Auto merge of #67648 - Mark-Simulacrum:rollup-dttcu1y, r=Mark-Simulacrum

Rollup of 12 pull requests

Successful merges:

 - #67112 (Refactor expression parsing thoroughly)
 - #67192 (Various const eval and pattern matching ICE fixes)
 - #67287 (typeck: note other end-point when checking range pats)
 - #67459 (prune ill-conceived BTreeMap iter_mut assertion and test its mutability)
 - #67576 (reuse `capacity` variable in slice::repeat)
 - #67602 (Use issue = "none" instead of "0" in intrinsics)
 - #67614 (Set callbacks globally)
 - #67617 (Remove `compiler_builtins_lib` documentation)
 - #67629 (Remove redundant link texts)
 - #67632 (Convert collapsed to shortcut reference links)
 - #67633 (Update .mailmap)
 - #67635 (Document safety of Path casting)

Failed merges:

r? @ghost

.mailmap

@@ -54,9 +54,10 @@ Chris C Cerami <chrisccerami@users.noreply.github.com> Chris C Cerami <chrisccer
 Chris Pressey <cpressey@gmail.com>
 Chris Thorn <chris@thorn.co> Chris Thorn <thorn@thoughtbot.com>
 Chris Vittal <christopher.vittal@gmail.com> Christopher Vittal <christopher.vittal@gmail.com>
-Christian Poveda <christianpoveda@protonmail.com> <cn.poveda.ruiz@gmail.com>
-Christian Poveda <christianpoveda@protonmail.com> <z1mvader@protonmail.com>
-Christian Poveda <christianpoveda@protonmail.com> <cpovedar@fnal.gov>
+Christian Poveda <git@christianpoveda.xyz> <christianpoveda@protonmail.com>
+Christian Poveda <git@christianpoveda.xyz> <cn.poveda.ruiz@gmail.com>
+Christian Poveda <git@christianpoveda.xyz> <z1mvader@protonmail.com>
+Christian Poveda <git@christianpoveda.xyz> <cpovedar@fnal.gov>
 Clark Gaebel <cg.wowus.cg@gmail.com> <cgaebel@mozilla.com>
 Clinton Ryan <clint.ryan3@gmail.com>
 Corey Richardson <corey@octayn.net> Elaine "See More" Nemo <corey@octayn.net>

CONTRIBUTING.md

@@ -367,7 +367,7 @@ labels to triage issues:
   to fix the issue.
 
 * The dark blue **final-comment-period** label marks bugs that are using the
-  RFC signoff functionality of [rfcbot][rfcbot] and are currently in the final
+  RFC signoff functionality of [rfcbot] and are currently in the final
   comment period.
 
 * Red, **I**-prefixed labels indicate the **importance** of the issue. The
@@ -385,7 +385,7 @@ labels to triage issues:
   label.
 
 * The gray **proposed-final-comment-period** label marks bugs that are using
-  the RFC signoff functionality of [rfcbot][rfcbot] and are currently awaiting
+  the RFC signoff functionality of [rfcbot] and are currently awaiting
   signoff of all team members in order to enter the final comment period.
 
 * Pink, **regression**-prefixed labels track regressions from stable to the

RELEASES.md

@@ -4951,10 +4951,10 @@ Stabilized APIs
 ---------------
 
 * [`std::panic`]
-* [`std::panic::catch_unwind`][] (renamed from `recover`)
-* [`std::panic::resume_unwind`][] (renamed from `propagate`)
-* [`std::panic::AssertUnwindSafe`][] (renamed from `AssertRecoverSafe`)
-* [`std::panic::UnwindSafe`][] (renamed from `RecoverSafe`)
+* [`std::panic::catch_unwind`] (renamed from `recover`)
+* [`std::panic::resume_unwind`] (renamed from `propagate`)
+* [`std::panic::AssertUnwindSafe`] (renamed from `AssertRecoverSafe`)
+* [`std::panic::UnwindSafe`] (renamed from `RecoverSafe`)
 * [`str::is_char_boundary`]
 * [`<*const T>::as_ref`]
 * [`<*mut T>::as_ref`]
@@ -5234,18 +5234,18 @@ Libraries
 ---------
 
