Auto merge of #87073 - jyn514:primitive-docs, r=GuillaumeGomez,jyn514

Fix rustdoc handling of primitive items

This is a complicated PR and does a lot of things. I'm willing to split it up a little more if it would help reviewing, but it would be tricky and I'd rather not unless it's necessary.

 ## What does this do?

- Fixes https://github.com/rust-lang/rust/issues/73423.
- Fixes https://github.com/rust-lang/rust/issues/79630. I'm not sure how to test this for the standard library explicitly, but you can see from some of the diffs from the `no_std` tests. I also tested it locally and it works correctly: ![image](https://user-images.githubusercontent.com/23638587/125214383-e1fdd000-e284-11eb-8048-76b5df958aad.png)
- Fixes https://github.com/rust-lang/rust/issues/83083.

## Why are these changes interconnected?

- Allowing anchors (https://github.com/rust-lang/rust/issues/83083) without fixing the online/offline problem (https://github.com/rust-lang/rust/issues/79630) will actually just silently discard the anchors, that's not a fix. The online/offline problem is directly related to the fragment hack; links need to go through `fn href()` to be fixed.
- Technically I could fix the online/offline problem without removing the error on anchors; I am willing to separate that out if it would be helpful for reviewing. However I can't fix the anchor problem without adding docs to core, since rustdoc needs all those primitives to have docs to avoid a fallback, and currently `#![no_std]` crates don't have docs for primitives. I also can't fix the online/offline problem without removing the fragment hack, since otherwise diffs like this will be wrong for some primitives but not others:
```diff
`@@` -385,7 +381,7 `@@` fn resolve_primitive_associated_item(
                         ty::AssocKind::Const => "associatedconstant",
                         ty::AssocKind::Type => "associatedtype",
                     };
-                    let fragment = format!("{}#{}.{}", prim_ty.as_sym(), out, item_name);
+                    let fragment = format!("{}.{}", out, item_name);
                     (Res::Primitive(prim_ty), fragment, Some((kind.as_def_kind(), item.def_id)))
                 })
         })
```
- Adding primitive docs to core without making any other change will cause links to go to `core` instead of `std`, even for crates with `extern crate std`. See "Breaking changes to doc(primitive)" below for why this is the case. That said, I could add some special casing to rustdoc at the same time that would let me separate this change from the others (it would fix https://github.com/rust-lang/rust/issues/73423 but still special-case intra-doc links). I'm willing to separate that out if helpful for reviewing.

### Add primitive documentation to libcore

This works by reusing the same `include!("primitive_docs.rs")` file in both core and std, and then special-casing links in core to use relative links instead of intra-doc links. This doesn't use purely intra-doc links because some of the primitive docs links to items only in std; this doesn't use purely relative links because that introduces new broken links when the docs are re-exported (e.g. String's `&str` deref impl, or Vec's slice deref impl).

Note that this copies the whole file to core, to avoid anyone compiling core to have to set `CARGO_PKG_NAME`. See https://rust-lang.zulipchat.com/#narrow/stream/122651-general/topic/Who.20should.20review.20changes.20to.20linkchecker.3F/near/249939598 for more context. It also adds a tidy check to make sure the two files are kept in sync.

### Fix inconsistent online/offline primitive docs

This does four things:
- Records modules with `doc(primitive)` in `cache.external_paths`. This is necessary for `href()` to find them later.
- Makes `cache.primitive_locations` available to the intra-doc link pass, by refactoring out a `PrimitiveType::primitive_locations` function that only uses `TyCtxt`.
- Special cases modules with `doc(primitive)` to be treated as always public for the purpose of links.
- Removes the fragment hack. cc `@notriddle,` I know you added some comments about this in the code (thank you for that!)

### Breaking changes to `doc(primitive)`

"Breaking" is a little misleading here - these are changes in behavior, none of them will cause code to fail to compile.

Let me preface this by saying I think stabilizing `doc(primitive)` was a uniquely terrible idea. As far as I can tell, it was stabilized by oversight; it's been stable since 1.0. No one should have need to use it except the standard library, and a crater run shows that in fact no one is using it: https://github.com/rust-lang/rust/pull/87050#issuecomment-886166706. I hope to actually make `doc(primitive)` a no-op unless you opt-in with a nightly feature, which will keep crates compiling without forcing rustdoc into trying to keep somewhat arbitrary behavior guarantees; but for now, this just subtly changes some of the behavior if you use `doc(primitive)` in a dependency.

That said, here are the changes:
-  Refactoring out `primitive_locations()` is technically a change in behavior, since it no longer looks for primitives in crates that were passed through `--extern`, but not used by the crate; however, that seems like such an unlikely edge case it's not worth dealing with.
- The precedence given to primitive locations is no longer just arbitrary, it can also be inconsistent from run to run. Let me explain that more: previously, primitive locations were sorted by the `CrateNum`; the comment on that sort said "Favor linking to as local extern as possible, so iterate all crates in reverse topological order." Unfortunately, that's not actually what CrateNum tracks: it measures the order crates are loaded, not the number of intermediate crates between that dependency and the root crate. It happened to work as intended before because the compiler injects `extern crate std;` at the top of every crate, which ensured it would have the first CrateNum other than the current, but every other CrateNum was completely arbitrary (for example, `core` often had a later CrateNum than `std`). This now removes the sort on CrateNum completely and special-cases core instead. In particular, if you depend on both `std` and a crate which defines a `doc(primitive)` module, it's arbitrary whether rustdoc will use the docs from std or the ones from the other crate. cc `@alexcrichton,` you wrote this originally.

cc `@rust-lang/rustdoc`
cc `@rust-lang/libs` for the addition to `core` (the commit you're interested in is https://github.com/rust-lang/rust/pull/87073/commits/91346c8293bb5f41d8e1d2ec9336433664652c53)

library/core/primitive_docs/box_into_raw.md

+../std/boxed/struct.Box.html#method.into_raw

library/core/primitive_docs/fs_file.md

+../std/fs/struct.File.html

library/core/primitive_docs/io_bufread.md

+../std/io/trait.BufRead.html

library/core/primitive_docs/io_read.md

+../std/io/trait.Read.html

library/core/primitive_docs/io_seek.md

+../std/io/trait.Seek.html

library/core/primitive_docs/io_write.md

+../std/io/trait.Write.html

library/core/primitive_docs/net_tosocketaddrs.md

+../std/net/trait.ToSocketAddrs.html

library/core/primitive_docs/process_exit.md

+../std/process/fn.exit.html

library/core/primitive_docs/string_string.md

+../std/string/struct.String.html

library/core/src/bool.rs

@@ -2,7 +2,8 @@
 
 #[lang = "bool"]
 impl bool {
-    /// Returns `Some(t)` if the `bool` is [`true`](keyword.true.html), or `None` otherwise.
+    /// Returns `Some(t)` if the `bool` is [`true`](../std/keyword.true.html),
+    /// or `None` otherwise.
     ///
     /// # Examples
     ///
@@ -18,7 +19,8 @@ pub fn then_some<T>(self, t: T) -> Option<T> {
         if self { Some(t) } else { None }
     }
 
