rust-for-linux.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Benoît du Garreau" <benoit@dugarreau.fr>
To: Oliver Mangold <oliver.mangold@pm.me>
Cc: "Benoît du Garreau" <bdgdlm@outlook.com>,
	"Andreas Hindborg" <a.hindborg@kernel.org>,
	"Miguel Ojeda" <ojeda@kernel.org>,
	"Alex Gaynor" <alex.gaynor@gmail.com>,
	"Boqun Feng" <boqun.feng@gmail.com>,
	"Gary Guo" <gary@garyguo.net>,
	"Björn Roy Baron" <bjorn3_gh@protonmail.com>,
	"Benno Lossin" <benno.lossin@proton.me>,
	"Alice Ryhl" <aliceryhl@google.com>,
	"Trevor Gross" <tmgross@umich.edu>,
	linux-kernel@vger.kernel.org, rust-for-linux@vger.kernel.org
Subject: Re: [PATCH] rust: adding UniqueRefCounted and UniqueRef types
Date: Sat,  1 Mar 2025 00:41:47 +0100	[thread overview]
Message-ID: <20250228234148.7270-1-benoit@dugarreau.fr> (raw)
In-Reply-To: <EiaQ-C0o3GMQQpw3jCnXUnNgph2WIJ5-Cm8P5N9OysIlDKYrjHNun5Ol4Q1FfVGw64k6TGCfUVBJK5r0_2eypg==@protonmail.internalid>

From: Benoît du Garreau <bdgdlm@outlook.com>

On Fri, 28 Feb 2025 14:43:03 +0000 Oliver Mangold <oliver.mangold@pm.me> wrote:

> Hi,
> 
> For usage with block-mq, we found that having a variant of ARef
> which is guaranteed to be unique being useful.
> As chances are it is useful in general, I implemented it
> as kernel::types::UniqueRef.
> The difference between ARef and UniqueRef
> is basically the same as between Arc and UniqueArc.
> 
> This v2 of the patch, addressing the issues raised by Andreas Hindborg.
> 
> On 250228 1417, Miguel Ojeda wrote:
> > 
> > I think should be caught by Clippy -- Oliver, please try building with
> > `CLIPPY=1` (we would like Clippy-clean builds as much as reasonably
> > possible),
> 
> Got it. This version should be okay for rustfmt, clippy and checkpatch :)
> 
> Best regards,
> 
> Oliver
> 
> ---
>  rust/kernel/types.rs | 153 +++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 153 insertions(+)
> 
> diff --git a/rust/kernel/types.rs b/rust/kernel/types.rs
> index 55ddd50e8aaa..72a973d9e1c7 100644
> --- a/rust/kernel/types.rs
> +++ b/rust/kernel/types.rs
> @@ -543,6 +543,12 @@ fn from(b: &T) -> Self {
>      }
>  }
>  
> +impl<T: UniqueRefCounted> From<UniqueRef<T>> for ARef<T> {
> +    fn from(b: UniqueRef<T>) -> Self {
> +        UniqueRefCounted::unique_to_shared(b)
> +    }
> +}
> +
>  impl<T: AlwaysRefCounted> Drop for ARef<T> {
>      fn drop(&mut self) {
>          // SAFETY: The type invariants guarantee that the `ARef` owns the reference we're about to
> @@ -551,6 +557,153 @@ fn drop(&mut self) {
>      }
>  }
>  
> +/// Types that are [`AlwaysRefCounted`] and can be safely converted to an [`UniqueRef`]
> +///
> +/// # Safety
> +///
> +/// Implementers must ensure that the methods of the trait
> +/// change the reference count of the underlying object such that:
> +/// - the uniqueness invariant is upheld, i.e. it is not possible
> +///   to obtain another reference by any means (other than through the [`UniqueRef`])
> +///   until the [`UniqueRef`] is dropped or converted to an [`ARef`].
> +/// - [`UniqueRefCounted::dec_ref`] correctly frees the underlying object.
> +/// - [`UniqueRefCounted::unique_to_shared`] set the reference count to the value
> +/// - that the returned [`ARef`] expects for an object with a single reference
> +///   in existence.
> +pub unsafe trait UniqueRefCounted: AlwaysRefCounted + Sized {
> +    /// Checks if the [`ARef`] is unique and convert it
> +    /// to an [`UniqueRef`] it that is that case.
> +    /// Otherwise it returns again an [`ARef`] to the same
> +    /// underlying object.
> +    fn try_shared_to_unique(this: ARef<Self>) -> Result<UniqueRef<Self>, ARef<Self>>;
> +    /// Converts the [`UniqueRef`] into an [`ARef`].
> +    fn unique_to_shared(this: UniqueRef<Self>) -> ARef<Self>;
> +    /// Decrements the reference count on the object when the [`UniqueRef`] is dropped.
> +    ///
> +    /// Frees the object when the count reaches zero.
> +    ///
> +    /// It defaults to [`AlwaysRefCounted::dec_ref`],
> +    /// but overriding it may be useful, e.g. in case of non-standard refcounting
> +    /// schemes.
> +    ///
> +    /// # Safety
> +    ///
> +    /// The same safety constraints as for [`AlwaysRefCounted::dec_ref`] apply,
> +    /// but as the reference is unique, it can be assumed that the function
> +    /// will not be called twice. In case the default implementation is not
> +    /// overridden, it has to be ensured that the call to [`AlwaysRefCounted::dec_ref`]
> +    /// can be used for an [`UniqueRef`], too.
> +    unsafe fn dec_ref(obj: NonNull<Self>) {
> +        // SAFETY: correct by function safety requirements
> +        unsafe { AlwaysRefCounted::dec_ref(obj) };
> +    }
> +}

