Re: [PATCH v10 03/13] rust: hrtimer: implement `HrTimerPointer` for `Arc`

[Andreas Hindborg <a.hindborg@kernel.org>]

"Benno Lossin" <benno.lossin@proton.me> writes:

> On Fri Mar 7, 2025 at 2:27 PM CET, Andreas Hindborg wrote:
>> "Benno Lossin" <benno.lossin@proton.me> writes:
>>
>>> On Fri Mar 7, 2025 at 11:11 AM CET, Andreas Hindborg wrote:
>>>> +impl<T> HrTimerPointer for Arc<T>
>>>> +where
>>>> +    T: 'static,
>>>> +    T: Send + Sync,
>>>> +    T: HasHrTimer<T>,
>>>> +    T: for<'a> HrTimerCallback<Pointer<'a> = Self>,
>>>> +    Arc<T>: for<'a> RawHrTimerCallback<CallbackTarget<'a> = ArcBorrow<'a, T>>,
>>>
>>> I don't understand why you need this bound here.
>>
>> This impl is applicable only when `Arc<T> has an implementation of
>> `RawTimerCallback` where CallbackTarget<'a> = ArcBorrow<'a, T>. I don't
>> want the impl to be available if that is not the case.
>
> The impl below has less strict other bounds than this one, so this bound
> doesn't change anything.
>
>> It's just an additional check.
>
> To me it's just additional noise.

I'll drop it then.

>
>>>> +{
>>>> +    type TimerHandle = ArcHrTimerHandle<T>;
>>>> +
>>>> +    fn start(self, expires: Ktime) -> ArcHrTimerHandle<T> {
>>>> +        // SAFETY:
>>>> +        //  - We keep `self` alive by wrapping it in a handle below.
>>>> +        //  - Since we generate the pointer passed to `start` from a valid
>>>> +        //    reference, it is a valid pointer.
>>>> +        unsafe { T::start(Arc::as_ptr(&self), expires) };
>>>> +        ArcHrTimerHandle { inner: self }
>>>> +    }
>>>> +}
>>>> +
>>>> +impl<T> RawHrTimerCallback for Arc<T>
>>>> +where
>>>> +    T: 'static,
>>>> +    T: HasHrTimer<T>,
>>>> +    T: for<'a> HrTimerCallback<Pointer<'a> = Self>,
>>>> +{
>>>> +    type CallbackTarget<'a> = ArcBorrow<'a, T>;
>>>> +
>>>> +    unsafe extern "C" fn run(ptr: *mut bindings::hrtimer) -> bindings::hrtimer_restart {
>>>> +        // `HrTimer` is `repr(C)`
>>>> +        let timer_ptr = ptr.cast::<super::HrTimer<T>>();
>>>> +
>>>> +        // SAFETY: By C API contract `ptr` is the pointer we passed when
>>>> +        // queuing the timer, so it is a `HrTimer<T>` embedded in a `T`.
>>>> +        let data_ptr = unsafe { T::timer_container_of(timer_ptr) };
>>>> +
>>>> +        // SAFETY: `data_ptr` points to the `T` that was used to queue the
>>>> +        // timer. This `T` is contained in an `Arc`.
>>>
>>> You're not justifying all safety requirements of `ArcBorrow::from_raw`.
>>
>> How is this:
>>
>>         // SAFETY:
>>         //  - `data_ptr` is derived form the pointer to the `T` that was used to
>>         //    queue the timer.
>>         //  - The `ArcTimerHandle` associated with this timer is guaranteed to
>>         //    be alive for the duration of the lifetime of `receiver`, so the
>
> There is no `receiver` in this context?

It's the value returned from the call, same line.

>
> Is the reason for the handle staying alive that when it is dropped, it
> calls `cancel` and that waits until the callback finishes? If so, did
> you write that down somewhere here?

Yes, it is in the safety requirement of the `HrTimerHandle` trait.
Should I add that? It becomes quite a story.

>
>>         //    refcount of the underlying `Arc` is guaranteed to be nonzero for
>>         //    the duration.
>>         //  - We own one refcount in the `ArcTimerHandle` associted with this
>>         //    timer, so it is not possible to get a `UniqueArc` to this
>>         //    allocation from other `Arc` clones.
>
> Otherwise this sounds good.

Cool.


Best regards,
Andreas Hindborg