Rollup of 4 pull requests

compiler/rustc_abi/src/layout.rs

@@ -812,7 +812,7 @@ impl<Cx: HasDataLayout> LayoutCalculator<Cx> {
         let (max, min) = largest_niche
             // We might have no inhabited variants, so pretend there's at least one.
             .unwrap_or((0, 0));
-        let (min_ity, signed) = discr_range_of_repr(min, max); //Integer::repr_discr(tcx, ty, &repr, min, max);
+        let (min_ity, signed) = discr_range_of_repr(min, max); //Integer::discr_range_of_repr(tcx, ty, &repr, min, max);
 
         let mut align = dl.aggregate_align;
         let mut max_repr_align = repr.align;

compiler/rustc_abi/src/layout/ty.rs

@@ -172,6 +172,8 @@ pub trait TyAbiInterface<'a, C>: Sized + std::fmt::Debug {
     fn is_tuple(this: TyAndLayout<'a, Self>) -> bool;
     fn is_unit(this: TyAndLayout<'a, Self>) -> bool;
     fn is_transparent(this: TyAndLayout<'a, Self>) -> bool;
+    /// See [`TyAndLayout::pass_indirectly_in_non_rustic_abis`] for details.
+    fn is_pass_indirectly_in_non_rustic_abis_flag_set(this: TyAndLayout<'a, Self>) -> bool;
 }
 
 impl<'a, Ty> TyAndLayout<'a, Ty> {
@@ -269,6 +271,30 @@ impl<'a, Ty> TyAndLayout<'a, Ty> {
         Ty::is_transparent(self)
     }
 
+    /// If this method returns `true`, then this type should always have a `PassMode` of
+    /// `Indirect { on_stack: false, .. }` when being used as the argument type of a function with a
+    /// non-Rustic ABI (this is true for structs annotated with the
+    /// `#[rustc_pass_indirectly_in_non_rustic_abis]` attribute).
+    ///
+    /// This is used to replicate some of the behaviour of C array-to-pointer decay; however unlike
+    /// C any changes the caller makes to the passed value will not be reflected in the callee, so
+    /// the attribute is only useful for types where observing the value in the caller after the
+    /// function call isn't allowed (a.k.a. `va_list`).
+    ///
+    /// This function handles transparent types automatically.
+    pub fn pass_indirectly_in_non_rustic_abis<C>(mut self, cx: &C) -> bool
+    where
+        Ty: TyAbiInterface<'a, C> + Copy,
+    {
+        while self.is_transparent()
+            && let Some((_, field)) = self.non_1zst_field(cx)
+        {
+            self = field;
+        }
+
+        Ty::is_pass_indirectly_in_non_rustic_abis_flag_set(self)
+    }
+
     /// Finds the one field that is not a 1-ZST.
     /// Returns `None` if there are multiple non-1-ZST fields or only 1-ZST-fields.
     pub fn non_1zst_field<C>(&self, cx: &C) -> Option<(FieldIdx, Self)>

compiler/rustc_abi/src/lib.rs

@@ -88,14 +88,17 @@ bitflags! {
         const IS_C               = 1 << 0;
         const IS_SIMD            = 1 << 1;
         const IS_TRANSPARENT     = 1 << 2;
-        // Internal only for now. If true, don't reorder fields.
-        // On its own it does not prevent ABI optimizations.
+        /// Internal only for now. If true, don't reorder fields.
+        /// On its own it does not prevent ABI optimizations.
         const IS_LINEAR          = 1 << 3;
-        // If true, the type's crate has opted into layout randomization.
-        // Other flags can still inhibit reordering and thus randomization.
-        // The seed stored in `ReprOptions.field_shuffle_seed`.
+        /// If true, the type's crate has opted into layout randomization.
+        /// Other flags can still inhibit reordering and thus randomization.
+        /// The seed stored in `ReprOptions.field_shuffle_seed`.
         const RANDOMIZE_LAYOUT   = 1 << 4;
-        // Any of these flags being set prevent field reordering optimisation.
+        /// If true, the type is always passed indirectly by non-Rustic ABIs.
+        /// See [`TyAndLayout::pass_indirectly_in_non_rustic_abis`] for details.
+        const PASS_INDIRECTLY_IN_NON_RUSTIC_ABIS = 1 << 5;
+        /// Any of these flags being set prevent field reordering optimisation.
         const FIELD_ORDER_UNOPTIMIZABLE   = ReprFlags::IS_C.bits()
                                  | ReprFlags::IS_SIMD.bits()
                                  | ReprFlags::IS_LINEAR.bits();
@@ -183,6 +186,11 @@ impl ReprOptions {
 
