Auto merge of rust-lang#116520 - Enselic:large-copy-into-fn, r=oli-obk

large_assignments: Lint on specific large args passed to functions

Requires lowering function call arg spans down to MIR, which is done in the second commit.

Part of rust-lang#83518

Also see
* https://rust-lang.zulipchat.com/#narrow/stream/182449-t-compiler.2Fhelp/topic/arg.20Spans.20for.20TerminatorKind.3A.3ACall.3F
* https://rust-lang.zulipchat.com/#narrow/stream/122651-general/topic/move_size_limit.20lint

r? `@oli-obk` (E-mentor)

compiler/rustc_borrowck/src/diagnostics/conflict_errors.rs

@@ -1257,7 +1257,10 @@ impl<'cx, 'tcx> MirBorrowckCtxt<'cx, 'tcx> {
                     return None;
                 };
                 debug!("checking call args for uses of inner_param: {:?}", args);
-                args.contains(&Operand::Move(inner_param)).then_some((loc, term))
+                args.iter()
+                    .map(|a| &a.node)
+                    .any(|a| a == &Operand::Move(inner_param))
+                    .then_some((loc, term))
             })
         else {
             debug!("no uses of inner_param found as a by-move call arg");
@@ -3242,7 +3245,8 @@ impl<'cx, 'tcx> MirBorrowckCtxt<'cx, 'tcx> {
                         assigned_to, args
                     );
                     for operand in args {
-                        let (Operand::Copy(assigned_from) | Operand::Move(assigned_from)) = operand
+                        let (Operand::Copy(assigned_from) | Operand::Move(assigned_from)) =
+                            &operand.node
                         else {
                             continue;
                         };

compiler/rustc_borrowck/src/diagnostics/explain_borrow.rs

@@ -691,7 +691,7 @@ impl<'cx, 'tcx> MirBorrowckCtxt<'cx, 'tcx> {
                         );
                         // Check if one of the arguments to this function is the target place.
                         let found_target = args.iter().any(|arg| {
-                            if let Operand::Move(place) = arg {
+                            if let Operand::Move(place) = arg.node {
                                 if let Some(potential) = place.as_local() {
                                     potential == target
                                 } else {

compiler/rustc_borrowck/src/diagnostics/mod.rs

@@ -23,6 +23,7 @@ use rustc_middle::ty::{self, Instance, Ty, TyCtxt};
 use rustc_middle::util::{call_kind, CallDesugaringKind};
 use rustc_mir_dataflow::move_paths::{InitLocation, LookupResult};
 use rustc_span::def_id::LocalDefId;
+use rustc_span::source_map::Spanned;
 use rustc_span::{symbol::sym, Span, Symbol, DUMMY_SP};
 use rustc_target::abi::{FieldIdx, VariantIdx};
 use rustc_trait_selection::infer::InferCtxtExt;
@@ -111,9 +112,9 @@ impl<'cx, 'tcx> MirBorrowckCtxt<'cx, 'tcx> {
                 debug!("add_moved_or_invoked_closure_note: id={:?}", id);
                 if Some(self.infcx.tcx.parent(id)) == self.infcx.tcx.lang_items().fn_once_trait() {
                     let closure = match args.first() {
-                        Some(Operand::Copy(place) | Operand::Move(place))
-                            if target == place.local_or_deref_local() =>
-                        {
+                        Some(Spanned {
+                            node: Operand::Copy(place) | Operand::Move(place), ..
+                        }) if target == place.local_or_deref_local() => {
                             place.local_or_deref_local().unwrap()
                         }
                         _ => return false,

compiler/rustc_borrowck/src/lib.rs

@@ -703,7 +703,7 @@ impl<'cx, 'tcx, R> rustc_mir_dataflow::ResultsVisitor<'cx, 'tcx, R> for MirBorro
             } => {
                 self.consume_operand(loc, (func, span), flow_state);
                 for arg in args {
-                    self.consume_operand(loc, (arg, span), flow_state);
+                    self.consume_operand(loc, (&arg.node, arg.span), flow_state);
                 }
                 self.mutate_place(loc, (*destination, span), Deep, flow_state);
             }

