[PATCH v1 0/3] rust: make various `alloc` functions `const fn`

[PATCH v1 0/3] rust: make various `alloc` functions `const fn`

[Onur Özkan]

This patch series makes various functions in the
`alloc` crate const fn.

Each patch corresponds to a different module within
the same `alloc` crate.

Onur Özkan (3):
  rust: make `allocator::aligned_size` a `const fn`
  rust: make `ArrayLayout::new_unchecked` a `const fn`
  rust: make `kvec::Vec` functions `const fn`

 rust/kernel/alloc/allocator.rs |  2 +-
 rust/kernel/alloc/kvec.rs      | 10 +++++-----
 rust/kernel/alloc/layout.rs    |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)

--
2.50.0

[PATCH v1 1/3] rust: make `allocator::aligned_size` a `const fn`

[Onur Özkan]

Makes `allocator::aligned_size` a `const fn` to allow
compile-time evaluation.

Signed-off-by: Onur Özkan <work@onurozkan.dev>
---
 rust/kernel/alloc/allocator.rs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/rust/kernel/alloc/allocator.rs b/rust/kernel/alloc/allocator.rs
index aa2dfa9dca4c..aa2957fd585b 100644
--- a/rust/kernel/alloc/allocator.rs
+++ b/rust/kernel/alloc/allocator.rs
@@ -44,7 +44,7 @@
 pub struct KVmalloc;

 /// Returns a proper size to alloc a new object aligned to `new_layout`'s alignment.
-fn aligned_size(new_layout: Layout) -> usize {
+const fn aligned_size(new_layout: Layout) -> usize {
     // Customized layouts from `Layout::from_size_align()` can have size < align, so pad first.
     let layout = new_layout.pad_to_align();

--
2.50.0

[PATCH v1 2/3] rust: make `ArrayLayout::new_unchecked` a `const fn`

[Onur Özkan]

Makes `ArrayLayout::new_unchecked` a `const fn` to allow
compile-time evaluation.

Signed-off-by: Onur Özkan <work@onurozkan.dev>
---
 rust/kernel/alloc/layout.rs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/rust/kernel/alloc/layout.rs b/rust/kernel/alloc/layout.rs
index 93ed514f7cc7..52cbf61c4539 100644
--- a/rust/kernel/alloc/layout.rs
+++ b/rust/kernel/alloc/layout.rs
@@ -80,7 +80,7 @@ pub const fn new(len: usize) -> Result<Self, LayoutError> {
     /// # Safety
     ///
     /// `len` must be a value, for which `len * size_of::<T>() <= isize::MAX` is true.
-    pub unsafe fn new_unchecked(len: usize) -> Self {
+    pub const unsafe fn new_unchecked(len: usize) -> Self {
         // INVARIANT: By the safety requirements of this function
         // `len * size_of::<T>() <= isize::MAX`.
         Self {
--
2.50.0

[PATCH v1 3/3] rust: make `kvec::Vec` functions `const fn`

[Onur Özkan]

Makes various `kvec::Vec` functions `const fn`
to allow compile-time evaluation.

Signed-off-by: Onur Özkan <work@onurozkan.dev>
---
 rust/kernel/alloc/kvec.rs | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/rust/kernel/alloc/kvec.rs b/rust/kernel/alloc/kvec.rs
index 1a0dd852a468..af6e2564a194 100644
--- a/rust/kernel/alloc/kvec.rs
+++ b/rust/kernel/alloc/kvec.rs
@@ -174,7 +174,7 @@ const fn is_zst() -> bool {

     /// Returns the number of elements that can be stored within the vector without allocating
     /// additional memory.
-    pub fn capacity(&self) -> usize {
+    pub const fn capacity(&self) -> usize {
         if const { Self::is_zst() } {
             usize::MAX
         } else {
@@ -184,7 +184,7 @@ pub fn capacity(&self) -> usize {

     /// Returns the number of elements stored within the vector.
     #[inline]
-    pub fn len(&self) -> usize {
+    pub const fn len(&self) -> usize {
         self.len
     }

