> Am 31.05.2018 um 13:47 schrieb Johan Hovold <jo...@kernel.org>: > > On Thu, May 31, 2018 at 11:52:18AM +0200, H. Nikolaus Schaller wrote: >> >>> Am 31.05.2018 um 10:52 schrieb Johan Hovold <jo...@kernel.org>: >>> >>> On Wed, May 30, 2018 at 07:38:22AM -0700, Richard Cochran wrote: >>>> On Wed, May 30, 2018 at 12:32:34PM +0200, Johan Hovold wrote: >>>>> Another possible extension is to add generic 1PPS support. >>>> >>>> There are two possibilities to consider. >>>> >>>> 1. If the PPS causes an interrupt, then it should hook into the PPS >>>> subsystem. >>> >>> Registering a PPS child device is what I had in mind for this. >> >> This seems to be duplicating functionality that is already solved by >> >> https://elixir.bootlin.com/linux/v4.17-rc7/source/Documentation/devicetree/bindings/pps/pps-gpio.txt >> >> and >> >> https://elixir.bootlin.com/linux/v4.17-rc7/source/drivers/pps/clients/pps-gpio.c >> >> Or what is bad with just using that? > > Using pps-gpio would not allow you to describe the hardware properly, > something which, for example, may be needed for power management (e.g. > to power on the GNSS receiver when the pps device is being accessed).
Yes, that is indeed a very valid reason to do it that way as the pps-gpio seems to assume an always-on impulse source. On the other hand it looks as if the pps framework can't tell the source when to power on/off because it does not notify when it is being accessed or not: https://elixir.bootlin.com/linux/v4.17-rc7/source/drivers/pps/pps.c#L305 BR, Nikolaus
