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::CoherentBox, a type that encapsulates a dma::Coherent
before its DMA address is exposed to the device. dma::CoherentBox can
guarantee exclusive access to the inner dma::Coherent and implement
Deref and DerefMut.

Once the memory is properly initialized, dma::CoherentBox can be
converted into a regular dma::Coherent.

Reviewed-by: Alice Ryhl <aliceryhl@google.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
---
 rust/kernel/dma.rs | 154 ++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 153 insertions(+), 1 deletion(-)

diff --git a/rust/kernel/dma.rs b/rust/kernel/dma.rs
index db645b01bdd0..cefb54f0424a 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 +358,152 @@ fn from(direction: DataDirection) -> Self {
     }
 }
 
+/// CPU-owned DMA allocation that can be converted into a device-shared [`Coherent`] object.
+///
+/// Unlike [`Coherent`], a [`CoherentBox`] is guaranteed to be fully owned by the CPU -- its DMA
+/// address is not exposed and it cannot be accessed by a device. This means it can safely be used
+/// like a normal boxed allocation (e.g. direct reads, writes, and mutable slices are all safe).
+///
+/// A typical use is to allocate a [`CoherentBox`], populate it with normal CPU access, and then
+/// convert it into a [`Coherent`] object to share it with the device.
+///
+/// # Examples
+///
+/// `CoherentBox<T>`:
+///
+/// ```
+/// # use kernel::device::{
+/// #     Bound,
+/// #     Device,
+/// # };
+/// use kernel::dma::{attrs::*,
+///     Coherent,
+///     CoherentBox,
+/// };
+///
+/// # fn test(dev: &Device<Bound>) -> Result {
+/// let mut dmem: CoherentBox<u64> = CoherentBox::zeroed(dev, GFP_KERNEL)?;
+/// *dmem = 42;
+/// let dmem: Coherent<u64> = dmem.into();
+/// # Ok::<(), Error>(()) }
+/// ```
+///
+/// `CoherentBox<[T]>`:
+///
+///
+/// ```
+/// # use kernel::device::{
+/// #     Bound,
+/// #     Device,
+/// # };
+/// use kernel::dma::{attrs::*,
+///     Coherent,
+///     CoherentBox,
+/// };
+///
+/// # fn test(dev: &Device<Bound>) -> Result {
+/// let mut dmem: CoherentBox<[u64]> = CoherentBox::zeroed_slice(dev, 4, GFP_KERNEL)?;
+/// dmem.fill(42);
+/// let dmem: Coherent<[u64]> = dmem.into();
+/// # Ok::<(), Error>(()) }
+/// ```
+pub struct CoherentBox<T: AsBytes + FromBytes + KnownSize + ?Sized>(Coherent<T>);
+
+impl<T: AsBytes + FromBytes> CoherentBox<[T]> {
+    /// [`CoherentBox`] variant of [`Coherent::zeroed_slice_with_attrs`].
+    #[inline]
+    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 [CoherentBox::zeroed_slice_with_attrs], but with `dma::Attrs(0)`.
+    #[inline]
+    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 = &raw 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> CoherentBox<T> {
+    /// Same as [`CoherentBox::zeroed_slice_with_attrs`], but for a single element.
+    #[inline]
+    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 [`CoherentBox::zeroed_slice`], but for a single element.
+    #[inline]
+    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 CoherentBox<T> {
+    type Target = T;
+
+    #[inline]
+    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 CoherentBox<T> {
+    #[inline]
+    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<CoherentBox<T>> for Coherent<T> {
+    #[inline]
+    fn from(value: CoherentBox<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
-- 
2.53.0

"Danilo Krummrich" <dakr@kernel.org> writes:

> 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::CoherentBox, a type that encapsulates a dma::Coherent
> before its DMA address is exposed to the device. dma::CoherentBox can
> guarantee exclusive access to the inner dma::Coherent and implement
> Deref and DerefMut.
>
> Once the memory is properly initialized, dma::CoherentBox can be
> converted into a regular dma::Coherent.
>
> Reviewed-by: Alice Ryhl <aliceryhl@google.com>
> Signed-off-by: Danilo Krummrich <dakr@kernel.org>
> ---
>  rust/kernel/dma.rs | 154 ++++++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 153 insertions(+), 1 deletion(-)
>
> diff --git a/rust/kernel/dma.rs b/rust/kernel/dma.rs
> index db645b01bdd0..cefb54f0424a 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 +358,152 @@ fn from(direction: DataDirection) -> Self {
>      }
>  }
>
> +/// CPU-owned DMA allocation that can be converted into a device-shared [`Coherent`] object.
> +///
> +/// Unlike [`Coherent`], a [`CoherentBox`] is guaranteed to be fully owned by the CPU -- its DMA
> +/// address is not exposed and it cannot be accessed by a device. This means it can safely be used
> +/// like a normal boxed allocation (e.g. direct reads, writes, and mutable slices are all safe).
> +///
> +/// A typical use is to allocate a [`CoherentBox`], populate it with normal CPU access, and then
> +/// convert it into a [`Coherent`] object to share it with the device.
> +///
> +/// # Examples
> +///
> +/// `CoherentBox<T>`:
> +///
> +/// ```
> +/// # use kernel::device::{
> +/// #     Bound,
> +/// #     Device,
> +/// # };
> +/// use kernel::dma::{attrs::*,
> +///     Coherent,
> +///     CoherentBox,
> +/// };
> +///
> +/// # fn test(dev: &Device<Bound>) -> Result {
> +/// let mut dmem: CoherentBox<u64> = CoherentBox::zeroed(dev, GFP_KERNEL)?;
> +/// *dmem = 42;
> +/// let dmem: Coherent<u64> = dmem.into();
> +/// # Ok::<(), Error>(()) }
> +/// ```
> +///
> +/// `CoherentBox<[T]>`:
> +///
> +///
> +/// ```
> +/// # use kernel::device::{
> +/// #     Bound,
> +/// #     Device,
> +/// # };
> +/// use kernel::dma::{attrs::*,
> +///     Coherent,
> +///     CoherentBox,
> +/// };
> +///
> +/// # fn test(dev: &Device<Bound>) -> Result {
> +/// let mut dmem: CoherentBox<[u64]> = CoherentBox::zeroed_slice(dev, 4, GFP_KERNEL)?;
> +/// dmem.fill(42);
> +/// let dmem: Coherent<[u64]> = dmem.into();
> +/// # Ok::<(), Error>(()) }
> +/// ```
> +pub struct CoherentBox<T: AsBytes + FromBytes + KnownSize + ?Sized>(Coherent<T>);
> +
> +impl<T: AsBytes + FromBytes> CoherentBox<[T]> {
> +    /// [`CoherentBox`] variant of [`Coherent::zeroed_slice_with_attrs`].
> +    #[inline]
> +    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 [CoherentBox::zeroed_slice_with_attrs], but with `dma::Attrs(0)`.
> +    #[inline]
> +    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.

