Re: [PATCH v2 06/10] rust: alloc: introduce the `BoxExt` trait

[Wedson Almeida Filho]

On Wed, 27 Mar 2024 at 14:09, Benno Lossin <benno.lossin@xxxxxxxxx> wrote:
>
> On 27.03.24 03:35, Wedson Almeida Filho wrote:
> > diff --git a/rust/kernel/alloc/allocator.rs b/rust/kernel/alloc/allocator.rs
> > index 01ad139e19bc..fc0439455faa 100644
> > --- a/rust/kernel/alloc/allocator.rs
> > +++ b/rust/kernel/alloc/allocator.rs
> > @@ -15,7 +15,11 @@
> >  ///
> >  /// - `ptr` can be either null or a pointer which has been allocated by this allocator.
> >  /// - `new_layout` must have a non-zero size.
> > -unsafe fn krealloc_aligned(ptr: *mut u8, new_layout: Layout, flags: bindings::gfp_t) -> *mut u8 {
> > +pub(crate) unsafe fn krealloc_aligned(
> > +    ptr: *mut u8,
> > +    new_layout: Layout,
> > +    flags: bindings::gfp_t,
>
> I think it is a good idea to use `Flags` here as well.

Yes, sounds like a good idea. I will do it in the `Flags` patch in v3.

>
> > +) -> *mut u8 {
> >      // Customized layouts from `Layout::from_size_align()` can have size < align, so pad first.
> >      let layout = new_layout.pad_to_align();
> >
> > diff --git a/rust/kernel/alloc/box_ext.rs b/rust/kernel/alloc/box_ext.rs
> > new file mode 100644
> > index 000000000000..aaa6a20d12b6
> > --- /dev/null
> > +++ b/rust/kernel/alloc/box_ext.rs
> > @@ -0,0 +1,62 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +
> > +//! Extensions to [`Box`] for fallible allocations.
> > +
> > +use super::Flags;
> > +use alloc::boxed::Box;
> > +use core::alloc::AllocError;
> > +use core::mem::MaybeUninit;
> > +use core::result::Result;
> > +
> > +/// Extensions to [`Box`].
> > +pub trait BoxExt<T>: Sized {
> > +    /// Allocates a new box.
> > +    ///
> > +    /// The allocation may fail, in which case an error is returned.
> > +    fn new(x: T, flags: Flags) -> Result<Self, AllocError>;
> > +
> > +    /// Allocates a new uninitialised box.
> > +    ///
> > +    /// The allocation may fail, in which case an error is returned.
> > +    fn new_uninit(flags: Flags) -> Result<Box<MaybeUninit<T>>, AllocError>;
> > +}
> > +
> > +impl<T> BoxExt<T> for Box<T> {
> > +    #[cfg(any(test, testlib))]
> > +    fn new(x: T, _flags: Flags) -> Result<Self, AllocError> {
> > +        Ok(Box::new(x))
> > +    }
> > +
> > +    #[cfg(not(any(test, testlib)))]
> > +    fn new(x: T, flags: Flags) -> Result<Self, AllocError> {
> > +        let ptr = if core::mem::size_of::<T>() == 0 {
> > +            core::ptr::NonNull::<T>::dangling().as_ptr()
> > +        } else {
> > +            let layout = core::alloc::Layout::new::<T>();
> > +
> > +            // SAFETY: Memory is being allocated (first arg is null). The only other source of
> > +            // safety issues is sleeping on atomic context, which is addressed by klint. Lastly,
> > +            // the type is not a SZT (checked above).
> > +            let ptr = unsafe {
> > +                super::allocator::krealloc_aligned(core::ptr::null_mut(), layout, flags.0)
> > +            };
> > +            if ptr.is_null() {
> > +                return Err(AllocError);
> > +            }
> > +
> > +            let ptr = ptr.cast::<T>();
> > +
> > +            // SAFETY: We just allocated the memory above, it is valid for write.
> > +            unsafe { ptr.write(x) };
> > +            ptr
> > +        };
> > +
> > +        // SAFETY: For non-zero-sized types, we allocate above using the global allocator. For
> > +        // zero-sized types, we use `NonNull::dangling`.
> > +        Ok(unsafe { Box::from_raw(ptr) })
> > +    }
> > +
> > +    fn new_uninit(flags: Flags) -> Result<Box<MaybeUninit<T>>, AllocError> {
> > +        <Box<_> as BoxExt<_>>::new(MaybeUninit::<T>::uninit(), flags)
>
> Note that the expression `MaybeUninit::<T>::uninit()` introduces a
> temporary which will be allocated on the stack. If the type `T` is very
> big, then this function will cause immediate stack overflows.
>
> I would recommend to implement `new` in terms of `new_uninit`. The
> `new_uninit` implementation can be almost the same as the current
> implementation of `new`, just remove the `ptr.write(x)` statement.

That's a good point.

For v3 I will implement `new_uninit` with `krealloc_aligned` and use
that to implement `new`.

> --
> Cheers,
> Benno
>
> > +    }
> > +}
>