Remove -Zoom=panic
There are major questions remaining about the reentrancy that this allows. It doesn't have any users on github outside of a single project that uses it in a panic=abort project to show backtraces. It can still be emulated through `#[alloc_error_handler]` or `set_alloc_error_hook` depending on if you use the standard library or not. And finally it makes it harder to do various improvements to the allocator shim.
With this PR the sole remaining symbol in the allocator shim that is not effectively emulating weak symbols is the symbol that prevents skipping the allocator shim on stable even when it would otherwise be empty because libstd + `#[global_allocator]` is used.
Closes https://github.com/rust-lang/rust/issues/43596
Fixes https://github.com/rust-lang/rust/issues/126683
This commit consolidates all changes, including the core logic fix for IntoIter::nth_back and the addition of the Miri regression test in `library/alloctests/tests/vec.rs`, to prevent Undefined Behavior (UB) when dealing with highly-aligned Zero-Sized Types.
alloc: Document panics when allocations will exceed max
Document panics in `String` and `Vec` due to capacity overflowing `isize::MAX`. Correct outdated claims of `usize::MAX` limit.
Fixes https://github.com/rust-lang/rust/issues/148598.
Ping `@lolbinarycat`
In `BTreeMap::eq`, do not compare the elements if the sizes are different.
Reverts rust-lang/rust#147101 in library/alloc/src/btree/
rust-lang/rust#147101 replaced some instances of code like `a.len() == b.len() && a.iter().eq(&b)` with just `a.iter().eq(&b)`, but the optimization that PR introduced only applies for `TrustedLen` iterators, and `BTreeMap`'s itertors are not `TrustedLen`, so this theoretically regressed perf for comparing large `BTreeMap`/`BTreeSet`s with unequal lengths but equal prefixes, (and also made it so that comparing two different-length `BTreeMap`/`BTreeSet`s with elements whose `PartialEq` impls that can panic now can panic, though this is not a "promised" behaviour either way (cc rust-lang/rust#149122))
Given that `TrustedLen` is an unsafe trait, I opted to not implement it for `BTreeMap`'s iterators, and instead just revert the change. If someone else wants to audit `BTreeMap`'s iterators to make sure they always return the right number of items (even in the face of incorrect user `Ord` impls) and then implement `TrustedLen` for them so that the optimization works for them, then this can be closed in favor of that (or if the perf regression is deemed too theoretical, this can be closed outright).
Example of theoretical perf regression: https://play.rust-lang.org/?version=beta&mode=release&edition=2024&gist=a37e3d61e6bf02669b251315c9a44fe2 (very rough estimates, using `Instant::elapsed`).
In release mode on stable the comparison takes ~23.68µs.
In release mode on beta/nightly the comparison takes ~48.351057ms.
There are major questions remaining about the reentrancy that this
allows. It doesn't have any users on github outside of a single project
that uses it in a panic=abort project to show backtraces. It
can still be emulated through #[alloc_error_handler] or
set_alloc_error_hook depending on if you use the standard library or
not. And finally it makes it harder to do various improvements to the
allocator shim.
Add `Box::clone_from_ref` and similar under `feature(clone_from_ref)`
Tracking issue: https://github.com/rust-lang/rust/issues/149075
Accepted ACP: https://github.com/rust-lang/libs-team/issues/483
This PR implements `clone_from_ref` (and `try_*` and `_*in` variants), to get a `Box<T>`, `Arc<T>`, or `Rc<T>` by cloning from a `&T` where `T: CloneToUninit`.
The "Implement..." commits replace some ad-hoc conversions with `clone_from_ref` variants, which can be split out to a separate PR if desired.
This PR will conflict with https://github.com/rust-lang/rust/pull/148769 due to usage of `Layout::dangling` (which that PR is renaming to `dangling_ptr`), so they should not be rolled up together, and the one which merges later will need to be amended.
alloc: fix `Debug` implementation of `ExtractIf`
I noticed this while reviewing rust-lang/rust#141032. Calling `get` on the inner `Vec` never returns `Some` as the `Vec`'s length is temporarily set to zero while the `ExtractIf` exists.
Stabilize vec_into_raw_parts
This stabilizes `Vec::into_raw_parts()` and `String::into_raw_parts()` per FCP in https://github.com/rust-lang/rust/issues/65816#issuecomment-3517630971. While this _does not_ stabilize `Vec::into_parts()`, I fixed up the examples that said they were waiting for `vec_into_raw_parts`. As `Vec::from_parts()` and `Vec::into_parts()` are covered by the same feature `box_vec_non_null`, any doctest that uses `Vec::from_parts()` can also use `Vec::into_parts()` (and same for allocator-aware versions).
Closesrust-lang/rust#65816
``@rustbot`` modify labels: +T-libs-api
Expose fmt::Arguments::from_str as unstable.
