Auto merge of #74219 - Mark-Simulacrum:beta-next, r=Mark-Simulacrum

[beta] next

Backports of:

* rustdoc: Fix doc aliases with crate filtering #73644
* rustdoc: Rename invalid_codeblock_attribute lint to be plural #74131
* rustc_lexer: Simplify shebang parsing once more #73596
* Perform obligation deduplication to avoid buggy `ExistentialMismatch` #73485
* Reorder order in which MinGW libs are linked to fix recent breakage #73184
* Change how compiler-builtins gets many CGUs #73136
* Fix wasm32 being broken due to a NodeJS version bump #73885

Author: bors

src/bootstrap/bin/rustc.rs

@@ -125,6 +125,20 @@ fn main() {
         } else {
             cmd.arg("-C").arg(format!("debug-assertions={}", debug_assertions));
         }
+
+        // For compiler-builtins we always use a high number of codegen units.
+        // The goal here is to place every single intrinsic into its own object
+        // file to avoid symbol clashes with the system libgcc if possible. Note
+        // that this number doesn't actually produce this many object files, we
+        // just don't create more than this number of object files.
+        //
+        // It's a bit of a bummer that we have to pass this here, unfortunately.
+        // Ideally this would be specified through an env var to Cargo so Cargo
+        // knows how many CGUs are for this specific crate, but for now
+        // per-crate configuration isn't specifiable in the environment.
+        if crate_name == Some("compiler_builtins") {
+            cmd.arg("-Ccodegen-units=10000");
+        }
     } else {
         // FIXME(rust-lang/cargo#5754) we shouldn't be using special env vars
         // here, but rather Cargo should know what flags to pass rustc itself.

src/ci/docker/wasm32/Dockerfile

@@ -25,7 +25,18 @@ RUN ln `which python3` /usr/bin/python
 
 ENV PATH=$PATH:/emsdk-portable
 ENV PATH=$PATH:/emsdk-portable/upstream/emscripten/
-ENV PATH=$PATH:/emsdk-portable/node/12.9.1_64bit/bin/
+
+# Rust's build system requires NodeJS to be in the path, but the directory in
+# which emsdk stores it contains the version number. This caused breakages in
+# the past when emsdk bumped the node version causing the path to point to a
+# missing directory.
+#
+# To avoid the problem this symlinks the latest NodeJs version available to
+# "latest", and adds that to the path.
+RUN ln -s /emsdk-portable/node/$(ls /emsdk-portable/node | sort -V | tail -n 1) \
+          /emsdk-portable/node/latest
+ENV PATH=$PATH:/emsdk-portable/node/latest/bin/
+
 ENV BINARYEN_ROOT=/emsdk-portable/upstream/
 ENV EMSDK=/emsdk-portable
 ENV EM_CONFIG=/emsdk-portable/.emscripten

src/librustc_lexer/src/lib.rs

@@ -239,21 +239,18 @@ pub enum Base {
 /// but shebang isn't a part of rust syntax.
 pub fn strip_shebang(input: &str) -> Option<usize> {
     // Shebang must start with `#!` literally, without any preceding whitespace.
-    if input.starts_with("#!") {
-        let input_tail = &input[2..];
-        // Shebang must have something non-whitespace after `#!` on the first line.
-        let first_line_tail = input_tail.lines().next()?;
-        if first_line_tail.contains(|c| !is_whitespace(c)) {
-            // Ok, this is a shebang but if the next non-whitespace token is `[` or maybe
-            // a doc comment (due to `TokenKind::(Line,Block)Comment` ambiguity at lexer level),
-            // then it may be valid Rust code, so consider it Rust code.
-            let next_non_whitespace_token = tokenize(input_tail).map(|tok| tok.kind).filter(|tok|
-                !matches!(tok, TokenKind::Whitespace | TokenKind::LineComment | TokenKind::BlockComment { .. })
-            ).next();
-            if next_non_whitespace_token != Some(TokenKind::OpenBracket) {
-                // No other choice than to consider this a shebang.
-                return Some(2 + first_line_tail.len());
-            }
+    // For simplicity we consider any line starting with `#!` a shebang,
+    // regardless of restrictions put on shebangs by specific platforms.
+    if let Some(input_tail) = input.strip_prefix("#!") {
+        // Ok, this is a shebang but if the next non-whitespace token is `[` or maybe
+        // a doc comment (due to `TokenKind::(Line,Block)Comment` ambiguity at lexer level),
+        // then it may be valid Rust code, so consider it Rust code.
+        let next_non_whitespace_token = tokenize(input_tail).map(|tok| tok.kind).find(|tok|
+            !matches!(tok, TokenKind::Whitespace | TokenKind::LineComment | TokenKind::BlockComment { .. })
+        );
+        if next_non_whitespace_token != Some(TokenKind::OpenBracket) {
+            // No other choice than to consider this a shebang.
+            return Some(2 + input_tail.lines().next().unwrap_or_default().len());
         }
     }
     None

