Hi Mark, On 19-12-17 12:58, Mark Brown wrote: > On Tue, Dec 17, 2019 at 08:35:33AM +0100, Marco Felsch wrote: > > On 19-12-16 11:44, Mark Brown wrote: > > > > What I'm saying is that I think the binding needs to explicitly talk > > > about that since at the minute it's really confusing reading it as it > > > is, it sounds very much like it's trying to override that in a chip > > > specific fashion as using gpiolib and the GPIO bindings for pinmuxing is > > > really quite unusual. > > > Hm.. I still think that we don't mux the pin to some special function. > > It is still a gpio input pin and if we don't request the pin we could > > read the input from user-space too and get a 'valid' value. Muxing would > > happen if we change the pad to so called _alternate_ function. Anyway, > > lets find a binding description: > > I don't think any of this makes much difference from a user point of > view. > > > IMHO this is very descriptive and needs no update. > > > description: > > - A GPIO reference to a local general purpose input, [1] calls it GPI. > > The DA9062 regulators can select between voltage-a/-b settings. > > Each regulator has a VBUCK*_GPI or VLDO*_GPI input to determine the > > active setting. In front of the VBUCK*_GPI/VLDO*_GPI input is a mux > > to select between different signal sources, valid sources are: the > > internal sequencer, GPI1, GPI2 and GPI3. See [1] table 63 for more > > information. Most the time the internal sequencer is fine but > > sometimes it is necessary to use the signal from the DA9062 GPI > > pads. This binding covers the second use case. > > Attention: Sharing the same GPI for other purposes or across multiple > > regulators is possible but the polarity setting must equal. > > This doesn't say anything about how the GPIO input is expected to be > controlled, for voltage setting any runtime control would need to be > done by the driver and it sounds like that's all that can be controlled. > The way this reads I'd expect one use of this to be for fast voltage > setting for example (you could even combine that with suspend sequencing > using the internal sequencer if you mux back to the sequencer during > suspend). The input signal is routed trough the da9062 gpio block to the regualtors. You can't set any voltage value using a gpio instead you decide which voltage setting is applied. The voltage values for runtime/suspend comes from the dt-data. No it's not just a fast switching option imagine the system suspend case where the cpu and soc voltage can be reduced to a very low value. Older soc's like the imx6 signaling this state by a hard wired gpio line because the soc and cpu cores don't work properly on such low voltage values. This is my use case and I can't use the sequencer. Regards, Marco
