Add a scheme for moving away from extern "rust-intrinsic" entirely

Author: oli-obk

compiler/rustc_borrowck/src/type_check/mod.rs

@@ -1648,7 +1648,7 @@ impl<'a, 'tcx> TypeChecker<'a, 'tcx> {
 
         let func_ty = func.ty(body, self.infcx.tcx);
         if let ty::FnDef(def_id, _) = *func_ty.kind() {
-            if let Some(sym::simd_shuffle) = self.tcx().intrinsic(def_id) {
+            if self.tcx().is_intrinsic(def_id, sym::simd_shuffle) {
                 if !matches!(args[2], Spanned { node: Operand::Constant(_), .. }) {
                     self.tcx().dcx().emit_err(SimdShuffleLastConst { span: term.source_info.span });
                 }

compiler/rustc_codegen_cranelift/src/abi/mod.rs

@@ -398,7 +398,17 @@ pub(crate) fn codegen_terminator_call<'tcx>(
                     Ok(()) => return,
                     // Unimplemented intrinsics must have a fallback body. The fallback body is obtained
                     // by converting the `InstanceDef::Intrinsic` to an `InstanceDef::Item`.
-                    Err(()) => Some(Instance::new(instance.def_id(), instance.args)),
+                    Err(()) => {
+                        let intrinsic = fx.tcx.intrinsic(instance.def_id()).unwrap();
+                        if intrinsic.must_be_overridden {
+                            span_bug!(
+                                source_info.span,
+                                "intrinsic {} must be overridden by codegen_cranelift, but isn't",
+                                intrinsic.name,
+                            );
+                        }
+                        Some(Instance::new(instance.def_id(), instance.args))
+                    }
                 }
             }
             InstanceDef::DropGlue(_, None) => {

compiler/rustc_codegen_ssa/src/back/symbol_export.rs

@@ -16,6 +16,7 @@ use rustc_middle::ty::{self, SymbolName, TyCtxt};
 use rustc_middle::ty::{GenericArgKind, GenericArgsRef};
 use rustc_middle::util::Providers;
 use rustc_session::config::{CrateType, OomStrategy};
+use rustc_span::sym;
 use rustc_target::spec::{SanitizerSet, TlsModel};
 
 pub fn threshold(tcx: TyCtxt<'_>) -> SymbolExportLevel {
@@ -81,6 +82,10 @@ fn reachable_non_generics_provider(tcx: TyCtxt<'_>, _: LocalCrate) -> DefIdMap<S
                 return library.kind.is_statically_included().then_some(def_id);
             }
 
+            if tcx.has_attr(def_id, sym::rustc_intrinsic_must_be_overridden) {
+                return None;
+            }
+
             // Only consider nodes that actually have exported symbols.
             match tcx.def_kind(def_id) {
                 DefKind::Fn | DefKind::Static(_) => {}

compiler/rustc_codegen_ssa/src/mir/block.rs

@@ -17,7 +17,7 @@ use rustc_middle::ty::layout::{HasTyCtxt, LayoutOf, ValidityRequirement};
 use rustc_middle::ty::print::{with_no_trimmed_paths, with_no_visible_paths};
 use rustc_middle::ty::{self, Instance, Ty};
 use rustc_session::config::OptLevel;
-use rustc_span::{source_map::Spanned, sym, Span, Symbol};
+use rustc_span::{source_map::Spanned, sym, Span};
 use rustc_target::abi::call::{ArgAbi, FnAbi, PassMode, Reg};
 use rustc_target::abi::{self, HasDataLayout, WrappingRange};
 use rustc_target::spec::abi::Abi;
@@ -672,7 +672,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
         &mut self,
         helper: &TerminatorCodegenHelper<'tcx>,
         bx: &mut Bx,
-        intrinsic: Option<Symbol>,
+        intrinsic: Option<ty::IntrinsicDef>,
         instance: Option<Instance<'tcx>>,
         source_info: mir::SourceInfo,
         target: Option<mir::BasicBlock>,
@@ -682,7 +682,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
         // Emit a panic or a no-op for `assert_*` intrinsics.
         // These are intrinsics that compile to panics so that we can get a message
         // which mentions the offending type, even from a const context.
-        let panic_intrinsic = intrinsic.and_then(|s| ValidityRequirement::from_intrinsic(s));
+        let panic_intrinsic = intrinsic.and_then(|i| ValidityRequirement::from_intrinsic(i.name));
         if let Some(requirement) = panic_intrinsic {
             let ty = instance.unwrap().args.type_at(0);
 