src/librustc_lint/lib.rs

@@ -61,7 +61,7 @@ use rustc_middle::ty::query::Providers;
 use rustc_middle::ty::TyCtxt;
 use rustc_session::lint::builtin::{
     BARE_TRAIT_OBJECTS, ELIDED_LIFETIMES_IN_PATHS, EXPLICIT_OUTLIVES_REQUIREMENTS,
-    INTRA_DOC_LINK_RESOLUTION_FAILURE, INVALID_CODEBLOCK_ATTRIBUTE, MISSING_DOC_CODE_EXAMPLES,
+    INTRA_DOC_LINK_RESOLUTION_FAILURE, INVALID_CODEBLOCK_ATTRIBUTES, MISSING_DOC_CODE_EXAMPLES,
     PRIVATE_DOC_TESTS,
 };
 use rustc_span::symbol::{Ident, Symbol};
@@ -301,7 +301,7 @@ fn register_builtins(store: &mut LintStore, no_interleave_lints: bool) {
     add_lint_group!(
         "rustdoc",
         INTRA_DOC_LINK_RESOLUTION_FAILURE,
-        INVALID_CODEBLOCK_ATTRIBUTE,
+        INVALID_CODEBLOCK_ATTRIBUTES,
         MISSING_DOC_CODE_EXAMPLES,
         PRIVATE_DOC_TESTS
     );

src/librustc_middle/ty/relate.rs

@@ -617,12 +617,22 @@ impl<'tcx> Relate<'tcx> for &'tcx ty::List<ty::ExistentialPredicate<'tcx>> {
         a: &Self,
         b: &Self,
     ) -> RelateResult<'tcx, Self> {
-        if a.len() != b.len() {
+        let tcx = relation.tcx();
+
+        // FIXME: this is wasteful, but want to do a perf run to see how slow it is.
+        // We need to perform this deduplication as we sometimes generate duplicate projections
+        // in `a`.
+        let mut a_v: Vec<_> = a.into_iter().collect();
+        let mut b_v: Vec<_> = b.into_iter().collect();
+        a_v.sort_by(|a, b| a.stable_cmp(tcx, b));
+        a_v.dedup();
+        b_v.sort_by(|a, b| a.stable_cmp(tcx, b));
+        b_v.dedup();
+        if a_v.len() != b_v.len() {
             return Err(TypeError::ExistentialMismatch(expected_found(relation, a, b)));
         }
 
-        let tcx = relation.tcx();
-        let v = a.iter().zip(b.iter()).map(|(ep_a, ep_b)| {
+        let v = a_v.into_iter().zip(b_v.into_iter()).map(|(ep_a, ep_b)| {
             use crate::ty::ExistentialPredicate::*;
             match (ep_a, ep_b) {
                 (Trait(ref a), Trait(ref b)) => Ok(Trait(relation.relate(a, b)?)),

src/librustc_mir/monomorphize/partitioning.rs

@@ -454,18 +454,11 @@ fn default_visibility(tcx: TyCtxt<'_>, id: DefId, is_generic: bool) -> Visibilit
 fn merge_codegen_units<'tcx>(
     tcx: TyCtxt<'tcx>,
     initial_partitioning: &mut PreInliningPartitioning<'tcx>,
-    mut target_cgu_count: usize,
+    target_cgu_count: usize,
 ) {
     assert!(target_cgu_count >= 1);
     let codegen_units = &mut initial_partitioning.codegen_units;
 
