rust-for-linux.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Alice Ryhl <aliceryhl@google.com>
To: Danilo Krummrich <dakr@kernel.org>
Cc: lorenzo.stoakes@oracle.com, vbabka@suse.cz,
	Liam.Howlett@oracle.com,  urezki@gmail.com, ojeda@kernel.org,
	alex.gaynor@gmail.com,  boqun.feng@gmail.com, gary@garyguo.net,
	bjorn3_gh@protonmail.com,  lossin@kernel.org,
	a.hindborg@kernel.org, tmgross@umich.edu,
	 abdiel.janulgue@gmail.com, acourbot@nvidia.com,
	 rust-for-linux@vger.kernel.org
Subject: Re: [PATCH v3 3/7] rust: alloc: implement VmallocPageIter
Date: Mon, 11 Aug 2025 09:14:30 +0000	[thread overview]
Message-ID: <aJm0duBjfWyZT3YJ@google.com> (raw)
In-Reply-To: <20250808181155.223504-4-dakr@kernel.org>

On Fri, Aug 08, 2025 at 08:10:17PM +0200, Danilo Krummrich wrote:
> Introduce the VmallocPageIter type; an instance of VmallocPageIter may
> be exposed by owners of vmalloc allocations to provide borrowed access
> to the backing pages of the vmalloc allocation.
> 
> For instance, this is useful to access and borrow the backing pages of
> allocation primitives, such as Box and Vec, backing a scatterlist.
> 
> Suggested-by: Alice Ryhl <aliceryhl@google.com>
> Signed-off-by: Danilo Krummrich <dakr@kernel.org>
> ---
>  rust/kernel/alloc/allocator.rs | 92 ++++++++++++++++++++++++++++++++++
>  1 file changed, 92 insertions(+)
> 
> diff --git a/rust/kernel/alloc/allocator.rs b/rust/kernel/alloc/allocator.rs
> index 2315f5063011..3ed2d5ff1e37 100644
> --- a/rust/kernel/alloc/allocator.rs
> +++ b/rust/kernel/alloc/allocator.rs
> @@ -10,6 +10,7 @@
>  
>  use super::Flags;
>  use core::alloc::Layout;
> +use core::marker::PhantomData;
>  use core::ptr;
>  use core::ptr::NonNull;
>  
> @@ -236,3 +237,94 @@ unsafe fn realloc(
>          unsafe { ReallocFunc::KVREALLOC.call(ptr, layout, old_layout, flags) }
>      }
>  }
> +
> +/// An [`Iterator`] of [`page::BorrowedPage`] items owned by a [`Vmalloc`] allocation.
> +///
> +/// # Guarantees
> +///
> +/// The pages iterated by the [`Iterator`] appear in the order as they are mapped in the CPU's
> +/// virtual address space ascendingly.
> +///
> +/// # Invariants
> +///
> +/// - `buf` is a valid pointer to the beginning of a [`Vmalloc`] allocation.
> +/// - `size` is the size of the [`Vmalloc`] allocation `buf` points to.
> +pub struct VmallocPageIter<'a> {
> +    /// The base address of the [`Vmalloc`] buffer.
> +    buf: NonNull<u8>,
> +    /// The size of the buffer pointed to by `buf` in bytes.
> +    size: usize,
> +    /// The current page index of the [`Iterator`].
> +    index: usize,

My understanding is that an implementation using two pointers that
repeatedly increments the starting pointer until it reaches the end
pointer is more efficient.

Alice

> +    _p: PhantomData<&'a u8>,

Semantically, I think it would make sense to write

	PhantomData<page::BorrowedPage<'a>>

here.

> +}
> +
> +impl<'a> Iterator for VmallocPageIter<'a> {

We can implement size_hint() too.