-    /// Returns `Some(f())` if the `bool` is [`true`](keyword.true.html), or `None` otherwise.
+    /// Returns `Some(f())` if the `bool` is [`true`](../std/keyword.true.html),
+    /// or `None` otherwise.
     ///
     /// # Examples
     ///

library/core/src/char/methods.rs

@@ -24,7 +24,7 @@ impl char {
     /// decoding error.
     ///
     /// It can occur, for example, when giving ill-formed UTF-8 bytes to
-    /// [`String::from_utf8_lossy`](string/struct.String.html#method.from_utf8_lossy).
+    /// [`String::from_utf8_lossy`](../std/string/struct.String.html#method.from_utf8_lossy).
     #[stable(feature = "assoc_char_consts", since = "1.52.0")]
     pub const REPLACEMENT_CHARACTER: char = '\u{FFFD}';
 
@@ -96,7 +96,7 @@ pub fn decode_utf16<I: IntoIterator<Item = u16>>(iter: I) -> DecodeUtf16<I::Into
     /// Converts a `u32` to a `char`.
     ///
     /// Note that all `char`s are valid [`u32`]s, and can be cast to one with
-    /// [`as`](keyword.as.html):
+    /// [`as`](../std/keyword.as.html):
     ///
     /// ```
     /// let c = '💯';
@@ -372,7 +372,7 @@ pub fn to_digit(self, radix: u32) -> Option<u32> {
     /// println!("\\u{{2764}}");
     /// ```
     ///
-    /// Using [`to_string`](string/trait.ToString.html#tymethod.to_string):
+    /// Using [`to_string`](../std/string/trait.ToString.html#tymethod.to_string):
     ///
     /// ```
     /// assert_eq!('❤'.escape_unicode().to_string(), "\\u{2764}");
@@ -448,7 +448,7 @@ pub(crate) fn escape_debug_ext(self, args: EscapeDebugExtArgs) -> EscapeDebug {
     /// println!("\\n");
     /// ```
     ///
-    /// Using [`to_string`](string/trait.ToString.html#tymethod.to_string):
+    /// Using [`to_string`](../std/string/trait.ToString.html#tymethod.to_string):
     ///
     /// ```
     /// assert_eq!('\n'.escape_debug().to_string(), "\\n");
@@ -502,7 +502,7 @@ pub fn escape_debug(self) -> EscapeDebug {
     /// println!("\\\"");
     /// ```
     ///
-    /// Using [`to_string`](string/trait.ToString.html#tymethod.to_string):
+    /// Using [`to_string`](../std/string/trait.ToString.html#tymethod.to_string):
     ///
     /// ```
     /// assert_eq!('"'.escape_default().to_string(), "\\\"");
@@ -937,7 +937,7 @@ pub fn is_numeric(self) -> bool {
     /// println!("i\u{307}");
     /// ```
     ///
-    /// Using [`to_string`](string/trait.ToString.html#tymethod.to_string):
+    /// Using [`to_string`](../std/string/trait.ToString.html#tymethod.to_string):
     ///
     /// ```
     /// assert_eq!('C'.to_lowercase().to_string(), "c");
@@ -1002,7 +1002,7 @@ pub fn to_lowercase(self) -> ToLowercase {
     /// println!("SS");
     /// ```
     ///
-    /// Using [`to_string`](string/trait.ToString.html#tymethod.to_string):
+    /// Using [`to_string`](../std/string/trait.ToString.html#tymethod.to_string):
     ///
     /// ```
     /// assert_eq!('c'.to_uppercase().to_string(), "C");

library/core/src/lib.rs

@@ -130,6 +130,7 @@
 #![feature(decl_macro)]
 #![feature(doc_cfg)]
 #![feature(doc_notable_trait)]
+#![feature(doc_primitive)]
 #![feature(exhaustive_patterns)]
 #![feature(extern_types)]
 #![feature(fundamental)]
@@ -355,3 +356,5 @@ pub mod arch {
         /* compiler built-in */
     }
 }
+
+include!("primitive_docs.rs");

library/core/src/slice/mod.rs

@@ -2257,9 +2257,9 @@ pub fn binary_search_by<'a, F>(&'a self, mut f: F) -> Result<usize, usize>
     /// assert!(match r { Ok(1..=4) => true, _ => false, });
     /// ```
     // Lint rustdoc::broken_intra_doc_links is allowed as `slice::sort_by_key` is
-    // in crate `alloc`, and as such doesn't exists yet when building `core`.
-    // links to downstream crate: #74481. Since primitives are only documented in
-    // libstd (#73423), this never leads to broken links in practice.
+    // in crate `alloc`, and as such doesn't exists yet when building `core`: #74481.
+    // This breaks links when slice is displayed in core, but changing it to use relative links
+    // would break when the item is re-exported. So allow the core links to be broken for now.
     #[allow(rustdoc::broken_intra_doc_links)]
     #[stable(feature = "slice_binary_search_by_key", since = "1.10.0")]
     #[inline]