-    if tcx.is_compiler_builtins(LOCAL_CRATE) {
-        // Compiler builtins require some degree of control over how mono items
-        // are partitioned into compilation units. Provide it by keeping the
-        // original partitioning when compiling the compiler builtins crate.
-        target_cgu_count = codegen_units.len();
-    }
-
     // Note that at this point in time the `codegen_units` here may not be in a
     // deterministic order (but we know they're deterministically the same set).
     // We want this merging to produce a deterministic ordering of codegen units

src/librustc_session/lint/builtin.rs

@@ -399,7 +399,7 @@ declare_lint! {
 }
 
 declare_lint! {
-    pub INVALID_CODEBLOCK_ATTRIBUTE,
+    pub INVALID_CODEBLOCK_ATTRIBUTES,
     Warn,
     "codeblock attribute looks a lot like a known one"
 }
@@ -585,7 +585,7 @@ declare_lint_pass! {
         UNSTABLE_NAME_COLLISIONS,
         IRREFUTABLE_LET_PATTERNS,
         INTRA_DOC_LINK_RESOLUTION_FAILURE,
-        INVALID_CODEBLOCK_ATTRIBUTE,
+        INVALID_CODEBLOCK_ATTRIBUTES,
         MISSING_CRATE_LEVEL_DOCS,
         MISSING_DOC_CODE_EXAMPLES,
         PRIVATE_DOC_TESTS,

src/librustc_target/spec/windows_gnu_base.rs

@@ -20,9 +20,9 @@ pub fn opts() -> TargetOptions {
     late_link_args.insert(
         LinkerFlavor::Gcc,
         vec![
+            "-lmsvcrt".to_string(),
             "-lmingwex".to_string(),
             "-lmingw32".to_string(),
-            "-lmsvcrt".to_string(),
             // mingw's msvcrt is a weird hybrid import library and static library.
             // And it seems that the linker fails to use import symbols from msvcrt
             // that are required from functions in msvcrt in certain cases. For example

src/librustdoc/core.rs

@@ -205,7 +205,7 @@ pub fn new_handler(
 
 /// This function is used to setup the lint initialization. By default, in rustdoc, everything
 /// is "allowed". Depending if we run in test mode or not, we want some of them to be at their
-/// default level. For example, the "INVALID_CODEBLOCK_ATTRIBUTE" lint is activated in both
+/// default level. For example, the "INVALID_CODEBLOCK_ATTRIBUTES" lint is activated in both
 /// modes.
 ///
 /// A little detail easy to forget is that there is a way to set the lint level for all lints
@@ -311,7 +311,7 @@ pub fn run_core(options: RustdocOptions) -> (clean::Crate, RenderInfo, RenderOpt
     let missing_doc_example = rustc_lint::builtin::MISSING_DOC_CODE_EXAMPLES.name;
     let private_doc_tests = rustc_lint::builtin::PRIVATE_DOC_TESTS.name;
     let no_crate_level_docs = rustc_lint::builtin::MISSING_CRATE_LEVEL_DOCS.name;
-    let invalid_codeblock_attribute_name = rustc_lint::builtin::INVALID_CODEBLOCK_ATTRIBUTE.name;
+    let invalid_codeblock_attribute_name = rustc_lint::builtin::INVALID_CODEBLOCK_ATTRIBUTES.name;
 
     // In addition to those specific lints, we also need to whitelist those given through
     // command line, otherwise they'll get ignored and we don't want that.

src/librustdoc/html/markdown.rs

@@ -657,7 +657,7 @@ impl<'a, 'b> ExtraInfo<'a, 'b> {
             (None, None) => return,
         };
         self.tcx.struct_span_lint_hir(
-            lint::builtin::INVALID_CODEBLOCK_ATTRIBUTE,
+            lint::builtin::INVALID_CODEBLOCK_ATTRIBUTES,
             hir_id,
             self.sp,
             |lint| {

src/librustdoc/html/static/main.js

@@ -1006,12 +1006,13 @@ function defocusSearchBar() {
                 var aliases = [];
                 var crateAliases = [];
                 var i;
-                if (filterCrates !== undefined &&
-                        ALIASES[filterCrates] &&
-                        ALIASES[filterCrates][query.search]) {
-                    for (i = 0; i < ALIASES[crate][query.search].length; ++i) {
-                        aliases.push(
-                            createAliasFromItem(searchIndex[ALIASES[filterCrates][query.search]]));
+                if (filterCrates !== undefined) {
+                    if (ALIASES[filterCrates] && ALIASES[filterCrates][query.search]) {
+                        for (i = 0; i < ALIASES[filterCrates][query.search].length; ++i) {
+                            aliases.push(
+                                createAliasFromItem(
+                                    searchIndex[ALIASES[filterCrates][query.search][i]]));
+                        }
                     }
                 } else {
                     Object.keys(ALIASES).forEach(function(crate) {

src/librustdoc/test.rs

@@ -46,7 +46,7 @@ pub struct TestOptions {
 pub fn run(options: Options) -> Result<(), String> {
     let input = config::Input::File(options.input.clone());
 
