Auto merge of #142771 - dianqk:mir-stmt-debuginfo, r=<try>

Introduce debuginfo to statements in MIR

try-job: aarch64-gnu
try-job: dist-loongarch64-linux
try-job: aarch64-apple
try-job: x86_64-msvc-*
try-job: aarch64-msvc-*

Author: rust-bors[bot]

compiler/rustc_codegen_gcc/src/debuginfo.rs

@@ -29,13 +29,24 @@ impl<'a, 'gcc, 'tcx> DebugInfoBuilderMethods for Builder<'a, 'gcc, 'tcx> {
         _variable_alloca: Self::Value,
         _direct_offset: Size,
         _indirect_offsets: &[Size],
-        _fragment: Option<Range<Size>>,
+        _fragment: &Option<Range<Size>>,
     ) {
         // FIXME(tempdragon): Not sure if this is correct, probably wrong but still keep it here.
         #[cfg(feature = "master")]
         _variable_alloca.set_location(_dbg_loc);
     }
 
+    fn dbg_var_value(
+        &mut self,
+        _dbg_var: Self::DIVariable,
+        _dbg_loc: Self::DILocation,
+        _value: Self::Value,
+        _direct_offset: Size,
+        _indirect_offsets: &[Size],
+        _fragment: &Option<Range<Size>>,
+    ) {
+    }
+
     fn insert_reference_to_gdb_debug_scripts_section_global(&mut self) {
         // TODO(antoyo): insert reference to gdb debug scripts section global.
     }

compiler/rustc_codegen_llvm/src/debuginfo/dwarf_const.rs

@@ -35,3 +35,6 @@ declare_constant!(DW_OP_plus_uconst: u64);
 /// Double-checked by a static assertion in `RustWrapper.cpp`.
 #[allow(non_upper_case_globals)]
 pub(crate) const DW_OP_LLVM_fragment: u64 = 0x1000;
+// It describes the actual value of a source variable which might not exist in registers or in memory.
+#[allow(non_upper_case_globals)]
+pub(crate) const DW_OP_stack_value: u64 = 0x9f;

compiler/rustc_codegen_llvm/src/debuginfo/mod.rs

@@ -156,7 +156,7 @@ impl<'ll> DebugInfoBuilderMethods for Builder<'_, 'll, '_> {
         variable_alloca: Self::Value,
         direct_offset: Size,
         indirect_offsets: &[Size],
-        fragment: Option<Range<Size>>,
+        fragment: &Option<Range<Size>>,
     ) {
         use dwarf_const::{DW_OP_LLVM_fragment, DW_OP_deref, DW_OP_plus_uconst};
 
@@ -187,7 +187,6 @@ impl<'ll> DebugInfoBuilderMethods for Builder<'_, 'll, '_> {
             llvm::LLVMDIBuilderCreateExpression(di_builder, addr_ops.as_ptr(), addr_ops.len())
         };
         unsafe {
-            // FIXME(eddyb) replace `llvm.dbg.declare` with `llvm.dbg.addr`.
             llvm::LLVMDIBuilderInsertDeclareRecordAtEnd(
                 di_builder,
                 variable_alloca,
@@ -199,6 +198,56 @@ impl<'ll> DebugInfoBuilderMethods for Builder<'_, 'll, '_> {
         };
     }
 
