Adds missing use core::ffi::c_int for when
sanitizer_cfi_normalize_integers is defined.

Some UI tests trigger behavior in rustc where it reads $CARGO and changes behavior if it exists.
To make the tests work that rely on it not being set, make sure it is not set.

By default, this is not set, but people may do weird hacks
that cause it to be set.

It has a single call site.

While a better approach would be to implement it for all ZSTs
which are `Copy` and have trivial `Clone`,
the last property cannot be detected for now.

Signed-off-by: Petr Portnov <me@progrm-jarvis.ru>

Signed-off-by: Petr Portnov <me@progrm-jarvis.ru>

Signed-off-by: Petr Portnov <me@progrm-jarvis.ru>

And remove `pub` from some local-only ones.

The majority of these aren't actually used, but I kept them anyway.

…kingjubilee

CFI: Add missing use core::ffi::c_int

Adds missing use core::ffi::c_int for when sanitizer_cfi_normalize_integers is defined.

Explicitly unset $CARGO for compiletest

Some UI tests trigger behavior in rustc where it reads $CARGO and changes behavior if it exists. To make the tests work that rely on it not being set, make sure it is not set.

By default, this is not set, but people may do weird hacks that cause it to be set.

closes rust-lang#118058

…iler-errors

`rustc_ty_utils` cleanups

Minor improvements I found while looking at this code.

r? ``@lcnr``

…-tuple, r=thomcc

feat: specialize `SpecFromElem` for `()`

# Description

This PR adds a specialization `SpecFromElem for ()` which allows to significantly reduce `vec![(), N]` time in debug builds (specifically, tests) turning it from observable $O(n)$ to $O(1)$.

# Observing the change

The problem this PR aims to fix explicitly is slow `vec![(), N]` on big `N`s which may appear in tests (see [Background section](#Background) for more details).

The minimal example to see the problem:

```rust
#![feature(test)]

extern crate test;

#[cfg(test)]
mod tests {
    const HUGE_SIZE: usize = i32::MAX as usize + 1;

    #[bench]
    fn bench_vec_literal(b: &mut test::Bencher) {
        b.iter(|| vec![(); HUGE_SIZE]);
    }

    #[bench]
    fn bench_vec_repeat(b: &mut test::Bencher) {
        b.iter(|| [(); 1].repeat(HUGE_SIZE));
    }
}
```
<details><summary>Output</summary>
<p>

```bash
cargo +nightly test -- --report-time -Zunstable-options
   Compiling huge-zst-vec-literal-bench v0.1.0 (/home/progrm_jarvis/RustroverProjects/huge-zst-vec-literal-bench)
    Finished test [unoptimized + debuginfo] target(s) in 0.31s
     Running unittests src/lib.rs (target/debug/deps/huge_zst_vec_literal_bench-e43b1ef287ba8b36)

running 2 tests
test tests::bench_vec_repeat  ... ok <0.000s>
test tests::bench_vec_literal ... ok <14.382s>

test result: ok. 2 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 14.38s

   Doc-tests huge-zst-vec-literal-bench

running 0 tests

test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
```
</p>
</details>

> [!IMPORTANT]
> This problem is only observable in Debug (unoptimized) builds, while Release (optimized) ones do not observe this problem. It still is worth fixing it, IMO, since the original scenario observes the problem in tests for which optimizations are disabled by default and it seems unreasonable to override this for the whole project while the problem is very local.

# Background

While working on a crate for a custom data format which has an `i32::MAX` limit on its list's sizes, I wrote the following test to ensure that this invariant is upheld:

```rust
#[test]
fn lists_should_have_i32_size() {
    assert!(
        RawNbtList::try_from(vec![(); i32::MAX as usize]).is_ok(),
        "lists should permit the size up to {}",
        i32::MAX
    );
    assert!(
        RawNbtList::try_from(vec![(); i32::MAX as usize + 1]).is_err(),
        "lists should have the size of at most {}",
        i32::MAX
    );
}
```

Soon I discovered that this takes $\approx 3--4s$ per assertion on my machine, almost all of which is spent on `vec![..]`.
While this would be logical for a non-ZST vector (which would require actual $O(n)$ allocation), here `()` was used intentionally considering that for ZSTs size-changing operations should anyway be $O(1)$ (at least from allocator perspective). Yet, this "overhead" is logical if we consider that in general case `clone()` (which is used by `Vec` literal) may have a non-trivial implementation and thus each element has to actually be visited (even if they occupy no space).

