Auto merge of #145043 - Zalathar:rollup-3dbvdrm, r=Zalathar

Rollup of 19 pull requests

Successful merges:

 - rust-lang/rust#137831 (Tweak auto trait errors)
 - rust-lang/rust#138689 (add nvptx_target_feature)
 - rust-lang/rust#140267 (implement continue_ok and break_ok for ControlFlow)
 - rust-lang/rust#143028 (emit `StorageLive` and schedule `StorageDead` for `let`-`else`'s bindings after matching)
 - rust-lang/rust#143764 (lower pattern bindings in the order they're written and base drop order on primary bindings' order)
 - rust-lang/rust#143808 (Port `#[should_panic]` to the new attribute parsing infrastructure )
 - rust-lang/rust#143906 (Miri: non-deterministic floating point operations in `foreign_items`)
 - rust-lang/rust#143929 (Mark all deprecation lints in name resolution as deny-by-default and report-in-deps)
 - rust-lang/rust#144133 (Stabilize const TypeId::of)
 - rust-lang/rust#144369 (Upgrade semicolon_in_expressions_from_macros from warn to deny)
 - rust-lang/rust#144439 (Introduce ModernIdent type to unify macro 2.0 hygiene handling)
 - rust-lang/rust#144473 (Address libunwind.a inconsistency issues in the bootstrap program)
 - rust-lang/rust#144601 (Allow `cargo fix` to partially apply `mismatched_lifetime_syntaxes`)
 - rust-lang/rust#144650 (Additional tce tests)
 - rust-lang/rust#144659 (bootstrap: refactor mingw dist and fix gnullvm)
 - rust-lang/rust#144682 (Stabilize `strict_overflow_ops`)
 - rust-lang/rust#145026 (Update books)
 - rust-lang/rust#145033 (Reimplement `print_region` in `type_name.rs`.)
 - rust-lang/rust#145040 (rustc-dev-guide subtree update)

Failed merges:

 - rust-lang/rust#143857 (Port #[macro_export] to the new attribute parsing infrastructure)

r? `@ghost`
`@rustbot` modify labels: rollup

compiler/rustc_codegen_ssa/src/target_features.rs

@@ -197,7 +197,10 @@ fn parse_rust_feature_flag<'a>(
 /// 2nd component of the return value, respectively).
 ///
 /// `target_base_has_feature` should check whether the given feature (a Rust feature name!) is
-/// enabled in the "base" target machine, i.e., without applying `-Ctarget-feature`.
+/// enabled in the "base" target machine, i.e., without applying `-Ctarget-feature`. Note that LLVM
+/// may consider features to be implied that we do not and vice-versa. We want `cfg` to be entirely
+/// consistent with Rust feature implications, and thus only consult LLVM to expand the target CPU
+/// to target features.
 ///
 /// We do not have to worry about RUSTC_SPECIFIC_FEATURES here, those are handled elsewhere.
 pub fn cfg_target_feature(
@@ -211,7 +214,15 @@ pub fn cfg_target_feature(
         .rust_target_features()
         .iter()
         .filter(|(feature, _, _)| target_base_has_feature(feature))
-        .map(|(feature, _, _)| Symbol::intern(feature))
+        .flat_map(|(base_feature, _, _)| {
+            // Expand the direct base feature into all transitively-implied features. Note that we
+            // cannot simply use the `implied` field of the tuple since that only contains
+            // directly-implied features.
+            //
+            // Iteration order is irrelevant because we're collecting into an `UnordSet`.
+            #[allow(rustc::potential_query_instability)]
+            sess.target.implied_target_features(base_feature).into_iter().map(|f| Symbol::intern(f))
+        })
         .collect();
 
     // Add enabled and remove disabled features.