On Wed, Aug 16, 2017 at 8:47 AM, Tony Lindgren <tony@xxxxxxxxxxx> wrote: > * Liam Breck <liam@xxxxxxxxxxxxxxxxx> [170812 12:00]: >> On Sat, Aug 12, 2017 at 9:42 AM, Sebastian Reichel <sre@xxxxxxxxxx> wrote: >> > On Fri, Aug 11, 2017 at 02:57:10PM -0700, Liam Breck wrote: >> >> >> +Other features: >> >> >> +- Use gpio-hog to set the OTG pin high to enable 500mA charge current on USB SDP port. >> >> > >> >> > Why doesn't the charger driver control this? >> >> >> >> This pin/gpio has to be high when USB is plugged in to make the >> >> charger chip negotiate 500mA from a SDP port. It should generally be >> >> configured that way even if the driver is not loaded, as the chip does >> >> not require a driver to function. Some boards may do that with a >> >> pull-up. Also the user may want to change this via sysfs and not have >> >> the driver "fix" it. >> > >> > drivers/power/supply/qcom_smbb.c exposes the otg feature as >> > regulator. I think in this case it could be exposed as >> > gpio-regulator. >> >> Hans is doing that here: >> https://patchwork.kernel.org/patch/9883717/ >> >> The datasheet labels one pin "OTG" but that's misleading; it really >> enables negotiation of 500mA input current, but does not turn on OTG >> mode. If this pin is low, it only prevents the driver from starting >> OTG mode via I2C msg. > > Yeah so before the device has enumerated this pin keeps the > max current at 100mA as required by the USB spec. This pin > should be only toggled after enumerating in bootloader or > by the kernel driver. I tink the USB PHY driver can also > be wired to control this line. > > Probably best to set it up as just a GPIO that the bq24190 > driver controls optionally. > > Then later on we should have some Linux generic way for > USB gadget subsystem to tell the charger that we have > enumerated, and then bq24190 driver can toggle this GPIO > to enable higher charge rates. I discovered a few months ago that setting this pin high after USB connect has NO effect on iinlim. It has to be high at connect to auto-negotiate 500mA. There is no point in making the driver control this pin; we already tried that. Here we are just documenting how the DT can enable this pin via gpio-hog. I believe the only way to defer 500mA negotiation is to leave this pin off and set iinlim via I2C msg to chip. We can do that in a gadget script via sysfs input_current_limit. Hans does not have to enable this pin to start OTG-host on his x86 device, perhaps because it is wired high there. Our script can set the linked gpio high if nec for OTG-host. -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