It would be great for this trait to only have a `is_unique` method, and that functions here
do the actual work. It would make it easier to implement and would avoid duplicating this
work.
Maybe this could even be a new method on `AlwaysRefCounted`? 

> +
> +/// An unique, owned reference to an [`AlwaysRefCounted`] object.
> +///
> +/// It works the same ways as [`ARef`] but ensures that the reference is unique
> +/// and thus can be dereferenced mutably.
> +///
> +/// # Invariants
> +///
> +/// - The pointer stored in `ptr` is non-null and valid for the lifetime of the [`UniqueRef`]
> +///   instance. In particular, the [`UniqueRef`] instance owns an increment
> +///   on the underlying object's reference count.
> +/// - No other references to the underlying object exist while the [`UniqueRef`] is live.

I think you meant "no other refcount" or "only references borrowed from this".

> +pub struct UniqueRef<T: UniqueRefCounted> {
> +    ptr: NonNull<T>,
> +    _p: PhantomData<T>,
> +}
> +
> +// SAFETY: It is safe to send `UniqueRef<T>` to another thread
> +// when the underlying `T` is `Sync` because
> +// it effectively means sharing `&T` (which is safe because `T` is `Sync`); additionally, it needs
> +// `T` to be `Send` because any thread that has an `UniqueRef<T>` may ultimately access `T` using a
> +// mutable reference, for example, when the reference count reaches zero and `T` is dropped.
> +unsafe impl<T: UniqueRefCounted + Sync + Send> Send for UniqueRef<T> {}

`UniqueRef` is essentially a `Box`, so it should have the same `Send`/`Sync` implementations. Here
I don't see how sending a `UniqueRef<T>` is sharing a `&T`.

> +
> +// SAFETY: It is safe to send `&UniqueRef<T>` to another thread when the underlying `T` is `Sync`
> +// because it effectively means sharing `&T` (which is safe because `T` is `Sync`); additionally,
> +// it needs `T` to be `Send` because any thread that has a `&UniqueRef<T>` may clone it and get an
> +// `UniqueRef<T>` on that thread, so the thread may ultimately access `T`
> +// using a mutable reference, for example, when the reference count reaches zero and `T` is dropped.
> +unsafe impl<T: UniqueRefCounted + Sync + Send> Sync for UniqueRef<T> {}

Same here: you can only get a `&T` from a `UniqueRef<T>`, definitely not clone it.

