Re: [PATCH 5/5] USB: Add Cypress c67x00 OTG controller driver to Kconfig and Makefiles

["Grant Likely" <grant.likely@secretlab.ca>]

On 11/24/07, David Brownell <david-b@pacbell.net> wrote:
> On Saturday 24 November 2007, Grant Likely wrote:
> > On 11/24/07, David Brownell <david-b@pacbell.net> wrote:
> > > On Friday 23 November 2007, Grant Likely wrote:
> > > > +config USB_C67X00_DRV
> > > > +tristate "Cypress C67x00 support"
> > > > +# only allowed to be =y if both USB!=m and USB_GADGET!=m
> > >
> > > This is wrong.  Remember that since this is a dual-role driver,
> > > there are exactly three possible driver modes ... and the driver
> > > must support only one of them at a time:
> >
> > Actually, this part has 2 serial in interface engines; either of which
> > can be in host, peripheral or OTG mode.  So; it is entirely relevant
> > for the driver to be doing both host and peripheral at the same time
> > with a single device.
>
> While conforming with USB specs?  Has that changed recently?

Let me rephrase; the silicon implements 2 USB engines with separate
pinouts.  They can be configured as 2 host controllers; 2 peripheral
controllers or 1 of each.  They share registers and the IRQ line from
the software interface, but they are distinctly separate USB busses.

This driver actually implements 2 sub devices.  Each sub device is
either bound to the host part of the driver or the gadget part.

>
> In the not-too-distant past I know there was explicit language
> saying that such configurations are not allowed.  (At least,
> in terms of external connectors.  A board using USB for internal
> peripheral connections would be OK.)
>
> Those issues can be deferred until peripheral support exists.
> At that point you'll need to get better solutions to those config
> issues...

ok

>
>
> > > Hmmm ... your patches don't include peripheral mode support yet.
> >
> > No, nobody has written it yet.
>
> Then what's it doing in that GIT tree you pointed me at?  :)

Sitting there being useless.  :-)  That driver doesn't actually work;
it's just a proof of concept placeholder (which is why I didn't
include it in my series)

>
>
> > > Either all these gadget hooks should vanish, or you should try
> > > to get the controller mode stuff right from the beginning.
> > >
> > > I've appended the relevant Kconfig bits from the musb_hdrc
> > > driver, which (despite some glitches) are pretty much correct.
> >
> > I'll drop the gadget hooks for now.  Most likely gadget and OTG
> > support will require more changes to this driver regardless, so it's
> > not a big deal.
>
> OK.  Then the first incaration of this will be as a host-only
> driver, in its own directory, and adding peripheral support will
> cause minor lunacy.  :)

:-)

g.

-- 
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.
grant.likely@secretlab.ca
(403) 399-0195