Re: [PATCH v3] phy: rcar-gen2 usb: Add Host/Function switching for USB0

[Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>]

On 07/13/2015 07:55 PM, Phil Edworthy wrote:

>>>>>>> Instead of statically selecting the PHY connection to either the
>>>>>>> USBHS (Function) or PCI0 (Host) IP blocks, this change allows the
>>>>>>> dts to specifiy gpio pins for the vbus and id signals. Additional

>>>>>>        These GPIOs don't have anything to do with the PHY, they're interfacing

>>>>       Perhaps that was too strong statement but nevertheless...

>>> Looking at your MAX3355 extcon driver, I can't see how it would work on it's
>>> own.

>>      I've added the 'extcon' support into the Renesas USBHS driver, so that it
>> should be able to sense the ID pin status.

>>> The system needs to sense vbus in order to determine that the board
>>> has been plugged into a USB Host.
>>
>>> Since the MAX3355 device doesn't
>>> directly provide any vbus signals,  this shouldn't be part of the extcon driver,
>>> so where should it be?
>>
>>      Sure, MAX3355 does provide voltage on Vbus, controlled by -OFFVBUS input.
>> It's just that the driver has only ID sensing support.
>>      Do you have the MAX3355 datasheet at all?
> Sorry, I meant that the max3355 device does not directly provide vbus signals
> suitable for input as a gpio.

    What about STATUS1/2? They are fed to GPIOs.

> Though as it provides the vbus signal to the usb
> connector, there is no reason this can't be used as a gpio input to rcar (just
> a level shifter needed ala the Koelsch board).

    Well, STATUS1/2 provide better info about Vbus, OTG-wise.

>>> On the other hand, the MAX3355 has a pair of status pins that can be used
>>> to get vbus instead.

>>      Sure, these bits reflect 4 OTG Vbus thresholds.

>>> If these pins aren't used for other functions, maybe
>>> it's better to use these in the extcon driver.

>>      They are not.

> My comments were based on the view that vbus output from the max3355
> device wasn't a signal that could be input to the rcar via gpio. Since it is, I agree
> that we should use that!

    I wasn't paying enough attention to the Koelsch scheme while replying, it 
seems. Yes, USBHS can see Vbus (almost) directly when SW6 is set to 2-3.

>>> My intention is to make the USB PHY driver listen for extcon events instead
>>> of directly accessing the ID and VBUS signals, but otherwise behave in the
>>> same way it currently does.

>>      I'm not sure the PHY driver should be interested in that...

> Then how is the PHY going to know when to switch between pci0 (host) and
> hsusb (function)?

    As I said, I planted the extcon hooks directly into USBHS (merged patch) 
and USB HCD code (unmerged, low quality hack).

>>> After reading some other threads, I also intend to set up a fixed regulator
>>> for the MAX3355 device to setup the shutdown and vbus enable pins.  I know
>>> that the vbus enable should really be controlled some other way depending
>>> on the role, but for the moment I think it's ok just to enable it always.

>>      It's OK only in the host mode.
>>      I don't think we need regulators. SHDN- is already supported by the driver
>> (though it only drives it high at start-up time), in fact it was the only reason
>> not to use gpio.

>>> Do you think that is the correct way to progress this?

>>      I didn't have a clear picture how to implement the OTG support at the time
>> of writing the MAX3355 driver; actually, I was tasked to only support ID p[in
>> sensing.

> Right, so the max3355 driver needs some changes to include vbus handling.
> Are you working on this, or should I take it on?

    No, not working yet, I have other things on my plate still...

>>      Note that there's some ongoing effort now on linux-usb to support OTG
>> functionality.

> Yes, I need to spend some time reading that list!

> Thanks
> Phil

WBR, Sergei