/usb/dwc2/gadget.c    | 29 +++++++++++++++++++++++++++++
>  3 files changed, 34 insertions(+)
> 

Please address Sashiko's feedback.

Thanks,
Thinh