[PATCH v2 0/3] rust: alloc: add Vec::resize and Vec::truncate

[PATCH v2 0/3] rust: alloc: add Vec::resize and Vec::truncate

[Andrew Ballance]

This patch series implements the Vec::truncate and Vec::resize methods
that were needed by the nova driver and removes the corresponding item
from their task list

changes in v2:
 - updated safety comments
 - fixed rustdoc comment spacing
 - reworded commit messages to be imperative
 - Link to v1: https://lore.kernel.org/rust-for-linux/20250315024235.5282-1-andrewjballance@gmail.com/

Andrew Ballance (3):
  rust: alloc: add Vec::truncate method
  rust: alloc: add Vec::resize method
  gpu: nova-core: remove completed Vec extentions from task list

 Documentation/gpu/nova/core/todo.rst | 10 -----
 rust/kernel/alloc/kvec.rs            | 62 ++++++++++++++++++++++++++++
 2 files changed, 62 insertions(+), 10 deletions(-)

-- 
2.48.1

[PATCH v2 1/3] rust: alloc: add Vec::truncate method

[Andrew Ballance]

implement the equivalent to the std's Vec::truncate
on the kernel's Vec type.

Signed-off-by: Andrew Ballance <andrewjballance@gmail.com>
---
 rust/kernel/alloc/kvec.rs | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

diff --git a/rust/kernel/alloc/kvec.rs b/rust/kernel/alloc/kvec.rs
index ae9d072741ce..18bcc59f0b38 100644
--- a/rust/kernel/alloc/kvec.rs
+++ b/rust/kernel/alloc/kvec.rs
@@ -452,6 +452,42 @@ pub fn reserve(&mut self, additional: usize, flags: Flags) -> Result<(), AllocEr
 
         Ok(())
     }
+
+    /// Shortens the vector, setting the length to `len` and drops the removed values.
+    /// If `len` is greater than or equal to the current length, this does nothing.
+    ///
+    /// This has no effect on the capacity and will not allocate.
+    /// # Examples
+    ///
+    /// ```
+    /// let mut v = kernel::kvec![1, 2, 3]?;
+    /// v.truncate(1);
+    /// assert_eq!(v.len(), 1);
+    /// assert_eq!(&v, &[1]);
+    ///
+    /// # Ok::<(), Error>(())
+    /// ```
+    pub fn truncate(&mut self, len: usize) {
+        if len >= self.len() {
+            return;
+        }
+
+        let drop_range = len..self.len();
+
+        // SAFETY: `drop_range` is a subrange of `[0, len)` by the bounds check above.
+        let ptr: *mut [T] = unsafe { self.get_unchecked_mut(drop_range) };
+
+        // SAFETY:
+        // - this will always shrink the vector because of the above bounds check
+        // - [`new_len`, `self.len`) will be dropped through the call to `drop_in_place` below
+        unsafe { self.set_len(len) };
+
+        // SAFETY:
+        // - the dropped values are valid `T`s by the type invariant
+        // - we are allowed to invalidate [`new_len`, `old_len`) because we just changed the
+        //   len, therefore we have exclusive access to [`new_len`, `old_len`)
+        unsafe { ptr::drop_in_place(ptr) };
+    }
 }
 
 impl<T: Clone, A: Allocator> Vec<T, A> {
-- 
2.48.1

[PATCH v2 2/3] rust: alloc: add Vec::resize method

[Andrew Ballance]

implement the equivalent of the rust std's Vec::resize
on the kernel's Vec type.

Signed-off-by: Andrew Ballance <andrewjballance@gmail.com>
---
 rust/kernel/alloc/kvec.rs | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/rust/kernel/alloc/kvec.rs b/rust/kernel/alloc/kvec.rs
index 18bcc59f0b38..eb6d40a1bf8b 100644
--- a/rust/kernel/alloc/kvec.rs
+++ b/rust/kernel/alloc/kvec.rs
@@ -554,6 +554,32 @@ pub fn from_elem(value: T, n: usize, flags: Flags) -> Result<Self, AllocError> {
 