Now that https://github.com/rust-lang/rust/pull/148789 is merged, we can have a fmt::Arguments::from_str. I don't know if we want to commit to always having an implementation that allows for this, but we can expose it as unstable for now so we can play with it.
Tracking issue: https://github.com/rust-lang/rust/issues/148905
stop specializing on `Copy`
fixes https://github.com/rust-lang/rust/issues/132442
`std` specializes on `Copy` to optimize certain library functions such as `clone_from_slice`. This is unsound, however, as the `Copy` implementation may not be always applicable because of lifetime bounds, which specialization does not take into account; the result being that values are copied even though they are not `Copy`. For instance, this code:
```rust
struct SometimesCopy<'a>(&'a Cell<bool>);
impl<'a> Clone for SometimesCopy<'a> {
fn clone(&self) -> Self {
self.0.set(true);
Self(self.0)
}
}
impl Copy for SometimesCopy<'static> {}
let clone_called = Cell::new(false);
// As SometimesCopy<'clone_called> is not 'static, this must run `clone`,
// setting the value to `true`.
let _ = [SometimesCopy(&clone_called)].clone();
assert!(clone_called.get());
```
should not panic, but does ([playground](https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=6be7a48cad849d8bd064491616fdb43c)).
To solve this, this PR introduces a new `unsafe` trait: `TrivialClone`. This trait may be implemented whenever the `Clone` implementation is equivalent to copying the value (so e.g. `fn clone(&self) -> Self { *self }`). Because of lifetime erasure, there is no way for the `Clone` implementation to observe lifetime bounds, meaning that even if the `TrivialClone` has stricter bounds than the `Clone` implementation, its invariant still holds. Therefore, it is sound to specialize on `TrivialClone`.
I've changed all `Copy` specializations in the standard library to specialize on `TrivialClone` instead. Unfortunately, the unsound `#[rustc_unsafe_specialization_marker]` attribute on `Copy` cannot be removed in this PR as `hashbrown` still depends on it. I'll make a PR updating `hashbrown` once this lands.
With `Copy` no longer being considered for specialization, this change alone would result in the standard library optimizations not being applied for user types unaware of `TrivialClone`. To avoid this and restore the optimizations in most cases, I have changed the expansion of `#[derive(Clone)]`: Currently, whenever both `Clone` and `Copy` are derived, the `clone` method performs a copy of the value. With this PR, the derive macro also adds a `TrivialClone` implementation to make this case observable using specialization. I anticipate that most users will use `#[derive(Clone, Copy)]` whenever both are applicable, so most users will still profit from the library optimizations.
Unfortunately, Hyrum's law applies to this PR: there are some popular crates which rely on the precise specialization behaviour of `core` to implement "specialization at home", e.g. [`libAFL`](89cff63702/libafl_bolts/src/tuples.rs (L27-L49)). I have no remorse for breaking such horrible code, but perhaps we should open other, better ways to satisfy their needs – for example by dropping the `'static` bound on `TypeId::of`...
Implement IsZero for ().
Implement default `IsZero` for all arrays, only returning true if the array is empty
(making the existing array impl for `IsZero` elements a specialization).
Optimize `IsZero::is_zero` for arrays of zero-sized `IsZero` elements.
add extend_front to VecDeque with specialization like extend
ACP: https://github.com/rust-lang/libs-team/issues/658
Tracking issue: rust-lang/rust#146975
_Text below was written before opening the ACP_
Feature was requested in rust-lang/rust#69939, I recently also needed it so decided to implement it as my first contribution to the Rust standard library. I plan on doing more but wanted to start with a small change.
Some questions I had (both on implementation and design) with answers:
- Q: `extend` allows iterators that yield `&T` where `T` is `Clone`, should extend_front do too?
A: No, users can use [`copied`](https://doc.rust-lang.org/std/iter/trait.Iterator.html#method.copied) and/or [`cloned`](https://doc.rust-lang.org/std/iter/trait.Iterator.html#method.cloned).
- Q: Does this need a whole new trait like Extend or only a method on `VecDeque`?
A: No, see ACP.
- Q: How do I deal with all the code duplication? Most code is similar to that of `extend`, maybe there is a nice way to factor out the code around `push_unchecked`/`push_front_unchecked`.
Will come back to this later.
- Q: Why are certain things behind feature gates, `cfg(not(test))` like `vec::IntoIter` here and `cfg(not(no_global_oom_handling))` like `Vec::extend_from_within`? (I am also looking at implementing `VecDeque::extend_from_within`)
A: See https://github.com/rust-lang/rust/pull/146861#pullrequestreview-3250163369
- Q: Should `extend_front` act like repeated pushes to the front of the queue? This reverses the order of the elements. Doing it different might incur an extra move if the iterator length is not known up front (where do you start placing elements in the buffer?).
A: `extend_front` acts like repeated pushes, `prepend` preserves the element order, see ACP or tracking issue.
