On Mon, Aug 01, 2011 at 05:29:13PM +0300, Heikki Krogerus wrote: > Hi guys, > > I want to hear you opinion before going ahead with this. These are > ideas for small architectural changes to the OTG utility we currently > have in Linux. > > The ideas I'm suggesting are these: > > > Separate USB transceivers from the OTG utility > > This would clarify things. Now we have these "otg drivers" that deal > with the controller and the transceiver. They are confusing and hard > to maintain (the langwell_otg.c as an example). Usually the > transceiver drivers are quite simple and the transceivers in most > cases completely autonomous. In most cases they also don't have to > deal with any of the OTG related tasks at all. If I remember > correctly, Felipe already suggested this some time ago. > > > Support for USB charging > > This is only an idea how I would do it. The utility has one > usb_charger per transceiver and the chargers will implement a power > supply. There are still issues that need to be addressed but nothing > too bad. I'm sure that you guys agree that it's time we had the > support for USB charging already. > > > OTG state machine provided by the utility > > I'm quite sure that you agree that this is a good idea. There is > nothing preventing us from doing it like this and there are so many > benefits that come with it. > > > We planned to implement a new platform independent driver for the > ChipIdea UDC, the one with four drivers in Linux at the moment, and > use these utilities with it. Since this is only adding new utility, > old drivers don't need to be converted immediately. > > I already have several questions, but before asking them, I really > would like to hear you are OK with this design. > > What do you guys think? Great ideas. So is it right that your OTG framework like below: otg | otg.c | vendor_otg.c | phy.c | vendor_utmi.c | vendor_ulpi.c gadget |vendor_udc.c host |ehci-vendor.c gadget/host's platform suspend/resume routine can call phy's suspend/resume. My questions are: - The vendor_utmi.c will call usb_register_transceiver to register a phy, how host matches a specific phys? At some platform, the otg host uses utmi, but the host only port uses external ulpi phys. - Usually, charger detect routine should be before bus enumeration due to pullup dp will be used to detect charger. How your usb_charger_work co-work with vendor_udc.c? > > > Heikki Krogerus (2): > usb: Add physical layer utility code > usb: otg: OTG 2.0 state machine > > drivers/usb/otg/otg.c | 787 +++++++++++++++++++++++++++++++++++++++++++++++ > drivers/usb/otg/phy.c | 376 ++++++++++++++++++++++ > include/linux/usb/otg.h | 120 +++++++ > include/linux/usb/phy.h | 227 ++++++++++++++ > 4 files changed, 1510 insertions(+), 0 deletions(-) > create mode 100644 drivers/usb/otg/phy.c > create mode 100644 include/linux/usb/phy.h > > -- > 1.7.4.1 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-usb" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html > -- Best Regards, Peter Chen -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
