On Thu Jan 22, 2026 at 2:44 PM CET, Maxime Ripard wrote:
> On Mon, Jan 19, 2026 at 03:37:17PM +0100, Danilo Krummrich wrote:
>> I don't see the issue with devm_regmap_init_mmio_clk()? It takes a reference
>> count of the clock and prepares it when called and unprepares the clk in
>> drops
>> its reference in regmap_mmio_free_context() called from the devres callback.
>>
>> That something we can easily do with the current API, no?
>
> The current one, yes. Doing that in the API suggested here would involve
> some boilerplate in all those drivers they don't have right now.
No, I did mean the API suggested here.
If you would implement something like devm_regmap_init_mmio_clk() in Rust with
this API, you'd have some object like
struct RegmapResource<T: Backend, R: Resource> {
map: Regmap<T>,
res: R,
}
and a concrete instance could have the following type
RegmapResource<MmIo, Clk<Prepared>>
So, eventually you could have:
fn devm_regmap_init_mmio_clk(dev: &Device<Bound>, name: &CStr, ...) ->
... {
let clk: Clk<Prepared> = Clk::get(dev, name)?;
let regmap = RegmapResource::new(..., clk);
Devres::new(dev, regmap)
}
Of course, we would never design the API in this way, as we have generic I/O
backends and register abstractions, and we'd also not have
devm_regmap_init_mmio_clk() as a constructor for a RegmapResource type, but you
get the idea.
