Re: [PATCH v5 6/6] rust: use strict provenance APIs

[Boqun Feng]

On Mon, Mar 17, 2025 at 02:04:34PM -0400, Tamir Duberstein wrote:
> On Mon, Mar 17, 2025 at 1:39 PM Boqun Feng <boqun.feng@xxxxxxxxx> wrote:
> >
> > On Mon, Mar 17, 2025 at 10:23:56AM -0400, Tamir Duberstein wrote:
> > [...]
> > > diff --git a/rust/kernel/lib.rs b/rust/kernel/lib.rs
> > > index fc6835cc36a3..c1b274c04a0f 100644
> > > --- a/rust/kernel/lib.rs
> > > +++ b/rust/kernel/lib.rs
> > > @@ -17,6 +17,11 @@
> > >  #![cfg_attr(not(CONFIG_RUSTC_HAS_COERCE_POINTEE), feature(coerce_unsized))]
> > >  #![cfg_attr(not(CONFIG_RUSTC_HAS_COERCE_POINTEE), feature(dispatch_from_dyn))]
> > >  #![cfg_attr(not(CONFIG_RUSTC_HAS_COERCE_POINTEE), feature(unsize))]
> > > +#![cfg_attr(
> > > +    CONFIG_RUSTC_HAS_STABLE_STRICT_PROVENANCE,
> > > +    feature(strict_provenance_lints),
> > > +    deny(fuzzy_provenance_casts, lossy_provenance_casts)
> > > +)]
> > >  #![feature(inline_const)]
> > >  #![feature(lint_reasons)]
> > >  // Stable in Rust 1.83
> > > @@ -25,6 +30,109 @@
> > >  #![feature(const_ptr_write)]
> > >  #![feature(const_refs_to_cell)]
> > >
> > > +#[cfg(CONFIG_RUSTC_HAS_STABLE_STRICT_PROVENANCE)]
> > > +#[allow(clippy::incompatible_msrv)]
> > > +mod strict_provenance {
> > > +    /// Gets the "address" portion of the pointer.
> > > +    ///
> > > +    /// See https://doc.rust-lang.org/stable/core/primitive.pointer.html#method.addr.
> > > +    #[inline]
> > > +    pub fn addr<T>(ptr: *const T) -> usize {
> > > +        ptr.addr()
> > > +    }
> > > +
> >
> > For addr(), I would just enable feature(strict_provenance) if
> > CONFIG_RUSTC_HAS_STABLE_STRICT_PROVENANCE=n, because that feature is
> > available for 1.78. Plus we may need with_addr() or map_addr() in the
> > future.
> 
> We still need these stubs to avoid `clippy::incompatible_msrv`, and
> we'll need those until MSRV is above 1.84.
> 

Hmm.. why? Clippy cannot work with unstable features?

Regards,
Boqun

> >
> > It saves the cost of maintaining our own *addr() and removing it when
> > we bump to a strict_provenance stablized version as minimal verision in
> > the future. Thoughts?
> >
> 
> I can do this by making this particular stub unconditional. I'll do that.