> +
> +impl<T: UniqueRefCounted> UniqueRef<T> {
> +    /// Creates a new instance of [`UniqueRef`].
> +    ///
> +    /// It takes over an increment of the reference count on the underlying object.
> +    ///
> +    /// # Safety
> +    ///
> +    /// Callers must ensure that the reference count is set to such a value
> +    /// that a call to [`UniqueRefCounted::dec_ref`] will release the underlying object
> +    /// in the way which is expected when the last reference is dropped.
> +    /// Callers must not use the underlying object anymore --
> +    /// it is only safe to do so via the newly created [`UniqueRef`].
> +    pub unsafe fn from_raw(ptr: NonNull<T>) -> Self {
> +        // INVARIANT: The safety requirements guarantee that the new instance now owns the
> +        // increment on the refcount.
> +        Self {
> +            ptr,
> +            _p: PhantomData,
> +        }
> +    }
> +
> +    /// Consumes the [`UniqueRef`], returning a raw pointer.
> +    ///
> +    /// This function does not change the refcount. After calling this function, the caller is
> +    /// responsible for the refcount previously managed by the [`UniqueRef`].
> +    pub fn into_raw(me: Self) -> NonNull<T> {
> +        ManuallyDrop::new(me).ptr
> +    }
> +}
> +
> +impl<T: UniqueRefCounted> Deref for UniqueRef<T> {
> +    type Target = T;
> +
> +    fn deref(&self) -> &Self::Target {
> +        // SAFETY: The type invariants guarantee that the object is valid.
> +        unsafe { self.ptr.as_ref() }
> +    }
> +}
> +
> +impl<T: UniqueRefCounted> DerefMut for UniqueRef<T> {
> +    fn deref_mut(&mut self) -> &mut Self::Target {
> +        // SAFETY: The type invariants guarantee that the object is valid.
> +        unsafe { self.ptr.as_mut() }
> +    }
> +}
> +
> +impl<T: UniqueRefCounted> From<&T> for UniqueRef<T> {
> +    /// Converts the [`UniqueRef`] into an [`ARef`]
> +    /// by calling [`UniqueRefCounted::unique_to_shared`] on it.
> +    fn from(b: &T) -> Self {
> +        b.inc_ref();
> +        // SAFETY: We just incremented the refcount above.
> +        unsafe { Self::from_raw(NonNull::from(b)) }
> +    }
> +}

This is wrong: the reference borrows from a refcount (as per `AlwaysRefCounted`), and this
method increments it once more. It cannot be unique when the function returns.
Actually the only way such conversion could be written is by cloning `T`, which is probably
not what we want.

> +
> +impl<T: UniqueRefCounted> TryFrom<ARef<T>> for UniqueRef<T> {
> +    type Error = ARef<T>;
> +    /// Tries to convert the [`ARef`] to an [`UniqueRef`]
> +    /// by calling [`UniqueRefCounted::try_shared_to_unique`].
> +    /// In case the [`ARef`] is not unique it returns again an [`ARef`] to the same
> +    /// underlying object.
> +    fn try_from(b: ARef<T>) -> Result<UniqueRef<T>, Self::Error> {
> +        UniqueRefCounted::try_shared_to_unique(b)
> +    }
> +}
> +
> +impl<T: UniqueRefCounted> Drop for UniqueRef<T> {
> +    fn drop(&mut self) {
> +        // SAFETY: The type invariants guarantee that the [`UniqueRef`] owns the reference
> +        // we're about to decrement.
> +        unsafe { UniqueRefCounted::dec_ref(self.ptr) };
> +    }
> +}
> +
>  /// A sum type that always holds either a value of type `L` or `R`.
>  ///
>  /// # Examples
> -- 
> 2.48.1
> 
> Signed-off-by: Oliver Mangold <oliver.mangold@pm.me>

Benoît du Garreau

  parent reply	other threads:[~2025-02-28 23:42 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-02-28 14:43 [PATCH] rust: adding UniqueRefCounted and UniqueRef types Oliver Mangold
2025-02-28 15:21 ` Miguel Ojeda
2025-02-28 15:54 ` Boqun Feng
2025-02-28 18:01 ` [PATCH v2] " Oliver Mangold
2025-02-28 18:09   ` Boqun Feng
2025-02-28 18:16     ` Boqun Feng
2025-02-28 18:29   ` Andreas Hindborg
2025-03-03 13:29     ` [PATCH v3] " Oliver Mangold
2025-03-03 14:22       ` Andreas Hindborg
2025-03-03 16:33         ` Oliver Mangold
2025-03-03 14:09   ` [PATCH v2] " Andreas Hindborg
2025-02-28 23:41 ` Benoît du Garreau [this message]
2025-03-01  8:06   ` [PATCH] " Oliver Mangold
  -- strict thread matches above, loose matches on Subject: below --
2025-02-28 16:06 Oliver Mangold
2025-02-28 16:21 ` Boqun Feng
2025-02-28 16:52 ` Miguel Ojeda

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20250228234148.7270-1-benoit@dugarreau.fr \
    --to=benoit@dugarreau.fr \
    --cc=a.hindborg@kernel.org \
    --cc=alex.gaynor@gmail.com \
    --cc=aliceryhl@google.com \
    --cc=bdgdlm@outlook.com \
    --cc=benno.lossin@proton.me \
    --cc=bjorn3_gh@protonmail.com \
    --cc=boqun.feng@gmail.com \
    --cc=gary@garyguo.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=ojeda@kernel.org \
    --cc=oliver.mangold@pm.me \
    --cc=rust-for-linux@vger.kernel.org \
    --cc=tmgross@umich.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).