> +    type Item = page::BorrowedPage<'a>;
> +
> +    fn next(&mut self) -> Option<Self::Item> {
> +        let offset = self.index.checked_mul(page::PAGE_SIZE)?;
> +
> +        // Even though `self.size()` may be smaller than `Self::page_count() * page::PAGE_SIZE`, it
> +        // is always a number between `(Self::page_count() - 1) * page::PAGE_SIZE` and
> +        // `Self::page_count() * page::PAGE_SIZE`, hence the check below is sufficient.
> +        if offset < self.size() {
> +            self.index += 1;
> +        } else {
> +            return None;
> +        }
> +
> +        // TODO: Use `NonNull::add()` instead, once the minimum supported compiler version is
> +        // bumped to 1.80 or later.
> +        //
> +        // SAFETY: `offset` is in the interval `[0, (self.page_count() - 1) * page::PAGE_SIZE]`,
> +        // hence the resulting pointer is guaranteed to be within the same allocation.
> +        let ptr = unsafe { self.buf.as_ptr().add(offset) };
> +
> +        // SAFETY: `ptr` is guaranteed to be non-null given that it is derived from `self.buf`.
> +        let ptr = unsafe { NonNull::new_unchecked(ptr) };
> +
> +        // SAFETY:
> +        // - `ptr` is a valid pointer to a `Vmalloc` allocation.
> +        // - `ptr` is valid for the duration of `'a`.
> +        Some(unsafe { Vmalloc::to_page(ptr) })
> +    }
> +}
> +
> +impl<'a> VmallocPageIter<'a> {
> +    /// Creates a new [`VmallocPageIter`] instance.
> +    ///
> +    /// # Safety
> +    ///
> +    /// - `buf` must be a pointer to the beginning of a [`Vmalloc`] allocation.
> +    /// - `buf` points to must at least be valid for the lifetime of `'a`.
> +    /// - `size` must be the size of the [`Vmalloc`] allocation `buf` points to.
> +    pub unsafe fn new(buf: NonNull<u8>, size: usize) -> Self {
> +        // INVARIANT: `buf` is a valid pointer to the beginning of a `Vmalloc` allocation by the
> +        // safety requirements of this function.
> +        Self {
> +            buf,
> +            size,
> +            index: 0,
> +            _p: PhantomData,
> +        }
> +    }
> +
> +    /// Returns the base address of the backing [`Vmalloc`] allocation.
> +    pub fn base_address(&self) -> NonNull<u8> {
> +        self.buf
> +    }
> +
> +    /// Returns the size of the backing [`Vmalloc`] allocation in bytes.
> +    ///
> +    /// Note that this is the size the [`Vmalloc`] allocation has been allocated with. Hence, this
> +    /// number may be smaller than `[`Self::page_count`] * [`page::PAGE_SIZE`]`.
> +    pub fn size(&self) -> usize {
> +        self.size
> +    }
> +
> +    /// Returns the number of pages owned by the backing [`Vmalloc`] allocation.
> +    pub fn page_count(&self) -> usize {
> +        self.size().div_ceil(page::PAGE_SIZE)
> +    }
> +}
> -- 
> 2.50.1
> 

  reply	other threads:[~2025-08-11  9:14 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-08-08 18:10 [PATCH v3 0/7] BorrowedPage, IntoPageIter and VmallocPageIter Danilo Krummrich
2025-08-08 18:10 ` [PATCH v3 1/7] rust: page: implement BorrowedPage Danilo Krummrich
2025-08-08 18:10 ` [PATCH v3 2/7] rust: alloc: vmalloc: implement Vmalloc::to_page() Danilo Krummrich
2025-08-11  9:11   ` Alice Ryhl
2025-08-08 18:10 ` [PATCH v3 3/7] rust: alloc: implement VmallocPageIter Danilo Krummrich
2025-08-11  9:14   ` Alice Ryhl [this message]
2025-08-11  9:30     ` Danilo Krummrich
2025-08-08 18:10 ` [PATCH v3 4/7] rust: page: define trait IntoPageIter Danilo Krummrich
2025-08-11  8:57   ` Alice Ryhl
2025-08-11  9:01     ` Danilo Krummrich
2025-08-11  9:16   ` Alice Ryhl
2025-08-11 10:38     ` Danilo Krummrich
2025-08-11 10:41       ` Alice Ryhl
2025-08-08 18:10 ` [PATCH v3 5/7] rust: alloc: kbox: implement IntoPageIter for VBox Danilo Krummrich
2025-08-08 18:10 ` [PATCH v3 6/7] rust: alloc: layout: implement ArrayLayout::size() Danilo Krummrich
2025-08-19 13:09   ` Daniel Almeida
2025-08-08 18:10 ` [PATCH v3 7/7] rust: alloc: kvec: implement IntoPageIter for VVec Danilo Krummrich

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=aJm0duBjfWyZT3YJ@google.com \
    --to=aliceryhl@google.com \
    --cc=Liam.Howlett@oracle.com \
    --cc=a.hindborg@kernel.org \
    --cc=abdiel.janulgue@gmail.com \
    --cc=acourbot@nvidia.com \
    --cc=alex.gaynor@gmail.com \
    --cc=bjorn3_gh@protonmail.com \
    --cc=boqun.feng@gmail.com \
    --cc=dakr@kernel.org \
    --cc=gary@garyguo.net \
    --cc=lorenzo.stoakes@oracle.com \
    --cc=lossin@kernel.org \
    --cc=ojeda@kernel.org \
    --cc=rust-for-linux@vger.kernel.org \
    --cc=tmgross@umich.edu \
    --cc=urezki@gmail.com \
    --cc=vbabka@suse.cz \
    /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).