From: "Gary Guo" <gary@garyguo.net>
To: "Danilo Krummrich" <dakr@kernel.org>, <aliceryhl@google.com>,
<acourbot@nvidia.com>, <ojeda@kernel.org>, <boqun@kernel.org>,
<gary@garyguo.net>, <bjorn3_gh@protonmail.com>,
<lossin@kernel.org>, <a.hindborg@kernel.org>, <tmgross@umich.edu>,
<abdiel.janulgue@gmail.com>, <daniel.almeida@collabora.com>,
<robin.murphy@arm.com>
Cc: <driver-core@lists.linux.dev>, <nouveau@lists.freedesktop.org>,
<dri-devel@lists.freedesktop.org>,
<rust-for-linux@vger.kernel.org>, <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 4/8] rust: dma: introduce dma::CoherentInit for memory initialization
Date: Thu, 12 Mar 2026 13:51:31 +0000 [thread overview]
Message-ID: <DH0UPPMZPLZO.9VPMF77QDPUT@garyguo.net> (raw)
In-Reply-To: <20260303162314.94363-5-dakr@kernel.org>
On Tue Mar 3, 2026 at 4:22 PM GMT, Danilo Krummrich wrote:
> Currently, dma::Coherent cannot safely provide (mutable) access to its
> underlying memory because the memory might be concurrently accessed by a
> DMA device. This makes it difficult to safely initialize the memory
> before handing it over to the hardware.
>
> Introduce dma::CoherentInit, a type that encapsulates a dma::Coherent
> before its DMA address is exposed to the device. dma::CoherentInit can
> guarantee exclusive access to the inner dma::Coherent and implement
> Deref and DerefMut.
>
> Once the memory is properly initialized, dma::CoherentInit can be
> converted into a regular dma::Coherent.
>
> Signed-off-by: Danilo Krummrich <dakr@kernel.org>
> ---
> rust/kernel/dma.rs | 153 ++++++++++++++++++++++++++++++++++++++++++++-
> 1 file changed, 152 insertions(+), 1 deletion(-)
>
> diff --git a/rust/kernel/dma.rs b/rust/kernel/dma.rs
> index 291fdea3b52b..79dd8717ac47 100644
> --- a/rust/kernel/dma.rs
> +++ b/rust/kernel/dma.rs
> @@ -20,7 +20,13 @@
> FromBytes, //
> }, //
> };
> -use core::ptr::NonNull;
> +use core::{
> + ops::{
> + Deref,
> + DerefMut, //
> + },
> + ptr::NonNull, //
> +};
>
> /// DMA address type.
> ///
> @@ -352,6 +358,151 @@ fn from(direction: DataDirection) -> Self {
> }
> }
>
> +/// Initializer type for [`Coherent`].
> +///
> +/// A [`Coherent`] object can't provide access to its memory as (mutable) slice safely, since it
> +/// can't fulfill the requirements for creating a slice. For instance, it is not valid to have a
> +/// (mutable) slice to of the memory of a [`Coherent`] while the memory might be accessed by a
> +/// device.
> +///
> +/// In contrast, this initializer type is able to fulfill the requirements to safely obtain a
> +/// (mutable) slice, as it neither provides access to the DMA address of the embedded [`Coherent`],
> +/// nor can it be used with the DMA projection accessors.
> +///
> +/// Once initialized, this type can be converted to a regular [`Coherent`] object.