@@ -195,7 +195,7 @@ pub fn len(&self) -> usize {
     /// - `additional` must be less than or equal to `self.capacity - self.len`.
     /// - All elements within the interval [`self.len`,`self.len + additional`) must be initialized.
     #[inline]
-    pub unsafe fn inc_len(&mut self, additional: usize) {
+    pub const unsafe fn inc_len(&mut self, additional: usize) {
         // Guaranteed by the type invariant to never underflow.
         debug_assert!(additional <= self.capacity() - self.len());
         // INVARIANT: By the safety requirements of this method this represents the exact number of
@@ -244,7 +244,7 @@ pub fn as_mut_ptr(&mut self) -> *mut T {
     /// Returns a raw pointer to the vector's backing buffer, or, if `T` is a ZST, a dangling raw
     /// pointer.
     #[inline]
-    pub fn as_ptr(&self) -> *const T {
+    pub const fn as_ptr(&self) -> *const T {
         self.ptr.as_ptr()
     }

@@ -260,7 +260,7 @@ pub fn as_ptr(&self) -> *const T {
     /// assert!(!v.is_empty());
     /// ```
     #[inline]
-    pub fn is_empty(&self) -> bool {
+    pub const fn is_empty(&self) -> bool {
         self.len() == 0
     }

--
2.50.0

Re: [PATCH v1 0/3] rust: make various `alloc` functions `const fn`

[Benno Lossin]

On Sun Jul 20, 2025 at 11:48 AM CEST, Onur Özkan wrote:
> This patch series makes various functions in the
> `alloc` crate const fn.
>
> Each patch corresponds to a different module within
> the same `alloc` crate.
>
> Onur Özkan (3):
>   rust: make `allocator::aligned_size` a `const fn`
>   rust: make `ArrayLayout::new_unchecked` a `const fn`
>   rust: make `kvec::Vec` functions `const fn`
>
>  rust/kernel/alloc/allocator.rs |  2 +-
>  rust/kernel/alloc/kvec.rs      | 10 +++++-----
>  rust/kernel/alloc/layout.rs    |  2 +-
>  3 files changed, 7 insertions(+), 7 deletions(-)

This looks sensible, any particular reason for why you need them const?

For the entire series:

Reviewed-by: Benno Lossin <lossin@kernel.org>

---
Cheers,
Benno

Re: [PATCH v1 0/3] rust: make various `alloc` functions `const fn`

[Onur Özkan]

On Sun, 20 Jul 2025 16:43:49 +0200
"Benno Lossin" <lossin@kernel.org> wrote:

> On Sun Jul 20, 2025 at 11:48 AM CEST, Onur Özkan wrote:
> > This patch series makes various functions in the
> > `alloc` crate const fn.
> >
> > Each patch corresponds to a different module within
> > the same `alloc` crate.
> >
> > Onur Özkan (3):
> >   rust: make `allocator::aligned_size` a `const fn`
> >   rust: make `ArrayLayout::new_unchecked` a `const fn`
> >   rust: make `kvec::Vec` functions `const fn`
> >
> >  rust/kernel/alloc/allocator.rs |  2 +-
> >  rust/kernel/alloc/kvec.rs      | 10 +++++-----
> >  rust/kernel/alloc/layout.rs    |  2 +-
> >  3 files changed, 7 insertions(+), 7 deletions(-)
> 
> This looks sensible, any particular reason for why you need them
> const?
> 
> For the entire series:
> 
> Reviewed-by: Benno Lossin <lossin@kernel.org>
> 
> ---
> Cheers,
> Benno

Personally, I don't have a specific reason. I thought the change is
harmless and might extend functionality for other people in the future.
It could also (although less likely) help the compiler optimize things
further.

Regards,
Onur

Re: [PATCH v1 0/3] rust: make various `alloc` functions `const fn`

[Miguel Ojeda]

On Sun, Jul 20, 2025 at 5:17 PM Onur Özkan <work@onurozkan.dev> wrote:
>
> Personally, I don't have a specific reason. I thought the change is
> harmless and might extend functionality for other people in the future.
> It could also (although less likely) help the compiler optimize things
> further.

I think it is OK -- even if we promise they are `const` and we have to
remove it in the future, it is fine, since there is no stable kernel
API. So that flexibility is another advantage of no promises there.

However, I am curious, in which cases it would help the compiler
optimize? The compiler already has the information on whether it could
actually be `const` and whether it can be evaluated at compile-time
and so on -- do you mean it has an effect on heuristics like inlining
or similar?

