Auto merge of #72813 - RalfJung:rollup-4ko6q8j, r=RalfJung

Rollup of 5 pull requests

Successful merges:

 - #72683 (from_u32_unchecked: check validity, and fix UB in Wtf8)
 - #72715 (Account for trailing comma when suggesting `where` clauses)
 - #72745 (generalize Borrow<[T]> for Interned<'tcx, List<T>>)
 - #72749 (Update stdarch submodule to latest head)
 - #72781 (Use `LocalDefId` instead of `NodeId` in `resolve_str_path_error`)

Failed merges:

r? @ghost

Author: bors

src/libcore/char/convert.rs

@@ -99,7 +99,7 @@ pub fn from_u32(i: u32) -> Option<char> {
 #[inline]
 #[stable(feature = "char_from_unchecked", since = "1.5.0")]
 pub unsafe fn from_u32_unchecked(i: u32) -> char {
-    transmute(i)
+    if cfg!(debug_assertions) { char::from_u32(i).unwrap() } else { transmute(i) }
 }
 
 #[stable(feature = "char_convert", since = "1.13.0")]
@@ -218,7 +218,7 @@ impl TryFrom<u32> for char {
             Err(CharTryFromError(()))
         } else {
             // SAFETY: checked that it's a legal unicode value
-            Ok(unsafe { from_u32_unchecked(i) })
+            Ok(unsafe { transmute(i) })
         }
     }
 }

src/libcore/char/methods.rs

@@ -593,16 +593,7 @@ impl char {
     #[stable(feature = "rust1", since = "1.0.0")]
     #[inline]
     pub fn len_utf8(self) -> usize {
-        let code = self as u32;
-        if code < MAX_ONE_B {
-            1
-        } else if code < MAX_TWO_B {
-            2
-        } else if code < MAX_THREE_B {
-            3
-        } else {
-            4
-        }
+        len_utf8(self as u32)
     }
 
     /// Returns the number of 16-bit code units this `char` would need if
@@ -670,36 +661,8 @@ impl char {
     #[stable(feature = "unicode_encode_char", since = "1.15.0")]
     #[inline]
     pub fn encode_utf8(self, dst: &mut [u8]) -> &mut str {
-        let code = self as u32;
-        let len = self.len_utf8();
-        match (len, &mut dst[..]) {
-            (1, [a, ..]) => {
-                *a = code as u8;
-            }
-            (2, [a, b, ..]) => {
-                *a = (code >> 6 & 0x1F) as u8 | TAG_TWO_B;
-                *b = (code & 0x3F) as u8 | TAG_CONT;
-            }
-            (3, [a, b, c, ..]) => {
-                *a = (code >> 12 & 0x0F) as u8 | TAG_THREE_B;
-                *b = (code >> 6 & 0x3F) as u8 | TAG_CONT;
-                *c = (code & 0x3F) as u8 | TAG_CONT;
-            }
-            (4, [a, b, c, d, ..]) => {
-                *a = (code >> 18 & 0x07) as u8 | TAG_FOUR_B;
-                *b = (code >> 12 & 0x3F) as u8 | TAG_CONT;
-                *c = (code >> 6 & 0x3F) as u8 | TAG_CONT;
-                *d = (code & 0x3F) as u8 | TAG_CONT;
-            }
-            _ => panic!(
-                "encode_utf8: need {} bytes to encode U+{:X}, but the buffer has {}",
-                len,
-                code,
-                dst.len(),
-            ),
-        };
-        // SAFETY: We just wrote UTF-8 content in, so converting to str is fine.
-        unsafe { from_utf8_unchecked_mut(&mut dst[..len]) }
+        // SAFETY: `char` is not a surrogate, so this is valid UTF-8.
+        unsafe { from_utf8_unchecked_mut(encode_utf8_raw(self as u32, dst)) }
     }
 
     /// Encodes this character as UTF-16 into the provided `u16` buffer,
@@ -739,28 +702,7 @@ impl char {
     #[stable(feature = "unicode_encode_char", since = "1.15.0")]
     #[inline]
     pub fn encode_utf16(self, dst: &mut [u16]) -> &mut [u16] {
-        let mut code = self as u32;
-        // SAFETY: each arm checks whether there are enough bits to write into
-        unsafe {
-            if (code & 0xFFFF) == code && !dst.is_empty() {
-                // The BMP falls through (assuming non-surrogate, as it should)
-                *dst.get_unchecked_mut(0) = code as u16;
-                slice::from_raw_parts_mut(dst.as_mut_ptr(), 1)
-            } else if dst.len() >= 2 {
-                // Supplementary planes break into surrogates.
-                code -= 0x1_0000;
-                *dst.get_unchecked_mut(0) = 0xD800 | ((code >> 10) as u16);
-                *dst.get_unchecked_mut(1) = 0xDC00 | ((code as u16) & 0x3FF);
-                slice::from_raw_parts_mut(dst.as_mut_ptr(), 2)
-            } else {
-                panic!(
-                    "encode_utf16: need {} units to encode U+{:X}, but the buffer has {}",
-                    from_u32_unchecked(code).len_utf16(),
-                    code,
-                    dst.len(),
-                )
-            }
-        }
+        encode_utf16_raw(self as u32, dst)
     }
 
