Re: [PATCH 1/3] rust: revocable: implement Revocable::access()

[Danilo Krummrich <dakr@kernel.org>]

On Sun, Apr 27, 2025 at 05:15:48PM +0000, Benno Lossin wrote:
> On Sun Apr 27, 2025 at 12:13 PM CEST, Danilo Krummrich wrote:
> > On Sun, Apr 27, 2025 at 08:37:00AM +0000, Benno Lossin wrote:
> >> On Sat Apr 26, 2025 at 11:18 PM CEST, Danilo Krummrich wrote:
> >> > On Sat, Apr 26, 2025 at 08:24:14PM +0000, Benno Lossin wrote:
> >> >> On Sat Apr 26, 2025 at 3:30 PM CEST, Danilo Krummrich wrote:
> >> >> > Implement an unsafe direct accessor for the data stored within the
> >> >> > Revocable.
> >> >> >
> >> >> > This is useful for cases where we can proof that the data stored within
> >> >> > the Revocable is not and cannot be revoked for the duration of the
> >> >> > lifetime of the returned reference.
> >> >> >
> >> >> > Signed-off-by: Danilo Krummrich <dakr@kernel.org>
> >> >> > ---
> >> >> > The explicit lifetimes in access() probably don't serve a practical
> >> >> > purpose, but I found them to be useful for documentation purposes.
> >> >> > ---
> >> >> >  rust/kernel/revocable.rs | 12 ++++++++++++
> >> >> >  1 file changed, 12 insertions(+)
> >> >> >
> >> >> > diff --git a/rust/kernel/revocable.rs b/rust/kernel/revocable.rs
> >> >> > index 971d0dc38d83..33535de141ce 100644
> >> >> > --- a/rust/kernel/revocable.rs
> >> >> > +++ b/rust/kernel/revocable.rs
> >> >> > @@ -139,6 +139,18 @@ pub fn try_access_with<R, F: FnOnce(&T) -> R>(&self, f: F) -> Option<R> {
> >> >> >          self.try_access().map(|t| f(&*t))
> >> >> >      }
> >> >> >  
> >> >> > +    /// Directly access the revocable wrapped object.
> >> >> > +    ///
> >> >> > +    /// # Safety
> >> >> > +    ///
> >> >> > +    /// The caller must ensure this [`Revocable`] instance hasn't been revoked and won't be revoked
> >> >> > +    /// for the duration of `'a`.
> >> >> 
> >> >> Ah I missed this in my other email, in case you want to directly refer
> >> >> to the lifetime, you should keep it defined. I would still remove the
> >> >> `'s` lifetime though.
> >> >> > +    pub unsafe fn access<'a, 's: 'a>(&'s self) -> &'a T {
> >> >> > +        // SAFETY: By the safety requirement of this function it is guaranteed that
> >> >> > +        // `self.data.get()` is a valid pointer to an instance of `T`.
> >> >> 
> >> >> I don't see how the "not-being revoked" state makes the `data` ptr be
> >> >> valid. Is that an invariant of `Revocable`? (it's not documented to have
> >> >> any invariants)
> >> >
> >> > What else makes it valid?
> >> 
> >> IMO an `# Invariants` section with the corresponding invariant that
> >> `data` is valid when `is_available` is true.
> >
> > Yeah, I agree that the # Invariants section is indeed missing and should be
> > fixed.
> >
> >> > AFAICS, try_access() and try_access_with_guard() argue the exact same way,
> >> > except that the reason for not being revoked is the atomic check and the RCU
> >> > read lock.
> >> 
> >> Just because other code is doing the same mistake doesn't make it
> >> correct. If I had reviewed the patch at that time I'm sure I would have
> >> pointed this out.
> >
> > I would say that try_access() and try_access_with_guard() are wrong, they rely
> 
> Did you mean to write `wouldn't`? Otherwise the second part doesn't
> match IMO.

Yes, I meant "wouldn't". :)

> 
> > on the correct thing, we just missed documenting the corresponding invariant.
> 
> Yeah it's not a behavior error, but since you agree that something
> should be fixed, there also is something that is 'wrong' :)
> 
> >> I opened an issue about this:
> >> 
> >>     https://github.com/Rust-for-Linux/linux/issues/1160
> >
> > Thanks for creating the issue!
> >
> > What do you suggest for this patch?
> 
> I don't mind if you take it with the lifetime changes, so
> 
> Reviewed-by: Benno Lossin <benno.lossin@proton.me>
> 
> But I'd like the invariant to be documented (maybe we should tag the
> issue with good-first-issue -- I don't actually think it is one, but
> maybe you disagree).

Yes, it should be documented; regarding the issue you created, I'd be fine
marking it as good-first-issue.

But I'd also be fine sending a fix for this myself outside the scope of this
series.