src/debuginfo/mod.rs

   1 mod emit;
   2 mod line_info;
   3
   4 use std::time::SystemTime;
   5
   6 use crate::prelude::*;
   7
   8 use rustc_span::{FileName, SourceFileHash, SourceFileHashAlgorithm};
   9
  10 use cranelift_codegen::ir::{StackSlots, ValueLabel, ValueLoc};
  11 use cranelift_codegen::isa::TargetIsa;
  12 use cranelift_codegen::ValueLocRange;
  13
  14 use gimli::write::{
  15     self, Address, AttributeValue, DwarfUnit, Expression, LineProgram, LineString, Location,
  16     LocationList, Range, RangeList, UnitEntryId, Writer, FileInfo,
  17 };
  18 use gimli::{Encoding, Format, LineEncoding, RunTimeEndian, X86_64};
  19
  20 pub(crate) use emit::{DebugReloc, DebugRelocName};
  21
  22 fn target_endian(tcx: TyCtxt<'_>) -> RunTimeEndian {
  23     use rustc_target::abi::Endian;
  24
  25     match tcx.data_layout.endian {
  26         Endian::Big => RunTimeEndian::Big,
  27         Endian::Little => RunTimeEndian::Little,
  28     }
  29 }
  30
  31 const MD5_LEN: usize = 16;
  32
  33 pub(crate) struct DebugContext<'tcx> {
  34     tcx: TyCtxt<'tcx>,
  35
  36     endian: RunTimeEndian,
  37     symbols: indexmap::IndexMap<FuncId, String>,
  38
  39     dwarf: DwarfUnit,
  40     unit_range_list: RangeList,
  41
  42     types: FxHashMap<Ty<'tcx>, UnitEntryId>,
  43 }
  44
  45 impl<'tcx> DebugContext<'tcx> {
  46     pub(crate) fn new(tcx: TyCtxt<'tcx>, address_size: u8) -> Self {
  47         let encoding = Encoding {
  48             format: Format::Dwarf32,
  49             // TODO: this should be configurable
  50             // macOS doesn't seem to support DWARF > 3
  51             // 5 version is required for md5 file hash
  52             version: 5,
  53             address_size,
  54         };
  55
  56         let mut dwarf = DwarfUnit::new(encoding);
  57
  58         // FIXME: how to get version when building out of tree?
  59         // Normally this would use option_env!("CFG_VERSION").
  60         let producer = format!("cg_clif (rustc {})", "unknown version");
  61         let comp_dir = tcx.sess.working_dir.0.to_string_lossy().into_owned();
  62         let (name, md5) = match tcx.sess.local_crate_source_file.clone() {
  63             Some(path) => {
  64                 let name = path.to_string_lossy().into_owned();
  65                 let hash = tcx.sess
  66                     .source_map()
  67                     .get_source_file(&FileName::Real(path))
  68                     .map(|f| f.src_hash)
  69                     .filter(|h| matches!(h, SourceFileHash { kind: SourceFileHashAlgorithm::Md5, .. }))
  70                     .map(|h| {
  71                         let mut buf = [0u8; MD5_LEN];
  72                         buf.copy_from_slice(h.hash_bytes());
  73                         buf
  74                     });
  75                 (name, hash)
  76             },
  77             None => (tcx.crate_name(LOCAL_CRATE).to_string(), None),
  78         };
  79
  80         let mut line_program = LineProgram::new(
  81             encoding,
  82             LineEncoding::default(),
  83             LineString::new(comp_dir.as_bytes(), encoding, &mut dwarf.line_strings),
  84             LineString::new(name.as_bytes(), encoding, &mut dwarf.line_strings),
  85             Some(FileInfo {
  86                 timestamp: SystemTime::now()
  87                     .duration_since(SystemTime::UNIX_EPOCH)
  88                     .map(|t| t.as_secs())
  89                     .unwrap_or(0),
  90                 size: 0,
  91                 md5: md5.unwrap_or_default(),
  92             }),
  93         );
  94         line_program.file_has_timestamp = true;
  95         line_program.file_has_md5 = md5.is_some();
  96
  97         dwarf.unit.line_program = line_program;
  98
  99         {
 100             let name = dwarf.strings.add(name);
 101             let comp_dir = dwarf.strings.add(comp_dir);
 102
 103             let root = dwarf.unit.root();
 104             let root = dwarf.unit.get_mut(root);