compiler/rustc_borrowck/src/polonius/loan_invalidations.rs

@@ -120,7 +120,7 @@ impl<'cx, 'tcx> Visitor<'tcx> for LoanInvalidationsGenerator<'cx, 'tcx> {
             } => {
                 self.consume_operand(location, func);
                 for arg in args {
-                    self.consume_operand(location, arg);
+                    self.consume_operand(location, &arg.node);
                 }
                 self.mutate_place(location, *destination, Deep);
             }

compiler/rustc_borrowck/src/type_check/mod.rs

@@ -36,6 +36,7 @@ use rustc_middle::ty::{
 };
 use rustc_middle::ty::{GenericArgsRef, UserArgs};
 use rustc_span::def_id::CRATE_DEF_ID;
+use rustc_span::source_map::Spanned;
 use rustc_span::symbol::sym;
 use rustc_span::{Span, DUMMY_SP};
 use rustc_target::abi::{FieldIdx, FIRST_VARIANT};
@@ -1359,7 +1360,7 @@ impl<'a, 'tcx> TypeChecker<'a, 'tcx> {
             TerminatorKind::Call { func, args, destination, call_source, target, .. } => {
                 self.check_operand(func, term_location);
                 for arg in args {
-                    self.check_operand(arg, term_location);
+                    self.check_operand(&arg.node, term_location);
                 }
 
                 let func_ty = func.ty(body, tcx);
@@ -1580,7 +1581,7 @@ impl<'a, 'tcx> TypeChecker<'a, 'tcx> {
         term: &Terminator<'tcx>,
         func: &Operand<'tcx>,
         sig: &ty::FnSig<'tcx>,
-        args: &[Operand<'tcx>],
+        args: &[Spanned<Operand<'tcx>>],
         term_location: Location,
         call_source: CallSource,
     ) {
@@ -1593,7 +1594,7 @@ impl<'a, 'tcx> TypeChecker<'a, 'tcx> {
             if self.tcx().is_intrinsic(def_id) {
                 match self.tcx().item_name(def_id) {
                     sym::simd_shuffle => {
-                        if !matches!(args[2], Operand::Constant(_)) {
+                        if !matches!(args[2], Spanned { node: Operand::Constant(_), .. }) {
                             self.tcx()
                                 .dcx()
                                 .emit_err(SimdShuffleLastConst { span: term.source_info.span });
@@ -1606,7 +1607,7 @@ impl<'a, 'tcx> TypeChecker<'a, 'tcx> {
         debug!(?func_ty);
 
         for (n, (fn_arg, op_arg)) in iter::zip(sig.inputs(), args).enumerate() {
-            let op_arg_ty = op_arg.ty(body, self.tcx());
+            let op_arg_ty = op_arg.node.ty(body, self.tcx());
 
             let op_arg_ty = self.normalize(op_arg_ty, term_location);
             let category = if call_source.from_hir_call() {

compiler/rustc_codegen_cranelift/src/abi/mod.rs

@@ -11,6 +11,7 @@ use cranelift_module::ModuleError;
 use rustc_middle::middle::codegen_fn_attrs::CodegenFnAttrFlags;
 use rustc_middle::ty::layout::FnAbiOf;
 use rustc_session::Session;
+use rustc_span::source_map::Spanned;
 use rustc_target::abi::call::{Conv, FnAbi};
 use rustc_target::spec::abi::Abi;
 
@@ -360,7 +361,7 @@ pub(crate) fn codegen_terminator_call<'tcx>(
     fx: &mut FunctionCx<'_, '_, 'tcx>,
     source_info: mir::SourceInfo,
     func: &Operand<'tcx>,
-    args: &[Operand<'tcx>],
+    args: &[Spanned<Operand<'tcx>>],
     destination: Place<'tcx>,
     target: Option<BasicBlock>,
 ) {
@@ -415,7 +416,7 @@ pub(crate) fn codegen_terminator_call<'tcx>(
 