@@ -818,7 +818,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
         // The arguments we'll be passing. Plus one to account for outptr, if used.
         let arg_count = fn_abi.args.len() + fn_abi.ret.is_indirect() as usize;
 
-        if intrinsic == Some(sym::caller_location) {
+        if matches!(intrinsic, Some(ty::IntrinsicDef { name: sym::caller_location, .. })) {
             return if let Some(target) = target {
                 let location =
                     self.get_caller_location(bx, mir::SourceInfo { span: fn_span, ..source_info });
@@ -838,7 +838,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
         }
 
         let instance = match intrinsic {
-            None | Some(sym::drop_in_place) => instance,
+            None | Some(ty::IntrinsicDef { name: sym::drop_in_place, .. }) => instance,
             Some(intrinsic) => {
                 let mut llargs = Vec::with_capacity(1);
                 let ret_dest = self.make_return_dest(
@@ -866,7 +866,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                         // third argument must be constant. This is
                         // checked by const-qualification, which also
                         // promotes any complex rvalues to constants.
-                        if i == 2 && intrinsic == sym::simd_shuffle {
+                        if i == 2 && intrinsic.name == sym::simd_shuffle {
                             if let mir::Operand::Constant(constant) = &arg.node {
                                 let (llval, ty) = self.simd_shuffle_indices(bx, constant);
                                 return OperandRef {
@@ -897,7 +897,16 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                         };
                     }
                     // Call the fallback body instead of generating the intrinsic code
-                    Err(()) => Some(Instance::new(instance.def_id(), instance.args)),
+                    Err(()) => {
+                        if intrinsic.must_be_overridden {
+                            span_bug!(
+                                span,
+                                "intrinsic {} must be overridden by codegen backend, but isn't",
+                                intrinsic.name,
+                            );
+                        }
+                        Some(Instance::new(instance.def_id(), instance.args))
+                    }
                 }
             }
         };

compiler/rustc_feature/src/builtin_attrs.rs

@@ -795,6 +795,10 @@ pub const BUILTIN_ATTRIBUTES: &[BuiltinAttribute] = &[
         rustc_intrinsic, Normal, template!(Word), ErrorFollowing,
         "the `#[rustc_intrinsic]` attribute is used to declare intrinsics with function bodies",
     ),
+    rustc_attr!(
+        rustc_intrinsic_must_be_overridden, Normal, template!(Word), ErrorFollowing,
+        "the `#[rustc_intrinsic_must_be_overridden]` attribute is used to declare intrinsics without real bodies",
+    ),
 
     // ==========================================================================
     // Internal attributes, Testing:

compiler/rustc_hir_analysis/src/check/check.rs

@@ -473,12 +473,12 @@ pub(crate) fn check_item_type(tcx: TyCtxt<'_>, def_id: LocalDefId) {
             check_enum(tcx, def_id);
         }
         DefKind::Fn => {
-            if let Some(name) = tcx.intrinsic(def_id) {
+            if let Some(i) = tcx.intrinsic(def_id) {
                 intrinsic::check_intrinsic_type(
                     tcx,
                     def_id,
                     tcx.def_ident_span(def_id).unwrap(),
-                    name,
+                    i.name,
                     Abi::Rust,
                 )
             }

compiler/rustc_hir_typeck/src/callee.rs

@@ -480,7 +480,7 @@ impl<'a, 'tcx> FnCtxt<'a, 'tcx> {
 
         if let Some(def_id) = def_id
             && self.tcx.def_kind(def_id) == hir::def::DefKind::Fn
-            && matches!(self.tcx.intrinsic(def_id), Some(sym::const_eval_select))
+            && self.tcx.is_intrinsic(def_id, sym::const_eval_select)
         {
             let fn_sig = self.resolve_vars_if_possible(fn_sig);
             for idx in 0..=1 {