> +///
> +/// # Examples
> +///
> +/// `CoherentInit<T>`:
> +///
> +/// ```
> +/// # use kernel::device::{
> +/// # Bound,
> +/// # Device,
> +/// # };
> +/// use kernel::dma::{attrs::*,
> +/// Coherent,
> +/// CoherentInit,
> +/// };
> +///
> +/// # fn test(dev: &Device<Bound>) -> Result {
> +/// let mut dmem: CoherentInit<u64> =
> +/// CoherentInit::zeroed_with_attrs(dev, GFP_KERNEL, DMA_ATTR_NO_WARN)?;
> +/// *dmem = 42;
> +/// let dmem: Coherent<u64> = dmem.into();
> +/// # Ok::<(), Error>(()) }
> +/// ```
> +///
> +/// `CoherentInit<[T]>`:
> +///
> +///
> +/// ```
> +/// # use kernel::device::{
> +/// # Bound,
> +/// # Device,
> +/// # };
> +/// use kernel::dma::{attrs::*,
> +/// Coherent,
> +/// CoherentInit,
> +/// };
> +///
> +/// # fn test(dev: &Device<Bound>) -> Result {
> +/// let mut dmem: CoherentInit<[u64]> =
> +/// CoherentInit::zeroed_slice_with_attrs(dev, 4, GFP_KERNEL, DMA_ATTR_NO_WARN)?;
> +/// dmem.fill(42);
> +/// let dmem: Coherent<[u64]> = dmem.into();
> +/// # Ok::<(), Error>(()) }
> +/// ```
> +pub struct CoherentInit<T: AsBytes + FromBytes + KnownSize + ?Sized>(Coherent<T>);
> +
> +impl<T: AsBytes + FromBytes> CoherentInit<[T]> {
> + /// Initializer variant of [`Coherent::zeroed_slice_with_attrs`].
> + pub fn zeroed_slice_with_attrs(
> + dev: &device::Device<Bound>,
> + count: usize,
> + gfp_flags: kernel::alloc::Flags,
> + dma_attrs: Attrs,
> + ) -> Result<Self> {
> + Coherent::zeroed_slice_with_attrs(dev, count, gfp_flags, dma_attrs).map(Self)
> + }
> +
> + /// Same as [CoherentInit::zeroed_slice_with_attrs], but with `dma::Attrs(0)`.
> + pub fn zeroed_slice(
> + dev: &device::Device<Bound>,
> + count: usize,
> + gfp_flags: kernel::alloc::Flags,
> + ) -> Result<Self> {
> + Self::zeroed_slice_with_attrs(dev, count, gfp_flags, Attrs(0))
> + }
> +
> + /// Initializes the element at `i` using the given initializer.
> + ///
> + /// Returns `EINVAL` if `i` is out of bounds.
> + pub fn init_at<E>(&mut self, i: usize, init: impl Init<T, E>) -> Result
> + where
> + Error: From<E>,
> + {
> + if i >= self.0.len() {
> + return Err(EINVAL);
> + }
> +
> + let ptr = core::ptr::from_mut(&mut self[i]);
> +
> + // SAFETY:
> + // - `ptr` is valid, properly aligned, and within this allocation.
> + // - `T: AsBytes + FromBytes` guarantees all bit patterns are valid, so partial writes on
> + // error cannot leave the element in an invalid state.
> + // - The DMA address has not been exposed yet, so there is no concurrent device access.
> + unsafe { init.__init(ptr)? };
> +
> + Ok(())
> + }
> +}
> +
> +impl<T: AsBytes + FromBytes> CoherentInit<T> {
> + /// Same as [`CoherentInit::zeroed_slice_with_attrs`], but for a single element.
> + pub fn zeroed_with_attrs(
> + dev: &device::Device<Bound>,
> + gfp_flags: kernel::alloc::Flags,
> + dma_attrs: Attrs,
> + ) -> Result<Self> {
> + Coherent::zeroed_with_attrs(dev, gfp_flags, dma_attrs).map(Self)
> + }
> +
> + /// Same as [`CoherentInit::zeroed_slice`], but for a single element.
> + pub fn zeroed(dev: &device::Device<Bound>, gfp_flags: kernel::alloc::Flags) -> Result<Self> {
> + Self::zeroed_with_attrs(dev, gfp_flags, Attrs(0))
> + }
> +}
> +
> +impl<T: AsBytes + FromBytes + KnownSize + ?Sized> Deref for CoherentInit<T> {
> + type Target = T;
> +
The deref impls should be `#[inline]`.
Some other methods could be as well.
Best,
Gary
> + fn deref(&self) -> &Self::Target {
> + // SAFETY:
> + // - We have not exposed the DMA address yet, so there can't be any concurrent access by a
> + // device.
> + // - We have exclusive access to `self.0`.
> + unsafe { self.0.as_ref() }
> + }
> +}
> +
> +impl<T: AsBytes + FromBytes + KnownSize + ?Sized> DerefMut for CoherentInit<T> {
> + fn deref_mut(&mut self) -> &mut Self::Target {
> + // SAFETY:
> + // - We have not exposed the DMA address yet, so there can't be any concurrent access by a
> + // device.
