On 3 June 2014 20:23, Mark Brown [off-list ref] wrote:

Well, we would need to look at what the drivers were doing and figure
out something sensible - it really depends why they're trying to set the
regulator and what would happen if it doesn't work.

For example, few cpufreq drivers are calling it during frequency
transition and are checking return value as well.. And fail if it failed.

One way out might be checking if pointer to regulator is valid or not
and only call it if pointer is not NULL..