Struct ManuallyDrop

pub struct ManuallyDrop<T>
where
    T: ?Sized,
{ /* private fields */ }

A wrapper to inhibit the compiler from automatically calling T’s destructor. This wrapper is 0-cost.

ManuallyDrop<T> is guaranteed to have the same layout and bit validity as T, and is subject to the same layout optimizations as T. As a consequence, it has no effect on the assumptions that the compiler makes about its contents. For example, initializing a ManuallyDrop<&mut T> with mem::zeroed is undefined behavior. If you need to handle uninitialized data, use MaybeUninit<T> instead.

Note that accessing the value inside a ManuallyDrop<T> is safe. This means that a ManuallyDrop<T> whose content has been dropped must not be exposed through a public safe API. Correspondingly, ManuallyDrop::drop is unsafe.

ManuallyDrop and drop order

Rust has a well-defined drop order of values. To make sure that fields or locals are dropped in a specific order, reorder the declarations such that the implicit drop order is the correct one.

It is possible to use ManuallyDrop to control the drop order, but this requires unsafe code and is hard to do correctly in the presence of unwinding.

For example, if you want to make sure that a specific field is dropped after the others, make it the last field of a struct:

struct Context;

struct Widget {
    children: Vec<Widget>,
    // `context` will be dropped after `children`.
    // Rust guarantees that fields are dropped in the order of declaration.
    context: Context,
}

Interaction with Box

Currently, if you have a ManuallyDrop<T>, where the type T is a Box or contains a Box inside, then dropping the T followed by moving the ManuallyDrop<T> is considered to be undefined behavior. That is, the following code causes undefined behavior:

use std::mem::ManuallyDrop;

let mut x = ManuallyDrop::new(Box::new(42));
unsafe {
    ManuallyDrop::drop(&mut x);
}
let y = x; // Undefined behavior!

This is likely to change in the future. In the meantime, consider using MaybeUninit instead.

Safety hazards when storing ManuallyDrop in a struct or an enum.

Special care is needed when all of the conditions below are met:

• A struct or enum contains a ManuallyDrop.
• The ManuallyDrop is not inside a union.
• The struct or enum is part of public API, or is stored in a struct or an enum that is part of public API.
• There is code that drops the contents of the ManuallyDrop field, and this code is outside the struct or enum’s Drop implementation.

In particular, the following hazards may occur:

Storing generic types

If the ManuallyDrop contains a client-supplied generic type, the client might provide a Box as that type. This would cause undefined behavior when the struct or enum is later moved, as mentioned in the previous section. For example, the following code causes undefined behavior:

use std::mem::ManuallyDrop;

pub struct BadOption<T> {
    // Invariant: Has been dropped iff `is_some` is false.
    value: ManuallyDrop<T>,
    is_some: bool,
}
impl<T> BadOption<T> {
    pub fn new(value: T) -> Self {
        Self { value: ManuallyDrop::new(value), is_some: true }
    }
    pub fn change_to_none(&mut self) {
        if self.is_some {
            self.is_some = false;
            unsafe {
                // SAFETY: `value` hasn't been dropped yet, as per the invariant
                // (This is actually unsound!)
                ManuallyDrop::drop(&mut self.value);
            }
        }
    }
}

// In another crate:

let mut option = BadOption::new(Box::new(42));
option.change_to_none();
let option2 = option; // Undefined behavior!

Deriving traits

Deriving Debug, Clone, PartialEq, PartialOrd, Ord, or Hash on the struct or enum could be unsound, since the derived implementations of these traits would access the ManuallyDrop field. For example, the following code causes undefined behavior:

use std::mem::ManuallyDrop;

// This derive is unsound in combination with the `ManuallyDrop::drop` call.
#[derive(Debug)]
pub struct Foo {
    value: ManuallyDrop<String>,
}
impl Foo {
    pub fn new() -> Self {
        let mut temp = Self {
            value: ManuallyDrop::new(String::from("Unsafe rust is hard."))
        };
        unsafe {
            // SAFETY: `value` hasn't been dropped yet.
            ManuallyDrop::drop(&mut temp.value);
        }
        temp
    }
}

// In another crate:

let foo = Foo::new();
println!("{:?}", foo); // Undefined behavior!

Implementations

impl<T> ManuallyDrop<T>

pub const fn new(value: T) -> ManuallyDrop<T>

Wrap a value to be manually dropped.