 * Stabilized APIs:
-  * [`str::encode_utf16`][] (renamed from `utf16_units`)
-  * [`str::EncodeUtf16`][] (renamed from `Utf16Units`)
+  * [`str::encode_utf16`] (renamed from `utf16_units`)
+  * [`str::EncodeUtf16`] (renamed from `Utf16Units`)
   * [`Ref::map`]
   * [`RefMut::map`]
   * [`ptr::drop_in_place`]
   * [`time::Instant`]
   * [`time::SystemTime`]
   * [`Instant::now`]
-  * [`Instant::duration_since`][] (renamed from `duration_from_earlier`)
+  * [`Instant::duration_since`] (renamed from `duration_from_earlier`)
   * [`Instant::elapsed`]
   * [`SystemTime::now`]
-  * [`SystemTime::duration_since`][] (renamed from `duration_from_earlier`)
+  * [`SystemTime::duration_since`] (renamed from `duration_from_earlier`)
   * [`SystemTime::elapsed`]
   * Various `Add`/`Sub` impls for `Time` and `SystemTime`
   * [`SystemTimeError`]
@@ -5432,8 +5432,8 @@ Libraries
 
 * Stabilized APIs
   * `Path`
-    * [`Path::strip_prefix`][] (renamed from relative_from)
-    * [`path::StripPrefixError`][] (new error type returned from strip_prefix)
+    * [`Path::strip_prefix`] (renamed from relative_from)
+    * [`path::StripPrefixError`] (new error type returned from strip_prefix)
   * `Ipv4Addr`
     * [`Ipv4Addr::is_loopback`]
     * [`Ipv4Addr::is_private`]
@@ -5646,7 +5646,7 @@ Libraries
 
 * Stabilized APIs:
   [`Read::read_exact`],
-  [`ErrorKind::UnexpectedEof`][] (renamed from `UnexpectedEOF`),
+  [`ErrorKind::UnexpectedEof`] (renamed from `UnexpectedEOF`),
   [`fs::DirBuilder`], [`fs::DirBuilder::new`],
   [`fs::DirBuilder::recursive`], [`fs::DirBuilder::create`],
   [`os::unix::fs::DirBuilderExt`],
@@ -5659,11 +5659,11 @@ Libraries
   [`collections::hash_set::HashSet::drain`],
   [`collections::binary_heap::Drain`],
   [`collections::binary_heap::BinaryHeap::drain`],
-  [`Vec::extend_from_slice`][] (renamed from `push_all`),
+  [`Vec::extend_from_slice`] (renamed from `push_all`),
   [`Mutex::get_mut`], [`Mutex::into_inner`], [`RwLock::get_mut`],
   [`RwLock::into_inner`],
-  [`Iterator::min_by_key`][] (renamed from `min_by`),
-  [`Iterator::max_by_key`][] (renamed from `max_by`).
+  [`Iterator::min_by_key`] (renamed from `min_by`),
+  [`Iterator::max_by_key`] (renamed from `max_by`).
 * The [core library][1.6co] is stable, as are most of its APIs.
 * [The `assert_eq!` macro supports arguments that don't implement
   `Sized`][1.6ae], such as arrays. In this way it behaves more like

src/doc/unstable-book/src/language-features/lang-items.md

@@ -188,11 +188,7 @@ pub extern fn rust_begin_panic(info: &PanicInfo) -> ! {
 
 In many cases, you may need to manually link to the `compiler_builtins` crate
 when building a `no_std` binary. You may observe this via linker error messages
-such as "```undefined reference to `__rust_probestack'```". Using this crate
-also requires enabling the library feature `compiler_builtins_lib`. You can read
-more about this [here][compiler-builtins-lib].
-
-[compiler-builtins-lib]: ../library-features/compiler-builtins-lib.md
+such as "```undefined reference to `__rust_probestack'```".
 
 ## More about the language items
 

src/liballoc/collections/binary_heap.rs

@@ -1094,7 +1094,7 @@ impl<T> FusedIterator for Iter<'_, T> {}
 
 /// An owning iterator over the elements of a `BinaryHeap`.
 ///
-/// This `struct` is created by the [`into_iter`] method on [`BinaryHeap`][`BinaryHeap`]
+/// This `struct` is created by the [`into_iter`] method on [`BinaryHeap`]
 /// (provided by the `IntoIterator` trait). See its documentation for more.
 ///
 /// [`into_iter`]: struct.BinaryHeap.html#method.into_iter

src/liballoc/collections/btree/map.rs

@@ -283,7 +283,7 @@ pub struct IterMut<'a, K: 'a, V: 'a> {
 
 /// An owning iterator over the entries of a `BTreeMap`.
 ///
-/// This `struct` is created by the [`into_iter`] method on [`BTreeMap`][`BTreeMap`]
+/// This `struct` is created by the [`into_iter`] method on [`BTreeMap`]
 /// (provided by the `IntoIterator` trait). See its documentation for more.
 ///
 /// [`into_iter`]: struct.BTreeMap.html#method.into_iter

src/liballoc/collections/btree/node.rs

@@ -493,7 +493,7 @@ impl<'a, K, V, Type> NodeRef<marker::Mut<'a>, K, V, Type> {
 
     /// Returns a raw ptr to avoid asserting exclusive access to the entire node.
     fn as_leaf_mut(&mut self) -> *mut LeafNode<K, V> {
-        // We are mutable, so we cannot be the root, so accessing this as a leaf is okay.
+        // We are mutable, so we cannot be the shared root, so accessing this as a leaf is okay.
         self.node.as_ptr()
     }
 