 105             root.set(
 106                 gimli::DW_AT_producer,
 107                 AttributeValue::StringRef(dwarf.strings.add(producer)),
 108             );
 109             root.set(
 110                 gimli::DW_AT_language,
 111                 AttributeValue::Language(gimli::DW_LANG_Rust),
 112             );
 113             root.set(gimli::DW_AT_name, AttributeValue::StringRef(name));
 114             root.set(gimli::DW_AT_comp_dir, AttributeValue::StringRef(comp_dir));
 115             root.set(
 116                 gimli::DW_AT_low_pc,
 117                 AttributeValue::Address(Address::Constant(0)),
 118             );
 119         }
 120
 121         DebugContext {
 122             tcx,
 123
 124             endian: target_endian(tcx),
 125             symbols: indexmap::IndexMap::new(),
 126
 127             dwarf,
 128             unit_range_list: RangeList(Vec::new()),
 129
 130             types: FxHashMap::default(),
 131         }
 132     }
 133
 134     fn dwarf_ty(&mut self, ty: Ty<'tcx>) -> UnitEntryId {
 135         if let Some(type_id) = self.types.get(ty) {
 136             return *type_id;
 137         }
 138
 139         let new_entry = |dwarf: &mut DwarfUnit, tag| dwarf.unit.add(dwarf.unit.root(), tag);
 140
 141         let primitive = |dwarf: &mut DwarfUnit, ate| {
 142             let type_id = new_entry(dwarf, gimli::DW_TAG_base_type);
 143             let type_entry = dwarf.unit.get_mut(type_id);
 144             type_entry.set(gimli::DW_AT_encoding, AttributeValue::Encoding(ate));
 145             type_id
 146         };
 147
 148         let name = format!("{}", ty);
 149         let layout = self.tcx.layout_of(ParamEnv::reveal_all().and(ty)).unwrap();
 150
 151         let type_id = match ty.kind {
 152             ty::Bool => primitive(&mut self.dwarf, gimli::DW_ATE_boolean),
 153             ty::Char => primitive(&mut self.dwarf, gimli::DW_ATE_UTF),
 154             ty::Uint(_) => primitive(&mut self.dwarf, gimli::DW_ATE_unsigned),
 155             ty::Int(_) => primitive(&mut self.dwarf, gimli::DW_ATE_signed),
 156             ty::Float(_) => primitive(&mut self.dwarf, gimli::DW_ATE_float),
 157             ty::Ref(_, pointee_ty, _mutbl)
 158             | ty::RawPtr(ty::TypeAndMut {
 159                 ty: pointee_ty,
 160                 mutbl: _mutbl,
 161             }) => {
 162                 let type_id = new_entry(&mut self.dwarf, gimli::DW_TAG_pointer_type);
 163
 164                 // Ensure that type is inserted before recursing to avoid duplicates
 165                 self.types.insert(ty, type_id);
 166
 167                 let pointee = self.dwarf_ty(pointee_ty);
 168
 169                 let type_entry = self.dwarf.unit.get_mut(type_id);
 170
 171                 //type_entry.set(gimli::DW_AT_mutable, AttributeValue::Flag(mutbl == rustc_hir::Mutability::Mut));
 172                 type_entry.set(gimli::DW_AT_type, AttributeValue::ThisUnitEntryRef(pointee));
 173
 174                 type_id
 175             }
 176             ty::Adt(adt_def, _substs) if adt_def.is_struct() && !layout.is_unsized() => {
 177                 let type_id = new_entry(&mut self.dwarf, gimli::DW_TAG_structure_type);
 178
 179                 // Ensure that type is inserted before recursing to avoid duplicates
 180                 self.types.insert(ty, type_id);
 181
 182                 let variant = adt_def.non_enum_variant();
 183
 184                 for (field_idx, field_def) in variant.fields.iter().enumerate() {
 185                     let field_offset = layout.fields.offset(field_idx);
 186                     let field_layout = layout.field(&layout::LayoutCx {
 187                         tcx: self.tcx,
 188                         param_env: ParamEnv::reveal_all(),
 189                     }, field_idx).unwrap();
 190
 191                     let field_type = self.dwarf_ty(field_layout.ty);
 192
 193                     let field_id = self.dwarf.unit.add(type_id, gimli::DW_TAG_member);
