Message-ID: <038748e0-22ce-4455-ba08-e8ae30e357df@proton.me>
Date: Thu, 01 Feb 2024 09:48:04 +0000
From: Benno Lossin <benno.lossin@...ton.me>
To: Alice Ryhl <aliceryhl@...gle.com>
Cc: Miguel Ojeda <ojeda@...nel.org>, Alex Gaynor <alex.gaynor@...il.com>, Wedson Almeida Filho <wedsonaf@...il.com>, Boqun Feng <boqun.feng@...il.com>, Gary Guo <gary@...yguo.net>, Björn Roy Baron <bjorn3_gh@...tonmail.com>, Andreas Hindborg <a.hindborg@...sung.com>, Peter Zijlstra <peterz@...radead.org>, Alexander Viro <viro@...iv.linux.org.uk>, Christian Brauner <brauner@...nel.org>, Greg Kroah-Hartman <gregkh@...uxfoundation.org>, Arve Hjønnevåg <arve@...roid.com>, Todd Kjos <tkjos@...roid.com>, Martijn Coenen <maco@...roid.com>, Joel Fernandes <joel@...lfernandes.org>, Carlos Llamas <cmllamas@...gle.com>, Suren Baghdasaryan <surenb@...gle.com>, Dan Williams <dan.j.williams@...el.com>, Kees Cook <keescook@...omium.org>, Matthew Wilcox <willy@...radead.org>, Thomas Gleixner <tglx@...utronix.de>, Daniel Xu <dxu@...uu.xyz>, linux-kernel@...r.kernel.org, rust-for-linux@...r.kernel.org, linux-fsdevel@...r.kernel.org
Subject: Re: [PATCH v3 1/9] rust: file: add Rust abstraction for `struct file`

On 01.02.24 10:41, Alice Ryhl wrote:
> On Thu, Feb 1, 2024 at 10:38 AM Benno Lossin <benno.lossin@...tonme> wrote:
>>
>> On 01.02.24 10:33, Alice Ryhl wrote:
>>> On Thu, Feb 1, 2024 at 10:31 AM Benno Lossin <benno.lossin@...ton.me> wrote:
>>>>
>>>> On 29.01.24 17:34, Alice Ryhl wrote:
>>>>> On Fri, Jan 26, 2024 at 4:04 PM Benno Lossin <benno.lossin@...ton.me> wrote:
>>>>>>> +///   closed.
>>>>>>> +/// * A light refcount must be dropped before returning to userspace.
>>>>>>> +#[repr(transparent)]
>>>>>>> +pub struct File(Opaque<bindings::file>);
>>>>>>> +
>>>>>>> +// SAFETY: By design, the only way to access a `File` is via an immutable reference or an `ARef`.
>>>>>>> +// This means that the only situation in which a `File` can be accessed mutably is when the
>>>>>>> +// refcount drops to zero and the destructor runs. It is safe for that to happen on any thread, so
>>>>>>> +// it is ok for this type to be `Send`.
>>>>>>
>>>>>> Technically, `drop` is never called for `File`, since it is only used
>>>>>> via `ARef<File>` which calls `dec_ref` instead. Also since it only contains
>>>>>> an `Opaque`, dropping it is a noop.
>>>>>> But what does `Send` mean for this type? Since it is used together with
>>>>>> `ARef`, being `Send` means that `File::dec_ref` can be called from any
>>>>>> thread. I think we are missing this as a safety requirement on
>>>>>> `AlwaysRefCounted`, do you agree?
>>>>>> I think the safety justification here could be (with the requirement added
>>>>>> to `AlwaysRefCounted`):
>>>>>>
>>>>>>         SAFETY:
>>>>>>         - `File::drop` can be called from any thread.
>>>>>>         - `File::dec_ref` can be called from any thread.
>>>>>
>>>>> This wording was taken from rust/kernel/task.rs. I think it's out of
>>>>> scope to reword it.
>>>>
>>>> Rewording the safety docs on `AlwaysRefCounted`, yes that is out of scope,
>>>> I was just checking if you agree that the current wording is incomplete.
>>>
>>> That's not what I meant. The wording of this safety comment is
>>> identical to the wording in other existing safety comments in the
>>> kernel, such as e.g. the one for `impl Send for Task`.
>>
>> Ah I see. But I still think changing it is better, since it would only get
>> shorter. The comment on `Task` can be fixed later.
>> Or do you want to keep consistency here? Because I would prefer to make
>> this right and then change `Task` later.
> 
> What would you like me to change it to?
> 
> For example:
> // SAFETY: It is okay to send references to a File across thread boundaries.

That would fit better as the safety comment for `Sync`, since
it refers to "references".

For `Send` I think this would be better:
// SAFETY:
// - `File::dec_ref` can be called from any thread.
// - It is okay to send ownership of `File` across thread boundaries.

-- 
Cheers,
Benno

Powered by blists - more mailing lists