Message-ID: <56ebda7b-c570-4dc6-8456-ab768d3a4b77@proton.me>
Date: Thu, 15 Aug 2024 13:24:47 +0000
From: Benno Lossin <benno.lossin@...ton.me>
To: Danilo Krummrich <dakr@...nel.org>
Cc: ojeda@...nel.org, alex.gaynor@...il.com, wedsonaf@...il.com, boqun.feng@...il.com, gary@...yguo.net, bjorn3_gh@...tonmail.com, a.hindborg@...sung.com, aliceryhl@...gle.com, akpm@...ux-foundation.org, daniel.almeida@...labora.com, faith.ekstrand@...labora.com, boris.brezillon@...labora.com, lina@...hilina.net, mcanal@...lia.com, zhiw@...dia.com, cjia@...dia.com, jhubbard@...dia.com, airlied@...hat.com, ajanulgu@...hat.com, lyude@...hat.com, linux-kernel@...r.kernel.org, rust-for-linux@...r.kernel.org, linux-mm@...ck.org
Subject: Re: [PATCH v5 09/26] rust: alloc: implement kernel `Box`

On 14.08.24 23:58, Danilo Krummrich wrote:
> On Wed, Aug 14, 2024 at 05:01:34PM +0000, Benno Lossin wrote:
>> On 12.08.24 20:22, Danilo Krummrich wrote:
>>> +/// The kernel's [`Box`] type - a heap allocation for a single value of type `T`.
>>> +///
>>> +/// This is the kernel's version of the Rust stdlib's `Box`. There are a couple of differences,
>>> +/// for example no `noalias` attribute is emitted and partially moving out of a `Box` is not
>>> +/// supported.
>>
>> I would add "But otherwise it works the same." (I don't know if there is
>> a comma needed after the "otherwise").
> 
> There are more differences we don't list here, and probably don't need to.
> Hence, saying that it otherwise works the same isn't correct.
> 
>> Also I remember that there was one more difference with a custom box
>> compared to the stdlib, but I forgot what that was, does someone else
>> remember? We should also put that here.
> 
> Obviously, there are also quite some API differences. For instance, `Box`
> generally requires two generics, value type and allocator, we take page flags
> and return a `Result`, where std just panics on failure.

Oh yeah that's true. The things listed above don't really refer to API
stuff, so I didn't consider that. How about changing "couple
differences" to "several differences"? Also adding that the APIs are
different would not hurt.

>>> +///
>>> +/// `Box` works with any of the kernel's allocators, e.g. [`super::allocator::Kmalloc`],
>>> +/// [`super::allocator::Vmalloc`] or [`super::allocator::KVmalloc`]. There are aliases for `Box`
>>> +/// with these allocators ([`KBox`], [`VBox`], [`KVBox`]).
>>> +///
>>> +/// When dropping a [`Box`], the value is also dropped and the heap memory is automatically freed.
>>> +///
>>> +/// # Examples
>>> +///
>>> +/// ```
>>> +/// let b = KBox::<u64>::new(24_u64, GFP_KERNEL)?;
>>> +///
>>> +/// assert_eq!(*b, 24_u64);
>>> +///
>>> +/// # Ok::<(), Error>(())
>>> +/// ```
>>> +///
>>> +/// ```
>>> +/// # use kernel::bindings;
>>> +///
>>> +/// const SIZE: usize = bindings::KMALLOC_MAX_SIZE as usize + 1;
>>> +/// struct Huge([u8; SIZE]);
>>> +///
>>> +/// assert!(KBox::<Huge>::new_uninit(GFP_KERNEL | __GFP_NOWARN).is_err());
>>> +/// ```
>>> +///
>>> +/// ```
>>> +/// # use kernel::bindings;
>>> +///
>>> +/// const SIZE: usize = bindings::KMALLOC_MAX_SIZE as usize + 1;
>>> +/// struct Huge([u8; SIZE]);
>>> +///
>>> +/// assert!(KVBox::<Huge>::new_uninit(GFP_KERNEL).is_ok());
>>> +/// ```
>>> +///
>>> +/// # Invariants
>>> +///
>>> +/// The [`Box`]' pointer always properly aligned and either points to memory allocated with `A` or,
>>
>> "pointer always properly" -> "pointer is properly"
>>
>>> +/// for zero-sized types, is a dangling pointer.
>>
>> I think this section would look nicer, if it were formatted using bullet
>> points (that way the bracketing of the "or" is also unambiguous).
>>
>> Additionally, this is missing that the pointer is valid for reads and
>> writes.
>>
>>> +pub struct Box<T: ?Sized, A: Allocator>(NonNull<T>, PhantomData<A>);
>>
>> Why no `repr(transparent)`?
> 
> I wasn't entirely sure whether that's OK with the additional `PhantomData`, but
> I think it is, gonna add it.

Yes it is fine, `repr(transparent)` requires that at most one field is
non-ZST, but the type can have as many ZST fields as it wants.
Otherwise the compiler will complain (there is no `unsafe` here, so just
adding it is completely fine).

>>> +
>>> +/// Type alias for `Box` with a `Kmalloc` allocator.
>>
>> I think we should add that this is only designed for small values.
> 
> I don't want duplicate the existing documentation around kmalloc and friends
> [1].
> 
> Maybe we can refer to the existing documentation somehow.
> 
> [1] https://www.kernel.org/doc/html/latest/core-api/memory-allocation.html

Oh great! With the C docs, I never know where to find them (is it in the
code and do they exist?). Yeah let's just link it.

>>> +///
>>> +/// # Examples
>>> +///
>>> +/// ```
>>> +/// let b = KBox::new(24_u64, GFP_KERNEL)?;
>>> +///
>>> +/// assert_eq!(*b, 24_u64);
>>> +///
>>> +/// # Ok::<(), Error>(())
>>> +/// ```
>>> +pub type KBox<T> = Box<T, super::allocator::Kmalloc>;
>>> +
>>> +/// Type alias for `Box` with a `Vmalloc` allocator.
>>
>> Same here, add that this is supposed to be used for big values (or is
>> this also a general-purpose allocator, just not guaranteeing that the
>> memory is physically contiguous? in that case I would document it
>> here and also on `Vmalloc`).
> 
> Same as above, I'd rather not duplicate that. But I'm happy to link things in,
> just not sure what's the best way doing it.

I took a look at the link and there is the "Selecting memory allocator"
section, but there isn't really just a vmalloc or kmalloc section, it is
rather stuff that we would put in the module documentation.
What I would write on these types would be what to use these boxes for.
eg large allocations, general purpose etc. I don't think that that is
easily accessible from the docs that you linked above.

---
Cheers,
Benno

Powered by blists - more mailing lists