 194                     let field_entry = self.dwarf.unit.get_mut(field_id);
 195
 196                     field_entry.set(gimli::DW_AT_name, AttributeValue::String(field_def.ident.as_str().to_string().into_bytes()));
 197                     field_entry.set(gimli::DW_AT_data_member_location, AttributeValue::Udata(field_offset.bytes()));
 198                     field_entry.set(gimli::DW_AT_type, AttributeValue::ThisUnitEntryRef(field_type));
 199                 }
 200
 201                 type_id
 202             }
 203             _ => new_entry(&mut self.dwarf, gimli::DW_TAG_structure_type),
 204         };
 205
 206         let type_entry = self.dwarf.unit.get_mut(type_id);
 207
 208         type_entry.set(gimli::DW_AT_name, AttributeValue::String(name.into_bytes()));
 209         type_entry.set(
 210             gimli::DW_AT_byte_size,
 211             AttributeValue::Udata(layout.size.bytes()),
 212         );
 213
 214         self.types.insert(ty, type_id);
 215
 216         type_id
 217     }
 218 }
 219
 220 pub(crate) struct FunctionDebugContext<'a, 'tcx> {
 221     debug_context: &'a mut DebugContext<'tcx>,
 222     entry_id: UnitEntryId,
 223     symbol: usize,
 224     instance: Instance<'tcx>,
 225     mir: &'tcx mir::Body<'tcx>,
 226 }
 227
 228 impl<'a, 'tcx> FunctionDebugContext<'a, 'tcx> {
 229     pub(crate) fn new(
 230         debug_context: &'a mut DebugContext<'tcx>,
 231         instance: Instance<'tcx>,
 232         func_id: FuncId,
 233         name: &str,
 234     ) -> Self {
 235         let mir = *debug_context.tcx.instance_mir(instance.def);
 236
 237         let (symbol, _) = debug_context.symbols.insert_full(func_id, name.to_string());
 238
 239         // FIXME: add to appropriate scope intead of root
 240         let scope = debug_context.dwarf.unit.root();
 241
 242         let entry_id = debug_context
 243             .dwarf
 244             .unit
 245             .add(scope, gimli::DW_TAG_subprogram);
 246         let entry = debug_context.dwarf.unit.get_mut(entry_id);
 247         let name_id = debug_context.dwarf.strings.add(name);
 248         entry.set(
 249             gimli::DW_AT_linkage_name,
 250             AttributeValue::StringRef(name_id),
 251         );
 252
 253         FunctionDebugContext {
 254             debug_context,
 255             entry_id,
 256             symbol,
 257             instance,
 258             mir,
 259         }
 260     }
 261
 262     fn define_local(&mut self, name: String, ty: Ty<'tcx>) -> UnitEntryId {
 263         let ty = self.debug_context.tcx.subst_and_normalize_erasing_regions(
 264             self.instance.substs,
 265             ty::ParamEnv::reveal_all(),
 266             &ty,
 267         );
 268         let dw_ty = self.debug_context.dwarf_ty(ty);
 269
 270         let var_id = self
 271             .debug_context
 272             .dwarf
 273             .unit
 274             .add(self.entry_id, gimli::DW_TAG_variable);
 275         let var_entry = self.debug_context.dwarf.unit.get_mut(var_id);
 276
 277         var_entry.set(gimli::DW_AT_name, AttributeValue::String(name.into_bytes()));
 278         var_entry.set(gimli::DW_AT_type, AttributeValue::ThisUnitEntryRef(dw_ty));
 279
 280         var_id
 281     }
 282
 283     pub(crate) fn define(
 284         &mut self,
 285         context: &Context,
 286         isa: &dyn TargetIsa,
 287         source_info_set: &indexmap::IndexSet<SourceInfo>,
 288         local_map: FxHashMap<mir::Local, CPlace<'tcx>>,
 289     ) {
 290         let end = self.create_debug_lines(context, isa, source_info_set);
 291
 292         self.debug_context
 293             .unit_range_list
 294             .0
 295             .push(Range::StartLength {
 296                 begin: Address::Symbol {
 297                     symbol: self.symbol,
 298                     addend: 0,
 299                 },
 300                 length: end as u64,
 301             });
 302
 303         // FIXME make it more reliable and implement scopes before re-enabling this.