Could you add an example for this function?


Reviewed-by: Andreas Hindborg <a.hindborg@kernel.org>


Best regards,
Andreas Hindborg

On Fri Mar 20, 2026 at 7:45 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::CoherentBox, a type that encapsulates a dma::Coherent
> before its DMA address is exposed to the device. dma::CoherentBox can
> guarantee exclusive access to the inner dma::Coherent and implement
> Deref and DerefMut.
>
> Once the memory is properly initialized, dma::CoherentBox can be
> converted into a regular dma::Coherent.
>
> Reviewed-by: Alice Ryhl <aliceryhl@google.com>
> Signed-off-by: Danilo Krummrich <dakr@kernel.org>

Reviewed-by: Gary Guo <gary@garyguo.net>

See some nits below:

> ---
>  rust/kernel/dma.rs | 154 ++++++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 153 insertions(+), 1 deletion(-)
>
> diff --git a/rust/kernel/dma.rs b/rust/kernel/dma.rs
> index db645b01bdd0..cefb54f0424a 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 +358,152 @@ fn from(direction: DataDirection) -> Self {
>      }
>  }
>  
> +/// CPU-owned DMA allocation that can be converted into a device-shared [`Coherent`] object.
> +///
> +/// Unlike [`Coherent`], a [`CoherentBox`] is guaranteed to be fully owned by the CPU -- its DMA
> +/// address is not exposed and it cannot be accessed by a device. This means it can safely be used
> +/// like a normal boxed allocation (e.g. direct reads, writes, and mutable slices are all safe).
> +///
> +/// A typical use is to allocate a [`CoherentBox`], populate it with normal CPU access, and then
> +/// convert it into a [`Coherent`] object to share it with the device.
> +///
> +/// # Examples
> +///
> +/// `CoherentBox<T>`:
> +///
> +/// ```
> +/// # use kernel::device::{
> +/// #     Bound,
> +/// #     Device,
> +/// # };
> +/// use kernel::dma::{attrs::*,
> +///     Coherent,
> +///     CoherentBox,
> +/// };
> +///
> +/// # fn test(dev: &Device<Bound>) -> Result {
> +/// let mut dmem: CoherentBox<u64> = CoherentBox::zeroed(dev, GFP_KERNEL)?;
> +/// *dmem = 42;
> +/// let dmem: Coherent<u64> = dmem.into();
> +/// # Ok::<(), Error>(()) }
> +/// ```
> +///
> +/// `CoherentBox<[T]>`:
> +///
> +///
> +/// ```
> +/// # use kernel::device::{
> +/// #     Bound,
> +/// #     Device,
> +/// # };
> +/// use kernel::dma::{attrs::*,
> +///     Coherent,
> +///     CoherentBox,
> +/// };
> +///
> +/// # fn test(dev: &Device<Bound>) -> Result {
> +/// let mut dmem: CoherentBox<[u64]> = CoherentBox::zeroed_slice(dev, 4, GFP_KERNEL)?;
> +/// dmem.fill(42);
> +/// let dmem: Coherent<[u64]> = dmem.into();
> +/// # Ok::<(), Error>(()) }
> +/// ```
> +pub struct CoherentBox<T: AsBytes + FromBytes + KnownSize + ?Sized>(Coherent<T>);

Similar to the changes I've made to `Coherent`, this can also just have
`KnownSize + ?Sized` bound on the struct, and only have those bounds on the
constructor.

This saves a quite bit of duplication where everything needs to say `T: AsBytes
+ FromBytes`.

Should be something fix-up-able on apply, or something left for future cleanup.

> +
> +impl<T: AsBytes + FromBytes> CoherentBox<[T]> {

this and ...

> +    /// [`CoherentBox`] variant of [`Coherent::zeroed_slice_with_attrs`].
> +    #[inline]
> +    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 [CoherentBox::zeroed_slice_with_attrs], but with `dma::Attrs(0)`.
> +    #[inline]
> +    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 = &raw 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> CoherentBox<T> {

this should be the only two places that need `AsBytes + FromBytes`.

Best,
Gary

> +    /// Same as [`CoherentBox::zeroed_slice_with_attrs`], but for a single element.
> +    #[inline]
> +    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 [`CoherentBox::zeroed_slice`], but for a single element.
> +    #[inline]
> +    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 CoherentBox<T> {
> +    type Target = T;
> +
> +    #[inline]
> +    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 CoherentBox<T> {
> +    #[inline]
> +    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<CoherentBox<T>> for Coherent<T> {
> +    #[inline]
> +    fn from(value: CoherentBox<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