Examples

use std::mem::ManuallyDrop;
let mut x = ManuallyDrop::new(String::from("Hello World!"));
x.truncate(5); // You can still safely operate on the value
assert_eq!(*x, "Hello");
// But `Drop` will not be run here

pub const fn into_inner(slot: ManuallyDrop<T>) -> T

Extracts the value from the ManuallyDrop container.

This allows the value to be dropped again.

Examples

use std::mem::ManuallyDrop;
let x = ManuallyDrop::new(Box::new(()));
let _: Box<()> = ManuallyDrop::into_inner(x); // This drops the `Box`.

pub unsafe fn take(slot: &mut ManuallyDrop<T>) -> T

Takes the value from the ManuallyDrop<T> container out.

This method is primarily intended for moving out values in drop. Instead of using ManuallyDrop::drop to manually drop the value, you can use this method to take the value and use it however desired.

Whenever possible, it is preferable to use into_inner instead, which prevents duplicating the content of the ManuallyDrop<T>.

Safety

This function semantically moves out the contained value without preventing further usage, leaving the state of this container unchanged. It is your responsibility to ensure that this ManuallyDrop is not used again.

impl<T> ManuallyDrop<T>

where T: ?Sized,

pub unsafe fn drop(slot: &mut ManuallyDrop<T>)

Manually drops the contained value.

This is exactly equivalent to calling ptr::drop_in_place with a pointer to the contained value. As such, unless the contained value is a packed struct, the destructor will be called in-place without moving the value, and thus can be used to safely drop pinned data.

If you have ownership of the value, you can use ManuallyDrop::into_inner instead.

Safety

This function runs the destructor of the contained value. Other than changes made by the destructor itself, the memory is left unchanged, and so as far as the compiler is concerned still holds a bit-pattern which is valid for the type T.

However, this “zombie” value should not be exposed to safe code, and this function should not be called more than once. To use a value after it’s been dropped, or drop a value multiple times, can cause Undefined Behavior (depending on what drop does). This is normally prevented by the type system, but users of ManuallyDrop must uphold those guarantees without assistance from the compiler.

Trait Implementations

impl<T> AsBytes for ManuallyDrop<T>

where T: AsBytes + ?Sized,

fn as_bytes(&self) -> &[u8]

Gets the bytes of this value.

fn as_bytes_mut(&mut self) -> &mut [u8]

where Self: FromBytes,

Gets the bytes of this value mutably.

fn write_to(&self, bytes: &mut [u8]) -> Option<()>

Writes a copy of self to bytes.

fn write_to_prefix(&self, bytes: &mut [u8]) -> Option<()>

Writes a copy of self to the prefix of bytes.

fn write_to_suffix(&self, bytes: &mut [u8]) -> Option<()>

Writes a copy of self to the suffix of bytes.

impl<T> Clone for ManuallyDrop<T>

where T: Clone + ?Sized,

fn clone(&self) -> ManuallyDrop<T>

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<T> Debug for ManuallyDrop<T>

where T: Debug + ?Sized,

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl<T> Default for ManuallyDrop<T>

where T: Default + ?Sized,

fn default() -> ManuallyDrop<T>

Returns the “default value” for a type.

impl<T> DerefMut for ManuallyDrop<T>

where T: ?Sized,

fn deref_mut(&mut self) -> &mut T

Mutably dereferences the value.

impl<T> FromBytes for ManuallyDrop<T>

where T: FromBytes + ?Sized,

fn ref_from(bytes: &[u8]) -> Option<&Self>

where Self: Sized,

Interprets the given bytes as a &Self without copying.

fn ref_from_prefix(bytes: &[u8]) -> Option<&Self>

where Self: Sized,

Interprets the prefix of the given bytes as a &Self without copying.

fn ref_from_suffix(bytes: &[u8]) -> Option<&Self>

where Self: Sized,

Interprets the suffix of the given bytes as a &Self without copying.

fn mut_from(bytes: &mut [u8]) -> Option<&mut Self>

where Self: Sized + AsBytes,

Interprets the given bytes as a &mut Self without copying.

fn mut_from_prefix(bytes: &mut [u8]) -> Option<&mut Self>

where Self: Sized + AsBytes,

Interprets the prefix of the given bytes as a &mut Self without copying.

