I am confused about the requirement of Pin<P> that
a value, once pinned, must remain pinned forever.
I saw the discussion thread on reddit but after reading through it I remain confused.
To me, it feels more natural if we require the pointee to be pinned when the Pin<P> object is alive, but can re-acquire the ability to move it again after the Pin<P> object is dropped.
As a concrete example,
use std::mem;
use std::pin::Pin;
fn move_pinned_ref<T>(mut a: T, mut b: T) {
unsafe {
let p: Pin<&mut T> = Pin::new_unchecked(&mut a);
// *I prefer* it to mean that the pointee `a` cannot move when `p` is in the scope.
// But *actually* it means the pointee `a` can never move again.
}
mem::swap(&mut a, &mut b);
// *I prefer* it to be valid because `p` has been dropped.
// But we *actually* have violated the pinning API contract that a value,
// once pinned, must remain pinned forever.
}
Can someone please point out why the design I prefer is problematic?
One important reason for this rule is that it allows self-referential structs to remain valid, even after the Pin is dropped. For example:
use std::marker::PhantomPinned;
use std::pin::Pin;
use std::ptr::NonNull;
struct Unmovable {
data: String,
slice: Option<NonNull<String>>,
_pin: PhantomPinned,
}
impl Unmovable {
fn update_slice(self: Pin<&mut Self>) {
unsafe {
let self_mut = Pin::get_unchecked_mut(self);
self_mut.slice = Some(NonNull::from(&mut self_mut.data));
}
}
fn print_slice(&self) {
if let Some(s) = self.slice {
unsafe {
println!("{}", s.as_ref());
}
}
}
}
fn main() {
let mut a = Unmovable {
data: "Hello, world!".into(),
slice: None,
_pin: PhantomPinned,
};
let p = unsafe { Pin::new_unchecked(&mut a) };
p.update_slice(); // This causes `p` to be dropped.
// If we move `a`, even after dropping the Pin:
let x2 = Box::new(a);
// Now x2 has a dangling pointer!
x2.print_slice() // <-- Undefined behavior!
}
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With