library/std/primitive_docs/box_into_raw.md

+Box::into_raw

library/std/primitive_docs/fs_file.md

+fs::File

library/std/primitive_docs/io_bufread.md

+io::BufRead

library/std/primitive_docs/io_read.md

+io::Read

library/std/primitive_docs/io_seek.md

+io::Seek

library/std/primitive_docs/io_write.md

+io::Write

library/std/primitive_docs/net_tosocketaddrs.md

+net::ToSocketAddrs

library/std/primitive_docs/process_exit.md

+process::exit

library/std/primitive_docs/string_string.md

+string::String

library/std/src/primitive_docs.rs

+// `library/{std,core}/src/primitive_docs.rs` should have the same contents.
+// These are different files so that relative links work properly without
+// having to have `CARGO_PKG_NAME` set, but conceptually they should always be the same.
 #[doc(primitive = "bool")]
 #[doc(alias = "true")]
 #[doc(alias = "false")]
@@ -20,12 +23,12 @@
 /// assert!(!bool_val);
 /// ```
 ///
-/// [`true`]: keyword.true.html
-/// [`false`]: keyword.false.html
+/// [`true`]: ../std/keyword.true.html
+/// [`false`]: ../std/keyword.false.html
 /// [`BitAnd`]: ops::BitAnd
 /// [`BitOr`]: ops::BitOr
 /// [`Not`]: ops::Not
-/// [`if`]: keyword.if.html
+/// [`if`]: ../std/keyword.if.html
 ///
 /// # Examples
 ///
@@ -103,7 +106,7 @@ mod prim_bool {}
 /// behaviour of the `!` type - expressions with type `!` will coerce into any other type.
 ///
 /// [`u32`]: prim@u32
-/// [`exit`]: process::exit
+#[doc = concat!("[`exit`]: ", include_str!("../primitive_docs/process_exit.md"))]
 ///
 /// # `!` and generics
 ///
@@ -188,7 +191,7 @@ mod prim_bool {}
 /// because `!` coerces to `Result<!, ConnectionError>` automatically.
 ///
 /// [`String::from_str`]: str::FromStr::from_str
-/// [`String`]: string::String
+#[doc = concat!("[`String`]: ", include_str!("../primitive_docs/string_string.md"))]
 /// [`FromStr`]: str::FromStr
 ///
 /// # `!` and traits
@@ -264,7 +267,7 @@ mod prim_bool {}
 /// `impl` for this which simply panics, but the same is true for any type (we could `impl
 /// Default` for (eg.) [`File`] by just making [`default()`] panic.)
 ///
-/// [`File`]: fs::File
+#[doc = concat!("[`File`]: ", include_str!("../primitive_docs/fs_file.md"))]
 /// [`Debug`]: fmt::Debug
 /// [`default()`]: Default::default
 ///
@@ -272,7 +275,6 @@ mod prim_bool {}
 mod prim_never {}
 
 #[doc(primitive = "char")]
-//
 /// A character type.
 ///
 /// The `char` type represents a single character. More specifically, since
@@ -304,7 +306,7 @@ mod prim_never {}
 /// assert_eq!(5, s.len() * std::mem::size_of::<u8>());
 /// ```
 ///
-/// [`String`]: string/struct.String.html
+#[doc = concat!("[`String`]: ", include_str!("../primitive_docs/string_string.md"))]
 ///
 /// As always, remember that a human intuition for 'character' might not map to
 /// Unicode's definitions. For example, despite looking similar, the 'é'
@@ -499,7 +501,7 @@ mod prim_unit {}
 /// [`null_mut`]: ptr::null_mut
 /// [`is_null`]: pointer::is_null
 /// [`offset`]: pointer::offset
-/// [`into_raw`]: Box::into_raw
+#[doc = concat!("[`into_raw`]: ", include_str!("../primitive_docs/box_into_raw.md"))]
 /// [`drop`]: mem::drop
 /// [`write`]: ptr::write
 #[stable(feature = "rust1", since = "1.0.0")]
@@ -581,9 +583,9 @@ mod prim_pointer {}
 /// # Editions
 ///
 /// Prior to Rust 1.53, arrays did not implement [`IntoIterator`] by value, so the method call
-/// `array.into_iter()` auto-referenced into a [slice iterator](slice::iter). Right now, the old behavior
-/// is preserved in the 2015 and 2018 editions of Rust for compatibility, ignoring
-/// `IntoIterator` by value. In the future, the behavior on the 2015 and 2018 edition
+/// `array.into_iter()` auto-referenced into a [slice iterator](slice::iter). Right now, the old
+/// behavior is preserved in the 2015 and 2018 editions of Rust for compatibility, ignoring
+/// [`IntoIterator`] by value. In the future, the behavior on the 2015 and 2018 edition
 /// might be made consistent to the behavior of later editions.
 ///
 /// ```rust,edition2018
@@ -1042,15 +1044,15 @@ mod prim_usize {}
 /// References, both shared and mutable.
 ///
 /// A reference represents a borrow of some owned value. You can get one by using the `&` or `&mut`
-/// operators on a value, or by using a [`ref`](keyword.ref.html) or
-/// <code>[ref](keyword.ref.html) [mut](keyword.mut.html)</code> pattern.
+/// operators on a value, or by using a [`ref`](../std/keyword.ref.html) or
+/// <code>[ref](../std/keyword.ref.html) [mut](../std/keyword.mut.html)</code> pattern.
 ///
 /// For those familiar with pointers, a reference is just a pointer that is assumed to be
 /// aligned, not null, and pointing to memory containing a valid value of `T` - for example,
 /// <code>&[bool]</code> can only point to an allocation containing the integer values `1`
-/// ([`true`](keyword.true.html)) or `0` ([`false`](keyword.false.html)), but creating a
-/// <code>&[bool]</code> that points to an allocation containing the value `3` causes
-/// undefined behaviour.
+/// ([`true`](../std/keyword.true.html)) or `0` ([`false`](../std/keyword.false.html)), but
+/// creating a <code>&[bool]</code> that points to an allocation containing
+/// the value `3` causes undefined behaviour.
 /// In fact, <code>[Option]\<&T></code> has the same memory representation as a
 /// nullable but aligned pointer, and can be passed across FFI boundaries as such.
 ///
@@ -1117,6 +1119,7 @@ mod prim_usize {}
 ///
 /// [`DerefMut`]: ops::DerefMut
 /// [`BorrowMut`]: borrow::BorrowMut
+/// [bool]: prim@bool
 ///
 /// The following traits are implemented on `&T` references if the underlying `T` also implements
 /// that trait:
@@ -1134,7 +1137,7 @@ mod prim_usize {}
 /// [`std::fmt`]: fmt
 /// ['Pointer`]: fmt::Pointer
 /// [`Hash`]: hash::Hash
-/// [`ToSocketAddrs`]: net::ToSocketAddrs
+#[doc = concat!("[`ToSocketAddrs`]: ", include_str!("../primitive_docs/net_tosocketaddrs.md"))]
 ///
 /// `&mut T` references get all of the above except `ToSocketAddrs`, plus the following, if `T`
 /// implements that trait:
@@ -1155,9 +1158,10 @@ mod prim_usize {}
 ///
 /// [`FusedIterator`]: iter::FusedIterator
 /// [`TrustedLen`]: iter::TrustedLen
-/// [`Seek`]: io::Seek
-/// [`BufRead`]: io::BufRead
-/// [`Read`]: io::Read
+#[doc = concat!("[`Seek`]: ", include_str!("../primitive_docs/io_seek.md"))]
+#[doc = concat!("[`BufRead`]: ", include_str!("../primitive_docs/io_bufread.md"))]
+#[doc = concat!("[`Read`]: ", include_str!("../primitive_docs/io_read.md"))]
+#[doc = concat!("[`io::Write`]: ", include_str!("../primitive_docs/io_write.md"))]
 ///
 /// Note that due to method call deref coercion, simply calling a trait method will act like they
 /// work on references as well as they do on owned values! The implementations described here are

src/librustdoc/clean/blanket_impl.rs

@@ -17,7 +17,7 @@ impl<'a, 'tcx> BlanketImplFinder<'a, 'tcx> {
         let param_env = self.cx.tcx.param_env(item_def_id);
         let ty = self.cx.tcx.type_of(item_def_id);
 