compiler/rustc_lint/src/builtin.rs

@@ -1227,7 +1227,7 @@ impl<'tcx> LateLintPass<'tcx> for MutableTransmutes {
         }
 
         fn def_id_is_transmute(cx: &LateContext<'_>, def_id: DefId) -> bool {
-            matches!(cx.tcx.intrinsic(def_id), Some(sym::transmute))
+            cx.tcx.is_intrinsic(def_id, sym::transmute)
         }
     }
 }

compiler/rustc_metadata/src/rmeta/decoder.rs

@@ -1744,7 +1744,7 @@ impl<'a, 'tcx> CrateMetadataRef<'a> {
         self.root.tables.attr_flags.get(self, index)
     }
 
-    fn get_intrinsic(self, index: DefIndex) -> Option<Symbol> {
+    fn get_intrinsic(self, index: DefIndex) -> Option<ty::IntrinsicDef> {
         self.root.tables.intrinsic.get(self, index).map(|d| d.decode(self))
     }
 

compiler/rustc_metadata/src/rmeta/mod.rs

@@ -375,7 +375,7 @@ macro_rules! define_tables {
 
 define_tables! {
 - defaulted:
-    intrinsic: Table<DefIndex, Option<LazyValue<Symbol>>>,
+    intrinsic: Table<DefIndex, Option<LazyValue<ty::IntrinsicDef>>>,
     is_macro_rules: Table<DefIndex, bool>,
     is_type_alias_impl_trait: Table<DefIndex, bool>,
     type_alias_is_lazy: Table<DefIndex, bool>,

compiler/rustc_middle/src/query/erase.rs

@@ -240,7 +240,7 @@ trivial! {
     Option<rustc_span::Span>,
     Option<rustc_target::spec::PanicStrategy>,
     Option<usize>,
-    Option<rustc_span::Symbol>,
+    Option<rustc_middle::ty::IntrinsicDef>,
     Result<(), rustc_errors::ErrorGuaranteed>,
     Result<(), rustc_middle::traits::query::NoSolution>,
     Result<rustc_middle::traits::EvaluationResult, rustc_middle::traits::OverflowError>,

compiler/rustc_middle/src/query/mod.rs

@@ -1748,7 +1748,7 @@ rustc_queries! {
         separate_provide_extern
     }
     /// Whether the function is an intrinsic
-    query intrinsic(def_id: DefId) -> Option<Symbol> {
+    query intrinsic(def_id: DefId) -> Option<rustc_middle::ty::IntrinsicDef> {
         desc { |tcx| "fetch intrinsic name if `{}` is an intrinsic", tcx.def_path_str(def_id) }
         separate_provide_extern
     }

compiler/rustc_middle/src/ty/intrinsic.rs

@@ -0,0 +1,17 @@
+use rustc_span::{def_id::DefId, Symbol};
+
+use super::TyCtxt;
+
+#[derive(Copy, Clone, Debug, Decodable, Encodable, HashStable)]
+pub struct IntrinsicDef {
+    pub name: Symbol,
+    /// Whether the intrinsic has no meaningful body and all backends need to shim all calls to it.
+    pub must_be_overridden: bool,
+}
+
+impl TyCtxt<'_> {
+    pub fn is_intrinsic(self, def_id: DefId, name: Symbol) -> bool {
+        let Some(i) = self.intrinsic(def_id) else { return false };
+        i.name == name
+    }
+}

compiler/rustc_middle/src/ty/mod.rs

@@ -30,6 +30,7 @@ pub use adt::*;
 pub use assoc::*;
 pub use generic_args::*;
 pub use generics::*;
+pub use intrinsic::IntrinsicDef;
 use rustc_ast as ast;
 use rustc_ast::node_id::NodeMap;
 use rustc_attr as attr;
@@ -147,6 +148,7 @@ mod generic_args;
 mod generics;
 mod impls_ty;
 mod instance;
+mod intrinsic;
 mod list;
 mod opaque_types;
 mod parameterized;

