Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
References: <87edn6do1b.fsf@metaspace.dk>
 <20230524111127.2491643-1-aliceryhl@google.com>
User-agent: mu4e 1.10.3; emacs 28.2.50
From:   Andreas Hindborg <nmi@metaspace.dk>
To:     Alice Ryhl <aliceryhl@google.com>
Cc:     alex.gaynor@gmail.com, benno.lossin@proton.me,
        bjorn3_gh@protonmail.com, boqun.feng@gmail.com, gary@garyguo.net,
        jiangshanlai@gmail.com, linux-kernel@vger.kernel.org,
        ojeda@kernel.org, patches@lists.linux.dev,
        rust-for-linux@vger.kernel.org, tj@kernel.org, wedsonaf@gmail.com
Subject: Re: [PATCH v1 3/7] rust: sync: add `Arc::{from_raw, into_raw}`
Date:   Thu, 25 May 2023 09:45:29 +0200
In-reply-to: <20230524111127.2491643-1-aliceryhl@google.com>
Message-ID: <87a5xsetq5.fsf@metaspace.dk>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk


Alice Ryhl <aliceryhl@google.com> writes:

> Andreas Hindborg <nmi@metaspace.dk> writes:
>> Alice Ryhl <aliceryhl@google.com> writes:
>>> +        // This preserves the metadata in the pointer, if any.
>>> +        let metadata = core::ptr::metadata(ptr as *const ArcInner<T>);
>> 
>> I can't follow this. How does this work? `ptr` was for field
>> `inner.data: T`, but we are casting to `ArcInner<T>`.
>> 
>>> +        let ptr = (ptr as *mut u8).wrapping_sub(val_offset) as *mut ();
>>> +        let ptr = core::ptr::from_raw_parts_mut(ptr, metadata);
>> 
>> Metadata was obtained from a pointer pointing to `inner.data`, we then
>> move it back to beginning of `ArcInner<T>` and then reconstruct the
>> potentially fat pointer with metadata from the pointer to `T`? How can
>> this be right?
>
> The metadata of a struct is always the metadata of its last field, so
> both `*mut T` and `*mut ArcInner<T>` have the same metadata. Because of
> that, moving the metadata over from one type to the other is ok.
>
> The reason that I cast to an `ArcInner<T>` pointer before calling
> `metadata` is because I get a type mismatch otherwise for the metadata,
> since the compiler doesn't unify the metadata types when the type is
> generic.

OK, cool. In that case, since this is common knowledge (is it?),
could you maybe include a link to the relevant documentation, or a
comment indicating why this is OK?

BR Andreas