Add debuginfo for statics

bjorn3 · bjorn3 · commit 4b80941ae36c · 2024-03-27T16:53:57.000Z
Most types are still unimplemented and get a placeholder byte array
instead.
diff --git a/src/constant.rs b/src/constant.rs
@@ -31,10 +31,16 @@ impl ConstantCx {
     }
 }
 
-pub(crate) fn codegen_static(tcx: TyCtxt<'_>, module: &mut dyn Module, def_id: DefId) {
+pub(crate) fn codegen_static(tcx: TyCtxt<'_>, module: &mut dyn Module, def_id: DefId) -> DataId {
     let mut constants_cx = ConstantCx::new();
     constants_cx.todo.push(TodoItem::Static(def_id));
     constants_cx.finalize(tcx, module);
+
+    data_id_for_static(
+        tcx, module, def_id, false,
+        // For a declaration the stated mutability doesn't matter.
+        false,
+    )
 }
 
 pub(crate) fn codegen_tls_ref<'tcx>(
diff --git a/src/debuginfo/emit.rs b/src/debuginfo/emit.rs
@@ -15,7 +15,6 @@ pub(super) fn address_for_func(func_id: FuncId) -> Address {
     Address::Symbol { symbol: symbol as usize, addend: 0 }
 }
 
-#[allow(dead_code)]
 pub(super) fn address_for_data(data_id: DataId) -> Address {
     let symbol = data_id.as_u32();
     assert!(symbol & 1 << 31 == 0);
diff --git a/src/debuginfo/mod.rs b/src/debuginfo/mod.rs
@@ -3,24 +3,28 @@
 mod emit;
 mod line_info;
 mod object;
+mod types;
 mod unwind;
 
 use cranelift_codegen::ir::Endianness;
 use cranelift_codegen::isa::TargetIsa;
+use cranelift_module::DataId;
 use gimli::write::{
     Address, AttributeValue, DwarfUnit, Expression, FileId, LineProgram, LineString, Range,
     RangeList, UnitEntryId,
 };
 use gimli::{AArch64, Encoding, Format, LineEncoding, Register, RiscV, RunTimeEndian, X86_64};
 use indexmap::IndexSet;
 use rustc_codegen_ssa::debuginfo::type_names;
+use rustc_hir::def::DefKind;
 use rustc_hir::def_id::DefIdMap;
 use rustc_session::Session;
 use rustc_span::{SourceFileHash, StableSourceFileId};
 
 pub(crate) use self::emit::{DebugReloc, DebugRelocName};
+pub(crate) use self::types::TypeDebugContext;
 pub(crate) use self::unwind::UnwindContext;
-use crate::debuginfo::emit::address_for_func;
+use crate::debuginfo::emit::{address_for_data, address_for_func};
 use crate::prelude::*;
 
 pub(crate) fn producer(sess: &Session) -> String {
@@ -35,6 +39,7 @@ pub(crate) struct DebugContext {
     created_files: FxHashMap<(StableSourceFileId, SourceFileHash), FileId>,
     stack_pointer_register: Register,
     namespace_map: DefIdMap<UnitEntryId>,
+    array_size_type: UnitEntryId,
 
     should_remap_filepaths: bool,
 }
@@ -129,13 +134,27 @@ impl DebugContext {
             root.set(gimli::DW_AT_low_pc, AttributeValue::Address(Address::Constant(0)));
         }
 
