Re: [PATCH v2] rust: pin-init: add `#[bind]` attribute to access previously initialized fields

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

On Wed Sep 10, 2025 at 5:06 PM CEST, Boqun Feng wrote:
> On Wed, Sep 10, 2025 at 04:59:11PM +0200, Benno Lossin wrote:
>> On Wed Sep 10, 2025 at 4:02 PM CEST, Boqun Feng wrote:
>> > On Wed, Sep 10, 2025 at 02:19:00PM +0200, Benno Lossin wrote:
>> >> On Wed Sep 10, 2025 at 12:40 PM CEST, Alice Ryhl wrote:
>> >> > On Wed, Sep 10, 2025 at 12:36 PM Benno Lossin <lossin@kernel.org> wrote:
>> >> >>
>> >> >> On Wed Sep 10, 2025 at 12:17 PM CEST, Alice Ryhl wrote:
>> >> >> > On Wed, Sep 10, 2025 at 12:07:53PM +0200, Benno Lossin wrote:
>> >> >> >> Assigning a field a value in an initializer macro can be marked with the
>> >> >> >> `#[bind]` attribute. Doing so creates a `let` binding with the same
>> >> >> >> name. This `let` binding has the type `Pin<&mut T>` if the field is
>> >> >> >> structurally pinned or `&mut T` otherwise (where `T` is the type of the
>> >> >> >> field).
>> >> >> >>
>> >> >> >> Signed-off-by: Benno Lossin <lossin@kernel.org>
>> >> >> >
>> >> >> > Is there a reason we can't apply this to all fields and avoid the
>> >> >> > attribute?
>> >> >>
>> >> >> Adding the attribute was due to Boqun's concern on v1 [1]. I think it
>> >> >> might be surprising too, but I'm also happy with no attribute.
>> >> >>
>> >> >> [1]: https://lore.kernel.org/all/aLshd0_C-1rh3FAg@tardis-2.local
>> >> >
>> >> > IMO the ideal is if it works without an attribute. Perhaps trying that
>> >> > in the kernel is a reasonable experiment to find out whether that's
>> >> > reasonable to do for the general language feature?
>> >> 
>> >> @Boqun what is your opinion on this?
>> >> 
>> >
>> > If we plan to make the in-place initializer language feature behave
>> > similar, as I asked here [1], then dropping `#[bind]` seems good to me.
>> 
>> I don't think we can promise how that language feature is going to look
>> like. It will definitely support accessing already initialized fields,
>> but in what form remains to be seen.
>> 
>
> Sure, but in kernel we are able to stay the same as whatever the
> language feature will be like, right?

Yes :)

> In other words, as long as we propose the same thing to the language
> feature and keep kernel code and language feature synced (presumbly
> there could be some more discussion on the syntax when presented to Rust
> commmunity), then I'm think it's fine.

I'll take v1 (with the in-tree fixes) then. Do you mind giving your RB
for that one? Thanks!

---
Cheers,
Benno