add module-level documentation for vec's in-place iteration

This commit is contained in:
The8472 2021-07-31 17:39:35 +02:00 committed by The 8472
parent 79b43b35be
commit a1a602adde
5 changed files with 168 additions and 20 deletions

View File

@ -1401,6 +1401,8 @@ impl<T> ExactSizeIterator for IntoIter<T> {
#[stable(feature = "fused", since = "1.26.0")]
impl<T> FusedIterator for IntoIter<T> {}
// In addition to the SAFETY invariants of the following three unsafe traits
// also refer to the vec::in_place_collect module documentation to get an overview
#[unstable(issue = "none", feature = "inplace_iteration")]
#[doc(hidden)]
unsafe impl<T> SourceIter for IntoIter<T> {

View File

@ -1,3 +1,132 @@
//! Inplace iterate-and-collect specialization for `Vec`
//!
//! The specialization in this module applies to iterators in the shape of
//! `source.adapter().adapter().adapter().collect::<Vec<U>>()`
//! where `source` is an owning iterator obtained from [`Vec<T>`], [`Box<[T]>`] (by conversion to `Vec`)
//! or [`BinaryHeap<T>`], the adapters each consume one or more items per step
//! (represented by [`InPlaceIterable`]), provide transitive access to `source` (via [`SourceIter`])
//! and thus the underlying allocation. And finally the layouts of `T` and `U` must
//! have the same size and alignment, this is currently ensured via const eval instead of trait
//! bounds.
//!
//! [`BinaryHeap<T>`]: crate::collections::BinaryHeap
//! [`Box<[T]>`]: crate::boxed::Box
//!
//! By extension some other collections which use `collect::Vec<_>()` internally in their
//! `FromIterator` implementation benefit from this too.
//!
//! Access to the underlying source goes through a further layer of indirection via the private
//! trait [`AsIntoIter`] to hide the implementation detail that other collections may use
//! `vec::IntoIter` internally.
//!
//! In-place iteration depends on the interaction of several unsafe traits, implementation
//! details of multiple parts in the iterator pipeline and often requires holistic reasoning
//! across multiple structs since iterators are executed cooperatively rather than having
//! a central evaluator/visitor struct executing all iterator components.
//!
//! # Reading from and writing to the same allocation
//!
//! By its nature collecting in place means that the reader and writer side of the iterator
//! use the same allocation. Since `fold()` and co. take a reference to the iterator for the
//! duration of the iteration that means we can't interleave the step of reading a value
//! and getting a reference to write to. Instead raw pointers must be used on the reader
//! and writer side.
//!
//! That writes never clobber a yet-to-be-read item is ensured by the [`InPlaceIterable`] requirements.
//!
//! # Layout constraints
//!
//! [`Allocator`] requires that `allocate()` and `deallocate()` have matching alignment and size.
//! Additionally this specialization doesn't make sense for ZSTs as there is no reallocation to
//! avoid and it would make pointer arithmetic more difficult.
//!
//! [`Allocator`]: core::alloc::Allocator
//!
//! # Drop- and panic-safety
//!
//! Iteration can panic, requiring dropping the already written parts but also the remainder of
//! the source. Iteration can also leave some source items unconsumed which must be dropped.
//! All those drops in turn can panic which then must either leak the allocation or abort to avoid
//! double-drops.
//!
//! These tasks are handled by [`InPlaceDrop`] and [`vec::IntoIter::forget_allocation_drop_remaining()`]
//!
//! [`vec::IntoIter::forget_allocation_drop_remaining()`]: super::IntoIter::forget_allocation_drop_remaining()
//!
//! # O(1) collect
//!
//! The main iteration itself is further specialized when the iterator implements
//! [`TrustedRandomAccessNoCoerce`] to let the optimizer see that it is a counted loop with a single
//! induction variable. This can turn some iterators into a noop, i.e. it reduces them from O(n) to
//! O(1). This particular optimization is quite fickle and doesn't always work, see [#79308]
//!
//! [#79308]: https://github.com/rust-lang/rust/issues/79308
//!
//! Since unchecked accesses through that trait do not advance the read pointer of `IntoIter`
//! this would interact unsoundly with the requirements about dropping the tail described above.
//! But since the normal `Drop` implementation of `IntoIter` would suffer from the same problem it
//! is only correct for `TrustedRandomAccessNoCoerce` to be implemented when the items don't
//! have a destructor. Thus that implicit requirement also makes the specialization safe to use for
//! in-place collection.
//!
//! # Adapter implementations
//!
//! The invariants for adapters are documented in [`SourceIter`] and [`InPlaceIterable`], but
//! getting them right can be rather subtle for multiple, sometimes non-local reasons.
//! For example `InPlaceIterable` would be valid to implement for [`Peekable`], except
//! that it is stateful, cloneable and `IntoIter`'s clone implementation shortens the underlying
