linux/rust/pin-init

pin-init

Note

This crate was originally named pinned-init, but the migration to pin-init is not yet complete. The legacy branch contains the current version of the pinned-init crate & the main branch already incorporates the rename to pin-init.

There are still some changes needed on the kernel side before the migration can be completed.

Library to safely and fallibly initialize pinned structs using in-place constructors.

Pinning is Rust's way of ensuring data does not move.

It also allows in-place initialization of big structs that would otherwise produce a stack overflow.

This library's main use-case is in Rust-for-Linux. Although this version can be used standalone.

There are cases when you want to in-place initialize a struct. For example when it is very big and moving it from the stack is not an option, because it is bigger than the stack itself. Another reason would be that you need the address of the object to initialize it. This stands in direct conflict with Rust's normal process of first initializing an object and then moving it into it's final memory location. For more information, see https://rust-for-linux.com/the-safe-pinned-initialization-problem.

This library allows you to do in-place initialization safely.

Nightly Needed for alloc feature

This library requires the allocator_api unstable feature when the alloc feature is enabled and thus this feature can only be used with a nightly compiler. When enabling the alloc feature, the user will be required to activate allocator_api as well.

The feature is enabled by default, thus by default pin-init will require a nightly compiler. However, using the crate on stable compilers is possible by disabling alloc. In practice this will require the std feature, because stable compilers have neither Box nor Arc in no-std mode.

Nightly needed for unsafe-pinned feature

This feature enables the Wrapper implementation on the unstable core::pin::UnsafePinned type. This requires the unsafe_pinned unstable feature and therefore a nightly compiler. Note that this feature is not enabled by default.

Overview

To initialize a struct with an in-place constructor you will need two things:

• an in-place constructor,
• a memory location that can hold your struct (this can be the stack, an Arc<T>, Box<T> or any other smart pointer that supports this library).

To get an in-place constructor there are generally three options:

• directly creating an in-place constructor using the [pin_init!] macro,
• a custom function/macro returning an in-place constructor provided by someone else,
• using the unsafe function [pin_init_from_closure()] to manually create an initializer.

Aside from pinned initialization, this library also supports in-place construction without pinning, the macros/types/functions are generally named like the pinned variants without the pin_ prefix.

Examples

Throughout the examples we will often make use of the CMutex type which can be found in ../examples/mutex.rs. It is essentially a userland rebuild of the struct mutex type from the Linux kernel. It also uses a wait list and a basic spinlock. Importantly the wait list requires it to be pinned to be locked and thus is a prime candidate for using this library.

Using the [pin_init!] macro

If you want to use [PinInit], then you will have to annotate your struct with #[[pin_data]]. It is a macro that uses #[pin] as a marker for structurally pinned fields. After doing this, you can then create an in-place constructor via [pin_init!]. The syntax is almost the same as normal struct initializers. The difference is that you need to write <- instead of : for fields that you want to initialize in-place.

use pin_init::{pin_data, pin_init, InPlaceInit};

#[pin_data]
struct Foo {
    #[pin]
    a: CMutex<usize>,
    b: u32,
}

let foo = pin_init!(Foo {
    a <- CMutex::new(42),
    b: 24,
});

foo now is of the type impl PinInit<Foo>. We can now use any smart pointer that we like (or just the stack) to actually initialize a Foo:

let foo: Result<Pin<Box<Foo>>, AllocError> = Box::pin_init(foo);

For more information see the [pin_init!] macro.

Using a custom function/macro that returns an initializer

Many types that use this library supply a function/macro that returns an initializer, because the above method only works for types where you can access the fields.

let mtx: Result<Pin<Arc<CMutex<usize>>>, _> = Arc::pin_init(CMutex::new(42));

To declare an init macro/function you just return an impl PinInit<T, E>:

#[pin_data]
struct DriverData {
    #[pin]
    status: CMutex<i32>,
    buffer: Box<[u8; 1_000_000]>,
}

impl DriverData {
    fn new() -> impl PinInit<Self, Error> {
        try_pin_init!(Self {
            status <- CMutex::new(0),
            buffer: Box::init(pin_init::init_zeroed())?,
        }? Error)
    }
}

Manual creation of an initializer

Often when working with primitives the previous approaches are not sufficient. That is where [pin_init_from_closure()] comes in. This unsafe function allows you to create a impl PinInit<T, E> directly from a closure. Of course you have to ensure that the closure actually does the initialization in the correct way. Here are the things to look out for (we are calling the parameter to the closure slot):

• when the closure returns Ok(()), then it has completed the initialization successfully, so slot now contains a valid bit pattern for the type T,
• when the closure returns Err(e), then the caller may deallocate the memory at slot, so you need to take care to clean up anything if your initialization fails mid-way,
• you may assume that slot will stay pinned even after the closure returns until drop of slot gets called.

use pin_init::{pin_data, pinned_drop, PinInit, PinnedDrop, pin_init_from_closure};
use core::{
    ptr::addr_of_mut,
    marker::PhantomPinned,
    cell::UnsafeCell,
    pin::Pin,
    mem::MaybeUninit,
};
mod bindings {
    #[repr(C)]
    pub struct foo {
        /* fields from C ... */
    }
    extern "C" {
        pub fn init_foo(ptr: *mut foo);
        pub fn destroy_foo(ptr: *mut foo);
        #[must_use = "you must check the error return code"]
        pub fn enable_foo(ptr: *mut foo, flags: u32) -> i32;
    }
}

/// # Invariants
///
/// `foo` is always initialized
#[pin_data(PinnedDrop)]
pub struct RawFoo {
    #[pin]
    _p: PhantomPinned,
    #[pin]
    foo: UnsafeCell<MaybeUninit<bindings::foo>>,
}

impl RawFoo {
    pub fn new(flags: u32) -> impl PinInit<Self, i32> {
        // SAFETY:
        // - when the closure returns `Ok(())`, then it has successfully initialized and
        //   enabled `foo`,
        // - when it returns `Err(e)`, then it has cleaned up before
        unsafe {
            pin_init_from_closure(move |slot: *mut Self| {
                // `slot` contains uninit memory, avoid creating a reference.
                let foo = addr_of_mut!((*slot).foo);
                let foo = UnsafeCell::raw_get(foo).cast::<bindings::foo>();

                // Initialize the `foo`
                bindings::init_foo(foo);

                // Try to enable it.
                let err = bindings::enable_foo(foo, flags);
                if err != 0 {
                    // Enabling has failed, first clean up the foo and then return the error.
                    bindings::destroy_foo(foo);
                    Err(err)
                } else {
                    // All fields of `RawFoo` have been initialized, since `_p` is a ZST.
                    Ok(())
                }
            })
        }
    }
}

#[pinned_drop]
impl PinnedDrop for RawFoo {
    fn drop(self: Pin<&mut Self>) {
        // SAFETY: Since `foo` is initialized, destroying is safe.
        unsafe { bindings::destroy_foo(self.foo.get().cast::<bindings::foo>()) };
    }
}

For more information on how to use [pin_init_from_closure()], take a look at the uses inside the kernel crate. The sync module is a good starting point.