@@ -514,33 +514,37 @@ impl<'a, K: 'a, V: 'a, Type> NodeRef<marker::Immut<'a>, K, V, Type> {
         // but we want to avoid that run-time check.
         // Instead, we create a slice pointing into the node whenever possible.
         // We can sometimes do this even for the shared root, as the slice will be
-        // empty.  We cannot *always* do this because if the type is too highly
-        // aligned, the offset of `keys` in a "full node" might be outside the bounds
-        // of the header!  So we do an alignment check first, that will be
-        // evaluated at compile-time, and only do any run-time check in the rare case
-        // that the alignment is very big.
-        if mem::align_of::<K>() > mem::align_of::<LeafNode<(), ()>>() && self.is_shared_root() {
+        // empty and `NodeHeader` contains an empty `keys_start` array.
+        // We cannot *always* do this because:
+        // - `keys_start` is not correctly typed because we want `NodeHeader`'s size to
+        //   not depend on the alignment of `K` (needed because `as_header` should be safe).
+        //   For this reason, `NodeHeader` has this `K2` parameter (that's usually `()`
+        //   and hence just adds a size-0-align-1 field, not affecting layout).
+        //   If the correctly typed header is more highly aligned than the allocated header,
+        //   we cannot transmute safely.
+        // - Even if we can transmute, the offset of a correctly typed `keys_start` might
+        //   be different and outside the bounds of the allocated header!
+        // So we do an alignment check and a size check first, that will be evaluated
+        // at compile-time, and only do any run-time check in the rare case that
+        // the compile-time checks signal danger.
+        if (mem::align_of::<NodeHeader<K, V, K>>() > mem::align_of::<NodeHeader<K, V>>()
+            || mem::size_of::<NodeHeader<K, V, K>>() != mem::size_of::<NodeHeader<K, V>>())
+            && self.is_shared_root()
+        {
             &[]
         } else {
-            // Thanks to the alignment check above, we know that `keys` will be
+            // If we are a `LeafNode<K, V>`, we can always transmute to
+            // `NodeHeader<K, V, K>` and `keys_start` always has the same offset
+            // as the actual `keys`.
+            // Thanks to the checks above, we know that we can transmute to
+            // `NodeHeader<K, V, K>` and that `keys_start` will be
             // in-bounds of some allocation even if this is the shared root!
             // (We might be one-past-the-end, but that is allowed by LLVM.)
-            // Getting the pointer is tricky though.  `NodeHeader` does not have a `keys`
-            // field because we want its size to not depend on the alignment of `K`
-            // (needed because `as_header` should be safe).  We cannot call `as_leaf`
-            // because we might be the shared root.
-            // For this reason, `NodeHeader` has this `K2` parameter (that's usually `()`
-            // and hence just adds a size-0-align-1 field, not affecting layout).
-            // We know that we can transmute `NodeHeader<K, V, ()>` to `NodeHeader<K, V, K>`
-            // because we did the alignment check above, and hence `NodeHeader<K, V, K>`
-            // is not bigger than `NodeHeader<K, V, ()>`!  Then we can use `NodeHeader<K, V, K>`
+            // Thus we can use `NodeHeader<K, V, K>`
             // to compute the pointer where the keys start.
             // This entire hack will become unnecessary once
             // <https://github.com/rust-lang/rfcs/pull/2582> lands, then we can just take a raw
             // pointer to the `keys` field of `*const InternalNode<K, V>`.
-
-            // This is a non-debug-assert because it can be completely compile-time evaluated.
-            assert!(mem::size_of::<NodeHeader<K, V>>() == mem::size_of::<NodeHeader<K, V, K>>());
             let header = self.as_header() as *const _ as *const NodeHeader<K, V, K>;
             let keys = unsafe { &(*header).keys_start as *const _ as *const K };
             unsafe { slice::from_raw_parts(keys, self.len()) }
@@ -549,7 +553,7 @@ impl<'a, K: 'a, V: 'a, Type> NodeRef<marker::Immut<'a>, K, V, Type> {
 
     fn into_val_slice(self) -> &'a [V] {
         debug_assert!(!self.is_shared_root());
-        // We cannot be the root, so `as_leaf` is okay
+        // We cannot be the shared root, so `as_leaf` is okay
         unsafe { slice::from_raw_parts(MaybeUninit::first_ptr(&self.as_leaf().vals), self.len()) }
     }
 