fn mut_from_suffix(bytes: &mut [u8]) -> Option<&mut Self>

where Self: Sized + AsBytes,

Interprets the suffix of the given bytes as a &mut Self without copying.

fn slice_from(bytes: &[u8]) -> Option<&[Self]>

where Self: Sized,

Interprets the given bytes as a &[Self] without copying.

fn slice_from_prefix(bytes: &[u8], count: usize) -> Option<(&[Self], &[u8])>

where Self: Sized,

Interprets the prefix of the given bytes as a &[Self] with length equal to count without copying.

fn slice_from_suffix(bytes: &[u8], count: usize) -> Option<(&[u8], &[Self])>

where Self: Sized,

Interprets the suffix of the given bytes as a &[Self] with length equal to count without copying.

fn mut_slice_from(bytes: &mut [u8]) -> Option<&mut [Self]>

where Self: Sized + AsBytes,

Interprets the given bytes as a &mut [Self] without copying.

fn mut_slice_from_prefix( bytes: &mut [u8], count: usize, ) -> Option<(&mut [Self], &mut [u8])>

where Self: Sized + AsBytes,

Interprets the prefix of the given bytes as a &mut [Self] with length equal to count without copying.

fn mut_slice_from_suffix( bytes: &mut [u8], count: usize, ) -> Option<(&mut [u8], &mut [Self])>

where Self: Sized + AsBytes,

Interprets the suffix of the given bytes as a &mut [Self] with length equal to count without copying.

fn read_from(bytes: &[u8]) -> Option<Self>

where Self: Sized,

Reads a copy of Self from bytes.

fn read_from_prefix(bytes: &[u8]) -> Option<Self>

where Self: Sized,

Reads a copy of Self from the prefix of bytes.

fn read_from_suffix(bytes: &[u8]) -> Option<Self>

where Self: Sized,

Reads a copy of Self from the suffix of bytes.

impl<T> FromZeroes for ManuallyDrop<T>

where T: FromZeroes + ?Sized,

fn zero(&mut self)

Overwrites self with zeroes.

fn new_zeroed() -> Self

where Self: Sized,

Creates an instance of Self from zeroed bytes.

impl<T> Hash for ManuallyDrop<T>

where T: Hash + ?Sized,

fn hash<__H>(&self, state: &mut __H)

where __H: Hasher,

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl<T> Ord for ManuallyDrop<T>

where T: Ord + ?Sized,

fn cmp(&self, other: &ManuallyDrop<T>) -> Ordering

This method returns an Ordering between self and other.

fn max(self, other: Self) -> Self

where Self: Sized,

Compares and returns the maximum of two values.

fn min(self, other: Self) -> Self

where Self: Sized,

Compares and returns the minimum of two values.

fn clamp(self, min: Self, max: Self) -> Self

where Self: Sized,

Restrict a value to a certain interval.

impl<T> PartialEq for ManuallyDrop<T>

where T: PartialEq + ?Sized,

fn eq(&self, other: &ManuallyDrop<T>) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl<T> PartialOrd for ManuallyDrop<T>

where T: PartialOrd + ?Sized,

fn partial_cmp(&self, other: &ManuallyDrop<T>) -> Option<Ordering>

This method returns an ordering between self and other values if one exists.

fn lt(&self, other: &Rhs) -> bool

Tests less than (for self and other) and is used by the < operator.

fn le(&self, other: &Rhs) -> bool

Tests less than or equal to (for self and other) and is used by the <= operator.

fn gt(&self, other: &Rhs) -> bool

Tests greater than (for self and other) and is used by the > operator.

fn ge(&self, other: &Rhs) -> bool

Tests greater than or equal to (for self and other) and is used by the >= operator.

impl<T> Deref for ManuallyDrop<T>

where T: ?Sized,

type Target = T

The resulting type after dereferencing.

fn deref(&self) -> &T

Dereferences the value.

impl<T> Copy for ManuallyDrop<T>

where T: Copy + ?Sized,

impl<T> DerefPure for ManuallyDrop<T>

where T: ?Sized,

impl<T> Eq for ManuallyDrop<T>

where T: Eq + ?Sized,

impl<T> StructuralPartialEq for ManuallyDrop<T>

where T: ?Sized,

impl<T> Unaligned for ManuallyDrop<T>

where T: Unaligned + ?Sized,