On 5/6/19 06:38, Mark Brown wrote: > On Fri, May 03, 2019 at 10:29:42AM +0200, Jorge Ramirez wrote: >> On 5/3/19 08:26, Mark Brown wrote: >>> On Thu, May 02, 2019 at 01:30:48PM +0200, Jorge Ramirez wrote: > >>> It seems a bit of a jump to add a new driver - it's just another >>> descriptor and ops structure isn't it? Though as ever with the Qualcomm >>> stuff this driver is pretty baroque which doesn't entirely help though I >>> think it's just another regulator type which there's already some >>> handling for. > >> So how do we move this forward? > >> To sum up his regulator needs to be able to bypass accesses to >> SPMI_COMMON_REG_VOLTAGE_RANGE and provide the range in some other way >> hence the change below > >> I can't find a simpler solution than this since the function does now >> what is supposed to do for all the regulator types supported in the driver > > The assumption that you need to have this regulator use functions that > use and provide ranges is the very thing I'm trying to get you to > change. It looks like these regulators just need their own > set_voltage_sel() and get_voltage_sel() then they can use the standard > linear range mapping functions (and pobably the set_voltage_time_sel() > needs fixing anyway for all the other regulators). Right, and I understand what you are asking, is just that I completely disagree with you. But moving on. Would you accept if I wrote a separate driver specific to pms405 or do you want me to integrate in qcom-spmi_regulator.c? I am asking because none of the ops will use the common functions (I wont be reusing much code from this qcom-spmi_regulator.c file) > > There's already some conditional code in the probe function for handling > different operations for the over current protection and SAW stuff, this > looks like it should fit in reasonably well. Usually this would be even > easier as probe functions are just data driven but for some reason more > than usual of this driver's data initializaiton is done dynamically. >