+    fn dbg_var_value(
+        &mut self,
+        dbg_var: &'ll DIVariable,
+        dbg_loc: &'ll DILocation,
+        value: Self::Value,
+        direct_offset: Size,
+        indirect_offsets: &[Size],
+        fragment: &Option<Range<Size>>,
+    ) {
+        use dwarf_const::{DW_OP_LLVM_fragment, DW_OP_deref, DW_OP_plus_uconst, DW_OP_stack_value};
+
+        // Convert the direct and indirect offsets and fragment byte range to address ops.
+        let mut addr_ops = SmallVec::<[u64; 8]>::new();
+
+        if direct_offset.bytes() > 0 {
+            addr_ops.push(DW_OP_plus_uconst);
+            addr_ops.push(direct_offset.bytes() as u64);
+            addr_ops.push(DW_OP_stack_value);
+        }
+        for &offset in indirect_offsets {
+            addr_ops.push(DW_OP_deref);
+            if offset.bytes() > 0 {
+                addr_ops.push(DW_OP_plus_uconst);
+                addr_ops.push(offset.bytes() as u64);
+            }
+        }
+        if let Some(fragment) = fragment {
+            // `DW_OP_LLVM_fragment` takes as arguments the fragment's
+            // offset and size, both of them in bits.
+            addr_ops.push(DW_OP_LLVM_fragment);
+            addr_ops.push(fragment.start.bits() as u64);
+            addr_ops.push((fragment.end - fragment.start).bits() as u64);
+        }
+
+        let di_builder = DIB(self.cx());
+        let addr_expr = unsafe {
+            llvm::LLVMDIBuilderCreateExpression(di_builder, addr_ops.as_ptr(), addr_ops.len())
+        };
+        unsafe {
+            llvm::LLVMDIBuilderInsertDbgValueRecordAtEnd(
+                di_builder,
+                value,
+                dbg_var,
+                addr_expr,
+                dbg_loc,
+                self.llbb(),
+            );
+        }
+    }
+
     fn set_dbg_loc(&mut self, dbg_loc: &'ll DILocation) {
         unsafe {
             llvm::LLVMSetCurrentDebugLocation2(self.llbuilder, dbg_loc);

compiler/rustc_codegen_llvm/src/llvm/ffi.rs

@@ -1991,6 +1991,15 @@ unsafe extern "C" {
         Block: &'ll BasicBlock,
     ) -> &'ll DbgRecord;
 
+    pub(crate) fn LLVMDIBuilderInsertDbgValueRecordAtEnd<'ll>(
+        Builder: &DIBuilder<'ll>,
+        Val: &'ll Value,
+        VarInfo: &'ll Metadata,
+        Expr: &'ll Metadata,
+        DebugLoc: &'ll Metadata,
+        Block: &'ll BasicBlock,
+    ) -> &'ll DbgRecord;
+
     pub(crate) fn LLVMDIBuilderCreateAutoVariable<'ll>(
         Builder: &DIBuilder<'ll>,
         Scope: &'ll Metadata,

compiler/rustc_codegen_ssa/src/mir/analyze.rs

@@ -260,6 +260,10 @@ impl<'a, 'b, 'tcx, Bx: BuilderMethods<'b, 'tcx>> Visitor<'tcx> for LocalAnalyzer
             PlaceContext::MutatingUse(MutatingUseContext::Yield) => bug!(),
         }
     }
+
+    fn visit_statement_debuginfo(&mut self, _: &mir::StmtDebugInfo<'tcx>, _: Location) {
+        // Debuginfo does not generate actual code.
+    }
 }
 
 #[derive(Copy, Clone, Debug, PartialEq, Eq)]

compiler/rustc_codegen_ssa/src/mir/block.rs

@@ -1320,6 +1320,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
             for statement in &data.statements {
                 self.codegen_statement(bx, statement);
             }
+            self.codegen_stmt_debuginfos(bx, &data.after_last_stmt_debuginfos);
 
             let merging_succ = self.codegen_terminator(bx, bb, data.terminator());
             if let MergingSucc::False = merging_succ {

compiler/rustc_codegen_ssa/src/mir/debuginfo.rs

@@ -253,6 +253,53 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
         spill_slot
     }
 
