Intended folder structure for subsystem bindings

Intended folder structure for subsystem bindings

[Philipp Stanner]

Hi,

I'm having a look into potentially creating bindings for the PCI
subsystem.

I was wondering what the Rust infrastructure people had in mind here.
Should there be one folder per subsystem in rust/bindings?

Thanks,
P.

Re: Intended folder structure for subsystem bindings

[Miguel Ojeda]

On Thu, Jan 18, 2024 at 10:26 AM Philipp Stanner <pstanner@redhat.com> wrote:
>
> I'm having a look into potentially creating bindings for the PCI
> subsystem.
>
> I was wondering what the Rust infrastructure people had in mind here.
> Should there be one folder per subsystem in rust/bindings?

Currently, abstractions go into the `kernel` crate at `rust/kernel`.
You would typically create a Rust module in that crate (which may be a
file or a folder).

In the future, we will have the ability to have Rust code in their
usual folders among the C code.

`rust/bindings` only holds the header list, the generated bits plus a
bit of glue.

Cheers,
Miguel

Re: Intended folder structure for subsystem bindings

[Trevor Gross]

On Thu, Jan 18, 2024 at 4:20 AM Philipp Stanner <pstanner@redhat.com> wrote:
>
> Hi,
>
> I'm having a look into potentially creating bindings for the PCI
> subsystem.

Not sure what exactly you are looking for but as a starting point,
Tomo has a basic abstraction at [1]. Based on the discussion at that
thread, it sounds like Andreas might be using it for the NVMe driver.

Do you have any specific project in mind? We can probably point you in
the right direction if you need any other abstractions.

- Trevor

[1]: https://github.com/Rust-for-Linux/linux/pull/856

> I was wondering what the Rust infrastructure people had in mind here.
> Should there be one folder per subsystem in rust/bindings?
>
> Thanks,
> P.
>
>