     /// Returns `true` if this `char` has the `Alphabetic` property.
@@ -1673,3 +1615,100 @@ impl char {
         }
     }
 }
+
+#[inline]
+fn len_utf8(code: u32) -> usize {
+    if code < MAX_ONE_B {
+        1
+    } else if code < MAX_TWO_B {
+        2
+    } else if code < MAX_THREE_B {
+        3
+    } else {
+        4
+    }
+}
+
+/// Encodes a raw u32 value as UTF-8 into the provided byte buffer,
+/// and then returns the subslice of the buffer that contains the encoded character.
+///
+/// Unlike `char::encode_utf8`, this method also handles codepoints in the surrogate range.
+/// (Creating a `char` in the surrogate range is UB.)
+/// The result is valid [generalized UTF-8] but not valid UTF-8.
+///
+/// [generalized UTF-8]: https://simonsapin.github.io/wtf-8/#generalized-utf8
+///
+/// # Panics
+///
+/// Panics if the buffer is not large enough.
+/// A buffer of length four is large enough to encode any `char`.
+#[unstable(feature = "char_internals", reason = "exposed only for libstd", issue = "none")]
+#[doc(hidden)]
+#[inline]
+pub fn encode_utf8_raw(code: u32, dst: &mut [u8]) -> &mut [u8] {
+    let len = len_utf8(code);
+    match (len, &mut dst[..]) {
+        (1, [a, ..]) => {
+            *a = code as u8;
+        }
+        (2, [a, b, ..]) => {
+            *a = (code >> 6 & 0x1F) as u8 | TAG_TWO_B;
+            *b = (code & 0x3F) as u8 | TAG_CONT;
+        }
+        (3, [a, b, c, ..]) => {
+            *a = (code >> 12 & 0x0F) as u8 | TAG_THREE_B;
+            *b = (code >> 6 & 0x3F) as u8 | TAG_CONT;
+            *c = (code & 0x3F) as u8 | TAG_CONT;
+        }
+        (4, [a, b, c, d, ..]) => {
+            *a = (code >> 18 & 0x07) as u8 | TAG_FOUR_B;
+            *b = (code >> 12 & 0x3F) as u8 | TAG_CONT;
+            *c = (code >> 6 & 0x3F) as u8 | TAG_CONT;
+            *d = (code & 0x3F) as u8 | TAG_CONT;
+        }
+        _ => panic!(
+            "encode_utf8: need {} bytes to encode U+{:X}, but the buffer has {}",
+            len,
+            code,
+            dst.len(),
+        ),
+    };
+    &mut dst[..len]
+}
+
+/// Encodes a raw u32 value as UTF-16 into the provided `u16` buffer,
+/// and then returns the subslice of the buffer that contains the encoded character.
+///
+/// Unlike `char::encode_utf16`, this method also handles codepoints in the surrogate range.
+/// (Creating a `char` in the surrogate range is UB.)
+///
+/// # Panics
+///
+/// Panics if the buffer is not large enough.
+/// A buffer of length 2 is large enough to encode any `char`.
+#[unstable(feature = "char_internals", reason = "exposed only for libstd", issue = "none")]
+#[doc(hidden)]
+#[inline]
+pub fn encode_utf16_raw(mut code: u32, dst: &mut [u16]) -> &mut [u16] {
+    // SAFETY: each arm checks whether there are enough bits to write into
+    unsafe {
+        if (code & 0xFFFF) == code && !dst.is_empty() {
+            // The BMP falls through
+            *dst.get_unchecked_mut(0) = code as u16;
+            slice::from_raw_parts_mut(dst.as_mut_ptr(), 1)
+        } else if dst.len() >= 2 {
+            // Supplementary planes break into surrogates.
+            code -= 0x1_0000;
+            *dst.get_unchecked_mut(0) = 0xD800 | ((code >> 10) as u16);
+            *dst.get_unchecked_mut(1) = 0xDC00 | ((code as u16) & 0x3FF);
+            slice::from_raw_parts_mut(dst.as_mut_ptr(), 2)
+        } else {
+            panic!(
+                "encode_utf16: need {} units to encode U+{:X}, but the buffer has {}",
+                from_u32_unchecked(code).len_utf16(),
+                code,
+                dst.len(),
+            )
+        }
+    }
+}

src/libcore/char/mod.rs

@@ -37,6 +37,12 @@ pub use self::decode::{decode_utf16, DecodeUtf16, DecodeUtf16Error};
 #[stable(feature = "unicode_version", since = "1.45.0")]
 pub use crate::unicode::UNICODE_VERSION;
 
