On Tue, Jun 17, 2025 at 03:54:19PM +0200, Danilo Krummrich wrote:
> On Tue, Jun 17, 2025 at 01:36:47PM +0000, Alice Ryhl wrote:
> > Since commit b20fbbc08a36 ("rust: check type of `$ptr` in
> > `container_of!`") we have enforced that the field pointer passed to
> > container_of! must match the declared field. This caused mismatches when
> > using a pointer to bindings::x for fields of type Opaque<bindings::x>.
> >
> > This situation encourages the user to simply pass field.cast() to the
> > container_of! macro, but this is not great because you might
> > accidentally pass a *mut bindings::y when the field type is
> > Opaque<bindings::x>, which would be wrong.
> >
> > To help catch this kind of mistake, add a new Opaque::from_raw that
> > wraps a raw pointer in Opaque without changing the inner type.
>
> The patch does more than that, it also adds a hint to container_of!() and
> fixes
> up two occurences. I feel like we should split it up.
I think they go together pretty naturally, but I can split it if you
insist.
> > + /// The opposite operation of [`Opaque::raw_get`].
> > + pub const fn from_raw(this: *const T) -> *const Self {
>
> Do we want to name this from_raw()? Usually from_raw() methods return either
> Self or &'a Self.
>
> Maybe something like cast_from() and rename raw_get() to cast_into()? I think
> the latter may be confusing anyways, since it sounds like it would do somthing
> with reference counting.
The name raw_get() mirrors the stdlib function UnsafeCell::raw_get().
The stdlib uses this naming because in Rust the word "get" normally has
nothing to do with reference counting - outside of the kernel, we use
"clone" for incrementing refcounts and nobody would ever call it "get".
That said, it may still be worth to rename the method. Thoughts?
Alice