//! allocation which means if the iterator has been peeked and then gets cloned there no longer is
//! enough room, thus breaking an invariant (#85322).
//!
//! [#85322]: https://github.com/rust-lang/rust/issues/85322
//! [`Peekable`]: core::iter::Peekable
//!
//!
//! # Examples
//!
//! Some cases that are optimized by this specialization, more can be found in the `Vec`
//! benchmarks:
//!
//! ```rust
//! # #[allow(dead_code)]
//! /// Converts a usize vec into an isize one.
//! pub fn cast(vec: Vec<usize>) -> Vec<isize> {
//! // Does not allocate, free or panic. On optlevel>=2 it does not loop.
//! // Of course this particular case could and should be written with `into_raw_parts` and
//! // `from_raw_parts` instead.
//! vec.into_iter().map(|u| u as isize).collect()
//! }
//! ```
//!
//! ```rust
//! # #[allow(dead_code)]
//! /// Drops remaining items in `src` and if the layouts of `T` and `U` match it
//! /// returns an empty Vec backed by the original allocation. Otherwise it returns a new
//! /// empty vec.
//! pub fn recycle_allocation<T, U>(src: Vec<T>) -> Vec<U> {
//! src.into_iter().filter_map(|_| None).collect()
//! }
//! ```
//!
//! ```rust
//! let vec = vec![13usize; 1024];
//! let _ = vec.into_iter()
//! .enumerate()
//! .filter_map(|(idx, val)| if idx % 2 == 0 { Some(val+idx) } else {None})
//! .collect::<Vec<_>>();
//!
//! // is equivalent to the following, but doesn't require bounds checks
//!
//! let mut vec = vec![13usize; 1024];
//! let mut write_idx = 0;
//! for idx in 0..vec.len() {
//! if idx % 2 == 0 {
//! vec[write_idx] = vec[idx] + idx;
//! write_idx += 1;
//! }
//! }
//! vec.truncate(write_idx);
//! ```
use core::iter::{InPlaceIterable, SourceIter, TrustedRandomAccessNoCoerce};
use core::mem::{self, ManuallyDrop};
use core::ptr::{self};
@ -16,11 +145,8 @@ where
I: Iterator<Item = T> + SourceIter<Source: AsIntoIter> + InPlaceIterableMarker,
{
default fn from_iter(mut iterator: I) -> Self {
// Additional requirements which cannot expressed via trait bounds. We rely on const eval
// instead:
// a) no ZSTs as there would be no allocation to reuse and pointer arithmetic would panic
// b) size match as required by Alloc contract
// c) alignments match as required by Alloc contract
// See "Layout constraints" section in the module documentation. We rely on const
// optimization here since these conditions currently cannot be expressed as trait bounds
if mem::size_of::<T>() == 0
|| mem::size_of::<T>()
!= mem::size_of::<<<I as SourceIter>::Source as AsIntoIter>::Item>()
@ -58,21 +184,13 @@ where
);
}
// drop any remaining values at the tail of the source
// but prevent drop of the allocation itself once IntoIter goes out of scope
// if the drop panics then we also leak any elements collected into dst_buf
// Drop any remaining values at the tail of the source but prevent drop of the allocation
// itself once IntoIter goes out of scope.
// If the drop panics then we also leak any elements collected into dst_buf.
//
// FIXME: Since `SpecInPlaceCollect::collect_in_place` above might use
// `__iterator_get_unchecked` internally, this call might be operating on
// a `vec::IntoIter` with incorrect internal state regarding which elements
// have already been “consumed”. However, the `TrustedRandomIteratorNoCoerce`
// implementation of `vec::IntoIter` is only present if the `Vec` elements
// dont have a destructor, so it doesnt matter if elements are “dropped multiple times”
// in this case.
// This argument technically currently lacks justification from the `# Safety` docs for
// `SourceIter`/`InPlaceIterable` and/or `TrustedRandomAccess`, so it might be possible that
// someone could inadvertently create new library unsoundness
// involving this `.forget_allocation_drop_remaining()` call.
// Note: This access to the source wouldn't be allowed by the TrustedRandomIteratorNoCoerce
// contract (used by SpecInPlaceCollect below). But see the "O(1) collect" section in the
// module documenttation why this is ok anyway.
src.forget_allocation_drop_remaining();
let vec = unsafe { Vec::from_raw_parts(dst_buf, len, cap) };
@ -155,7 +273,21 @@ where
}
}
// internal helper trait for in-place iteration specialization.
/// Internal helper trait for in-place iteration specialization.
///
/// Currently this is only implemented by [`vec::IntoIter`] - returning a reference to itself - and
/// [`binary_heap::IntoIter`] which returns a reference to its inner representation.
///
/// Since this is an internal trait it hides the implementation detail `binary_heap::IntoIter`
/// uses `vec::IntoIter` internally.
///
/// [`vec::IntoIter`]: super::IntoIter
/// [`binary_heap::IntoIter`]: crate::collections::binary_heap::IntoIter
///
/// # Safety
///
/// In-place iteration relies on implementation details of `vec::IntoIter`, most importantly that
/// it does not create references to the whole allocation during iteration, only raw pointers
#[rustc_specialization_trait]
pub(crate) unsafe trait AsIntoIter {
type Item;