Thanks!

Cheers,
Miguel

Re: [PATCH v1 0/3] rust: make various `alloc` functions `const fn`

[Onur Özkan]

On Sun, 20 Jul 2025 17:42:43 +0200
Miguel Ojeda <miguel.ojeda.sandonis@gmail.com> wrote:

> On Sun, Jul 20, 2025 at 5:17 PM Onur Özkan <work@onurozkan.dev> wrote:
> >
> > Personally, I don't have a specific reason. I thought the change is
> > harmless and might extend functionality for other people in the
> > future. It could also (although less likely) help the compiler
> > optimize things further.
> 
> I think it is OK -- even if we promise they are `const` and we have to
> remove it in the future, it is fine, since there is no stable kernel
> API. So that flexibility is another advantage of no promises there.
> 
> However, I am curious, in which cases it would help the compiler
> optimize? The compiler already has the information on whether it could
> actually be `const` and whether it can be evaluated at compile-time
> and so on -- do you mean it has an effect on heuristics like inlining
> or similar?

I thought it had effects similar to inlining, but after digging into the
assembly output of a simple program (with and without `const`
expressions) and reading some related discussions [1], it seems I was
wrong about it, sorry.

[1]: https://users.rust-lang.org/t/the-effects-of-const-fn/48303

Regards,
Onur

Re: [PATCH v1 0/3] rust: make various `alloc` functions `const fn`

[Alice Ryhl]

On Sun, Jul 20, 2025 at 12:48:35PM +0300, Onur Özkan wrote:
> This patch series makes various functions in the
> `alloc` crate const fn.
> 
> Each patch corresponds to a different module within
> the same `alloc` crate.
> 
> Onur Özkan (3):
>   rust: make `allocator::aligned_size` a `const fn`
>   rust: make `ArrayLayout::new_unchecked` a `const fn`
>   rust: make `kvec::Vec` functions `const fn`
> 
>  rust/kernel/alloc/allocator.rs |  2 +-
>  rust/kernel/alloc/kvec.rs      | 10 +++++-----
>  rust/kernel/alloc/layout.rs    |  2 +-
>  3 files changed, 7 insertions(+), 7 deletions(-)
> 
> --
> 2.50.0

Reviewed-by: Alice Ryhl <aliceryhl@google.com>

Re: [PATCH v1 0/3] rust: make various `alloc` functions `const fn`

[Miguel Ojeda]

On Mon, Jul 21, 2025 at 8:14 AM Onur Özkan <work@onurozkan.dev> wrote:
>
> I thought it had effects similar to inlining, but after digging into the
> assembly output of a simple program (with and without `const`
> expressions) and reading some related discussions [1], it seems I was
> wrong about it, sorry.

No worries at all, and thanks for taking a look into it!

I think Danilo will take this eventually through `alloc`, but it
sounds good to me.

Cheers,
Miguel

Re: [PATCH v1 0/3] rust: make various `alloc` functions `const fn`

[Danilo Krummrich]

On 7/20/25 11:48 AM, Onur Özkan wrote:
>    rust: make `allocator::aligned_size` a `const fn`

I dropped this patch, since it would conflict with [1].

>    rust: make `ArrayLayout::new_unchecked` a `const fn`
>    rust: make `kvec::Vec` functions `const fn`

Applied to alloc-next, thanks!

[1] https://lore.kernel.org/all/20250731154919.4132-2-dakr@kernel.org/

Re: [PATCH v1 1/3] rust: make `allocator::aligned_size` a `const fn`

[Miguel Ojeda]

On Sun, Jul 20, 2025 at 11:49 AM Onur Özkan <work@onurozkan.dev> wrote:
>
> Makes `allocator::aligned_size` a `const fn` to allow
> compile-time evaluation.
>
> Signed-off-by: Onur Özkan <work@onurozkan.dev>

For future reference: we wouldn't be able to enable it as-is without
the unstable feature anyway (which Danilo noticed when applied his
version of this) -- if it is needed in the future, the patch is at:

    https://lore.kernel.org/all/CANiq72mUXy6AYkwCW_kO3ikjNBc5pLzXw0+fXFGmYum0tGmw1g@mail.gmail.com/

Cheers,
Miguel