-    let invalid_codeblock_attribute_name = rustc_lint::builtin::INVALID_CODEBLOCK_ATTRIBUTE.name;
+    let invalid_codeblock_attribute_name = rustc_lint::builtin::INVALID_CODEBLOCK_ATTRIBUTES.name;
 
     // In addition to those specific lints, we also need to whitelist those given through
     // command line, otherwise they'll get ignored and we don't want that.

src/test/codegen-units/partitioning/compiler-builtins.rs

@@ -0,0 +1,9 @@
+// exact-check
+
+const QUERY = 'true';
+
+const FILTER_CRATE = 'some_other_crate';
+
+const EXPECTED = {
+    'others': [],
+};

src/test/rustdoc-js/doc-alias-filter-out.js

@@ -0,0 +1,4 @@
+#![feature(doc_alias)]
+
+#[doc(alias = "true")]
+pub struct Foo;

src/test/rustdoc-js/doc-alias-filter-out.rs

@@ -0,0 +1,17 @@
+// exact-check
+
+const QUERY = 'true';
+
+const FILTER_CRATE = 'doc_alias_filter';
+
+const EXPECTED = {
+    'others': [
+        {
+            'path': 'doc_alias_filter',
+            'name': 'Foo',
+            'alias': 'true',
+            'href': '../doc_alias_filter/struct.Foo.html',
+            'is_alias': true
+        },
+    ],
+};

src/test/rustdoc-js/doc-alias-filter.js

@@ -0,0 +1,7 @@
+#![feature(doc_alias)]
+
+#[doc(alias = "true")]
+pub struct Foo;
+
+#[doc(alias = "false")]
+pub struct Bar;

src/test/rustdoc-js/doc-alias-filter.rs

@@ -1,6 +1,6 @@
 // compile-flags:--test
 
-#![deny(invalid_codeblock_attribute)]
+#![deny(invalid_codeblock_attributes)]
 
 /// foo
 ///

src/test/rustdoc-ui/check-attr-test.rs

@@ -11,8 +11,8 @@ error: unknown attribute `compile-fail`. Did you mean `compile_fail`?
 note: the lint level is defined here
  --> $DIR/check-attr-test.rs:3:9
   |
-3 | #![deny(invalid_codeblock_attribute)]
-  |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+3 | #![deny(invalid_codeblock_attributes)]
+  |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
   = help: the code block will either not be tested if not marked as a rust one or won't fail if it compiles successfully
 
 error: unknown attribute `compilefail`. Did you mean `compile_fail`?

src/test/rustdoc-ui/check-attr-test.stderr

@@ -1,4 +1,4 @@
-#![deny(invalid_codeblock_attribute)]
+#![deny(invalid_codeblock_attributes)]
 
 /// foo
 //~^ ERROR

src/test/rustdoc-ui/check-attr.rs

@@ -13,8 +13,8 @@ LL | | /// ```
 note: the lint level is defined here
   --> $DIR/check-attr.rs:1:9
    |
-LL | #![deny(invalid_codeblock_attribute)]
-   |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+LL | #![deny(invalid_codeblock_attributes)]
+   |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    = help: the code block will either not be tested if not marked as a rust one or won't fail if it compiles successfully
 
 error: unknown attribute `compilefail`. Did you mean `compile_fail`?

src/test/rustdoc-ui/check-attr.stderr

@@ -0,0 +1,26 @@
+// check-pass
+trait Service {
+    type S;
+}
+
+trait Framing {
+    type F;
+}
+
+impl Framing for () {
+    type F = ();
+}
+
+trait HttpService<F: Framing>: Service<S = F::F> {}
+
+type BoxService = Box<dyn HttpService<(), S = ()>>;
+
+fn build_server<F: FnOnce() -> BoxService>(_: F) {}
+
+fn make_server<F: Framing>() -> Box<dyn HttpService<F, S = F::F>> {
+    unimplemented!()
+}
+
+fn main() {
+    build_server(|| make_server())
+}

src/test/ui/issues/issue-59326.rs

@@ -0,0 +1,4 @@
+#!
+
+// check-pass
+fn main() {}

src/test/ui/parser/shebang/shebang-empty.rs

@@ -0,0 +1,5 @@
+#!    
+
+// check-pass
+// ignore-tidy-end-whitespace
+fn main() {}