-        debug!("get_blanket_impls({:?})", ty);
+        trace!("get_blanket_impls({:?})", ty);
         let mut impls = Vec::new();
         for &trait_def_id in self.cx.tcx.all_traits(()).iter() {
             if !self.cx.cache.access_levels.is_public(trait_def_id)
@@ -28,9 +28,10 @@ impl<'a, 'tcx> BlanketImplFinder<'a, 'tcx> {
             // NOTE: doesn't use `for_each_relevant_impl` to avoid looking at anything besides blanket impls
             let trait_impls = self.cx.tcx.trait_impls_of(trait_def_id);
             for &impl_def_id in trait_impls.blanket_impls() {
-                debug!(
+                trace!(
                     "get_blanket_impls: Considering impl for trait '{:?}' {:?}",
-                    trait_def_id, impl_def_id
+                    trait_def_id,
+                    impl_def_id
                 );
                 let trait_ref = self.cx.tcx.impl_trait_ref(impl_def_id).unwrap();
                 let is_param = matches!(trait_ref.self_ty().kind(), ty::Param(_));
@@ -50,9 +51,11 @@ impl<'a, 'tcx> BlanketImplFinder<'a, 'tcx> {
                         // FIXME(eddyb) ignoring `obligations` might cause false positives.
                         drop(obligations);
 
-                        debug!(
+                        trace!(
                             "invoking predicate_may_hold: param_env={:?}, trait_ref={:?}, ty={:?}",
-                            param_env, trait_ref, ty
+                            param_env,
+                            trait_ref,
+                            ty
                         );
                         let predicates = self
                             .cx

src/librustdoc/clean/inline.rs

@@ -447,9 +447,9 @@ fn merge_attrs(
     }
 
     let (merged_attrs, cfg) = merge_attrs(cx, parent_module.into(), load_attrs(cx, did), attrs);
-    debug!("merged_attrs={:?}", merged_attrs);
+    trace!("merged_attrs={:?}", merged_attrs);
 
-    debug!("build_impl: impl {:?} for {:?}", trait_.def_id(), for_.def_id());
+    trace!("build_impl: impl {:?} for {:?}", trait_.def_id(), for_.def_id());
     ret.push(clean::Item::from_def_id_and_attrs_and_parts(
         did,
         None,

src/librustdoc/clean/mod.rs

@@ -1406,7 +1406,7 @@ fn normalize(cx: &mut DocContext<'tcx>, ty: Ty<'_>) -> Option<Ty<'tcx>> {
 
 impl<'tcx> Clean<Type> for Ty<'tcx> {
     fn clean(&self, cx: &mut DocContext<'_>) -> Type {
-        debug!("cleaning type: {:?}", self);
+        trace!("cleaning type: {:?}", self);
         let ty = normalize(cx, self).unwrap_or(self);
         match *ty.kind() {
             ty::Never => Never,

src/librustdoc/clean/types.rs

@@ -461,60 +461,20 @@ pub fn from_def_id_and_attrs_and_parts(
             .map_or(&[][..], |v| v.as_slice())
             .iter()
             .filter_map(|ItemLink { link: s, link_text, did, ref fragment }| {
-                match did {
-                    Some(did) => {
-                        if let Ok((mut href, ..)) = href(*did, cx) {
-                            if let Some(ref fragment) = *fragment {
-                                href.push('#');
-                                href.push_str(fragment);
-                            }
-                            Some(RenderedLink {
-                                original_text: s.clone(),
-                                new_text: link_text.clone(),
-                                href,
-                            })
-                        } else {
-                            None
-                        }
-                    }
-                    // FIXME(83083): using fragments as a side-channel for
-                    // primitive names is very unfortunate
-                    None => {
-                        let relative_to = &cx.current;
-                        if let Some(ref fragment) = *fragment {
-                            let url = match cx.cache().extern_locations.get(&self.def_id.krate()) {
-                                Some(&ExternalLocation::Local) => {
-                                    if relative_to[0] == "std" {
-                                        let depth = relative_to.len() - 1;
-                                        "../".repeat(depth)
-                                    } else {
-                                        let depth = relative_to.len();
-                                        format!("{}std/", "../".repeat(depth))
-                                    }
-                                }
-                                Some(ExternalLocation::Remote(ref s)) => {
-                                    format!("{}/std/", s.trim_end_matches('/'))
-                                }
-                                Some(ExternalLocation::Unknown) | None => {
-                                    format!("{}/std/", crate::DOC_RUST_LANG_ORG_CHANNEL)
-                                }
-                            };
-                            // This is a primitive so the url is done "by hand".
-                            let tail = fragment.find('#').unwrap_or_else(|| fragment.len());
-                            Some(RenderedLink {
-                                original_text: s.clone(),
-                                new_text: link_text.clone(),
-                                href: format!(
-                                    "{}primitive.{}.html{}",
-                                    url,
-                                    &fragment[..tail],
-                                    &fragment[tail..]
-                                ),
-                            })
-                        } else {
-                            panic!("This isn't a primitive?!");
-                        }
+                debug!(?did);
+                if let Ok((mut href, ..)) = href(*did, cx) {
+                    debug!(?href);
+                    if let Some(ref fragment) = *fragment {
+                        href.push('#');
+                        href.push_str(fragment);
                     }
+                    Some(RenderedLink {
+                        original_text: s.clone(),
+                        new_text: link_text.clone(),
+                        href,
+                    })
+                } else {
+                    None
                 }
             })
             .collect()
@@ -531,18 +491,10 @@ pub fn from_def_id_and_attrs_and_parts(
             .get(&self.def_id)
             .map_or(&[][..], |v| v.as_slice())
             .iter()
-            .filter_map(|ItemLink { link: s, link_text, did, fragment }| {
-                // FIXME(83083): using fragments as a side-channel for
-                // primitive names is very unfortunate
-                if did.is_some() || fragment.is_some() {
-                    Some(RenderedLink {
-                        original_text: s.clone(),
-                        new_text: link_text.clone(),
-                        href: String::new(),
-                    })
-                } else {
-                    None
-                }
+            .map(|ItemLink { link: s, link_text, .. }| RenderedLink {
+                original_text: s.clone(),
+                new_text: link_text.clone(),
+                href: String::new(),
             })
             .collect()
     }
@@ -963,7 +915,7 @@ fn from_iter<T>(iter: T) -> Self
     crate other_attrs: Vec<ast::Attribute>,
 }
 