         Ok(v)
     }
+
+    /// Resizes the [`Vec`] so that `len` is equal to `new_len`.
+    ///
+    /// If `new_len` is smaller than `len`, the `Vec` is [`Vec::truncate`]d.
+    /// If `new_len` is larger, each new slot is filled with clones of `value`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// let mut v = kernel::kvec![1, 2, 3]?;
+    /// v.resize(1, 42, GFP_KERNEL)?;
+    /// assert_eq!(&v, &[1]);
+    ///
+    /// v.resize(3, 42, GFP_KERNEL)?;
+    /// assert_eq!(&v, &[1, 42, 42]);
+    ///
+    /// # Ok::<(), Error>(())
+    /// ```
+    pub fn resize(&mut self, new_len: usize, value: T, flags: Flags) -> Result<(), AllocError> {
+        if new_len > self.len() {
+            self.extend_with(new_len - self.len(), value, flags)
+        } else {
+            self.truncate(new_len);
+            Ok(())
+        }
+    }
 }
 
 impl<T, A> Drop for Vec<T, A>
-- 
2.48.1

[PATCH v2 3/3] gpu: nova-core: remove completed Vec extentions from task list

[Andrew Ballance]

The requested Vec methods have been implemented thus, removes
the completed item from the nova task list

Signed-off-by: Andrew Ballance <andrewjballance@gmail.com>
---
 Documentation/gpu/nova/core/todo.rst | 10 ----------
 1 file changed, 10 deletions(-)

diff --git a/Documentation/gpu/nova/core/todo.rst b/Documentation/gpu/nova/core/todo.rst
index ca08377d3b73..234d753d3eac 100644
--- a/Documentation/gpu/nova/core/todo.rst
+++ b/Documentation/gpu/nova/core/todo.rst
@@ -190,16 +190,6 @@ Rust abstraction for debugfs APIs.
 | Reference: Export GSP log buffers
 | Complexity: Intermediate
 
-Vec extensions
---------------
-
-Implement ``Vec::truncate`` and ``Vec::resize``.
-
-Currently this is used for some experimental code to parse the vBIOS.
-
-| Reference vBIOS support
-| Complexity: Beginner
-
 GPU (general)
 =============
 
-- 
2.48.1

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Benno Lossin]

On Sun Mar 16, 2025 at 12:16 PM CET, Andrew Ballance wrote:
> implement the equivalent of the rust std's Vec::resize
> on the kernel's Vec type.
>
> Signed-off-by: Andrew Ballance <andrewjballance@gmail.com>

Reviewed-by: Benno Lossin <benno.lossin@proton.me>

---
Cheers,
Benno

> ---
>  rust/kernel/alloc/kvec.rs | 26 ++++++++++++++++++++++++++
>  1 file changed, 26 insertions(+)

Re: [PATCH v2 1/3] rust: alloc: add Vec::truncate method

[Danilo Krummrich]

On Sun, Mar 16, 2025 at 06:16:42AM -0500, Andrew Ballance wrote:
> implement the equivalent to the std's Vec::truncate
> on the kernel's Vec type.
> 
> Signed-off-by: Andrew Ballance <andrewjballance@gmail.com>
> ---
>  rust/kernel/alloc/kvec.rs | 36 ++++++++++++++++++++++++++++++++++++
>  1 file changed, 36 insertions(+)
> 
> diff --git a/rust/kernel/alloc/kvec.rs b/rust/kernel/alloc/kvec.rs
> index ae9d072741ce..18bcc59f0b38 100644
> --- a/rust/kernel/alloc/kvec.rs
> +++ b/rust/kernel/alloc/kvec.rs
> @@ -452,6 +452,42 @@ pub fn reserve(&mut self, additional: usize, flags: Flags) -> Result<(), AllocEr
>  
>          Ok(())
>      }
> +
> +    /// Shortens the vector, setting the length to `len` and drops the removed values.
> +    /// If `len` is greater than or equal to the current length, this does nothing.
> +    ///
> +    /// This has no effect on the capacity and will not allocate.
> +    /// # Examples
> +    ///
> +    /// ```
> +    /// let mut v = kernel::kvec![1, 2, 3]?;
> +    /// v.truncate(1);
> +    /// assert_eq!(v.len(), 1);
> +    /// assert_eq!(&v, &[1]);
> +    ///
> +    /// # Ok::<(), Error>(())
> +    /// ```
> +    pub fn truncate(&mut self, len: usize) {
> +        if len >= self.len() {
> +            return;
> +        }
> +
> +        let drop_range = len..self.len();
> +
> +        // SAFETY: `drop_range` is a subrange of `[0, len)` by the bounds check above.
> +        let ptr: *mut [T] = unsafe { self.get_unchecked_mut(drop_range) };
> +
> +        // SAFETY:
> +        // - this will always shrink the vector because of the above bounds check
> +        // - [`new_len`, `self.len`) will be dropped through the call to `drop_in_place` below

We've just figured out that this part is not needed after all, sorry for the
inconvenience. No need to resend for this though, I can remove this line when
applying the patch.