In my specific case, the following (contextual) equivalent solved the issue:

```rust
#[test]
fn lists_should_have_i32_size() {
    assert!(
        RawNbtList::try_from([(); 1].repeat(i32::MAX as usize)).is_ok(),
        "lists should permit the size up to {}",
        i32::MAX
    );
    assert!(
        RawNbtList::try_from([(); 1].repeat(i32::MAX as usize + 1)).is_err(),
        "lists should have the size of at most {}",
        i32::MAX
    );
}
```

which works since `repeat` explicitly uses `T: Copy` and so does not have to think about non-trivial `Clone`.

But it still may be counter-intuitive for users to observe such long time on the "canonical" vec literal thus the PR.

# Generic solution

This change is explicitly non-generic. Initially I considered it possible to implement in generically, but this would require the specialization to have the following type requirements:
- ✅ the type must be a ZST: easily done via
  ```rust
  if core::mem::size_of::<T>() == 0 {
    todo!("specialization")
  }
  ```
  or
  ```rust
  use core::mem::SizedTypeProperties;
  if T::IS_ZST {
    todo!("specialization")
  }
  ```
- :white_check_mark`: the type must implement `Copy`: implementable non-conflictable via a separate specialization:
  ```rust
  trait IsCopyZst: Sized {
    fn is_copy_zst() -> bool;
  }
  impl<T> IsCopyZst for T {
    fn is_copy_zst() -> bool {
        false
    }
  }
  impl<T: Copy> IsCopyZst for T {
    fn is_copy_zst() -> bool {
        Self::IS_ZST
    }
  }
  ```
- ❌ the type should have a trivial `Clone` implementation: since `vec![t; n]` is specified to use `clone()`, we can only use this "performance optimization" when we are guaranteed that `clone()` does nothing except for copying.

The latter is the real blocker for a generic fix since I am unaware of any way to get this information in a compiler-guaranteed way.

While there may be a fix for this (my friend `@CertainLach` has suggested a potential solution by an perma-unstable fn in `Clone` like `is_trivially_cloneable()` defaulting to `false` and only overridable by `rustc` on derive), this is surely out of this PRs scope.

Update books

## rust-lang/book

2 commits in 5b6c1ceaa62ecbd6caef08df39b33b3938e99deb..71352deb20727b4dda9ebfe8182709d5bf17dfea
2023-11-09 14:49:45 UTC to 2023-11-09 14:49:16 UTC

- Fixed 'Devtools' link (rust-lang/book#3770)
- Fix mdBook links (rust-lang/book#3769)

## rust-lang/rust-by-example

7 commits in 311b84962016b28c75525c86e7b3f49fd9101a39..a6581246f96837113968c02187db24f742af3908
2023-11-18 21:45:20 UTC to 2023-11-07 22:32:53 UTC

- rename `y` to `_y` to get the correct compile error (rust-lang/rust-by-example#1769)
- fix test name in cargo/test.md (rust-lang/rust-by-example#1768)
- Various minor edits for typo fixes, formatting fixes, and clarifications (rust-lang/rust-by-example#1765)
- Update closures.md to correct a typo (rust-lang/rust-by-example#1763)
- Link to the Bulgarian translation (rust-lang/rust-by-example#1764)
- Fix asm example explanation for `inlateout` usage (22.1 Inline Assembly) (rust-lang/rust-by-example#1766)
- Update index.md: Added descriptions for the 'leftover' points (rust-lang/rust-by-example#1767)

## rust-lang/rustc-dev-guide

3 commits in 77dbe5782b2488af3bb489ad702eaff438f465bf..ddb8b1309f9e905804cea1e248a4572fed6b464b
2023-11-18 21:08:13 UTC to 2023-11-08 14:43:50 UTC

- Add link for unsize.md (rust-lang/rustc-dev-guide#1825)
- Fix typo in contribution walkthrough (rust-lang/rustc-dev-guide#1824)
- Update documentation for coverage tests (rust-lang/rustc-dev-guide#1823)

…piler-errors

Fix occurrences of old fn names in comment and tracing

…errors

`rustc_hir` cleanups

Just some improvements I found while looking at this code.

r? `@WaffleLapkin`