1 use super::address_transform::AddressTransform; 2 use super::attr::{clone_die_attributes, FileAttributeContext}; 3 use super::expression::compile_expression; 4 use super::line_program::clone_line_program; 5 use super::range_info_builder::RangeInfoBuilder; 6 use super::refs::{PendingDebugInfoRefs, PendingUnitRefs, UnitRefsMap}; 7 use super::utils::{add_internal_types, append_vmctx_info}; 8 use super::DebugInputContext; 9 use crate::debug::{Compilation, Reader}; 10 use anyhow::{Context, Error}; 11 use cranelift_codegen::ir::Endianness; 12 use cranelift_codegen::isa::TargetIsa; 13 use gimli::write; 14 use gimli::{AttributeValue, DebuggingInformationEntry, Unit}; 15 use std::collections::HashSet; 16 use wasmtime_environ::StaticModuleIndex; 17 use wasmtime_versioned_export_macros::versioned_stringify_ident; 18 19 struct InheritedAttr<T> { 20 stack: Vec<(usize, T)>, 21 } 22 23 impl<T> InheritedAttr<T> { 24 fn new() -> Self { 25 InheritedAttr { stack: Vec::new() } 26 } 27 28 fn update(&mut self, depth: usize) { 29 while !self.stack.is_empty() && self.stack.last().unwrap().0 >= depth { 30 self.stack.pop(); 31 } 32 } 33 34 fn push(&mut self, depth: usize, value: T) { 35 self.stack.push((depth, value)); 36 } 37 38 fn top(&self) -> Option<&T> { 39 self.stack.last().map(|entry| &entry.1) 40 } 41 42 fn is_empty(&self) -> bool { 43 self.stack.is_empty() 44 } 45 } 46 47 fn get_base_type_name( 48 type_entry: &DebuggingInformationEntry<Reader<'_>>, 49 unit: &Unit<Reader<'_>, usize>, 50 context: &DebugInputContext<Reader<'_>>, 51 ) -> Result<String, Error> { 52 // FIXME remove recursion. 53 if let Some(AttributeValue::UnitRef(ref offset)) = type_entry.attr_value(gimli::DW_AT_type)? { 54 let mut entries = unit.entries_at_offset(*offset)?; 55 entries.next_entry()?; 56 if let Some(die) = entries.current() { 57 if let Some(AttributeValue::DebugStrRef(str_offset)) = 58 die.attr_value(gimli::DW_AT_name)? 59 { 60 return Ok(String::from( 61 context.debug_str.get_str(str_offset)?.to_string()?, 62 )); 63 } 64 match die.tag() { 65 gimli::DW_TAG_const_type => { 66 return Ok(format!("const {}", get_base_type_name(die, unit, context)?)); 67 } 68 gimli::DW_TAG_pointer_type => { 69 return Ok(format!("{}*", get_base_type_name(die, unit, context)?)); 70 } 71 gimli::DW_TAG_reference_type => { 72 return Ok(format!("{}&", get_base_type_name(die, unit, context)?)); 73 } 74 gimli::DW_TAG_array_type => { 75 return Ok(format!("{}[]", get_base_type_name(die, unit, context)?)); 76 } 77 _ => (), 78 } 79 } 80 } 81 Ok(String::from("??")) 82 } 83 84 enum WebAssemblyPtrKind { 85 Reference, 86 Pointer, 87 } 88 89 /// Replaces WebAssembly pointer type DIE with the wrapper 90 /// which natively represented by offset in a Wasm memory. 91 /// 92 /// `pointer_type_entry` is a DW_TAG_pointer_type entry (e.g. `T*`), 93 /// which refers its base type (e.g. `T`), or is a 94 /// DW_TAG_reference_type (e.g. `T&`). 95 /// 96 /// The generated wrapper is a structure that contains only the 97 /// `__ptr` field. The utility operators overloads is added to 98 /// provide better debugging experience. 