Re: [PATCH v3 1/1] rust: introduce abstractions for fwctl

[Danilo Krummrich]

On Tue Mar 3, 2026 at 9:15 PM CET, Jason Gunthorpe wrote:
> Anyhow is someone waiting for me to do something with it?

I still have this on my list for review, I can probably do a pass tomorrow. :)

>> +impl<T: Operations> Device<T> {
>> +    /// Allocate a new fwctl device with embedded driver data.
>> +    ///
>> +    /// Returns an [`ARef`] that can be passed to [`Registration::new()`]
>> +    /// to make the device visible to userspace. The caller may inspect or
>> +    /// configure the device between allocation and registration.
>> +    pub fn new(
>> +        parent: &device::Device<device::Bound>,
>> +        data: impl PinInit<T::DeviceData, Error>,
>> +    ) -> Result<ARef<Self>> {
>> +        let ops = core::ptr::from_ref::<bindings::fwctl_ops>(&VTable::<T>::VTABLE).cast_mut();
>> +
>> +        // SAFETY: `_fwctl_alloc_device` allocates `size` bytes via kzalloc and
>> +        // initialises the embedded fwctl_device. `ops` points to a static vtable
>> +        // that outlives the device. `parent` is bound.
>> +        let raw = unsafe {
>> +            bindings::_fwctl_alloc_device(parent.as_raw(), ops, core::mem::size_of::<Self>())
>> +        };
>> +
>> +        if raw.is_null() {
>> +            return Err(ENOMEM);
>> +        }
>> +
>> +        // CAST: Device<T> is repr(C) with fwctl_device at offset 0.
>> +        let this = raw as *mut Self;
>
> Should this have some helper? It looks a bit fragile, in C we have a
> static_assert on offsetof to prevent errors.

We could add a helper with something like this:

	    const {
	        assert!(
	            core::mem::offset_of!(Self, dev) == 0,
	            "struct fwctl_device must be at offset 0"
	        )
	    };

But we could also just include this in this constructor.