Looks like our replies crossed paths...
On Thu, Feb 09, 2023 at 06:39:41PM -0800, Jerry Zhang wrote:
> ok I was actually able to get it working! setting
> snps,bus-suspend-enable; and adding this simple change
>
> diff --git a/drivers/usb/dwc3/dwc3-msm.c b/drivers/usb/dwc3/dwc3-msm.c
> index 801381de3769..09229d25b39a 100644
> --- a/drivers/usb/dwc3/dwc3-msm.c
> +++ b/drivers/usb/dwc3/dwc3-msm.c
> @@ -6331,11 +6331,11 @@ static void handle_state_peripheral(struct
> dwc3_msm *mdwc, bool *work)
>
> static void handle_state_peripheral_suspend(struct dwc3_msm *mdwc)
> {
> struct dwc3 *dwc = platform_get_drvdata(mdwc->dwc3);
>
> - if (!test_bit(B_SESS_VLD, &mdwc->inputs)) {
> + if (test_bit(B_SESS_VLD, &mdwc->inputs)) {
> dev_dbg(mdwc->dev, "BSUSP: !bsv\n");
> mdwc->drd_state = DRD_STATE_IDLE;
> cancel_delayed_work_sync(&mdwc->sdp_check);
> dwc3_otg_start_peripheral(mdwc, 0);
> } else if (!test_bit(B_SUSPEND, &mdwc->inputs)) {
>
> allows the device to successfully enumerate after a host reboot. Can
> you provide some feedback on the correctness of this patch?
You're inverting the conditional logic of the if check here so it's
probably not correct. But (treading carefully here as I just mentioned I
trying not to deep dive downstream...) in doing so you are forcing the
driver to call dwc3_otg_start_peripheral(0) call, and lo and behold it
would end up calling dwc3_overide_vbus_status(false). The msm-dwc3
state machine then probably takes you back to call start_peripheral(1)
again.
Can you try to do this more directly by toggling
dwc3_override_vbus_status() off then on again as I suggested in my
other reply?
Thanks,
Jack
