Re: [PATCH 3/4] rust: devres: get rid of Devres' inner Arc

["Benno Lossin" <lossin@kernel.org>]

On Sun Jun 22, 2025 at 2:08 PM CEST, Danilo Krummrich wrote:
> On Sun, Jun 22, 2025 at 09:05:51AM +0200, Benno Lossin wrote:
>> On Thu Jun 12, 2025 at 4:51 PM CEST, Danilo Krummrich wrote:
>> > +#[pinned_drop]
>> > +impl<T> PinnedDrop for Devres<T> {
>> > +    fn drop(self: Pin<&mut Self>) {
>> >          // SAFETY: When `drop` runs, it is guaranteed that nobody is accessing the revocable data
>> >          // anymore, hence it is safe not to wait for the grace period to finish.
>> > -        if unsafe { self.0.data.revoke_nosync() } {
>> > -            // We revoked `self.0.data` before the devres action did, hence try to remove it.
>> > -            if !DevresInner::remove_action(&self.0) {
>> > +        if unsafe { self.data.revoke_nosync() } {
>> > +            // We revoked `self.data` before the devres action did, hence try to remove it.
>> > +            if !self.remove_action() {
>> >                  // We could not remove the devres action, which means that it now runs concurrently,
>> > -                // hence signal that `self.0.data` has been revoked successfully.
>> > -                self.0.revoke.complete_all();
>> > +                // hence signal that `self.data` has been revoked by us successfully.
>> > +                self.revoke.complete_all();
>> > +
>> > +                // Wait for `Self::devres_callback` to be done using this object.
>> > +                self.devm.wait_for_completion();
>> >              }
>> > +        } else {
>> > +            // `Self::devres_callback` revokes `self.data` for us, hence wait for it to be done
>> > +            // using this object.
>> > +            self.devm.wait_for_completion();
>> 
>> I don't understand this change, maybe it's best to move that into a
>> separate commit?
>
> We can't do that, without this change the code would be incorrect.
>
> What happens here is that, if drop() races with devres_callback() we have to
> make drop() wait until devres_callback() is completed, because otherwise
> devres_callback() might experience a use-after-free.
>
> Previoulsly this has been taken care of by Arc<DevresInner>, which C devres held
> a reference of.

Yeah I understand it now, the diff was adding too much noise and looking
at it directly was helpful :)

Theoretically, you could add it in a commit before removing the Arc, but
probably not worth it.

---
Cheers,
Benno