 304         if false {
 305             let value_labels_ranges = context.build_value_labels_ranges(isa).unwrap();
 306
 307             for (local, _local_decl) in self.mir.local_decls.iter_enumerated() {
 308                 let var_id = self.define_local(format!("{:?}", local), &self.mir.local_decls[local].ty);
 309
 310                 let location = place_location(
 311                     self,
 312                     isa,
 313                     context,
 314                     &local_map,
 315                     &value_labels_ranges,
 316                     Place {
 317                         local,
 318                         projection: ty::List::empty(),
 319                     },
 320                 );
 321
 322                 let var_entry = self.debug_context.dwarf.unit.get_mut(var_id);
 323                 var_entry.set(gimli::DW_AT_location, location);
 324             }
 325         }
 326
 327         // FIXME create locals for all entries in mir.var_debug_info
 328     }
 329 }
 330
 331 fn place_location<'a, 'tcx>(
 332     func_debug_ctx: &mut FunctionDebugContext<'a, 'tcx>,
 333     isa: &dyn TargetIsa,
 334     context: &Context,
 335     local_map: &FxHashMap<mir::Local, CPlace<'tcx>>,
 336     #[allow(rustc::default_hash_types)]
 337     value_labels_ranges: &std::collections::HashMap<ValueLabel, Vec<ValueLocRange>>,
 338     place: Place<'tcx>,
 339 ) -> AttributeValue {
 340     assert!(place.projection.is_empty()); // FIXME implement them
 341
 342     match local_map[&place.local].inner() {
 343         CPlaceInner::Var(local) => {
 344             let value_label = cranelift_codegen::ir::ValueLabel::from_u32(local.as_u32());
 345             if let Some(value_loc_ranges) = value_labels_ranges.get(&value_label) {
 346                 let loc_list = LocationList(
 347                     value_loc_ranges
 348                         .iter()
 349                         .map(|value_loc_range| Location::StartEnd {
 350                             begin: Address::Symbol {
 351                                 symbol: func_debug_ctx.symbol,
 352                                 addend: i64::from(value_loc_range.start),
 353                             },
 354                             end: Address::Symbol {
 355                                 symbol: func_debug_ctx.symbol,
 356                                 addend: i64::from(value_loc_range.end),
 357                             },
 358                             data: Expression(
 359                                 translate_loc(isa, value_loc_range.loc, &context.func.stack_slots).unwrap(),
 360                             ),
 361                         })
 362                         .collect(),
 363                 );
 364                 let loc_list_id = func_debug_ctx.debug_context.dwarf.unit.locations.add(loc_list);
 365
 366                 AttributeValue::LocationListRef(loc_list_id)
 367             } else {
 368                 // FIXME set value labels for unused locals
 369
 370                 AttributeValue::Exprloc(Expression(vec![]))
 371             }
 372         }
 373         CPlaceInner::Addr(_, _) => {
 374             // FIXME implement this (used by arguments and returns)
 375
 376             AttributeValue::Exprloc(Expression(vec![]))
 377
 378             // For PointerBase::Stack:
 379             //AttributeValue::Exprloc(Expression(translate_loc(ValueLoc::Stack(*stack_slot), &context.func.stack_slots).unwrap()))
 380         }
 381     }
 382 }
 383
 384 // Adapted from https://github.com/CraneStation/wasmtime/blob/5a1845b4caf7a5dba8eda1fef05213a532ed4259/crates/debug/src/transform/expression.rs#L59-L137
 385 fn translate_loc(isa: &dyn TargetIsa, loc: ValueLoc, stack_slots: &StackSlots) -> Option<Vec<u8>> {
 386     match loc {
 387         ValueLoc::Reg(reg) => {
 388             let machine_reg = isa.map_dwarf_register(reg).unwrap();
 389             assert!(machine_reg <= 32); // FIXME
 390             Some(vec![gimli::constants::DW_OP_reg0.0 + machine_reg as u8])
 391         }
 392         ValueLoc::Stack(ss) => {
 393             if let Some(ss_offset) = stack_slots[ss].offset {
 394                 let endian = gimli::RunTimeEndian::Little;
 395                 let mut writer = write::EndianVec::new(endian);
 396                 writer
 397                     .write_u8(gimli::constants::DW_OP_breg0.0 + X86_64::RBP.0 as u8)
 398                     .expect("bp wr");
 399                 writer.write_sleb128(ss_offset as i64 + 16).expect("ss wr");
 400                 let buf = writer.into_vec();
 401                 return Some(buf);
 402             }
 403             None
 404         }
 405         _ => None,
 406     }
 407 }