@@ -567,9 +571,11 @@ impl<'a, K: 'a, V: 'a, Type> NodeRef<marker::Mut<'a>, K, V, Type> {
     }
 
     fn into_key_slice_mut(mut self) -> &'a mut [K] {
-        // Same as for `into_key_slice` above, we try to avoid a run-time check
-        // (the alignment comparison will usually be performed at compile-time).
-        if mem::align_of::<K>() > mem::align_of::<LeafNode<(), ()>>() && self.is_shared_root() {
+        // Same as for `into_key_slice` above, we try to avoid a run-time check.
+        if (mem::align_of::<NodeHeader<K, V, K>>() > mem::align_of::<NodeHeader<K, V>>()
+            || mem::size_of::<NodeHeader<K, V, K>>() != mem::size_of::<NodeHeader<K, V>>())
+            && self.is_shared_root()
+        {
             &mut []
         } else {
             unsafe {

src/liballoc/collections/btree/set.rs

@@ -83,7 +83,7 @@ impl<T: fmt::Debug> fmt::Debug for Iter<'_, T> {
 
 /// An owning iterator over the items of a `BTreeSet`.
 ///
-/// This `struct` is created by the [`into_iter`] method on [`BTreeSet`][`BTreeSet`]
+/// This `struct` is created by the [`into_iter`] method on [`BTreeSet`]
 /// (provided by the `IntoIterator` trait). See its documentation for more.
 ///
 /// [`BTreeSet`]: struct.BTreeSet.html

src/liballoc/collections/linked_list.rs

@@ -105,7 +105,7 @@ impl<T: fmt::Debug> fmt::Debug for IterMut<'_, T> {
 
 /// An owning iterator over the elements of a `LinkedList`.
 ///
-/// This `struct` is created by the [`into_iter`] method on [`LinkedList`][`LinkedList`]
+/// This `struct` is created by the [`into_iter`] method on [`LinkedList`]
 /// (provided by the `IntoIterator` trait). See its documentation for more.
 ///
 /// [`into_iter`]: struct.LinkedList.html#method.into_iter

src/liballoc/collections/vec_deque.rs

@@ -2474,7 +2474,7 @@ impl<T> FusedIterator for IterMut<'_, T> {}
 
 /// An owning iterator over the elements of a `VecDeque`.
 ///
-/// This `struct` is created by the [`into_iter`] method on [`VecDeque`][`VecDeque`]
+/// This `struct` is created by the [`into_iter`] method on [`VecDeque`]
 /// (provided by the `IntoIterator` trait). See its documentation for more.
 ///
 /// [`into_iter`]: struct.VecDeque.html#method.into_iter

src/liballoc/fmt.rs

@@ -330,7 +330,7 @@
 //!
 //! Additionally, the return value of this function is [`fmt::Result`] which is a
 //! type alias of [`Result`]`<(), `[`std::fmt::Error`]`>`. Formatting implementations
-//! should ensure that they propagate errors from the [`Formatter`][`Formatter`] (e.g., when
+//! should ensure that they propagate errors from the [`Formatter`] (e.g., when
 //! calling [`write!`]). However, they should never return errors spuriously. That
 //! is, a formatting implementation must and may only return an error if the
 //! passed-in [`Formatter`] returns an error. This is because, contrary to what

src/liballoc/slice.rs

@@ -450,7 +450,8 @@ impl<T> [T] {
         // and `rem` is the remaining part of `n`.
 
         // Using `Vec` to access `set_len()`.
-        let mut buf = Vec::with_capacity(self.len().checked_mul(n).expect("capacity overflow"));
+        let capacity = self.len().checked_mul(n).expect("capacity overflow");
+        let mut buf = Vec::with_capacity(capacity);
 
         // `2^expn` repetition is done by doubling `buf` `expn`-times.
         buf.extend(self);
@@ -476,7 +477,7 @@ impl<T> [T] {
 
         // `rem` (`= n - 2^expn`) repetition is done by copying
         // first `rem` repetitions from `buf` itself.
-        let rem_len = self.len() * n - buf.len(); // `self.len() * rem`
+        let rem_len = capacity - buf.len(); // `self.len() * rem`
         if rem_len > 0 {
             // `buf.extend(buf[0 .. rem_len])`:
             unsafe {
@@ -487,8 +488,7 @@ impl<T> [T] {
                     rem_len,
                 );
                 // `buf.len() + rem_len` equals to `buf.capacity()` (`= self.len() * n`).
-                let buf_cap = buf.capacity();
-                buf.set_len(buf_cap);
+                buf.set_len(capacity);
             }
         }
         buf