+        let array_size_type = dwarf.unit.add(dwarf.unit.root(), gimli::DW_TAG_base_type);
+        let array_size_type_entry = dwarf.unit.get_mut(array_size_type);
+        array_size_type_entry.set(
+            gimli::DW_AT_name,
+            AttributeValue::StringRef(dwarf.strings.add("__ARRAY_SIZE_TYPE__")),
+        );
+        array_size_type_entry
+            .set(gimli::DW_AT_encoding, AttributeValue::Encoding(gimli::DW_ATE_unsigned));
+        array_size_type_entry.set(
+            gimli::DW_AT_byte_size,
+            AttributeValue::Udata(isa.frontend_config().pointer_bytes().into()),
+        );
+
         DebugContext {
             endian,
             dwarf,
             unit_range_list: RangeList(Vec::new()),
             created_files: FxHashMap::default(),
             stack_pointer_register,
             namespace_map: DefIdMap::default(),
+            array_size_type,
             should_remap_filepaths,
         }
     }
@@ -231,6 +250,62 @@ impl DebugContext {
             source_loc_set: IndexSet::new(),
         }
     }
+
+    // Adapted from https://github.com/rust-lang/rust/blob/10a7aa14fed9b528b74b0f098c4899c37c09a9c7/compiler/rustc_codegen_llvm/src/debuginfo/metadata.rs#L1288-L1346
+    pub(crate) fn define_static<'tcx>(
+        &mut self,
+        tcx: TyCtxt<'tcx>,
+        type_dbg: &mut TypeDebugContext<'tcx>,
+        def_id: DefId,
+        data_id: DataId,
+    ) {
+        let DefKind::Static { nested, .. } = tcx.def_kind(def_id) else { bug!() };
+        if nested {
+            return;
+        }
+
+        let scope = self.item_namespace(tcx, tcx.parent(def_id));
+
+        let span = tcx.def_span(def_id);
+        let (file_id, line, _column) = self.get_span_loc(tcx, span, span);
+
+        let static_type = Instance::mono(tcx, def_id).ty(tcx, ty::ParamEnv::reveal_all());
+        let static_layout = tcx.layout_of(ty::ParamEnv::reveal_all().and(static_type)).unwrap();
+        // FIXME use the actual type layout
+        let type_id = self.debug_type(tcx, type_dbg, static_type);
+
+        let name = tcx.item_name(def_id);
+        let linkage_name = tcx.symbol_name(Instance::mono(tcx, def_id)).name;
+
+        let entry_id = self.dwarf.unit.add(scope, gimli::DW_TAG_variable);
+        let entry = self.dwarf.unit.get_mut(entry_id);
+        let linkage_name_id = if name.as_str() != linkage_name {
+            Some(self.dwarf.strings.add(linkage_name))
+        } else {
+            None
+        };
+        let name_id = self.dwarf.strings.add(name.as_str());
+
+        entry.set(gimli::DW_AT_name, AttributeValue::StringRef(name_id));
+        entry.set(gimli::DW_AT_type, AttributeValue::UnitRef(type_id));
+
+        if tcx.is_reachable_non_generic(def_id) {
+            entry.set(gimli::DW_AT_external, AttributeValue::FlagPresent);
+        }
+
+        entry.set(gimli::DW_AT_decl_file, AttributeValue::FileIndex(Some(file_id)));
+        entry.set(gimli::DW_AT_decl_line, AttributeValue::Udata(line));
+
+        entry.set(gimli::DW_AT_alignment, AttributeValue::Udata(static_layout.align.pref.bytes()));
+
+        let mut expr = Expression::new();
+        expr.op_addr(address_for_data(data_id));
+        entry.set(gimli::DW_AT_location, AttributeValue::Exprloc(expr));
+
+        if let Some(linkage_name_id) = linkage_name_id {
+            entry.set(gimli::DW_AT_linkage_name, AttributeValue::StringRef(linkage_name_id));
+        }
+    }
 }
 
