[PATCH v3 1/3] rust: clk: use the type-state pattern

Boris Brezillon boris.brezillon at collabora.com
Tue Feb 3 01:47:31 PST 2026


On Tue, 3 Feb 2026 10:09:13 +0100
Boris Brezillon <boris.brezillon at collabora.com> wrote:

> Hello Daniel,
> 
> On Mon, 2 Feb 2026 17:10:38 +0100
> Boris Brezillon <boris.brezillon at collabora.com> wrote:
> 
> > > > -#[pin_data(PinnedDrop)]
> > > > +#[pin_data]
> > > >  pub(crate) struct TyrData {
> > > >      pub(crate) pdev: ARef<platform::Device>,
> > > >  
> > > > @@ -92,13 +92,9 @@ fn probe(
> > > >          pdev: &platform::Device<Core>,
> > > >          _info: Option<&Self::IdInfo>,
> > > >      ) -> impl PinInit<Self, Error> {
> > > > -        let core_clk = Clk::get(pdev.as_ref(), Some(c_str!("core")))?;
> > > > -        let stacks_clk = OptionalClk::get(pdev.as_ref(), Some(c_str!("stacks")))?;
> > > > -        let coregroup_clk = OptionalClk::get(pdev.as_ref(), Some(c_str!("coregroup")))?;
> > > > -
> > > > -        core_clk.prepare_enable()?;
> > > > -        stacks_clk.prepare_enable()?;
> > > > -        coregroup_clk.prepare_enable()?;
> > > > +        let core_clk = Clk::<Enabled>::get(pdev.as_ref(), Some(c_str!("core")))?;      
> > > 
> > > Ah, more turbofish.. I'd really want to avoid them if possible.
> > > 
> > > Any disadvantage on just ask the user to chain `.get().prepare_enable()?`? This
> > > way it is also clear that some action is performed.    
> > 
> > I've just disc  
> 
> Sorry, I've hit the reply button before I had finished writing my
> answer. So I was about to say that I had started writing something
> similar without knowing this series existed, and I feel like we'd don't
> really need those prepare_enable() shortcuts that exist in C. We might
> has well just go:
> 
> 	Clk::get(dev, Some(c_str!("core"))).prepare()?.enable()?;

If we want this pattern to work, we also need:

    impl<State: ClkState> From<Error<State>> for kernel::error::Error {
        fn from(e: Error<State>) -> kernel::error::Error {
            e.error
        }
    }

> 
> and have the following variant-specofoc functions
> 
> - Clk<Unprepared>::get[_optional]() (no get on Prepared and Enabled
>   variants)
> - Clk<Unprepared>::prepare()
> - Clk<Prepared>::{enable,unprepare}()
> - Clk<Enabled>::{disable}()
> 
> Regards,
> 
> Boris
> 




More information about the linux-riscv mailing list