     /// Returns the discriminant type, given these `repr` options.
     /// This must only be called on enums!
+    ///
+    /// This is the "typeck type" of the discriminant, which is effectively the maximum size:
+    /// discriminant values will be wrapped to fit (with a lint). Layout can later decide to use a
+    /// smaller type for the tag that stores the discriminant at runtime and that will work just
+    /// fine, it just induces casts when getting/setting the discriminant.
     pub fn discr_type(&self) -> IntegerType {
         self.int.unwrap_or(IntegerType::Pointer(true))
     }

compiler/rustc_attr_parsing/src/attributes/codegen_attrs.rs

@@ -676,3 +676,12 @@ impl<S: Stage> SingleAttributeParser<S> for SanitizeParser {
         Some(AttributeKind::Sanitize { on_set, off_set, span: cx.attr_span })
     }
 }
+
+pub(crate) struct RustcPassIndirectlyInNonRusticAbisParser;
+
+impl<S: Stage> NoArgsAttributeParser<S> for RustcPassIndirectlyInNonRusticAbisParser {
+    const PATH: &[Symbol] = &[sym::rustc_pass_indirectly_in_non_rustic_abis];
+    const ON_DUPLICATE: OnDuplicate<S> = OnDuplicate::Error;
+    const ALLOWED_TARGETS: AllowedTargets = AllowedTargets::AllowList(&[Allow(Target::Struct)]);
+    const CREATE: fn(Span) -> AttributeKind = AttributeKind::RustcPassIndirectlyInNonRusticAbis;
+}

compiler/rustc_attr_parsing/src/context.rs

@@ -20,8 +20,9 @@ use crate::attributes::allow_unstable::{
 use crate::attributes::body::CoroutineParser;
 use crate::attributes::codegen_attrs::{
     ColdParser, CoverageParser, ExportNameParser, ForceTargetFeatureParser, NakedParser,
-    NoMangleParser, ObjcClassParser, ObjcSelectorParser, OptimizeParser, SanitizeParser,
-    TargetFeatureParser, TrackCallerParser, UsedParser,
+    NoMangleParser, ObjcClassParser, ObjcSelectorParser, OptimizeParser,
+    RustcPassIndirectlyInNonRusticAbisParser, SanitizeParser, TargetFeatureParser,
+    TrackCallerParser, UsedParser,
 };
 use crate::attributes::confusables::ConfusablesParser;
 use crate::attributes::crate_level::{
@@ -243,6 +244,7 @@ attribute_parsers!(
         Single<WithoutArgs<PubTransparentParser>>,
         Single<WithoutArgs<RustcCoherenceIsCoreParser>>,
         Single<WithoutArgs<RustcMainParser>>,
+        Single<WithoutArgs<RustcPassIndirectlyInNonRusticAbisParser>>,
         Single<WithoutArgs<SpecializationTraitParser>>,
         Single<WithoutArgs<StdInternalSymbolParser>>,
         Single<WithoutArgs<TrackCallerParser>>,

compiler/rustc_feature/src/builtin_attrs.rs

@@ -657,6 +657,12 @@ pub static BUILTIN_ATTRIBUTES: &[BuiltinAttribute] = &[
         template!(Word, "https://doc.rust-lang.org/reference/attributes/codegen.html#the-naked-attribute"),
         WarnFollowing, EncodeCrossCrate::No
     ),
+    // See `TyAndLayout::pass_indirectly_in_non_rustic_abis` for details.
+    rustc_attr!(
+        rustc_pass_indirectly_in_non_rustic_abis, Normal, template!(Word), ErrorFollowing,
+        EncodeCrossCrate::No,
+        "types marked with `#[rustc_pass_indirectly_in_non_rustic_abis]` are always passed indirectly by non-Rustic abis."
+    ),
 
     // Limits:
     ungated!(

compiler/rustc_hir/src/attrs/data_structures.rs

@@ -679,6 +679,9 @@ pub enum AttributeKind {
     /// Represents `#[rustc_object_lifetime_default]`.
     RustcObjectLifetimeDefault,
 