> +        unsafe { self.set_len(len) };
> +
> +        // SAFETY:
> +        // - the dropped values are valid `T`s by the type invariant
> +        // - we are allowed to invalidate [`new_len`, `old_len`) because we just changed the
> +        //   len, therefore we have exclusive access to [`new_len`, `old_len`)
> +        unsafe { ptr::drop_in_place(ptr) };
> +    }
>  }
>  
>  impl<T: Clone, A: Allocator> Vec<T, A> {
> -- 
> 2.48.1
> 

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Tamir Duberstein]

On Sun, Mar 16, 2025 at 7:17 AM Andrew Ballance
<andrewjballance@gmail.com> wrote:
>
> implement the equivalent of the rust std's Vec::resize
> on the kernel's Vec type.
>
> Signed-off-by: Andrew Ballance <andrewjballance@gmail.com>
> ---
>  rust/kernel/alloc/kvec.rs | 26 ++++++++++++++++++++++++++
>  1 file changed, 26 insertions(+)
>
> diff --git a/rust/kernel/alloc/kvec.rs b/rust/kernel/alloc/kvec.rs
> index 18bcc59f0b38..eb6d40a1bf8b 100644
> --- a/rust/kernel/alloc/kvec.rs
> +++ b/rust/kernel/alloc/kvec.rs
> @@ -554,6 +554,32 @@ pub fn from_elem(value: T, n: usize, flags: Flags) -> Result<Self, AllocError> {
>
>          Ok(v)
>      }
> +
> +    /// Resizes the [`Vec`] so that `len` is equal to `new_len`.
> +    ///
> +    /// If `new_len` is smaller than `len`, the `Vec` is [`Vec::truncate`]d.
> +    /// If `new_len` is larger, each new slot is filled with clones of `value`.
> +    ///
> +    /// # Examples
> +    ///
> +    /// ```
> +    /// let mut v = kernel::kvec![1, 2, 3]?;
> +    /// v.resize(1, 42, GFP_KERNEL)?;
> +    /// assert_eq!(&v, &[1]);
> +    ///
> +    /// v.resize(3, 42, GFP_KERNEL)?;
> +    /// assert_eq!(&v, &[1, 42, 42]);
> +    ///
> +    /// # Ok::<(), Error>(())
> +    /// ```
> +    pub fn resize(&mut self, new_len: usize, value: T, flags: Flags) -> Result<(), AllocError> {
> +        if new_len > self.len() {
> +            self.extend_with(new_len - self.len(), value, flags)
> +        } else {
> +            self.truncate(new_len);
> +            Ok(())
> +        }
> +    }

You can avoid underflow checking in debug builds by using `checked_sub`:

        match new_len.checked_sub(self.len()) {
            Some(n) => self.extend_with(n, value, flags),
            None => {
                self.truncate(new_len);
                Ok(())
            }
        }

>  }
>
>  impl<T, A> Drop for Vec<T, A>
> --
> 2.48.1
>
>

Either way:

Reviewed-by: Tamir Duberstein <tamird@gmail.com>

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Benno Lossin]

On Tue Mar 18, 2025 at 9:12 PM CET, Tamir Duberstein wrote:
> On Sun, Mar 16, 2025 at 7:17 AM Andrew Ballance
> <andrewjballance@gmail.com> wrote:
>> +    pub fn resize(&mut self, new_len: usize, value: T, flags: Flags) -> Result<(), AllocError> {
>> +        if new_len > self.len() {
>> +            self.extend_with(new_len - self.len(), value, flags)
>> +        } else {
>> +            self.truncate(new_len);
>> +            Ok(())
>> +        }
>> +    }
>
> You can avoid underflow checking in debug builds by using `checked_sub`:

`checked_sub` doesn't only avoid underflow in debug builds, but rather
in all builds. But the code below is a good suggestion.

---
Cheers,
Benno

>         match new_len.checked_sub(self.len()) {
>             Some(n) => self.extend_with(n, value, flags),
>             None => {
>                 self.truncate(new_len);
>                 Ok(())
>             }
>         }
>
>>  }
>>
>>  impl<T, A> Drop for Vec<T, A>
>> --
>> 2.48.1
>>
>>
>
> Either way:
>
> Reviewed-by: Tamir Duberstein <tamird@gmail.com>

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Tamir Duberstein]

On Tue, Mar 18, 2025 at 8:50 PM Benno Lossin <benno.lossin@proton.me> wrote:
>
> On Tue Mar 18, 2025 at 9:12 PM CET, Tamir Duberstein wrote:
> > On Sun, Mar 16, 2025 at 7:17 AM Andrew Ballance
> > <andrewjballance@gmail.com> wrote:
> >> +    pub fn resize(&mut self, new_len: usize, value: T, flags: Flags) -> Result<(), AllocError> {
> >> +        if new_len > self.len() {
> >> +            self.extend_with(new_len - self.len(), value, flags)
> >> +        } else {
> >> +            self.truncate(new_len);
> >> +            Ok(())
> >> +        }
> >> +    }
> >
> > You can avoid underflow checking in debug builds by using `checked_sub`:
>
> `checked_sub` doesn't only avoid underflow in debug builds, but rather
> in all builds. But the code below is a good suggestion.