compiler/rustc_middle/src/ty/parameterized.rs

@@ -75,6 +75,7 @@ trivially_parameterized_over_tcx! {
     ty::Visibility<DefIndex>,
     ty::adjustment::CoerceUnsizedInfo,
     ty::fast_reject::SimplifiedType,
+    ty::IntrinsicDef,
     rustc_ast::Attribute,
     rustc_ast::DelimArgs,
     rustc_ast::expand::StrippedCfgItem<rustc_hir::def_id::DefIndex>,

compiler/rustc_middle/src/ty/util.rs

@@ -18,7 +18,7 @@ use rustc_hir::def_id::{CrateNum, DefId, LocalDefId};
 use rustc_index::bit_set::GrowableBitSet;
 use rustc_macros::HashStable;
 use rustc_session::Limit;
-use rustc_span::{sym, Symbol};
+use rustc_span::sym;
 use rustc_target::abi::{Integer, IntegerType, Primitive, Size};
 use rustc_target::spec::abi::Abi;
 use smallvec::SmallVec;
@@ -1539,11 +1539,14 @@ pub fn is_doc_notable_trait(tcx: TyCtxt<'_>, def_id: DefId) -> bool {
 }
 
 /// Determines whether an item is an intrinsic by Abi. or by whether it has a `rustc_intrinsic` attribute
-pub fn intrinsic(tcx: TyCtxt<'_>, def_id: LocalDefId) -> Option<Symbol> {
+pub fn intrinsic(tcx: TyCtxt<'_>, def_id: LocalDefId) -> Option<ty::IntrinsicDef> {
     if matches!(tcx.fn_sig(def_id).skip_binder().abi(), Abi::RustIntrinsic | Abi::PlatformIntrinsic)
         || tcx.has_attr(def_id, sym::rustc_intrinsic)
     {
-        Some(tcx.item_name(def_id.into()))
+        Some(ty::IntrinsicDef {
+            name: tcx.item_name(def_id.into()),
+            must_be_overridden: tcx.has_attr(def_id, sym::rustc_intrinsic_must_be_overridden),
+        })
     } else {
         None
     }

compiler/rustc_mir_dataflow/src/rustc_peek.rs

@@ -202,7 +202,7 @@ impl PeekCall {
             &terminator.kind
         {
             if let ty::FnDef(def_id, fn_args) = *func.const_.ty().kind() {
-                if tcx.intrinsic(def_id)? != sym::rustc_peek {
+                if tcx.intrinsic(def_id)?.name != sym::rustc_peek {
                     return None;
                 }
 

compiler/rustc_mir_transform/src/cross_crate_inline.rs

@@ -9,6 +9,7 @@ use rustc_middle::query::Providers;
 use rustc_middle::ty::TyCtxt;
 use rustc_session::config::InliningThreshold;
 use rustc_session::config::OptLevel;
+use rustc_span::sym;
 
 pub fn provide(providers: &mut Providers) {
     providers.cross_crate_inlinable = cross_crate_inlinable;
@@ -22,6 +23,10 @@ fn cross_crate_inlinable(tcx: TyCtxt<'_>, def_id: LocalDefId) -> bool {
         return false;
     }
 
+    if tcx.has_attr(def_id, sym::rustc_intrinsic_must_be_overridden) {
+        return false;
+    }
+
     // This just reproduces the logic from Instance::requires_inline.
     match tcx.def_kind(def_id) {
         DefKind::Ctor(..) | DefKind::Closure => return true,

compiler/rustc_mir_transform/src/instsimplify.rs

@@ -305,8 +305,8 @@ fn resolve_rust_intrinsic<'tcx>(
     func_ty: Ty<'tcx>,
 ) -> Option<(Symbol, GenericArgsRef<'tcx>)> {
     if let ty::FnDef(def_id, args) = *func_ty.kind() {
-        let name = tcx.intrinsic(def_id)?;
-        return Some((name, args));
+        let intrinsic = tcx.intrinsic(def_id)?;
+        return Some((intrinsic.name, args));
     }
     None
 }

