Re: [PATCH] rust: alloc: use `spare_capacity_mut` to reduce unsafe

[Danilo Krummrich <dakr@kernel.org>]

On Mon, Mar 17, 2025 at 05:22:15PM +0000, Benno Lossin wrote:
> On Mon Mar 17, 2025 at 6:09 PM CET, Danilo Krummrich wrote:
> > On Mon, Mar 17, 2025 at 10:39:05AM -0400, Tamir Duberstein wrote:
> >> On Mon, Mar 17, 2025 at 10:34 AM Benno Lossin <benno.lossin@proton.me> wrote:
> >> > On Mon Mar 17, 2025 at 12:42 PM CET, Tamir Duberstein wrote:
> >> > > Use `spare_capacity_mut` in the implementation of `push` to reduce the
> >> > > use of `unsafe`. Both methods were added in commit 2aac4cd7dae3 ("rust:
> >> > > alloc: implement kernel `Vec` type").
> >> > >
> >> > > Signed-off-by: Tamir Duberstein <tamird@gmail.com>
> >> > > ---
> >> > >  rust/kernel/alloc/kvec.rs | 11 ++---------
> >> > >  1 file changed, 2 insertions(+), 9 deletions(-)
> >> > >
> >> > > diff --git a/rust/kernel/alloc/kvec.rs b/rust/kernel/alloc/kvec.rs
> >> > > index ae9d072741ce..d2bc3d02179e 100644
> >> > > --- a/rust/kernel/alloc/kvec.rs
> >> > > +++ b/rust/kernel/alloc/kvec.rs
> >> > > @@ -285,15 +285,8 @@ pub fn spare_capacity_mut(&mut self) -> &mut [MaybeUninit<T>] {
> >> > >      pub fn push(&mut self, v: T, flags: Flags) -> Result<(), AllocError> {
> >> > >          self.reserve(1, flags)?;
> >> > >
> >> > > -        // SAFETY:
> >> > > -        // - `self.len` is smaller than `self.capacity` and hence, the resulting pointer is
> >> > > -        //   guaranteed to be part of the same allocated object.
> >> > > -        // - `self.len` can not overflow `isize`.
> >> > > -        let ptr = unsafe { self.as_mut_ptr().add(self.len) };
> >> > > -
> >> > > -        // SAFETY:
> >> > > -        // - `ptr` is properly aligned and valid for writes.
> >> > > -        unsafe { core::ptr::write(ptr, v) };
> >> > > +        // The call to `reserve` was successful so the spare capacity is at least 1.
> >> > > +        self.spare_capacity_mut()[0].write(v);
> >> >
> >> > I think the code uses unsafe to avoid a bounds check, but I'm not 100%
> >> > sure. Danilo might remember more info.
> >
> > Yes, that was the justification to use unsafe calls instead.
> >
> > (This may also justify keeping dec_len() unsafe, since otherwise it would
> > introduce an additional boundary check for pop().)
> 
> If we use saturating_sub then we don't need a bounds check (at least on
> non-debug builds), right? 

	fn dec_len(&mut self, count: usize) -> &mut [T] {
	    self.len = self.len.saturating_sub(count);

	    // Potentially broken, since maybe `count > self.len`, hence need an
	    // additional check.
	    unsafe { slice::from_raw_parts_mut(self.as_mut_ptr().add(self.len), count) }
	}