+    /// Represents `#[rustc_pass_indirectly_in_non_rustic_abis]`
+    RustcPassIndirectlyInNonRusticAbis(Span),
+
     /// Represents `#[rustc_simd_monomorphize_lane_limit = "N"]`.
     RustcSimdMonomorphizeLaneLimit(Limit),
 

compiler/rustc_hir/src/attrs/encode_cross_crate.rs

@@ -91,6 +91,7 @@ impl AttributeKind {
             RustcLayoutScalarValidRangeStart(..) => Yes,
             RustcMain => No,
             RustcObjectLifetimeDefault => No,
+            RustcPassIndirectlyInNonRusticAbis(..) => No,
             RustcSimdMonomorphizeLaneLimit(..) => Yes, // Affects layout computation, which needs to work cross-crate
             Sanitize { .. } => No,
             ShouldPanic { .. } => No,

compiler/rustc_hir_analysis/src/check/check.rs

@@ -782,7 +782,7 @@ pub(crate) fn check_item_type(tcx: TyCtxt<'_>, def_id: LocalDefId) -> Result<(),
             tcx.ensure_ok().generics_of(def_id);
             tcx.ensure_ok().type_of(def_id);
             tcx.ensure_ok().predicates_of(def_id);
-            crate::collect::lower_enum_variant_types(tcx, def_id.to_def_id());
+            crate::collect::lower_enum_variant_types(tcx, def_id);
             check_enum(tcx, def_id);
             check_variances_for_type_defn(tcx, def_id);
         }

compiler/rustc_hir_analysis/src/collect.rs

@@ -19,7 +19,7 @@ use std::cell::Cell;
 use std::iter;
 use std::ops::Bound;
 
-use rustc_abi::ExternAbi;
+use rustc_abi::{ExternAbi, Size};
 use rustc_ast::Recovered;
 use rustc_data_structures::fx::{FxHashSet, FxIndexMap};
 use rustc_data_structures::unord::UnordMap;
@@ -605,32 +605,70 @@ pub(super) fn lower_variant_ctor(tcx: TyCtxt<'_>, def_id: LocalDefId) {
     tcx.ensure_ok().predicates_of(def_id);
 }
 
-pub(super) fn lower_enum_variant_types(tcx: TyCtxt<'_>, def_id: DefId) {
+pub(super) fn lower_enum_variant_types(tcx: TyCtxt<'_>, def_id: LocalDefId) {
     let def = tcx.adt_def(def_id);
     let repr_type = def.repr().discr_type();
     let initial = repr_type.initial_discriminant(tcx);
     let mut prev_discr = None::<Discr<'_>>;
+    // Some of the logic below relies on `i128` being able to hold all c_int and c_uint values.
+    assert!(tcx.sess.target.c_int_width < 128);
+    let mut min_discr = i128::MAX;
+    let mut max_discr = i128::MIN;
 
     // fill the discriminant values and field types
     for variant in def.variants() {
         let wrapped_discr = prev_discr.map_or(initial, |d| d.wrap_incr(tcx));
-        prev_discr = Some(
-            if let ty::VariantDiscr::Explicit(const_def_id) = variant.discr {
-                def.eval_explicit_discr(tcx, const_def_id).ok()
-            } else if let Some(discr) = repr_type.disr_incr(tcx, prev_discr) {
-                Some(discr)
-            } else {
+        let cur_discr = if let ty::VariantDiscr::Explicit(const_def_id) = variant.discr {
+            def.eval_explicit_discr(tcx, const_def_id).ok()
+        } else if let Some(discr) = repr_type.disr_incr(tcx, prev_discr) {
+            Some(discr)
+        } else {
+            let span = tcx.def_span(variant.def_id);
+            tcx.dcx().emit_err(errors::EnumDiscriminantOverflowed {
+                span,
+                discr: prev_discr.unwrap().to_string(),
+                item_name: tcx.item_ident(variant.def_id),
+                wrapped_discr: wrapped_discr.to_string(),
+            });
+            None
+        }
+        .unwrap_or(wrapped_discr);
+
+        if def.repr().c() {
+            let c_int = Size::from_bits(tcx.sess.target.c_int_width);
+            let c_uint_max = i128::try_from(c_int.unsigned_int_max()).unwrap();
+            // c_int is a signed type, so get a proper signed version of the discriminant
+            let discr_size = cur_discr.ty.int_size_and_signed(tcx).0;
+            let discr_val = discr_size.sign_extend(cur_discr.val);
+            min_discr = min_discr.min(discr_val);
+            max_discr = max_discr.max(discr_val);
+
+            // The discriminant range must either fit into c_int or c_uint.
+            if !(min_discr >= c_int.signed_int_min() && max_discr <= c_int.signed_int_max())
+                && !(min_discr >= 0 && max_discr <= c_uint_max)
+            {
                 let span = tcx.def_span(variant.def_id);
-                tcx.dcx().emit_err(errors::EnumDiscriminantOverflowed {
+                let msg = if discr_val < c_int.signed_int_min() || discr_val > c_uint_max {
+                    "`repr(C)` enum discriminant does not fit into C `int` nor into C `unsigned int`"
+                } else if discr_val < 0 {
+                    "`repr(C)` enum discriminant does not fit into C `unsigned int`, and a previous discriminant does not fit into C `int`"
+                } else {
+                    "`repr(C)` enum discriminant does not fit into C `int`, and a previous discriminant does not fit into C `unsigned int`"
+                };
+                tcx.node_span_lint(
+                    rustc_session::lint::builtin::REPR_C_ENUMS_LARGER_THAN_INT,
+                    tcx.local_def_id_to_hir_id(def_id),
                     span,
-                    discr: prev_discr.unwrap().to_string(),
-                    item_name: tcx.item_ident(variant.def_id),
-                    wrapped_discr: wrapped_discr.to_string(),
-                });
-                None
+                    |d| {
+                        d.primary_message(msg)
+                        .note("`repr(C)` enums with big discriminants are non-portable, and their size in Rust might not match their size in C")
+                        .help("use `repr($int_ty)` instead to explicitly set the size of this enum");
+                    }
+                );
             }
-            .unwrap_or(wrapped_discr),
-        );
+        }
+
+        prev_discr = Some(cur_discr);
 