99 /// 100 /// Wrappers of pointer and reference types are identical except for 101 /// their name -- they are formatted and accessed from a debugger 102 /// the same way. 103 /// 104 /// Notice that "resolve_vmctx_memory_ptr" is external/builtin 105 /// subprogram that is not part of Wasm code. 106 fn replace_pointer_type( 107 parent_id: write::UnitEntryId, 108 kind: WebAssemblyPtrKind, 109 comp_unit: &mut write::Unit, 110 wp_die_id: write::UnitEntryId, 111 pointer_type_entry: &DebuggingInformationEntry<Reader<'_>>, 112 unit: &Unit<Reader<'_>, usize>, 113 context: &DebugInputContext<Reader<'_>>, 114 out_strings: &mut write::StringTable, 115 pending_die_refs: &mut PendingUnitRefs, 116 ) -> Result<write::UnitEntryId, Error> { 117 const WASM_PTR_LEN: u8 = 4; 118 119 macro_rules! add_tag { 120 ($parent_id:ident, $tag:expr => $die:ident as $die_id:ident { $($a:path = $v:expr),* }) => { 121 let $die_id = comp_unit.add($parent_id, $tag); 122 #[allow(unused_variables)] 123 let $die = comp_unit.get_mut($die_id); 124 $( $die.set($a, $v); )* 125 }; 126 } 127 128 // Build DW_TAG_structure_type for the wrapper: 129 // .. DW_AT_name = "WebAssemblyPtrWrapper<T>", 130 // .. DW_AT_byte_size = 4, 131 let name = match kind { 132 WebAssemblyPtrKind::Pointer => format!( 133 "WebAssemblyPtrWrapper<{}>", 134 get_base_type_name(pointer_type_entry, unit, context)? 135 ), 136 WebAssemblyPtrKind::Reference => format!( 137 "WebAssemblyRefWrapper<{}>", 138 get_base_type_name(pointer_type_entry, unit, context)? 139 ), 140 }; 141 add_tag!(parent_id, gimli::DW_TAG_structure_type => wrapper_die as wrapper_die_id { 142 gimli::DW_AT_name = write::AttributeValue::StringRef(out_strings.add(name.as_str())), 143 gimli::DW_AT_byte_size = write::AttributeValue::Data1(WASM_PTR_LEN) 144 }); 145 146 // Build DW_TAG_pointer_type for `WebAssemblyPtrWrapper<T>*`: 147 // .. DW_AT_type = <wrapper_die> 148 add_tag!(parent_id, gimli::DW_TAG_pointer_type => wrapper_ptr_type as wrapper_ptr_type_id { 149 gimli::DW_AT_type = write::AttributeValue::UnitRef(wrapper_die_id) 150 }); 151 152 let base_type_id = pointer_type_entry.attr_value(gimli::DW_AT_type)?; 153 // Build DW_TAG_reference_type for `T&`: 154 // .. DW_AT_type = <base_type> 155 add_tag!(parent_id, gimli::DW_TAG_reference_type => ref_type as ref_type_id {}); 156 if let Some(AttributeValue::UnitRef(ref offset)) = base_type_id { 157 pending_die_refs.insert(ref_type_id, gimli::DW_AT_type, *offset); 158 } 159 160 // Build DW_TAG_pointer_type for `T*`: 161 // .. DW_AT_type = <base_type> 162 add_tag!(parent_id, gimli::DW_TAG_pointer_type => ptr_type as ptr_type_id {}); 163 if let Some(AttributeValue::UnitRef(ref offset)) = base_type_id { 164 pending_die_refs.insert(ptr_type_id, gimli::DW_AT_type, *offset); 165 } 166 167 // Build wrapper_die's DW_TAG_template_type_parameter: 168 // .. DW_AT_name = "T" 169 // .. DW_AT_type = <base_type> 170 add_tag!(wrapper_die_id, gimli::DW_TAG_template_type_parameter => t_param_die as t_param_die_id { 171 gimli::DW_AT_name = write::AttributeValue::StringRef(out_strings.add("T")) 172 }); 173 if let Some(AttributeValue::UnitRef(ref offset)) = base_type_id { 174 pending_die_refs.insert(t_param_die_id, gimli::DW_AT_type, *offset); 175 } 176 177 // Build wrapper_die's DW_TAG_member for `__ptr`: 178 // .. DW_AT_name = "__ptr" 179 // .. DW_AT_type = <wp_die> 180 // .. DW_AT_location = 0 181 add_tag!(wrapper_die_id, gimli::DW_TAG_member => m_die as m_die_id { 182 gimli::DW_AT_name = write::AttributeValue::StringRef(out_strings.add("__ptr")), 183 gimli::DW_AT_type = write::AttributeValue::UnitRef(wp_die_id), 184 gimli::DW_AT_data_member_location = write::AttributeValue::Data1(0) 185 }); 186 187 // Build wrapper_die's DW_TAG_subprogram for `ptr()`: 188 // .. DW_AT_linkage_name = "resolve_vmctx_memory_ptr" 189 // .. DW_AT_name = "ptr" 190 // .. DW_AT_type = <ptr_type> 191 // .. DW_TAG_formal_parameter 192 // .. .. DW_AT_type = <wrapper_ptr_type> 193 // .. .. DW_AT_artificial = 1 194 add_tag!