Re: [PATCH 1/2] regulators: Add definition of regulator_set_voltage_time() for !CONFIG_REGULATOR

On 2 June 2014 15:32, Mark Brown broonie@kernel.org wrote:

No, think about what you're changing here. You're changing the stub - the stub has a regulator_get() which always succeeeds.

Right, things might start to break with the change to regulator_set_voltage()..

When I compare this to clk-APIs, the dummy implementations always pass and so we are allowed to send NULL clk to any routine (the way we can do it here, probably to simply code)..

Now, why do we want to return -EINVAL from set_voltage here ? Similar routines in clk-API are returning 0 and even clk_get_rate() returns zero, unlike in regulators, as we return -EINVAL..

Not sure which of these frameworks is doing the right thing.

What do you suggest.