> + // - We have exclusive access to `self.0`.
> + unsafe { self.0.as_mut() }
> + }
> +}
> +
> +impl<T: AsBytes + FromBytes + KnownSize + ?Sized> From<CoherentInit<T>> for Coherent<T> {
> + fn from(value: CoherentInit<T>) -> Self {
> + value.0
> + }
> +}
> +
> /// An abstraction of the `dma_alloc_coherent` API.
> ///
> /// This is an abstraction around the `dma_alloc_coherent` API which is used to allocate and map
next prev parent reply other threads:[~2026-03-12 13:51 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <9A1gWen63AbJmi06Y_vu_qF1S86nj3fcQEil9RzLAn8QkIYpCOrLMNO7JcgIP0BccPHhzaYaQ_te1S_gKuHRgg==@protonmail.internalid>
2026-03-03 16:22 ` [PATCH 0/8] dma::Coherent & dma::CoherentInit API Danilo Krummrich
2026-03-03 16:22 ` [PATCH 1/8] rust: dma: use "kernel vertical" style for imports Danilo Krummrich
2026-03-03 17:12 ` Gary Guo
2026-03-03 16:22 ` [PATCH 2/8] rust: dma: add generalized container for types other than slices Danilo Krummrich
2026-03-17 6:50 ` Alice Ryhl
2026-03-17 12:56 ` Gary Guo
2026-03-20 13:58 ` Alexandre Courbot
2026-03-03 16:22 ` [PATCH 3/8] rust: dma: add zeroed constructor to `Coherent` Danilo Krummrich
2026-03-17 6:47 ` Alice Ryhl
2026-03-20 14:35 ` Alexandre Courbot
2026-03-03 16:22 ` [PATCH 4/8] rust: dma: introduce dma::CoherentInit for memory initialization Danilo Krummrich
2026-03-12 13:51 ` Gary Guo [this message]
2026-03-17 6:52 ` Alice Ryhl
2026-03-17 14:40 ` Danilo Krummrich
2026-03-17 14:43 ` Alice Ryhl
2026-03-20 14:35 ` Alexandre Courbot
2026-03-20 13:51 ` Danilo Krummrich
2026-03-03 16:22 ` [PATCH 5/8] rust: dma: add Coherent:init() and Coherent::init_with_attrs() Danilo Krummrich
2026-03-03 16:22 ` [PATCH 6/8] gpu: nova-core: use Coherent::init to initialize GspFwWprMeta Danilo Krummrich
2026-03-03 17:21 ` Gary Guo
2026-03-20 14:36 ` Alexandre Courbot
2026-03-03 16:22 ` [PATCH 7/8] gpu: nova-core: convert Gsp::new() to use CoherentInit Danilo Krummrich
2026-03-03 17:33 ` Gary Guo
2026-03-03 16:22 ` [PATCH 8/8] gpu: nova-core: convert to new dma::Coherent API Danilo Krummrich
2026-03-20 14:36 ` Alexandre Courbot
2026-03-20 15:05 ` Gary Guo
2026-03-24 12:33 ` [PATCH 0/8] dma::Coherent & dma::CoherentInit API Andreas Hindborg
2026-03-24 12:36 ` Danilo Krummrich
2026-03-24 12:46 ` Andreas Hindborg
2026-03-24 12:51 ` 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=DH0UPPMZPLZO.9VPMF77QDPUT@garyguo.net \
--to=gary@garyguo.net \
--cc=a.hindborg@kernel.org \
--cc=abdiel.janulgue@gmail.com \
--cc=acourbot@nvidia.com \
--cc=aliceryhl@google.com \
--cc=bjorn3_gh@protonmail.com \
--cc=boqun@kernel.org \
--cc=dakr@kernel.org \
--cc=daniel.almeida@collabora.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=driver-core@lists.linux.dev \
--cc=linux-kernel@vger.kernel.org \
--cc=lossin@kernel.org \
--cc=nouveau@lists.freedesktop.org \
--cc=ojeda@kernel.org \
--cc=robin.murphy@arm.com \
--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