Yes, I know :)

I included that language because the underflow check is likely
optimized away in release builds.

Tamir

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Benno Lossin]

On Wed Mar 19, 2025 at 2:42 PM CET, Tamir Duberstein wrote:
> On Tue, Mar 18, 2025 at 8:50 PM Benno Lossin <benno.lossin@proton.me> wrote:
>>
>> On Tue Mar 18, 2025 at 9:12 PM CET, Tamir Duberstein wrote:
>> > On Sun, Mar 16, 2025 at 7:17 AM Andrew Ballance
>> > <andrewjballance@gmail.com> wrote:
>> >> +    pub fn resize(&mut self, new_len: usize, value: T, flags: Flags) -> Result<(), AllocError> {
>> >> +        if new_len > self.len() {
>> >> +            self.extend_with(new_len - self.len(), value, flags)
>> >> +        } else {
>> >> +            self.truncate(new_len);
>> >> +            Ok(())
>> >> +        }
>> >> +    }
>> >
>> > You can avoid underflow checking in debug builds by using `checked_sub`:
>>
>> `checked_sub` doesn't only avoid underflow in debug builds, but rather
>> in all builds. But the code below is a good suggestion.
>
> Yes, I know :)
>
> I included that language because the underflow check is likely
> optimized away in release builds.

If the function is inlined and the compiler can argue that `new_len >
self.len()`, then yes, but otherwise I'm pretty sure it won't be
optimized away.

Also if it is optimized away, then the check was still "executed", so I
find it a bit misleading to say "in debug builds" (making it sound like
it wouldn't do it in non-debug builds).

---
Cheers,
Benno

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Tamir Duberstein]

On Wed, Mar 19, 2025 at 10:34 AM Benno Lossin <benno.lossin@proton.me> wrote:
>
> On Wed Mar 19, 2025 at 2:42 PM CET, Tamir Duberstein wrote:
> > On Tue, Mar 18, 2025 at 8:50 PM Benno Lossin <benno.lossin@proton.me> wrote:
> >>
> >> On Tue Mar 18, 2025 at 9:12 PM CET, Tamir Duberstein wrote:
> >> > On Sun, Mar 16, 2025 at 7:17 AM Andrew Ballance
> >> > <andrewjballance@gmail.com> wrote:
> >> >> +    pub fn resize(&mut self, new_len: usize, value: T, flags: Flags) -> Result<(), AllocError> {
> >> >> +        if new_len > self.len() {
> >> >> +            self.extend_with(new_len - self.len(), value, flags)
> >> >> +        } else {
> >> >> +            self.truncate(new_len);
> >> >> +            Ok(())
> >> >> +        }
> >> >> +    }
> >> >
> >> > You can avoid underflow checking in debug builds by using `checked_sub`:
> >>
> >> `checked_sub` doesn't only avoid underflow in debug builds, but rather
> >> in all builds. But the code below is a good suggestion.
> >
> > Yes, I know :)
> >
> > I included that language because the underflow check is likely
> > optimized away in release builds.
>
> If the function is inlined and the compiler can argue that `new_len >
> self.len()`, then yes, but otherwise I'm pretty sure it won't be
> optimized away.
>
> Also if it is optimized away, then the check was still "executed", so I
> find it a bit misleading to say "in debug builds" (making it sound like
> it wouldn't do it in non-debug builds).

If we're talking about the same thing then I think we're both wrong
and the correct phrasing would have been: "you can avoid underflow
checking when CONFIG_RUST_OVERFLOW_CHECKS=y by using `checked_sub`". I
was referring to the underflow check implicit in `new_len -
self.len()`.

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Miguel Ojeda]

On Wed, Mar 19, 2025 at 4:59 PM Tamir Duberstein <tamird@gmail.com> wrote:
>
> If we're talking about the same thing then I think we're both wrong
> and the correct phrasing would have been: "you can avoid underflow
> checking when CONFIG_RUST_OVERFLOW_CHECKS=y by using `checked_sub`". I
> was referring to the underflow check implicit in `new_len -
> self.len()`.