-#[derive(Clone, Debug, Default, PartialEq, Eq, Hash)]
+#[derive(Clone, Debug, PartialEq, Eq, Hash)]
 /// A link that has not yet been rendered.
 ///
 /// This link will be turned into a rendered link by [`Item::links`].
@@ -975,7 +927,7 @@ fn from_iter<T>(iter: T) -> Self
     /// This may not be the same as `link` if there was a disambiguator
     /// in an intra-doc link (e.g. \[`fn@f`\])
     pub(crate) link_text: String,
-    pub(crate) did: Option<DefId>,
+    pub(crate) did: DefId,
     /// The url fragment to append to the link
     pub(crate) fragment: Option<String>,
 }
@@ -1802,6 +1754,39 @@ impl PrimitiveType {
             Never => sym::never,
         }
     }
+
+    /// Returns the DefId of the module with `doc(primitive)` for this primitive type.
+    /// Panics if there is no such module.
+    ///
+    /// This gives precedence to primitives defined in the current crate, and deprioritizes primitives defined in `core`,
+    /// but otherwise, if multiple crates define the same primitive, there is no guarantee of which will be picked.
+    /// In particular, if a crate depends on both `std` and another crate that also defines `doc(primitive)`, then
+    /// it's entirely random whether `std` or the other crate is picked. (no_std crates are usually fine unless multiple dependencies define a primitive.)
+    crate fn primitive_locations(tcx: TyCtxt<'_>) -> &FxHashMap<PrimitiveType, DefId> {
+        static PRIMITIVE_LOCATIONS: OnceCell<FxHashMap<PrimitiveType, DefId>> = OnceCell::new();
+        PRIMITIVE_LOCATIONS.get_or_init(|| {
+            let mut primitive_locations = FxHashMap::default();
+            // NOTE: technically this misses crates that are only passed with `--extern` and not loaded when checking the crate.
+            // This is a degenerate case that I don't plan to support.
+            for &crate_num in tcx.crates(()) {
+                let e = ExternalCrate { crate_num };
+                let crate_name = e.name(tcx);
+                debug!(?crate_num, ?crate_name);
+                for &(def_id, prim) in &e.primitives(tcx) {
+                    // HACK: try to link to std instead where possible
+                    if crate_name == sym::core && primitive_locations.contains_key(&prim) {
+                        continue;
+                    }
+                    primitive_locations.insert(prim, def_id);
+                }
+            }
+            let local_primitives = ExternalCrate { crate_num: LOCAL_CRATE }.primitives(tcx);
+            for (def_id, prim) in local_primitives {
+                primitive_locations.insert(prim, def_id);
+            }
+            primitive_locations
+        })
+    }
 }
 
 impl From<ast::IntTy> for PrimitiveType {

src/librustdoc/formats/cache.rs

@@ -6,7 +6,7 @@
 use rustc_middle::ty::TyCtxt;
 use rustc_span::symbol::sym;
 
-use crate::clean::{self, GetDefId, ItemId};
+use crate::clean::{self, GetDefId, ItemId, PrimitiveType};
 use crate::config::RenderOptions;
 use crate::fold::DocFolder;
 use crate::formats::item_type::ItemType;
@@ -159,17 +159,16 @@ impl Cache {
             self.external_paths.insert(e.def_id(), (vec![name.to_string()], ItemType::Module));
         }
 
-        // Cache where all known primitives have their documentation located.
-        //
-        // Favor linking to as local extern as possible, so iterate all crates in
-        // reverse topological order.
-        for &e in krate.externs.iter().rev() {
-            for &(def_id, prim) in &e.primitives(tcx) {
-                self.primitive_locations.insert(prim, def_id);
-            }
-        }
-        for &(def_id, prim) in &krate.primitives {
-            self.primitive_locations.insert(prim, def_id);
+        // FIXME: avoid this clone (requires implementing Default manually)
+        self.primitive_locations = PrimitiveType::primitive_locations(tcx).clone();
+        for (prim, &def_id) in &self.primitive_locations {
+            let crate_name = tcx.crate_name(def_id.krate);
+            // Recall that we only allow primitive modules to be at the root-level of the crate.
+            // If that restriction is ever lifted, this will have to include the relative paths instead.
+            self.external_paths.insert(
+                def_id,
+                (vec![crate_name.to_string(), prim.as_sym().to_string()], ItemType::Primitive),
+            );
         }
 
         krate = CacheBuilder { tcx, cache: self }.fold_crate(krate);

src/librustdoc/html/format.rs

@@ -509,7 +509,11 @@ fn to_module_fqp(shortty: ItemType, fqp: &[String]) -> &[String] {
         if shortty == ItemType::Module { fqp } else { &fqp[..fqp.len() - 1] }
     }
 
-    if !did.is_local() && !cache.access_levels.is_public(did) && !cache.document_private {
+    if !did.is_local()
+        && !cache.access_levels.is_public(did)
+        && !cache.document_private
+        && !cache.primitive_locations.values().any(|&id| id == did)
+    {
         return Err(HrefError::Private);
     }
 
@@ -517,6 +521,7 @@ fn to_module_fqp(shortty: ItemType, fqp: &[String]) -> &[String] {
     let (fqp, shortty, mut url_parts) = match cache.paths.get(&did) {
         Some(&(ref fqp, shortty)) => (fqp, shortty, {
             let module_fqp = to_module_fqp(shortty, fqp);
+            debug!(?fqp, ?shortty, ?module_fqp);
             href_relative_parts(module_fqp, relative_to)
         }),
         None => {
@@ -548,6 +553,7 @@ fn to_module_fqp(shortty: ItemType, fqp: &[String]) -> &[String] {
             url_parts.insert(0, root);
         }
     }
+    debug!(?url_parts);
     let last = &fqp.last().unwrap()[..];
     let filename;
     match shortty {
@@ -742,7 +748,7 @@ fn fmt_type<'cx>(
     use_absolute: bool,
     cx: &'cx Context<'_>,
 ) -> fmt::Result {
-    debug!("fmt_type(t = {:?})", t);
+    trace!("fmt_type(t = {:?})", t);
 
     match *t {
         clean::Generic(name) => write!(f, "{}", name),

src/librustdoc/json/conversions.rs

@@ -30,9 +30,7 @@ pub(super) fn convert_item(&self, item: clean::Item) -> Option<Item> {
             .get(&item.def_id)
             .into_iter()
             .flatten()
-            .filter_map(|clean::ItemLink { link, did, .. }| {
-                did.map(|did| (link.clone(), from_item_id(did.into())))
-            })
+            .map(|clean::ItemLink { link, did, .. }| (link.clone(), from_item_id((*did).into())))
             .collect();
         let docs = item.attrs.collapsed_doc_value();
         let attrs = item

src/librustdoc/passes/collect_intra_doc_links.rs

@@ -14,7 +14,7 @@
 };
 use rustc_hir::def_id::{CrateNum, DefId};
 use rustc_middle::ty::TyCtxt;
-use rustc_middle::{bug, ty};
+use rustc_middle::{bug, span_bug, ty};
 use rustc_resolve::ParentScope;
 use rustc_session::lint::Lint;
 use rustc_span::hygiene::{MacroKind, SyntaxContext};
@@ -98,14 +98,10 @@ fn name(self, tcx: TyCtxt<'_>) -> Symbol {
         }
     }
 