+    // Indicates that local is set to a new value. The `layout` and `projection` are used to
+    // calculate the offset.
+    pub(crate) fn debug_new_val_to_local(
+        &self,
+        bx: &mut Bx,
+        local: mir::Local,
+        base: PlaceRef<'tcx, Bx::Value>,
+        projection: &[mir::PlaceElem<'tcx>],
+    ) {
+        let full_debug_info = bx.sess().opts.debuginfo == DebugInfo::Full;
+        if !full_debug_info {
+            return;
+        }
+
+        let vars = match &self.per_local_var_debug_info {
+            Some(per_local) => &per_local[local],
+            None => return,
+        };
+
+        let DebugInfoOffset { direct_offset, indirect_offsets, result: _ } =
+            calculate_debuginfo_offset(bx, projection, base.layout);
+        for var in vars.iter() {
+            let Some(dbg_var) = var.dbg_var else {
+                continue;
+            };
+            let Some(dbg_loc) = self.dbg_loc(var.source_info) else {
+                continue;
+            };
+            bx.dbg_var_value(
+                dbg_var,
+                dbg_loc,
+                base.val.llval,
+                direct_offset,
+                &indirect_offsets,
+                &var.fragment,
+            );
+        }
+    }
+
+    pub(crate) fn debug_poison_to_local(&self, bx: &mut Bx, local: mir::Local) {
+        let ty = self.monomorphize(self.mir.local_decls[local].ty);
+        let layout = bx.cx().layout_of(ty);
+        let to_backend_ty = bx.cx().immediate_backend_type(layout);
+        let place_ref = PlaceRef::new_sized(bx.cx().const_poison(to_backend_ty), layout);
+        self.debug_new_val_to_local(bx, local, place_ref, &[]);
+    }
+
     /// Apply debuginfo and/or name, after creating the `alloca` for a local,
     /// or initializing the local with an operand (whichever applies).
     pub(crate) fn debug_introduce_local(&self, bx: &mut Bx, local: mir::Local) {
@@ -424,7 +471,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                 alloca.val.llval,
                 Size::ZERO,
                 &[Size::ZERO],
-                var.fragment,
+                &var.fragment,
             );
         } else {
             bx.dbg_var_addr(
@@ -433,7 +480,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                 base.val.llval,
                 direct_offset,
                 &indirect_offsets,
-                var.fragment,
+                &var.fragment,
             );
         }
     }
@@ -455,7 +502,7 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                 let base = FunctionCx::spill_operand_to_stack(operand, Some(name), bx);
                 bx.clear_dbg_loc();
 
-                bx.dbg_var_addr(dbg_var, dbg_loc, base.val.llval, Size::ZERO, &[], fragment);
+                bx.dbg_var_addr(dbg_var, dbg_loc, base.val.llval, Size::ZERO, &[], &fragment);
             }
         }
     }

compiler/rustc_codegen_ssa/src/mir/operand.rs

@@ -71,16 +71,23 @@ pub enum OperandValue<V> {
 }
 
 impl<V: CodegenObject> OperandValue<V> {
+    /// Return the data pointer and optional metadata as backend values
+    /// if this value can be treat as a pointer.
+    pub(crate) fn try_pointer_parts(self) -> Option<(V, Option<V>)> {
+        match self {
+            OperandValue::Immediate(llptr) => Some((llptr, None)),
+            OperandValue::Pair(llptr, llextra) => Some((llptr, Some(llextra))),
+            OperandValue::Ref(_) | OperandValue::ZeroSized => None,
+        }
+    }
+
     /// Treat this value as a pointer and return the data pointer and
     /// optional metadata as backend values.
     ///
     /// If you're making a place, use [`Self::deref`] instead.
     pub(crate) fn pointer_parts(self) -> (V, Option<V>) {
-        match self {
-            OperandValue::Immediate(llptr) => (llptr, None),
-            OperandValue::Pair(llptr, llextra) => (llptr, Some(llextra)),
-            _ => bug!("OperandValue cannot be a pointer: {self:?}"),
-        }
+        self.try_pointer_parts()
+            .unwrap_or_else(|| bug!("OperandValue cannot be a pointer: {self:?}"))
     }
 
     /// Treat this value as a pointer and return the place to which it points.

compiler/rustc_codegen_ssa/src/mir/statement.rs

@@ -1,4 +1,4 @@
-use rustc_middle::mir::{self, NonDivergingIntrinsic};
+use rustc_middle::mir::{self, NonDivergingIntrinsic, StmtDebugInfo};
 use rustc_middle::span_bug;
 use tracing::instrument;
 