`checked_sub` always checks (if not optimized away). The config option
is about the implicit one.

Do you mean avoiding panics?

Cheers,
Miguel

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Tamir Duberstein]

On Wed, Mar 19, 2025 at 12:06 PM Miguel Ojeda
<miguel.ojeda.sandonis@gmail.com> wrote:
>
> On Wed, Mar 19, 2025 at 4:59 PM Tamir Duberstein <tamird@gmail.com> wrote:
> >
> > If we're talking about the same thing then I think we're both wrong
> > and the correct phrasing would have been: "you can avoid underflow
> > checking when CONFIG_RUST_OVERFLOW_CHECKS=y by using `checked_sub`". I
> > was referring to the underflow check implicit in `new_len -
> > self.len()`.
>
> `checked_sub` always checks (if not optimized away). The config option
> is about the implicit one.
>
> Do you mean avoiding panics?

No, I meant avoiding the check. The existing code already explicitly
checks `new_len > self.len()` before evaluating `new_len -
self.len()`. This means the check occurs twice. `checked_sub` reduces
the number of checks by 1. Perhaps my wording could have been clearer
("avoid *an* underflow check").

Tamir

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Miguel Ojeda]

On Wed, Mar 19, 2025 at 5:13 PM Tamir Duberstein <tamird@gmail.com> wrote:
>
> No, I meant avoiding the check. The existing code already explicitly
> checks `new_len > self.len()` before evaluating `new_len -
> self.len()`. This means the check occurs twice. `checked_sub` reduces
> the number of checks by 1. Perhaps my wording could have been clearer
> ("avoid *an* underflow check").

Ah, you mean in the function you suggested, I see.

I think it they all may compile down to the same thing, whether
overflows checks are enabled or not, and whether the version in the
patch or `checked_sub` is used or not. At least in a quick Compiler
Explorer test it seems so, but I didn't check in an actual kernel
build.

The implicit check is gated behind the other one, so that one can be
removed, even if values are unknown -- we always have optimizations
enabled, even under "debug" builds (assuming "debug" means overflow
checking enabled).

Cheers,
Miguel

Re: [PATCH v2 2/3] rust: alloc: add Vec::resize method

[Tamir Duberstein]

On Wed, Mar 19, 2025 at 12:43 PM Miguel Ojeda
<miguel.ojeda.sandonis@gmail.com> wrote:
>
> On Wed, Mar 19, 2025 at 5:13 PM Tamir Duberstein <tamird@gmail.com> wrote:
> >
> > No, I meant avoiding the check. The existing code already explicitly
> > checks `new_len > self.len()` before evaluating `new_len -
> > self.len()`. This means the check occurs twice. `checked_sub` reduces
> > the number of checks by 1. Perhaps my wording could have been clearer
> > ("avoid *an* underflow check").
>
> Ah, you mean in the function you suggested, I see.
>
> I think it they all may compile down to the same thing, whether
> overflows checks are enabled or not, and whether the version in the
> patch or `checked_sub` is used or not. At least in a quick Compiler
> Explorer test it seems so, but I didn't check in an actual kernel
> build.
>
> The implicit check is gated behind the other one, so that one can be
> removed, even if values are unknown -- we always have optimizations
> enabled, even under "debug" builds (assuming "debug" means overflow
> checking enabled).

Yeah, this elision is what I was trying to allude to in the original
wording. I still think the suggested code is simpler, but gave my RB
either way.

Re: [PATCH v2 0/3] rust: alloc: add Vec::resize and Vec::truncate

[Danilo Krummrich]

On 3/16/25 12:16 PM, Andrew Ballance wrote:
> This patch series implements the Vec::truncate and Vec::resize methods
> that were needed by the nova driver and removes the corresponding item
> from their task list
> 
> changes in v2:
>   - updated safety comments
>   - fixed rustdoc comment spacing
>   - reworded commit messages to be imperative
>   - Link to v1: https://lore.kernel.org/rust-for-linux/20250315024235.5282-1-andrewjballance@gmail.com/
> 
> Andrew Ballance (3):
>    rust: alloc: add Vec::truncate method
>    rust: alloc: add Vec::resize method

With the following changes, applied to rust/alloc-next, thanks!

   - Rewrote safety comment of set_len().
   - Use checked_sub(), as suggested by Tamir.

>    gpu: nova-core: remove completed Vec extentions from task list

Applied to nova-next, thanks!

>   Documentation/gpu/nova/core/todo.rst | 10 -----
>   rust/kernel/alloc/kvec.rs            | 62 ++++++++++++++++++++++++++++
>   2 files changed, 62 insertions(+), 10 deletions(-)
>