 impl FunctionDebugContext {
diff --git a/src/debuginfo/types.rs b/src/debuginfo/types.rs
@@ -0,0 +1,162 @@
+// Adapted from https://github.com/rust-lang/rust/blob/10a7aa14fed9b528b74b0f098c4899c37c09a9c7/compiler/rustc_codegen_llvm/src/debuginfo/metadata.rs
+
+use gimli::write::{AttributeValue, UnitEntryId};
+use rustc_codegen_ssa::debuginfo::type_names;
+use rustc_data_structures::fx::FxHashMap;
+use rustc_middle::ty::{self, Ty, TyCtxt};
+
+use crate::{has_ptr_meta, DebugContext};
+
+#[derive(Default)]
+pub(crate) struct TypeDebugContext<'tcx> {
+    type_map: FxHashMap<Ty<'tcx>, UnitEntryId>,
+}
+
+/// Returns from the enclosing function if the type debuginfo node with the given
+/// unique ID can be found in the type map.
+macro_rules! return_if_type_created_in_meantime {
+    ($type_dbg:expr, $ty:expr) => {
+        if let Some(&type_id) = $type_dbg.type_map.get(&$ty) {
+            return type_id;
+        }
+    };
+}
+
+impl DebugContext {
+    pub(crate) fn debug_type<'tcx>(
+        &mut self,
+        tcx: TyCtxt<'tcx>,
+        type_dbg: &mut TypeDebugContext<'tcx>,
+        ty: Ty<'tcx>,
+    ) -> UnitEntryId {
+        if let Some(&type_id) = type_dbg.type_map.get(&ty) {
+            return type_id;
+        }
+
+        let type_id = match ty.kind() {
+            ty::Never | ty::Bool | ty::Char | ty::Int(_) | ty::Uint(_) | ty::Float(_) => {
+                self.basic_type(tcx, ty)
+            }
+            ty::Tuple(elems) if elems.is_empty() => self.basic_type(tcx, ty),
+            ty::Array(elem_ty, len) => self.array_type(
+                tcx,
+                type_dbg,
+                *elem_ty,
+                len.eval_target_usize(tcx, ty::ParamEnv::reveal_all()),
+            ),
+            // ty::Slice(_) | ty::Str
+            // ty::Dynamic
+            // ty::Foreign
+            ty::RawPtr(pointee_type, _) | ty::Ref(_, pointee_type, _) => {
+                self.pointer_type(tcx, type_dbg, ty, *pointee_type)
+            }
+            // ty::Adt(def, args) if def.is_box() && args.get(1).map_or(true, |arg| cx.layout_of(arg.expect_ty()).is_1zst())
+            // ty::FnDef(..) | ty::FnPtr(..)
+            // ty::Closure(..)
+            // ty::Adt(def, ..)
+            // ty::Tuple(_)
+            // ty::Param(_)
+            // FIXME implement remaining types and add unreachable!() to the fallback branch
+            _ => self.placeholder_for_type(tcx, type_dbg, ty),
+        };
+
+        type_dbg.type_map.insert(ty, type_id);
+
+        type_id
+    }
+
+    fn basic_type<'tcx>(&mut self, tcx: TyCtxt<'tcx>, ty: Ty<'tcx>) -> UnitEntryId {
+        let (name, encoding) = match ty.kind() {
+            ty::Never => ("!", gimli::DW_ATE_unsigned),
+            ty::Tuple(elems) if elems.is_empty() => ("()", gimli::DW_ATE_unsigned),
+            ty::Bool => ("bool", gimli::DW_ATE_boolean),
+            ty::Char => ("char", gimli::DW_ATE_UTF),
+            ty::Int(int_ty) => (int_ty.name_str(), gimli::DW_ATE_signed),
+            ty::Uint(uint_ty) => (uint_ty.name_str(), gimli::DW_ATE_unsigned),
+            ty::Float(float_ty) => (float_ty.name_str(), gimli::DW_ATE_float),
+            _ => unreachable!(),
+        };
+
+        let type_id = self.dwarf.unit.add(self.dwarf.unit.root(), gimli::DW_TAG_base_type);
+        let type_entry = self.dwarf.unit.get_mut(type_id);
+        type_entry.set(gimli::DW_AT_name, AttributeValue::StringRef(self.dwarf.strings.add(name)));
+        type_entry.set(gimli::DW_AT_encoding, AttributeValue::Encoding(encoding));
+        type_entry.set(
+            gimli::DW_AT_byte_size,
+            AttributeValue::Udata(
+                tcx.layout_of(ty::ParamEnv::reveal_all().and(ty)).expect("FIXME").size.bytes(),
+            ),
+        );
+
+        type_id
+    }
+
+    fn array_type<'tcx>(
+        &mut self,
+        tcx: TyCtxt<'tcx>,
+        type_dbg: &mut TypeDebugContext<'tcx>,
+        elem_ty: Ty<'tcx>,
+        len: u64,
+    ) -> UnitEntryId {
+        let elem_dw_ty = self.debug_type(tcx, type_dbg, elem_ty);
+
+        return_if_type_created_in_meantime!(type_dbg, elem_ty);
+
+        let array_type_id = self.dwarf.unit.add(self.dwarf.unit.root(), gimli::DW_TAG_array_type);
+        let array_type_entry = self.dwarf.unit.get_mut(array_type_id);
+        array_type_entry.set(gimli::DW_AT_type, AttributeValue::UnitRef(elem_dw_ty));
+
+        let subrange_id = self.dwarf.unit.add(array_type_id, gimli::DW_TAG_subrange_type);
+        let subrange_entry = self.dwarf.unit.get_mut(subrange_id);
+        subrange_entry.set(gimli::DW_AT_type, AttributeValue::UnitRef(self.array_size_type));
+        subrange_entry.set(gimli::DW_AT_lower_bound, AttributeValue::Udata(0));
+        subrange_entry.set(gimli::DW_AT_count, AttributeValue::Udata(len));
+
+        array_type_id
+    }
+
+    fn pointer_type<'tcx>(
+        &mut self,
+        tcx: TyCtxt<'tcx>,
+        type_dbg: &mut TypeDebugContext<'tcx>,
+        ptr_type: Ty<'tcx>,
+        pointee_type: Ty<'tcx>,
+    ) -> UnitEntryId {
+        let pointee_dw_ty = self.debug_type(tcx, type_dbg, pointee_type);
+
+        return_if_type_created_in_meantime!(type_dbg, ptr_type);
+
+        let name = type_names::compute_debuginfo_type_name(tcx, ptr_type, true);
+
+        if !has_ptr_meta(tcx, ptr_type) {
+            let pointer_type_id =
+                self.dwarf.unit.add(self.dwarf.unit.root(), gimli::DW_TAG_pointer_type);
+            let pointer_entry = self.dwarf.unit.get_mut(pointer_type_id);
+            pointer_entry.set(gimli::DW_AT_type, AttributeValue::UnitRef(pointee_dw_ty));
+            pointer_entry
+                .set(gimli::DW_AT_name, AttributeValue::StringRef(self.dwarf.strings.add(name)));
+
+            pointer_type_id
+        } else {
+            // FIXME implement debuginfo for fat pointers
+            self.placeholder_for_type(tcx, type_dbg, ptr_type)
+        }
+    }
+
+    fn placeholder_for_type<'tcx>(
+        &mut self,
+        tcx: TyCtxt<'tcx>,
+        type_dbg: &mut TypeDebugContext<'tcx>,
+        ty: Ty<'tcx>,
+    ) -> UnitEntryId {
+        self.debug_type(
+            tcx,
+            type_dbg,
+            Ty::new_array(
+                tcx,
+                tcx.types.u8,
+                tcx.layout_of(ty::ParamEnv::reveal_all().and(ty)).unwrap().size.bytes(),
+            ),
+        )
+    }
+}
diff --git a/src/driver/aot.rs b/src/driver/aot.rs
@@ -23,6 +23,7 @@ use rustc_session::config::{DebugInfo, OutFileName, OutputFilenames, OutputType}
 use rustc_session::Session;
 