View File

@ -99,6 +99,9 @@ impl<T, A: Allocator> IntoIter<T, A> {
/// (&mut into_iter).for_each(core::mem::drop);
/// unsafe { core::ptr::write(&mut into_iter, Vec::new().into_iter()); }
/// ```
///
/// This method is used by in-place iteration, refer to the vec::in_place_collect
/// documentation for an overview.
#[cfg(not(no_global_oom_handling))]
pub(super) fn forget_allocation_drop_remaining(&mut self) {
let remaining = self.as_raw_mut_slice();
@ -325,6 +328,8 @@ unsafe impl<#[may_dangle] T, A: Allocator> Drop for IntoIter<T, A> {
}
}
// In addition to the SAFETY invariants of the following three unsafe traits
// also refer to the vec::in_place_collect module documentation to get an overview
#[unstable(issue = "none", feature = "inplace_iteration")]
#[doc(hidden)]
unsafe impl<T, A: Allocator> InPlaceIterable for IntoIter<T, A> {}

View File

@ -78,6 +78,11 @@ pub use self::zip::zip;
/// The trait is unsafe because implementers must uphold additional safety properties.
/// See [`as_inner`] for details.
///
/// The primary use of this trait is in-place iteration. Refer to the [`vec::in_place_collect`]
/// module documentation for more information.
///
/// [`vec::in_place_collect`]: ../../../../alloc/vec/in_place_collect/index.html
///
/// # Examples
///
/// Retrieving a partially consumed source:

View File

@ -51,6 +51,10 @@ unsafe impl<I: TrustedLen + ?Sized> TrustedLen for &mut I {}
/// in its place, assuming structural constraints of the source allow such an insertion.
/// In other words this trait indicates that an iterator pipeline can be collected in place.
///
/// The primary use of this trait is in-place iteration. Refer to the [`vec::in_place_collect`]
/// module documentation for more information.
///
/// [`vec::in_place_collect`]: ../../../../alloc/vec/in_place_collect/index.html
/// [`SourceIter`]: crate::iter::SourceIter
/// [`next()`]: Iterator::next
/// [`try_fold()`]: Iterator::try_fold