-    fn def_id(self) -> DefId {
-        self.opt_def_id().expect("called def_id() on a primitive")
-    }
-
-    fn opt_def_id(self) -> Option<DefId> {
+    fn def_id(self, tcx: TyCtxt<'_>) -> DefId {
         match self {
-            Res::Def(_, id) => Some(id),
-            Res::Primitive(_) => None,
+            Res::Def(_, id) => id,
+            Res::Primitive(prim) => *PrimitiveType::primitive_locations(tcx).get(&prim).unwrap(),
         }
     }
 
@@ -237,10 +233,7 @@ enum AnchorFailure {
     /// link, Rustdoc disallows having a user-specified anchor.
     ///
     /// Most of the time this is fine, because you can just link to the page of
-    /// the item if you want to provide your own anchor. For primitives, though,
-    /// rustdoc uses the anchor as a side channel to know which page to link to;
-    /// it doesn't show up in the generated link. Ideally, rustdoc would remove
-    /// this limitation, allowing you to link to subheaders on primitives.
+    /// the item if you want to provide your own anchor.
     RustdocAnchorConflict(Res),
 }
 
@@ -388,7 +381,7 @@ fn resolve_primitive_associated_item(
                         ty::AssocKind::Const => "associatedconstant",
                         ty::AssocKind::Type => "associatedtype",
                     };
-                    let fragment = format!("{}#{}.{}", prim_ty.as_sym(), out, item_name);
+                    let fragment = format!("{}.{}", out, item_name);
                     (Res::Primitive(prim_ty), fragment, Some((kind.as_def_kind(), item.def_id)))
                 })
         })
@@ -475,14 +468,6 @@ fn resolve<'path>(
                     return handle_variant(self.cx, res, extra_fragment);
                 }
                 // Not a trait item; just return what we found.
-                Res::Primitive(ty) => {
-                    if extra_fragment.is_some() {
-                        return Err(ErrorKind::AnchorFailure(
-                            AnchorFailure::RustdocAnchorConflict(res),
-                        ));
-                    }
-                    return Ok((res, Some(ty.as_sym().to_string())));
-                }
                 _ => return Ok((res, extra_fragment.clone())),
             }
         }
@@ -517,6 +502,8 @@ fn resolve<'path>(
                 let (res, fragment, side_channel) =
                     self.resolve_associated_item(ty_res, item_name, ns, module_id)?;
                 let result = if extra_fragment.is_some() {
+                    // NOTE: can never be a primitive since `side_channel.is_none()` only when `res`
+                    // is a trait (and the side channel DefId is always an associated item).
                     let diag_res = side_channel.map_or(res, |(k, r)| Res::Def(k, r));
                     Err(ErrorKind::AnchorFailure(AnchorFailure::RustdocAnchorConflict(diag_res)))
                 } else {
@@ -1152,7 +1139,7 @@ fn resolve_link(
             module_id = DefId { krate, index: CRATE_DEF_INDEX };
         }
 
-        let (mut res, mut fragment) = self.resolve_with_disambiguator_cached(
+        let (mut res, fragment) = self.resolve_with_disambiguator_cached(
             ResolutionInfo {
                 module_id,
                 dis: disambiguator,
@@ -1174,16 +1161,7 @@ fn resolve_link(
             if let Some(prim) = resolve_primitive(path_str, TypeNS) {
                 // `prim@char`
                 if matches!(disambiguator, Some(Disambiguator::Primitive)) {
-                    if fragment.is_some() {
-                        anchor_failure(
-                            self.cx,
-                            diag_info,
-                            AnchorFailure::RustdocAnchorConflict(prim),
-                        );
-                        return None;
-                    }
                     res = prim;
-                    fragment = Some(prim.name(self.cx.tcx).to_string());
                 } else {
                     // `[char]` when a `char` module is in scope
                     let candidates = vec![res, prim];
@@ -1303,12 +1281,17 @@ fn resolve_link(
                     }
                 }
 
-                Some(ItemLink { link: ori_link.link, link_text, did: None, fragment })
+                Some(ItemLink {
+                    link: ori_link.link,
+                    link_text,
+                    did: res.def_id(self.cx.tcx),
+                    fragment,
+                })
             }
             Res::Def(kind, id) => {
                 verify(kind, id)?;
                 let id = clean::register_res(self.cx, rustc_hir::def::Res::Def(kind, id));
-                Some(ItemLink { link: ori_link.link, link_text, did: Some(id), fragment })
+                Some(ItemLink { link: ori_link.link, link_text, did: id, fragment })
             }
         }
     }
@@ -2069,8 +2052,11 @@ fn anchor_failure(cx: &DocContext<'_>, diag_info: DiagnosticInfo<'_>, failure: A
             diag.span_label(sp, "invalid anchor");
         }
         if let AnchorFailure::RustdocAnchorConflict(Res::Primitive(_)) = failure {
-            diag.note("this restriction may be lifted in a future release");
-            diag.note("see https://github.com/rust-lang/rust/issues/83083 for more information");
+            if let Some(sp) = sp {
+                span_bug!(sp, "anchors should be allowed now");
+            } else {
+                bug!("anchors should be allowed now");
+            }
         }
     });
 }
@@ -2198,10 +2184,11 @@ fn handle_variant(
     use rustc_middle::ty::DefIdTree;
 
     if extra_fragment.is_some() {
+        // NOTE: `res` can never be a primitive since this function is only called when `tcx.def_kind(res) == DefKind::Variant`.
         return Err(ErrorKind::AnchorFailure(AnchorFailure::RustdocAnchorConflict(res)));
     }
     cx.tcx
-        .parent(res.def_id())
+        .parent(res.def_id(cx.tcx))
         .map(|parent| {
             let parent_def = Res::Def(DefKind::Enum, parent);
             let variant = cx.tcx.expect_variant_res(res.as_hir_res().unwrap());

src/test/rustdoc-js-std/multi-query.js

@@ -2,8 +2,9 @@ const QUERY = 'str,u8';
 
 const EXPECTED = {
     'others': [
-        { 'path': 'std', 'name': 'str' },
-        { 'path': 'std', 'name': 'u8' },
-        { 'path': 'std::ffi', 'name': 'CStr' },
+        { 'path': 'std', 'name': 'str', 'href': '../std/primitive.str.html' },
+        { 'path': 'std', 'name': 'u8', 'href': '../std/primitive.u8.html'  },
+        { 'path': 'std', 'name': 'str', 'href': '../std/str/index.html' },
+        { 'path': 'std', 'name': 'u8', 'href': '../std/u8/index.html' },
     ],
 };