 use crate::concurrency_limiter::{ConcurrencyLimiter, ConcurrencyLimiterToken};
+use crate::debuginfo::TypeDebugContext;
 use crate::global_asm::GlobalAsmConfig;
 use crate::{prelude::*, BackendConfig};
 
@@ -460,6 +461,7 @@ fn module_codegen(
                 tcx.sess.opts.debuginfo != DebugInfo::None,
                 cgu_name,
             );
+            let mut type_dbg = TypeDebugContext::default();
             super::predefine_mono_items(tcx, &mut module, &mono_items);
             let mut codegened_functions = vec![];
             for (mono_item, _) in mono_items {
@@ -475,7 +477,10 @@ fn module_codegen(
                         codegened_functions.push(codegened_function);
                     }
                     MonoItem::Static(def_id) => {
-                        crate::constant::codegen_static(tcx, &mut module, def_id)
+                        let data_id = crate::constant::codegen_static(tcx, &mut module, def_id);
+                        if let Some(debug_context) = &mut cx.debug_context {
+                            debug_context.define_static(tcx, &mut type_dbg, def_id, data_id);
+                        }
                     }
                     MonoItem::GlobalAsm(item_id) => {
                         crate::global_asm::codegen_global_asm_item(

Original file line number	Diff line number	Diff line change
`@@ -31,10 +31,16 @@ impl ConstantCx {`
`31`	`31`	`}`
`32`	`32`	`}`
`33`	`33`
`34`		`-pub(crate) fn codegen_static(tcx: TyCtxt<'_>, module: &mut dyn Module, def_id: DefId) {`
	`34`	`+pub(crate) fn codegen_static(tcx: TyCtxt<'_>, module: &mut dyn Module, def_id: DefId) -> DataId {`
`35`	`35`	`let mut constants_cx = ConstantCx::new();`
`36`	`36`	`constants_cx.todo.push(TodoItem::Static(def_id));`
`37`	`37`	`constants_cx.finalize(tcx, module);`
	`38`	`+`
	`39`	`+ data_id_for_static(`
	`40`	`+ tcx, module, def_id, false,`
	`41`	`+ // For a declaration the stated mutability doesn't matter.`
	`42`	`+ false,`
	`43`	`+ )`
`38`	`44`	`}`
`39`	`45`
`40`	`46`	`pub(crate) fn codegen_tls_ref<'tcx>(`
Original file line number	Diff line number	Diff line change
`@@ -15,7 +15,6 @@ pub(super) fn address_for_func(func_id: FuncId) -> Address {`
`15`	`15`	`Address::Symbol { symbol: symbol as usize, addend: 0 }`
`16`	`16`	`}`
`17`	`17`
`18`		`-#[allow(dead_code)]`
`19`	`18`	`pub(super) fn address_for_data(data_id: DataId) -> Address {`
`20`	`19`	`let symbol = data_id.as_u32();`
`21`	`20`	`assert!(symbol & 1 << 31 == 0);`