Re: [PATCH v3] rust: regulator: add a bare minimum regulator abstraction

["Benno Lossin" <lossin@kernel.org>]

On Wed May 14, 2025 at 12:16 PM CEST, Mark Brown wrote:
> On Wed, May 14, 2025 at 11:37:46AM +0200, Benno Lossin wrote:
>> On Wed May 14, 2025 at 9:46 AM CEST, Mark Brown wrote:
>> > On Tue, May 13, 2025 at 10:01:05PM +0200, Benno Lossin wrote:
>
>> >> This isn't fully clear what it's supposed to mean to me. Maybe mention
>> >> the `regulator_enable` function?
>
>> > I suspect this is adequately clear to someone with the domain specific
>> > knowledge required to be using the API.
>
>> I still think it's useful to name the exact function that is meant by
>> "enabled".
>
> It's not clear to me that it's helpful to have to refer to the C API, as
> opposed to just being free standing.

To me it would be much more clear if the function were named.

>> >> Why don't we drop the refcount if the `regulator_disable` call fails?
>
>> > If you fail to disable the regulator then the underlying C code won't
>> > drop it's reference count.
>
>> So if it fails, the regulator should stay alive indefinitely? Would be
>> useful to explain that in the comment above the `ManuallyDrop`.
>
> Practically speaking if the regulator disable fails the system is having
> an extremely bad time and the actual state of the regulator is not clear.
> Users might want to try some attempt at retrying, one of which could
> possibly succeed in future, but realistically if this happens there's
> something fairly catastrophic going on.  Some critical users might want
> to care and have a good idea what makes sense for them, but probably the
> majority of users of the API aren't going to have a good strategy here.

Makes sense. So does `regulator_disable` take ownership of the refcount?
If yes, then just put that in the comment above the `ManuallyDrop` & in
the `Drop` impl of `EnabledRegulator`.

---
Cheers,
Benno