Re: [PATCH V9 09/17] rust: cpu: Add from_cpu()

[Danilo Krummrich <dakr@kernel.org>]

On Fri, Apr 11, 2025 at 12:18:26PM -0400, Yury Norov wrote:
> On Fri, Apr 11, 2025 at 04:25:08PM +0530, Viresh Kumar wrote:
> > +/// Creates a new instance of CPU's device.
> > +///
> > +/// # Safety
> > +///
> > +/// Reference counting is not implemented for the CPU device in the C code. When a CPU is
> > +/// hot-unplugged, the corresponding CPU device is unregistered, but its associated memory
> > +/// is not freed.
> > +///
> > +/// Callers must ensure that the CPU device is not used after it has been unregistered.
> > +/// This can be achieved, for example, by registering a CPU hotplug notifier and removing
> > +/// any references to the CPU device within the notifier's callback.
> > +pub unsafe fn from_cpu(cpu: u32) -> Result<&'static Device> {
> > +    // SAFETY: The pointer returned by `get_cpu_device()`, if not `NULL`, is a valid pointer to
> > +    // a `struct device` and is never freed by the C code.
> > +    let ptr = unsafe { bindings::get_cpu_device(cpu) };

You don't need to justify the return value here, you need to justify why cpu is
a valid value to call get_cpu_device() with.

> > +    if ptr.is_null() {
> > +        return Err(ENODEV);
> > +    }
> > +
> > +    // SAFETY: The pointer returned by `get_cpu_device()`, if not `NULL`, is a valid pointer to
> > +    // a `struct device` and is never freed by the C code.

Here, the comment is good.

> If you think that writing the same thing twice would help in SAFETY
> department, I'd recommend you to write it for a couple more times.

Fixing the above, should also fix the duplication. :)

> > +    Ok(unsafe { Device::as_ref(ptr) })