+// perma-unstable re-exports
+#[unstable(feature = "char_internals", reason = "exposed only for libstd", issue = "none")]
+pub use self::methods::encode_utf16_raw;
+#[unstable(feature = "char_internals", reason = "exposed only for libstd", issue = "none")]
+pub use self::methods::encode_utf8_raw;
+
 use crate::fmt::{self, Write};
 use crate::iter::FusedIterator;
 

src/librustc_hir/hir.rs

@@ -524,6 +524,13 @@ impl WhereClause<'_> {
     pub fn span_for_predicates_or_empty_place(&self) -> Span {
         self.span
     }
+
+    /// `Span` where further predicates would be suggested, accounting for trailing commas, like
+    ///  in `fn foo<T>(t: T) where T: Foo,` so we don't suggest two trailing commas.
+    pub fn tail_span_for_suggestion(&self) -> Span {
+        let end = self.span_for_predicates_or_empty_place().shrink_to_hi();
+        self.predicates.last().map(|p| p.span()).unwrap_or(end).shrink_to_hi().to(end)
+    }
 }
 
 /// A single predicate in a where-clause.

src/librustc_middle/ty/context.rs

@@ -1971,32 +1971,8 @@ impl<'tcx, T: Hash> Hash for Interned<'tcx, List<T>> {
     }
 }
 
-impl<'tcx> Borrow<[Ty<'tcx>]> for Interned<'tcx, List<Ty<'tcx>>> {
-    fn borrow<'a>(&'a self) -> &'a [Ty<'tcx>] {
-        &self.0[..]
-    }
-}
-
-impl<'tcx> Borrow<[CanonicalVarInfo]> for Interned<'tcx, List<CanonicalVarInfo>> {
-    fn borrow(&self) -> &[CanonicalVarInfo] {
-        &self.0[..]
-    }
-}
-
-impl<'tcx> Borrow<[GenericArg<'tcx>]> for Interned<'tcx, InternalSubsts<'tcx>> {
-    fn borrow<'a>(&'a self) -> &'a [GenericArg<'tcx>] {
-        &self.0[..]
-    }
-}
-
-impl<'tcx> Borrow<[ProjectionKind]> for Interned<'tcx, List<ProjectionKind>> {
-    fn borrow(&self) -> &[ProjectionKind] {
-        &self.0[..]
-    }
-}
-
-impl<'tcx> Borrow<[PlaceElem<'tcx>]> for Interned<'tcx, List<PlaceElem<'tcx>>> {
-    fn borrow(&self) -> &[PlaceElem<'tcx>] {
+impl<'tcx, T> Borrow<[T]> for Interned<'tcx, List<T>> {
+    fn borrow<'a>(&'a self) -> &'a [T] {
         &self.0[..]
     }
 }
@@ -2007,34 +1983,12 @@ impl<'tcx> Borrow<RegionKind> for Interned<'tcx, RegionKind> {
     }
 }
 
-impl<'tcx> Borrow<[ExistentialPredicate<'tcx>]>
-    for Interned<'tcx, List<ExistentialPredicate<'tcx>>>
-{
-    fn borrow<'a>(&'a self) -> &'a [ExistentialPredicate<'tcx>] {
-        &self.0[..]
-    }
-}
-
-impl<'tcx> Borrow<[Predicate<'tcx>]> for Interned<'tcx, List<Predicate<'tcx>>> {
-    fn borrow<'a>(&'a self) -> &'a [Predicate<'tcx>] {
-        &self.0[..]
-    }
-}
-
 impl<'tcx> Borrow<Const<'tcx>> for Interned<'tcx, Const<'tcx>> {
     fn borrow<'a>(&'a self) -> &'a Const<'tcx> {
         &self.0
     }
 }
 
-impl<'tcx> Borrow<[traits::ChalkEnvironmentClause<'tcx>]>
-    for Interned<'tcx, List<traits::ChalkEnvironmentClause<'tcx>>>
-{
-    fn borrow<'a>(&'a self) -> &'a [traits::ChalkEnvironmentClause<'tcx>] {
-        &self.0[..]
-    }
-}
-
 impl<'tcx> Borrow<PredicateKind<'tcx>> for Interned<'tcx, PredicateKind<'tcx>> {
     fn borrow<'a>(&'a self) -> &'a PredicateKind<'tcx> {
         &self.0

src/librustc_middle/ty/diagnostics.rs

@@ -7,7 +7,6 @@ use rustc_errors::{Applicability, DiagnosticBuilder};
 use rustc_hir as hir;
 use rustc_hir::def_id::DefId;
 use rustc_hir::{QPath, TyKind, WhereBoundPredicate, WherePredicate};
-use rustc_span::{BytePos, Span};
 
 impl<'tcx> TyS<'tcx> {
     /// Similar to `TyS::is_primitive`, but also considers inferred numeric values to be primitive.
@@ -221,24 +220,11 @@ pub fn suggest_constraining_type_param(
             }
         }
 
