Re: [PATCH v4 1/6] rust: bitmap: add MAX_LEN and NO_ALLOC_MAX_LEN constants

[Yury Norov]

On Mon, Nov 10, 2025 at 02:20:17PM +0000, Alice Ryhl wrote:
> On Mon, Nov 10, 2025 at 08:59:36AM -0500, Tamir Duberstein wrote:
> > On Mon, Nov 10, 2025 at 8:06 AM Alice Ryhl <aliceryhl@xxxxxxxxxx> wrote:
> > >
> > > To avoid hard-coding these values in drivers, define constants for them
> > > that drivers can reference.
> > >
> > > Acked-by: Danilo Krummrich <dakr@xxxxxxxxxx>
> > > Reviewed-by: Burak Emir <bqe@xxxxxxxxxx>
> > > Signed-off-by: Alice Ryhl <aliceryhl@xxxxxxxxxx>
> > > ---
> > >  rust/kernel/bitmap.rs | 16 +++++++++++-----
> > >  1 file changed, 11 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/rust/kernel/bitmap.rs b/rust/kernel/bitmap.rs
> > > index aa8fc7bf06fc99865ae755d8694e4bec3dc8e7f0..15fa23b45054b9272415fcc000e3e3b52c74d7c1 100644
> > > --- a/rust/kernel/bitmap.rs
> > > +++ b/rust/kernel/bitmap.rs
> > > @@ -149,14 +149,14 @@ macro_rules! bitmap_assert_return {
> > >  ///
> > >  /// # Invariants
> > >  ///
> > > -/// * `nbits` is `<= i32::MAX` and never changes.
> > > +/// * `nbits` is `<= MAX_LEN`.
> > >  /// * if `nbits <= bindings::BITS_PER_LONG`, then `repr` is a `usize`.
> > 
> > Should this and other references to bindings::BITS_PER_LONG be
> > `NO_ALLOC_MAX_LEN` instead?
> 
> Ah yeah it probably makes sense to update this in a bunch of places.

Yes, please. 

NO_ALLOC sounds a bit weird in exported API. Maybe NBITS_INPLACE
or similar?

Also, at this point we're really close to:

   pub const NBITS_INPLACE: usize = CONFIG_NBITS_INPLACE;

   union BitmapRepr {
       bitmap: [usize, BITS_TO_LONGS(NBITS_INPLACE)]
       ptr: NonNull<usize>,
   }

That would be a very useful addition for some particular scenarios,
I believe. Even if you don't want to make it configurable, let's
keep this option in mind?

Thanks,
Yury