     let extra_args = &args[fn_sig.inputs().skip_binder().len()..];
     let extra_args = fx.tcx.mk_type_list_from_iter(
-        extra_args.iter().map(|op_arg| fx.monomorphize(op_arg.ty(fx.mir, fx.tcx))),
+        extra_args.iter().map(|op_arg| fx.monomorphize(op_arg.node.ty(fx.mir, fx.tcx))),
     );
     let fn_abi = if let Some(instance) = instance {
         RevealAllLayoutCx(fx.tcx).fn_abi_of_instance(instance, extra_args)
@@ -440,10 +441,10 @@ pub(crate) fn codegen_terminator_call<'tcx>(
     // Unpack arguments tuple for closures
     let mut args = if fn_sig.abi() == Abi::RustCall {
         let (self_arg, pack_arg) = match args {
-            [pack_arg] => (None, codegen_call_argument_operand(fx, pack_arg)),
+            [pack_arg] => (None, codegen_call_argument_operand(fx, &pack_arg.node)),
             [self_arg, pack_arg] => (
-                Some(codegen_call_argument_operand(fx, self_arg)),
-                codegen_call_argument_operand(fx, pack_arg),
+                Some(codegen_call_argument_operand(fx, &self_arg.node)),
+                codegen_call_argument_operand(fx, &pack_arg.node),
             ),
             _ => panic!("rust-call abi requires one or two arguments"),
         };
@@ -463,7 +464,7 @@ pub(crate) fn codegen_terminator_call<'tcx>(
         }
         args
     } else {
-        args.iter().map(|arg| codegen_call_argument_operand(fx, arg)).collect::<Vec<_>>()
+        args.iter().map(|arg| codegen_call_argument_operand(fx, &arg.node)).collect::<Vec<_>>()
     };
 
     // Pass the caller location for `#[track_caller]`.

compiler/rustc_codegen_cranelift/src/intrinsics/llvm.rs

@@ -7,7 +7,7 @@ pub(crate) fn codegen_llvm_intrinsic_call<'tcx>(
     fx: &mut FunctionCx<'_, '_, 'tcx>,
     intrinsic: &str,
     generic_args: GenericArgsRef<'tcx>,
-    args: &[mir::Operand<'tcx>],
+    args: &[Spanned<mir::Operand<'tcx>>],
     ret: CPlace<'tcx>,
     target: Option<BasicBlock>,
     span: Span,

compiler/rustc_codegen_cranelift/src/intrinsics/llvm_aarch64.rs

@@ -7,7 +7,7 @@ pub(crate) fn codegen_aarch64_llvm_intrinsic_call<'tcx>(
     fx: &mut FunctionCx<'_, '_, 'tcx>,
     intrinsic: &str,
     _args: GenericArgsRef<'tcx>,
-    args: &[mir::Operand<'tcx>],
+    args: &[Spanned<mir::Operand<'tcx>>],
     ret: CPlace<'tcx>,
     target: Option<BasicBlock>,
 ) {

compiler/rustc_codegen_cranelift/src/intrinsics/llvm_x86.rs

@@ -11,7 +11,7 @@ pub(crate) fn codegen_x86_llvm_intrinsic_call<'tcx>(
     fx: &mut FunctionCx<'_, '_, 'tcx>,
     intrinsic: &str,
     _args: GenericArgsRef<'tcx>,
-    args: &[mir::Operand<'tcx>],
+    args: &[Spanned<mir::Operand<'tcx>>],
     ret: CPlace<'tcx>,
     target: Option<BasicBlock>,
     span: Span,
@@ -175,9 +175,9 @@ pub(crate) fn codegen_x86_llvm_intrinsic_call<'tcx>(
                 [x, y, kind] => (x, y, kind),
                 _ => bug!("wrong number of args for intrinsic {intrinsic}"),
             };
-            let x = codegen_operand(fx, x);
-            let y = codegen_operand(fx, y);
-            let kind = match kind {
+            let x = codegen_operand(fx, &x.node);
+            let y = codegen_operand(fx, &y.node);
+            let kind = match &kind.node {
                 Operand::Constant(const_) => crate::constant::eval_mir_constant(fx, const_).0,
                 Operand::Copy(_) | Operand::Move(_) => unreachable!("{kind:?}"),
             };
@@ -287,8 +287,8 @@ pub(crate) fn codegen_x86_llvm_intrinsic_call<'tcx>(
                 [a, b] => (a, b),
                 _ => bug!("wrong number of args for intrinsic {intrinsic}"),
             };
-            let a = codegen_operand(fx, a);
-            let b = codegen_operand(fx, b);
+            let a = codegen_operand(fx, &a.node);
+            let b = codegen_operand(fx, &b.node);
 