@@ -8,6 +8,7 @@ use crate::traits::*;
 impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
     #[instrument(level = "debug", skip(self, bx))]
     pub(crate) fn codegen_statement(&mut self, bx: &mut Bx, statement: &mir::Statement<'tcx>) {
+        self.codegen_stmt_debuginfos(bx, &statement.debuginfos);
         self.set_debug_loc(bx, statement.source_info);
         match statement.kind {
             mir::StatementKind::Assign(box (ref place, ref rvalue)) => {
@@ -101,4 +102,49 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
             | mir::StatementKind::Nop => {}
         }
     }
+
+    pub(crate) fn codegen_stmt_debuginfo(&mut self, bx: &mut Bx, debuginfo: &StmtDebugInfo<'tcx>) {
+        match debuginfo {
+            StmtDebugInfo::AssignRef(dest, place) => {
+                let local_ref = match self.locals[place.local] {
+                    // For an rvalue like `&(_1.1)`, when `BackendRepr` is `BackendRepr::Memory`, we allocate a block of memory to this place.
+                    // The place is an indirect pointer, we can refer to it directly.
+                    LocalRef::Place(place_ref) => Some((place_ref, place.projection.as_slice())),
+                    // For an rvalue like `&((*_1).1)`, we are calculating the address of `_1.1`.
+                    // The deref projection is no-op here.
+                    LocalRef::Operand(operand_ref) if place.is_indirect_first_projection() => {
+                        Some((operand_ref.deref(bx.cx()), &place.projection[1..]))
+                    }
+                    // For an rvalue like `&1`, when `BackendRepr` is `BackendRepr::Scalar`,
+                    // we cannot get the address.
+                    // N.B. `non_ssa_locals` returns that this is an SSA local.
+                    LocalRef::Operand(_) => None,
+                    LocalRef::UnsizedPlace(_) | LocalRef::PendingOperand => None,
+                }
+                .filter(|(_, projection)| {
+                    // Drop unsupported projections.
+                    projection.iter().all(|p| p.can_use_in_debuginfo())
+                });
+                if let Some((base, projection)) = local_ref {
+                    self.debug_new_val_to_local(bx, *dest, base, projection);
+                } else {
+                    // If the address cannot be calculated, use poison to indicate that the value has been optimized out.
+                    self.debug_poison_to_local(bx, *dest);
+                }
+            }
+            StmtDebugInfo::InvalidAssign(local) => {
+                self.debug_poison_to_local(bx, *local);
+            }
+        }
+    }
+
+    pub(crate) fn codegen_stmt_debuginfos(
+        &mut self,
+        bx: &mut Bx,
+        debuginfos: &[StmtDebugInfo<'tcx>],
+    ) {
+        for debuginfo in debuginfos {
+            self.codegen_stmt_debuginfo(bx, debuginfo);
+        }
+    }
 }

compiler/rustc_codegen_ssa/src/traits/debuginfo.rs

@@ -77,7 +77,19 @@ pub trait DebugInfoBuilderMethods: BackendTypes {
         indirect_offsets: &[Size],
         // Byte range in the `dbg_var` covered by this fragment,
         // if this is a fragment of a composite `DIVariable`.
-        fragment: Option<Range<Size>>,
+        fragment: &Option<Range<Size>>,
+    );
+    fn dbg_var_value(
+        &mut self,
+        dbg_var: Self::DIVariable,
+        dbg_loc: Self::DILocation,
+        value: Self::Value,
+        direct_offset: Size,
+        // NB: each offset implies a deref (i.e. they're steps in a pointer chain).
+        indirect_offsets: &[Size],
+        // Byte range in the `dbg_var` covered by this fragment,
+        // if this is a fragment of a composite `DIVariable`.
+        fragment: &Option<Range<Size>>,
     );
     fn set_dbg_loc(&mut self, dbg_loc: Self::DILocation);
     fn clear_dbg_loc(&mut self);