On Fri, Jan 23, 2015 at 09:26:56AM +0200, Roger Quadros wrote:
> Felipe,
> 
> On 22/01/15 22:29, Felipe Balbi wrote:
> > On Thu, Jan 22, 2015 at 04:57:56PM +0200, Roger Quadros wrote:
> >>
> >> On 22/01/15 15:32, Roger Quadros wrote:
> >>> Felipe,
> >>>
> >>> On 20/01/15 21:02, Felipe Balbi wrote:
> >>>> On Mon, Jan 19, 2015 at 07:52:17PM +0200, Roger Quadros wrote:
> >>>>> Hi,
> >>>>>
> >>>>> On DRA7 EVMs the USB ID pin is connected to a GPIO line. The USB drivers
> >>>>> (dwc3 + dwc3-omap) depend on extcon framework to get the USB cable state
> >>>>> (USB or USB-Host) to put the controller in the right mode.
> >>>>>
> >>>>> There were earlier attempts [1] to get this working by trying to patch up
> >>>>> the existing GPIO extcon driver.
> >>>>>
> >>>>> This series attemts to take a different approach by introducing a new
> >>>>> USB specific extcon driver to handle the USB ID GPIO pin and
> >>>>> interpret a right USB cable state.
> >>>>>
> >>>>> The reasoning to introduce this new driver is:
> >>>>> 1) The existing GPIO extcon driver doesn't understand USB cable states
> >>>>> and it can't handle more than one cable per instance.
> >>>>>    
> >>>>> For the USB case we need to handle at least 2 cable states.
> >>>>>     a) USB (attach/detach)
> >>>>>     b) USB-Host (attach/detach)
> >>>>> and could possible include more states like
> >>>>>     c) Fast-charger (attach/detach)
> >>>>>     d) Slow-charger (attach/detach)
> >>>>>     
> >>>>> 2) This USB specific driver can be easily updated in the future to
> >>>>> handle VBUS events, or charger detect events, in case it happens
> >>>>> to be available on GPIO for any platform.
> >>>>>
> >>>>> 3) The DT implementation is very easy. You just need one extcon node per USB
> >>>>> instead of one extcon node per cable state as in case of [1].
> >>>>>
> >>>>> 4) The cable state string doesn't need to be encoded in the device tree
> >>>>> as in case of [1].
> >>>>>
> >>>>> 5) With only ID event available, you can simulate a USB-peripheral attach
> >>>>> when USB-Host is detacted instead of hacking the USB driver to do the same.
> >>>>>
> >>>>> Tested on DRA7-evm and DRA72-evm.
> >>>>
> >>>> while at that, you might want to patch X15 too.
> >>>>
> >>> USB2 port is meant for peripheral use only. ID pin from USB port is not connected to GPIO.
> >>>
> >>
> >> OK answering myself here :).
> >> Peripheral mode doesn't work on x15-bb as the USB driver (dwc3-omap)
> >> doesn't set the mailbox correctly even when dwc3 node is set as otg = "peripheral".
> >>
> >> Looks like we need to implement usb-gpio-extcon for x15 even though ID is hard coded.
> > 
> > right, another option is to have dwc3-omap read the child's DTS to check
> > dr_mode and hardcode things based on that.
> > 
> I think that option is better as it doesn't require a GPIO line to be
> reserved for ID when USB is not meant for dual-role use.

yeah, let's implement that for 3.21.

-- 
balbi