             // Based on the pseudocode at https://github.com/rust-lang/stdarch/blob/1cfbca8b38fd9b4282b2f054f61c6ca69fc7ce29/crates/core_arch/src/x86/avx2.rs#L2319-L2332
             let zero = fx.bcx.ins().iconst(types::I8, 0);
@@ -325,9 +325,9 @@ pub(crate) fn codegen_x86_llvm_intrinsic_call<'tcx>(
                 [a, b, imm8] => (a, b, imm8),
                 _ => bug!("wrong number of args for intrinsic {intrinsic}"),
             };
-            let a = codegen_operand(fx, a);
-            let b = codegen_operand(fx, b);
-            let imm8 = codegen_operand(fx, imm8).load_scalar(fx);
+            let a = codegen_operand(fx, &a.node);
+            let b = codegen_operand(fx, &b.node);
+            let imm8 = codegen_operand(fx, &imm8.node).load_scalar(fx);
 
             let a_low = a.value_typed_lane(fx, fx.tcx.types.u128, 0).load_scalar(fx);
             let a_high = a.value_typed_lane(fx, fx.tcx.types.u128, 1).load_scalar(fx);
@@ -956,14 +956,14 @@ pub(crate) fn codegen_x86_llvm_intrinsic_call<'tcx>(
             let b = b.load_scalar(fx);
             let lb = lb.load_scalar(fx);
 
-            let imm8 = if let Some(imm8) = crate::constant::mir_operand_get_const_val(fx, &args[4])
-            {
-                imm8
-            } else {
-                fx.tcx
-                    .dcx()
-                    .span_fatal(span, "Index argument for `_mm_cmpestri` is not a constant");
-            };
+            let imm8 =
+                if let Some(imm8) = crate::constant::mir_operand_get_const_val(fx, &args[4].node) {
+                    imm8
+                } else {
+                    fx.tcx
+                        .dcx()
+                        .span_fatal(span, "Index argument for `_mm_cmpestri` is not a constant");
+                };
 
             let imm8 = imm8.try_to_u8().unwrap_or_else(|_| panic!("kind not scalar: {:?}", imm8));
 
@@ -1009,14 +1009,14 @@ pub(crate) fn codegen_x86_llvm_intrinsic_call<'tcx>(
             let b = b.load_scalar(fx);
             let lb = lb.load_scalar(fx);
 
-            let imm8 = if let Some(imm8) = crate::constant::mir_operand_get_const_val(fx, &args[4])
-            {
-                imm8
-            } else {
-                fx.tcx
-                    .dcx()
-                    .span_fatal(span, "Index argument for `_mm_cmpestrm` is not a constant");
-            };
+            let imm8 =
+                if let Some(imm8) = crate::constant::mir_operand_get_const_val(fx, &args[4].node) {
+                    imm8
+                } else {
+                    fx.tcx
+                        .dcx()
+                        .span_fatal(span, "Index argument for `_mm_cmpestrm` is not a constant");
+                };
 
             let imm8 = imm8.try_to_u8().unwrap_or_else(|_| panic!("kind not scalar: {:?}", imm8));
 
@@ -1056,15 +1056,15 @@ pub(crate) fn codegen_x86_llvm_intrinsic_call<'tcx>(
             let a = a.load_scalar(fx);
             let b = b.load_scalar(fx);
 
-            let imm8 = if let Some(imm8) = crate::constant::mir_operand_get_const_val(fx, &args[2])
-            {
-                imm8
-            } else {
-                fx.tcx.dcx().span_fatal(
-                    span,
-                    "Index argument for `_mm_clmulepi64_si128` is not a constant",
-                );
-            };
+            let imm8 =
+                if let Some(imm8) = crate::constant::mir_operand_get_const_val(fx, &args[2].node) {
+                    imm8
+                } else {
+                    fx.tcx.dcx().span_fatal(
+                        span,
+                        "Index argument for `_mm_clmulepi64_si128` is not a constant",
+                    );
+                };
 
             let imm8 = imm8.try_to_u8().unwrap_or_else(|_| panic!("kind not scalar: {:?}", imm8));
 
@@ -1093,15 +1093,15 @@ pub(crate) fn codegen_x86_llvm_intrinsic_call<'tcx>(
 
             let a = a.load_scalar(fx);
 
