Re: [PATCH] usb: dwc3: turn off VBUS when leaving host mode

[Bin Liu <b-liu@ti.com>]

On Tue, Dec 10, 2019 at 01:50:58PM +0200, Felipe Balbi wrote:
> 
> Hi,
> 
> Bin Liu <b-liu@ti.com> writes:
> >> >>> VBUS should be turned off when leaving the host mode.
> >> >>> Set GCTL_PRTCAP to device mode in teardown to de-assert DRVVBUS pin to
> >> >>> turn off VBUS power.
> >> >>>
> >> >>> Fixes: 5f94adfeed97 ("usb: dwc3: core: refactor mode initialization to its own function")
> >> >>> Signed-off-by: Bin Liu <b-liu@ti.com>
> >> >>> ---
> >> >>>   drivers/usb/dwc3/core.c | 1 +
> >> >>>   1 file changed, 1 insertion(+)
> >> >>>
> >> >>> diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
> >> >>> index 97d6ae3c4df2..76ac9cd54e64 100644
> >> >>> --- a/drivers/usb/dwc3/core.c
> >> >>> +++ b/drivers/usb/dwc3/core.c
> >> >>> @@ -1201,6 +1201,7 @@ static void dwc3_core_exit_mode(struct dwc3 *dwc)
> >> >>>   		break;
> >> >>>   	case USB_DR_MODE_HOST:
> >> >>>   		dwc3_host_exit(dwc);
> >> >>> +		dwc3_set_prtcap(dwc, DWC3_GCTL_PRTCAP_DEVICE);
> >> >> 
> >> >> seems like this should be done as part of dwc3_host_exit()
> >> >> 
> >> >
> >> > That wouldn't work well with OTG case as dwc3_host_exit(dwc) is
> >> > called when switching roles and we don't want PRTCAP
> >> > to change from DWC3_GCTL_PRTCAP_OTG.
> >> 
> >> if (port != OTG)
> >>   set_prtcap(DEVICE)
> >
> > During init, the PRTCAP is set in dwc3_core_init_mode() besides
> > dwc3_{host,gadget,drd}_init(). So for tearing down setting it in
> > dwc3_core_exit_mode() would make the code logic symmetric and easy to
> > understand.
> 
> Fair enough, makes sense.
> 
> > Also it turns out that setting PRTCAP is required for OTG mode too to
> 
> But, then, how do we go back to OTG?

This new setting (PRTCAP = DEVICE) is in dwc3_core_exit_mode() which is
called in platform driver .remove(). So PRTCAP will be set to OTG for
dr_mode = 'otg' in dwc3_core_init_mode() at the next time in the
platform driver .probe().

> 
> > de-assert DRVVBUS. If left GCTL[PRTCAP] to OTG, grounding the ID pin
> > would make the controller to assert DRVVBUS without any software
> > involved. So the fix should be the following. Please let me know you
> > comments.
> 
> But that's expected, no? If port is OTG, then it must obey ID pin rules
> specified by OTG. IIRC, dwc3 is OTG 2.x compliant, at least the version
> TI uses.

It is expected when the usb drivers are bind to the controller. I am
trying to solve the issue that DRVVBUS is still asserted on both host
port and otg port with ID pin grounded when the drivers are unbind from
the controller (after modprobe -r dwc3-omap, DRVVBUS should be
de-asserted).

-Bin.