-        let where_clause_span = generics.where_clause.span_for_predicates_or_empty_place();
-        // Account for `fn foo<T>(t: T) where T: Foo,` so we don't suggest two trailing commas.
-        let mut trailing_comma = false;
-        if let Ok(snippet) = tcx.sess.source_map().span_to_snippet(where_clause_span) {
-            trailing_comma = snippet.ends_with(',');
-        }
-        let where_clause_span = if trailing_comma {
-            let hi = where_clause_span.hi();
-            Span::new(hi - BytePos(1), hi, where_clause_span.ctxt())
-        } else {
-            where_clause_span.shrink_to_hi()
-        };
-
         match &param_spans[..] {
             &[&param_span] => suggest_restrict(param_span.shrink_to_hi()),
             _ => {
                 err.span_suggestion_verbose(
-                    where_clause_span,
+                    generics.where_clause.tail_span_for_suggestion(),
                     &msg_restrict_type_further,
                     format!(", {}: {}", param_name, constraint),
                     Applicability::MachineApplicable,

src/librustc_resolve/lib.rs

@@ -2902,7 +2902,7 @@ impl<'a> Resolver<'a> {
         span: Span,
         path_str: &str,
         ns: Namespace,
-        module_id: NodeId,
+        module_id: LocalDefId,
     ) -> Result<(ast::Path, Res), ()> {
         let path = if path_str.starts_with("::") {
             ast::Path {
@@ -2922,10 +2922,7 @@ impl<'a> Resolver<'a> {
                     .collect(),
             }
         };
-        let module = self.block_map.get(&module_id).copied().unwrap_or_else(|| {
-            let def_id = self.definitions.local_def_id(module_id);
-            self.module_map.get(&def_id).copied().unwrap_or(self.graph_root)
-        });
+        let module = self.module_map.get(&module_id).copied().unwrap_or(self.graph_root);
         let parent_scope = &ParentScope::module(module);
         let res = self.resolve_ast_path(&path, ns, parent_scope).map_err(|_| ())?;
         Ok((path, res))

src/librustc_trait_selection/traits/error_reporting/suggestions.rs

@@ -170,7 +170,7 @@ pub trait InferCtxtExt<'tcx> {
 
 fn predicate_constraint(generics: &hir::Generics<'_>, pred: String) -> (Span, String) {
     (
-        generics.where_clause.span_for_predicates_or_empty_place().shrink_to_hi(),
+        generics.where_clause.tail_span_for_suggestion(),
         format!(
             "{} {}",
             if !generics.where_clause.predicates.is_empty() { "," } else { " where" },

src/librustdoc/core.rs

@@ -1,4 +1,3 @@
-use rustc_ast::ast::CRATE_NODE_ID;
 use rustc_attr as attr;
 use rustc_data_structures::fx::{FxHashMap, FxHashSet};
 use rustc_data_structures::sync::{self, Lrc};
@@ -7,7 +6,7 @@ use rustc_errors::emitter::{Emitter, EmitterWriter};
 use rustc_errors::json::JsonEmitter;
 use rustc_feature::UnstableFeatures;
 use rustc_hir::def::Namespace::TypeNS;
-use rustc_hir::def_id::{CrateNum, DefId, DefIndex, LOCAL_CRATE};
+use rustc_hir::def_id::{CrateNum, DefId, DefIndex, LocalDefId, CRATE_DEF_INDEX, LOCAL_CRATE};
 use rustc_hir::HirId;
 use rustc_interface::interface;
 use rustc_middle::middle::cstore::CrateStore;
@@ -390,7 +389,12 @@ pub fn run_core(options: RustdocOptions) -> (clean::Crate, RenderInfo, RenderOpt
                 resolver.borrow_mut().access(|resolver| {
                     for extern_name in &extern_names {
                         resolver
-                            .resolve_str_path_error(DUMMY_SP, extern_name, TypeNS, CRATE_NODE_ID)
+                            .resolve_str_path_error(
+                                DUMMY_SP,
+                                extern_name,
+                                TypeNS,
+                                LocalDefId { local_def_index: CRATE_DEF_INDEX },
+                            )
                             .unwrap_or_else(|()| {
                                 panic!("Unable to resolve external crate {}", extern_name)
                             });