src/test/rustdoc-ui/intra-doc/anchors.rs

@@ -37,13 +37,3 @@ pub fn bar() {}
 /// Damn enum's variants: [Enum::A#whatever].
 //~^ ERROR `Enum::A#whatever` contains an anchor
 pub fn enum_link() {}
-
-/// Primitives?
-///
-/// [u32#hello]
-//~^ ERROR `u32#hello` contains an anchor
-pub fn x() {}
-
-/// [prim@usize#x]
-//~^ ERROR `prim@usize#x` contains an anchor
-pub mod usize {}

src/test/rustdoc-ui/intra-doc/anchors.stderr

-error: `prim@usize#x` contains an anchor, but links to builtin types are already anchored
-  --> $DIR/anchors.rs:47:6
-   |
-LL | /// [prim@usize#x]
-   |      ^^^^^^^^^^--
-   |                |
-   |                invalid anchor
-   |
-note: the lint level is defined here
-  --> $DIR/anchors.rs:1:9
-   |
-LL | #![deny(rustdoc::broken_intra_doc_links)]
-   |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
-   = note: this restriction may be lifted in a future release
-   = note: see https://github.com/rust-lang/rust/issues/83083 for more information
-
 error: `Foo::f#hola` contains an anchor, but links to fields are already anchored
   --> $DIR/anchors.rs:25:15
    |
@@ -21,6 +5,12 @@ LL | /// Or maybe [Foo::f#hola].
    |               ^^^^^^-----
    |                     |
    |                     invalid anchor
+   |
+note: the lint level is defined here
+  --> $DIR/anchors.rs:1:9
+   |
+LL | #![deny(rustdoc::broken_intra_doc_links)]
+   |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 error: `hello#people#!` contains multiple anchors
   --> $DIR/anchors.rs:31:28
@@ -38,16 +28,5 @@ LL | /// Damn enum's variants: [Enum::A#whatever].
    |                                   |
    |                                   invalid anchor
 
-error: `u32#hello` contains an anchor, but links to builtin types are already anchored
-  --> $DIR/anchors.rs:43:6
-   |
-LL | /// [u32#hello]
-   |      ^^^------
-   |         |
-   |         invalid anchor
-   |
-   = note: this restriction may be lifted in a future release
-   = note: see https://github.com/rust-lang/rust/issues/83083 for more information
-
-error: aborting due to 5 previous errors
+error: aborting due to 3 previous errors
 

src/test/rustdoc/auxiliary/issue-15318.rs

+// no-prefer-dynamic
 // compile-flags: -Cmetadata=aux
-
+#![crate_type = "rlib"]
 #![doc(html_root_url = "http://example.com/")]
+#![feature(lang_items)]
+#![no_std]
+
+#[lang = "eh_personality"]
+fn foo() {}
+
+#[panic_handler]
+fn bar(_: &core::panic::PanicInfo) -> ! { loop {} }
 
 /// dox
 #[doc(primitive = "pointer")]

src/test/rustdoc/auxiliary/primitive-doc.rs

 // compile-flags: --crate-type lib --edition 2018
 
+#![feature(no_core)]
+#![no_core]
+
 #[doc(primitive = "usize")]
 /// This is the built-in type `usize`.
 mod usize {

src/test/rustdoc/cross-crate-primitive-doc.rs

 // aux-build:primitive-doc.rs
 // compile-flags: --extern-html-root-url=primitive_doc=../ -Z unstable-options
+// ignore-windows
 
-#![no_std]
+#![feature(no_core)]
+#![no_core]
 
 extern crate primitive_doc;
 
 // @has 'cross_crate_primitive_doc/fn.foo.html' '//a[@href="../primitive_doc/primitive.usize.html"]' 'usize'
+// @has 'cross_crate_primitive_doc/fn.foo.html' '//a[@href="../primitive_doc/primitive.usize.html"]' 'link'
+/// [link](usize)
 pub fn foo() -> usize { 0 }

src/test/rustdoc/intra-doc/anchors.rs

@@ -10,3 +10,15 @@
 ///
 /// To link to [Something#Anchor!]
 pub struct SomeOtherType;
+
+/// Primitives?
+///
+/// [u32#hello]
+// @has anchors/fn.x.html
+// @has - '//a/@href' '{{channel}}/std/primitive.u32.html#hello'
+pub fn x() {}
+
+/// [prim@usize#x]
+// @has anchors/usize/index.html
+// @has - '//a/@href' '{{channel}}/std/primitive.usize.html#x'
+pub mod usize {}

src/test/rustdoc/intra-doc/auxiliary/my-core.rs

@@ -2,6 +2,10 @@
 #![no_core]
 #![crate_type="rlib"]
 
+#[doc(primitive = "char")]
+/// Some char docs
+mod char {}
+
 #[lang = "char"]
 impl char {
     pub fn len_utf8(self) -> usize {

src/test/rustdoc/intra-doc/prim-methods-external-core.rs

@@ -9,8 +9,8 @@
 #![crate_type = "rlib"]
 
 // @has prim_methods_external_core/index.html
-// @has - '//*[@id="main"]//a[@href="{{channel}}/std/primitive.char.html"]' 'char'
-// @has - '//*[@id="main"]//a[@href="{{channel}}/std/primitive.char.html#method.len_utf8"]' 'char::len_utf8'
+// @has - '//*[@id="main"]//a[@href="../my_core/primitive.char.html"]' 'char'
+// @has - '//*[@id="main"]//a[@href="../my_core/primitive.char.html#method.len_utf8"]' 'char::len_utf8'
 
 //! A [`char`] and its [`char::len_utf8`].
 

src/test/rustdoc/intra-doc/prim-methods-local.rs

@@ -5,10 +5,13 @@
 
 
 // @has prim_methods_local/index.html
-// @has - '//*[@id="main"]//a[@href="{{channel}}/std/primitive.char.html"]' 'char'
-// @has - '//*[@id="main"]//a[@href="{{channel}}/std/primitive.char.html#method.len_utf8"]' 'char::len_utf8'
+// @has - '//*[@id="main"]//a[@href="primitive.char.html"]' 'char'
+// @has - '//*[@id="main"]//a[@href="primitive.char.html#method.len_utf8"]' 'char::len_utf8'
 
-//! A [`char`] and its [`char::len_utf8`].
+//! A [prim@`char`] and its [`char::len_utf8`].
+
+#[doc(primitive = "char")]
+mod char {}
 