compiler/rustc_mir_transform/src/lib.rs

@@ -165,7 +165,7 @@ fn remap_mir_for_const_eval_select<'tcx>(
                 fn_span,
                 ..
             } if let ty::FnDef(def_id, _) = *const_.ty().kind()
-                && matches!(tcx.intrinsic(def_id), Some(sym::const_eval_select)) =>
+                && tcx.is_intrinsic(def_id, sym::const_eval_select) =>
             {
                 let [tupled_args, called_in_const, called_at_rt]: [_; 3] =
                     std::mem::take(args).try_into().unwrap();
@@ -637,6 +637,12 @@ fn optimized_mir(tcx: TyCtxt<'_>, did: LocalDefId) -> &Body<'_> {
 }
 
 fn inner_optimized_mir(tcx: TyCtxt<'_>, did: LocalDefId) -> Body<'_> {
+    if let Some(attr) = tcx.get_attr(did, sym::rustc_intrinsic_must_be_overridden) {
+        span_bug!(
+            attr.span,
+            "this intrinsic must be overridden by the codegen backend, it has no meaningful body",
+        )
+    }
     if tcx.is_constructor(did.to_def_id()) {
         // There's no reason to run all of the MIR passes on constructors when
         // we can just output the MIR we want directly. This also saves const

compiler/rustc_mir_transform/src/lower_intrinsics.rs

@@ -14,9 +14,9 @@ impl<'tcx> MirPass<'tcx> for LowerIntrinsics {
             if let TerminatorKind::Call { func, args, destination, target, .. } =
                 &mut terminator.kind
                 && let ty::FnDef(def_id, generic_args) = *func.ty(local_decls, tcx).kind()
-                && let Some(intrinsic_name) = tcx.intrinsic(def_id)
+                && let Some(intrinsic) = tcx.intrinsic(def_id)
             {
-                match intrinsic_name {
+                match intrinsic.name {
                     sym::unreachable => {
                         terminator.kind = TerminatorKind::Unreachable;
                     }
@@ -94,7 +94,7 @@ impl<'tcx> MirPass<'tcx> for LowerIntrinsics {
                             lhs = args.next().unwrap();
                             rhs = args.next().unwrap();
                         }
-                        let bin_op = match intrinsic_name {
+                        let bin_op = match intrinsic.name {
                             sym::wrapping_add => BinOp::Add,
                             sym::wrapping_sub => BinOp::Sub,
                             sym::wrapping_mul => BinOp::Mul,
@@ -125,7 +125,7 @@ impl<'tcx> MirPass<'tcx> for LowerIntrinsics {
                                 lhs = args.next().unwrap();
                                 rhs = args.next().unwrap();
                             }
-                            let bin_op = match intrinsic_name {
+                            let bin_op = match intrinsic.name {
                                 sym::add_with_overflow => BinOp::Add,
                                 sym::sub_with_overflow => BinOp::Sub,
                                 sym::mul_with_overflow => BinOp::Mul,
@@ -144,7 +144,7 @@ impl<'tcx> MirPass<'tcx> for LowerIntrinsics {
                     sym::size_of | sym::min_align_of => {
                         if let Some(target) = *target {
                             let tp_ty = generic_args.type_at(0);
-                            let null_op = match intrinsic_name {
+                            let null_op = match intrinsic.name {
                                 sym::size_of => NullOp::SizeOf,
                                 sym::min_align_of => NullOp::AlignOf,
                                 _ => bug!("unexpected intrinsic"),

compiler/rustc_monomorphize/src/collector.rs

@@ -1009,6 +1009,11 @@ fn should_codegen_locally<'tcx>(tcx: TyCtxt<'tcx>, instance: &Instance<'tcx>) ->
         return false;
     }
 
+    if tcx.has_attr(def_id, sym::rustc_intrinsic_must_be_overridden) {
+        // These are implemented by backends directly and have no meaningful body.
+        return false;
+    }
+
     if def_id.is_local() {
         // Local items cannot be referred to locally without monomorphizing them locally.
         return true;

compiler/rustc_span/src/symbol.rs

@@ -1412,6 +1412,7 @@ symbols! {
         rustc_inherit_overflow_checks,
         rustc_insignificant_dtor,
         rustc_intrinsic,
+        rustc_intrinsic_must_be_overridden,
         rustc_layout,
         rustc_layout_scalar_valid_range_end,
         rustc_layout_scalar_valid_range_start,