         for f in &variant.fields {
             tcx.ensure_ok().generics_of(f.did);

compiler/rustc_infer/src/infer/snapshot/fudge.rs

@@ -1,3 +1,4 @@
+use std::fmt::Debug;
 use std::ops::Range;
 
 use rustc_data_structures::{snapshot_vec as sv, unify as ut};
@@ -84,11 +85,12 @@ impl<'tcx> InferCtxt<'tcx> {
     /// the actual types (`?T`, `Option<?T>`) -- and remember that
     /// after the snapshot is popped, the variable `?T` is no longer
     /// unified.
-    #[instrument(skip(self, f), level = "debug")]
+    #[instrument(skip(self, f), level = "debug", ret)]
     pub fn fudge_inference_if_ok<T, E, F>(&self, f: F) -> Result<T, E>
     where
         F: FnOnce() -> Result<T, E>,
         T: TypeFoldable<TyCtxt<'tcx>>,
+        E: Debug,
     {
         let variable_lengths = self.variable_lengths();
         let (snapshot_vars, value) = self.probe(|_| {

compiler/rustc_lint/src/types.rs

@@ -10,7 +10,7 @@ use rustc_span::{Span, Symbol, sym};
 use tracing::debug;
 use {rustc_ast as ast, rustc_hir as hir};
 
-mod improper_ctypes; // these filed do the implementation for ImproperCTypesDefinitions,ImproperCTypesDeclarations
+mod improper_ctypes; // these files do the implementation for ImproperCTypesDefinitions,ImproperCTypesDeclarations
 pub(crate) use improper_ctypes::ImproperCTypesLint;
 
 use crate::lints::{
@@ -25,7 +25,6 @@ use crate::lints::{
 use crate::{LateContext, LateLintPass, LintContext};
 
 mod literal;
-
 use literal::{int_ty_range, lint_literal, uint_ty_range};
 