 #[lang = "char"]
 impl char {

src/test/rustdoc/intra-link-prim-self.rs

@@ -7,8 +7,8 @@
 /// [Self::f]
 /// [Self::MAX]
 // @has intra_link_prim_self/primitive.usize.html
-// @has - '//a[@href="{{channel}}/std/primitive.usize.html#method.f"]' 'Self::f'
-// @has - '//a[@href="{{channel}}/std/primitive.usize.html#associatedconstant.MAX"]' 'Self::MAX'
+// @has - '//a[@href="primitive.usize.html#method.f"]' 'Self::f'
+// @has - '//a[@href="primitive.usize.html#associatedconstant.MAX"]' 'Self::MAX'
 impl usize {
     /// Some docs
     pub fn f() {}

src/test/rustdoc/issue-15318-2.rs

 // aux-build:issue-15318.rs
 // ignore-cross-compile
+#![no_std]
 
 extern crate issue_15318;
 

src/test/rustdoc/no_std-primitive.rs

+#![no_std]
+
+/// Link to [intra-doc link][u8]
+// @has 'no_std_primitive/fn.foo.html' '//a[@href="{{channel}}/core/primitive.u8.html"]' 'intra-doc link'
+// @has - '//a[@href="{{channel}}/core/primitive.u8.html"]' 'u8'
+pub fn foo() -> u8 {}

src/test/rustdoc/primitive/no_std.rs

+#![no_std]
+#![deny(warnings)]
+#![deny(rustdoc::broken_intra_doc_links)]
+
+// @has no_std/fn.foo.html '//a/[@href="{{channel}}/core/primitive.u8.html"]' 'u8'
+// @has no_std/fn.foo.html '//a/[@href="{{channel}}/core/primitive.u8.html"]' 'primitive link'
+/// Link to [primitive link][u8]
+pub fn foo() -> u8 {}
+
+// Test that all primitives can be linked to.
+/// [isize] [i8] [i16] [i32] [i64] [i128]
+/// [usize] [u8] [u16] [u32] [u64] [u128]
+/// [f32] [f64]
+/// [char] [bool] [str] [slice] [array] [tuple] [unit]
+/// [pointer] [reference] [fn] [never]
+pub fn bar() {}

src/tools/linkchecker/main.rs

@@ -30,6 +30,7 @@
 // If at all possible you should use intra-doc links to avoid linkcheck issues. These
 // are cases where that does not work
 // [(generated_documentation_page, &[broken_links])]
+#[rustfmt::skip]
 const LINKCHECK_EXCEPTIONS: &[(&str, &[&str])] = &[
     // These try to link to std::collections, but are defined in alloc
     // https://github.com/rust-lang/rust/issues/74481
@@ -37,6 +38,19 @@
     ("std/collections/btree_set/struct.BTreeSet.html", &["#insert-and-complex-keys"]),
     ("alloc/collections/btree_map/struct.BTreeMap.html", &["#insert-and-complex-keys"]),
     ("alloc/collections/btree_set/struct.BTreeSet.html", &["#insert-and-complex-keys"]),
+
+    // These try to link to various things in std, but are defined in core.
+    // The docs in std::primitive use proper intra-doc links, so these seem fine to special-case.
+    // Most these are broken because liballoc uses `#[lang_item]` magic to define things on
+    // primitives that aren't available in core.
+    ("alloc/slice/trait.Join.html", &["#method.join"]),
+    ("alloc/slice/trait.Concat.html", &["#method.concat"]),
+    ("alloc/slice/index.html", &["#method.concat", "#method.join"]),
+    ("alloc/vec/struct.Vec.html", &["#method.sort_by_key", "#method.sort_by_cached_key"]),
+    ("core/primitive.str.html", &["#method.to_ascii_uppercase", "#method.to_ascii_lowercase"]),
+    ("core/primitive.slice.html", &["#method.to_ascii_uppercase", "#method.to_ascii_lowercase",
+                                    "core/slice::sort_by_key", "core\\slice::sort_by_key",
+                                    "#method.sort_by_cached_key"]),
 ];
 
 #[rustfmt::skip]
@@ -376,6 +390,10 @@ fn check(&mut self, file: &Path, report: &mut Report) {
 
     /// Load a file from disk, or from the cache if available.
     fn load_file(&mut self, file: &Path, report: &mut Report) -> (String, &FileEntry) {
+        // https://docs.microsoft.com/en-us/windows/win32/debug/system-error-codes--0-499-
+        #[cfg(windows)]
+        const ERROR_INVALID_NAME: i32 = 123;
+
         let pretty_path =
             file.strip_prefix(&self.root).unwrap_or(&file).to_str().unwrap().to_string();
 
@@ -392,6 +410,14 @@ fn load_file(&mut self, file: &Path, report: &mut Report) -> (String, &FileEntry
                 }
                 Err(e) if e.kind() == ErrorKind::NotFound => FileEntry::Missing,
                 Err(e) => {
+                    // If a broken intra-doc link contains `::`, on windows, it will cause `ERROR_INVALID_NAME` rather than `NotFound`.
+                    // Explicitly check for that so that the broken link can be allowed in `LINKCHECK_EXCEPTIONS`.
+                    #[cfg(windows)]
+                    if e.raw_os_error() == Some(ERROR_INVALID_NAME)
+                        && file.as_os_str().to_str().map_or(false, |s| s.contains("::"))
+                    {
+                        return FileEntry::Missing;
+                    }
                     panic!("unexpected read error for {}: {}", file.display(), e);
                 }
             });

src/tools/tidy/src/lib.rs

@@ -46,6 +46,7 @@
 pub mod extdeps;
 pub mod features;
 pub mod pal;
+pub mod primitive_docs;
 pub mod style;
 pub mod target_specific_tests;
 pub mod ui_tests;

src/tools/tidy/src/main.rs

@@ -71,6 +71,7 @@ fn main() {
 
         // Checks that only make sense for the std libs.
         check!(pal, &library_path);
+        check!(primitive_docs, &library_path);
 
         // Checks that need to be done for both the compiler and std libraries.
         check!(unit_tests, &src_path);

src/tools/tidy/src/primitive_docs.rs

+//! Tidy check to make sure `library/{std,core}/src/primitive_docs.rs` are the same file.  These are
+//! different files so that relative links work properly without having to have `CARGO_PKG_NAME`
+//! set, but conceptually they should always be the same.
+
+use std::path::Path;
+
+pub fn check(library_path: &Path, bad: &mut bool) {
+    let std_name = "std/src/primitive_docs.rs";
+    let core_name = "core/src/primitive_docs.rs";
+    let std_contents = std::fs::read_to_string(library_path.join(std_name))
+        .unwrap_or_else(|e| panic!("failed to read library/{}: {}", std_name, e));
+    let core_contents = std::fs::read_to_string(library_path.join(core_name))
+        .unwrap_or_else(|e| panic!("failed to read library/{}: {}", core_name, e));
+    if std_contents != core_contents {
+        tidy_error!(bad, "library/{} and library/{} have different contents", core_name, std_name);
+    }
+}