-            let imm8 = if let Some(imm8) = crate::constant::mir_operand_get_const_val(fx, &args[1])
-            {
-                imm8
-            } else {
-                fx.tcx.dcx().span_fatal(
-                    span,
-                    "Index argument for `_mm_aeskeygenassist_si128` is not a constant",
-                );
-            };
+            let imm8 =
+                if let Some(imm8) = crate::constant::mir_operand_get_const_val(fx, &args[1].node) {
+                    imm8
+                } else {
+                    fx.tcx.dcx().span_fatal(
+                        span,
+                        "Index argument for `_mm_aeskeygenassist_si128` is not a constant",
+                    );
+                };
 
             let imm8 = imm8.try_to_u8().unwrap_or_else(|_| panic!("kind not scalar: {:?}", imm8));
 

compiler/rustc_codegen_cranelift/src/intrinsics/mod.rs

@@ -5,7 +5,7 @@ macro_rules! intrinsic_args {
     ($fx:expr, $args:expr => ($($arg:tt),*); $intrinsic:expr) => {
         #[allow(unused_parens)]
         let ($($arg),*) = if let [$($arg),*] = $args {
-            ($(codegen_operand($fx, $arg)),*)
+            ($(codegen_operand($fx, &($arg).node)),*)
         } else {
             $crate::intrinsics::bug_on_incorrect_arg_count($intrinsic);
         };
@@ -22,6 +22,7 @@ use rustc_middle::ty;
 use rustc_middle::ty::layout::{HasParamEnv, ValidityRequirement};
 use rustc_middle::ty::print::{with_no_trimmed_paths, with_no_visible_paths};
 use rustc_middle::ty::GenericArgsRef;
+use rustc_span::source_map::Spanned;
 use rustc_span::symbol::{kw, sym, Symbol};
 
 pub(crate) use self::llvm::codegen_llvm_intrinsic_call;
@@ -263,7 +264,7 @@ fn bool_to_zero_or_max_uint<'tcx>(
 pub(crate) fn codegen_intrinsic_call<'tcx>(
     fx: &mut FunctionCx<'_, '_, 'tcx>,
     instance: Instance<'tcx>,
-    args: &[mir::Operand<'tcx>],
+    args: &[Spanned<mir::Operand<'tcx>>],
     destination: CPlace<'tcx>,
     target: Option<BasicBlock>,
     source_info: mir::SourceInfo,
@@ -301,7 +302,7 @@ pub(crate) fn codegen_intrinsic_call<'tcx>(
 fn codegen_float_intrinsic_call<'tcx>(
     fx: &mut FunctionCx<'_, '_, 'tcx>,
     intrinsic: Symbol,
-    args: &[mir::Operand<'tcx>],
+    args: &[Spanned<mir::Operand<'tcx>>],
     ret: CPlace<'tcx>,
 ) -> bool {
     let (name, arg_count, ty, clif_ty) = match intrinsic {
@@ -353,18 +354,21 @@ fn codegen_float_intrinsic_call<'tcx>(
     let (a, b, c);
     let args = match args {
         [x] => {
-            a = [codegen_operand(fx, x).load_scalar(fx)];
+            a = [codegen_operand(fx, &x.node).load_scalar(fx)];
             &a as &[_]
         }
         [x, y] => {
-            b = [codegen_operand(fx, x).load_scalar(fx), codegen_operand(fx, y).load_scalar(fx)];
+            b = [
+                codegen_operand(fx, &x.node).load_scalar(fx),
+                codegen_operand(fx, &y.node).load_scalar(fx),
+            ];
             &b
         }
         [x, y, z] => {
             c = [
-                codegen_operand(fx, x).load_scalar(fx),
-                codegen_operand(fx, y).load_scalar(fx),
-                codegen_operand(fx, z).load_scalar(fx),
+                codegen_operand(fx, &x.node).load_scalar(fx),
+                codegen_operand(fx, &y.node).load_scalar(fx),
+                codegen_operand(fx, &z.node).load_scalar(fx),
             ];
             &c
         }
@@ -422,7 +426,7 @@ fn codegen_regular_intrinsic_call<'tcx>(
     instance: Instance<'tcx>,
     intrinsic: Symbol,
     generic_args: GenericArgsRef<'tcx>,
-    args: &[mir::Operand<'tcx>],
+    args: &[Spanned<mir::Operand<'tcx>>],
     ret: CPlace<'tcx>,
     destination: Option<BasicBlock>,
     source_info: mir::SourceInfo,