 declare_lint! {

compiler/rustc_lint_defs/src/builtin.rs

@@ -86,6 +86,7 @@ declare_lint_pass! {
         REFINING_IMPL_TRAIT_INTERNAL,
         REFINING_IMPL_TRAIT_REACHABLE,
         RENAMED_AND_REMOVED_LINTS,
+        REPR_C_ENUMS_LARGER_THAN_INT,
         REPR_TRANSPARENT_NON_ZST_FIELDS,
         RUST_2021_INCOMPATIBLE_CLOSURE_CAPTURES,
         RUST_2021_INCOMPATIBLE_OR_PATTERNS,
@@ -5213,3 +5214,52 @@ declare_lint! {
     Warn,
     r#"detects when a function annotated with `#[inline(always)]` and `#[target_feature(enable = "..")]` is inlined into a caller without the required target feature"#,
 }
+
+declare_lint! {
+    /// The `repr_c_enums_larger_than_int` lint detects `repr(C)` enums with discriminant
+    /// values that do not fit into a C `int` or `unsigned int`.
+    ///
+    /// ### Example
+    ///
+    /// ```rust,ignore (only errors on 64bit)
+    /// #[repr(C)]
+    /// enum E {
+    ///     V = 9223372036854775807, // i64::MAX
+    /// }
+    /// ```
+    ///
+    /// This will produce:
+    ///
+    /// ```text
+    /// error: `repr(C)` enum discriminant does not fit into C `int` nor into C `unsigned int`
+    ///   --> $DIR/repr-c-big-discriminant1.rs:16:5
+    ///    |
+    /// LL |     A = 9223372036854775807, // i64::MAX
+    ///    |     ^
+    ///    |
+    ///    = note: `repr(C)` enums with big discriminants are non-portable, and their size in Rust might not match their size in C
+    ///    = help: use `repr($int_ty)` instead to explicitly set the size of this enum
+    /// ```
+    ///
+    /// ### Explanation
+    ///
+    /// In C, enums with discriminants that do not all fit into an `int` or all fit into an
+    /// `unsigned int` are a portability hazard: such enums are only permitted since C23, and not
+    /// supported e.g. by MSVC.
+    ///
+    /// Furthermore, Rust interprets the discriminant values of `repr(C)` enums as expressions of
+    /// type `isize`. This makes it impossible to implement the C23 behavior of enums where the enum
+    /// discriminants have no predefined type and instead the enum uses a type large enough to hold
+    /// all discriminants.
+    ///
+    /// Therefore, `repr(C)` enums in Rust require that either all discriminants to fit into a C
+    /// `int` or they all fit into an `unsigned int`.
+    pub REPR_C_ENUMS_LARGER_THAN_INT,
+    Warn,
+    "repr(C) enums with discriminant values that do not fit into a C int",
+    @future_incompatible = FutureIncompatibleInfo {
+        reason: FutureIncompatibilityReason::FutureReleaseError,
+        reference: "issue #124403 <https://github.com/rust-lang/rust/issues/124403>",
+        report_in_deps: false,
+    };
+}

compiler/rustc_middle/src/ty/layout.rs

@@ -3,7 +3,7 @@ use std::{cmp, fmt};
 
 use rustc_abi::{
     AddressSpace, Align, ExternAbi, FieldIdx, FieldsShape, HasDataLayout, LayoutData, PointeeInfo,
-    PointerKind, Primitive, ReprOptions, Scalar, Size, TagEncoding, TargetDataLayout,
+    PointerKind, Primitive, ReprFlags, ReprOptions, Scalar, Size, TagEncoding, TargetDataLayout,
     TyAbiInterface, VariantIdx, Variants,
 };
 use rustc_error_messages::DiagMessage;
@@ -72,7 +72,10 @@ impl abi::Integer {
     /// signed discriminant range and `#[repr]` attribute.
     /// N.B.: `u128` values above `i128::MAX` will be treated as signed, but
     /// that shouldn't affect anything, other than maybe debuginfo.
-    fn repr_discr<'tcx>(
+    ///
+    /// This is the basis for computing the type of the *tag* of an enum (which can be smaller than
+    /// the type of the *discriminant*, which is determined by [`ReprOptions::discr_type`]).
+    fn discr_range_of_repr<'tcx>(
         tcx: TyCtxt<'tcx>,
         ty: Ty<'tcx>,
         repr: &ReprOptions,
@@ -108,7 +111,8 @@ impl abi::Integer {
             abi::Integer::I8
         };
 
-        // Pick the smallest fit.
+        // Pick the smallest fit. Prefer unsigned; that matches clang in cases where this makes a
+        // difference (https://godbolt.org/z/h4xEasW1d) so it is crucial for repr(C).
         if unsigned_fit <= signed_fit {
             (cmp::max(unsigned_fit, at_least), false)
         } else {
@@ -1173,6 +1177,11 @@ where
     fn is_transparent(this: TyAndLayout<'tcx>) -> bool {
         matches!(this.ty.kind(), ty::Adt(def, _) if def.repr().transparent())
     }
+
+    /// See [`TyAndLayout::pass_indirectly_in_non_rustic_abis`] for details.
+    fn is_pass_indirectly_in_non_rustic_abis_flag_set(this: TyAndLayout<'tcx>) -> bool {
+        matches!(this.ty.kind(), ty::Adt(def, _) if def.repr().flags.contains(ReprFlags::PASS_INDIRECTLY_IN_NON_RUSTIC_ABIS))
+    }
 }
 
 /// Calculates whether a function's ABI can unwind or not.