(wrapper_die_id, gimli::DW_TAG_subprogram => deref_op_die as deref_op_die_id { 195 gimli::DW_AT_linkage_name = write::AttributeValue::StringRef(out_strings.add(versioned_stringify_ident!(resolve_vmctx_memory_ptr))), 196 gimli::DW_AT_name = write::AttributeValue::StringRef(out_strings.add("ptr")), 197 gimli::DW_AT_type = write::AttributeValue::UnitRef(ptr_type_id) 198 }); 199 add_tag!(deref_op_die_id, gimli::DW_TAG_formal_parameter => deref_op_this_param as deref_op_this_param_id { 200 gimli::DW_AT_type = write::AttributeValue::UnitRef(wrapper_ptr_type_id), 201 gimli::DW_AT_artificial = write::AttributeValue::Flag(true) 202 }); 203 204 // Build wrapper_die's DW_TAG_subprogram for `operator*`: 205 // .. DW_AT_linkage_name = "resolve_vmctx_memory_ptr" 206 // .. DW_AT_name = "operator*" 207 // .. DW_AT_type = <ref_type> 208 // .. DW_TAG_formal_parameter 209 // .. .. DW_AT_type = <wrapper_ptr_type> 210 // .. .. DW_AT_artificial = 1 211 add_tag!(wrapper_die_id, gimli::DW_TAG_subprogram => deref_op_die as deref_op_die_id { 212 gimli::DW_AT_linkage_name = write::AttributeValue::StringRef(out_strings.add(versioned_stringify_ident!(resolve_vmctx_memory_ptr))), 213 gimli::DW_AT_name = write::AttributeValue::StringRef(out_strings.add("operator*")), 214 gimli::DW_AT_type = write::AttributeValue::UnitRef(ref_type_id) 215 }); 216 add_tag!(deref_op_die_id, gimli::DW_TAG_formal_parameter => deref_op_this_param as deref_op_this_param_id { 217 gimli::DW_AT_type = write::AttributeValue::UnitRef(wrapper_ptr_type_id), 218 gimli::DW_AT_artificial = write::AttributeValue::Flag(true) 219 }); 220 221 // Build wrapper_die's DW_TAG_subprogram for `operator->`: 222 // .. DW_AT_linkage_name = "resolve_vmctx_memory_ptr" 223 // .. DW_AT_name = "operator->" 224 // .. DW_AT_type = <ptr_type> 225 // .. DW_TAG_formal_parameter 226 // .. .. DW_AT_type = <wrapper_ptr_type> 227 // .. .. DW_AT_artificial = 1 228 add_tag!(wrapper_die_id, gimli::DW_TAG_subprogram => deref_op_die as deref_op_die_id { 229 gimli::DW_AT_linkage_name = write::AttributeValue::StringRef(out_strings.add(versioned_stringify_ident!(resolve_vmctx_memory_ptr))), 230 gimli::DW_AT_name = write::AttributeValue::StringRef(out_strings.add("operator->")), 231 gimli::DW_AT_type = write::AttributeValue::UnitRef(ptr_type_id) 232 }); 233 add_tag!(deref_op_die_id, gimli::DW_TAG_formal_parameter => deref_op_this_param as deref_op_this_param_id { 234 gimli::DW_AT_type = write::AttributeValue::UnitRef(wrapper_ptr_type_id), 235 gimli::DW_AT_artificial = write::AttributeValue::Flag(true) 236 }); 237 238 Ok(wrapper_die_id) 239 } 240 241 fn is_dead_code(entry: &DebuggingInformationEntry<Reader<'_>>) -> bool { 242 const TOMBSTONE: u64 = u32::MAX as u64; 243 244 match entry.attr_value(gimli::DW_AT_low_pc) { 245 Ok(Some(AttributeValue::Addr(addr))) => addr == TOMBSTONE, 246 _ => false, 247 } 248 } 249 250 pub(crate) fn clone_unit( 251 compilation: &mut Compilation<'_>, 252 module: StaticModuleIndex, 253 unit: &Unit<Reader<'_>, usize>, 254 split_unit: Option<&Unit<Reader<'_>, usize>>, 255 split_dwarf: Option<&gimli::Dwarf<Reader<'_>>>, 256 context: &DebugInputContext<Reader<'_>>, 257 addr_tr: &AddressTransform, 258 out_encoding: gimli::Encoding, 259 out_units: &mut write::UnitTable, 260 out_strings: &mut write::StringTable, 261 translated: &mut HashSet<usize>, 262 isa: &dyn TargetIsa, 263 ) -> Result<Option<(write::UnitId, UnitRefsMap, PendingDebugInfoRefs)>, Error> { 264 let mut die_ref_map = UnitRefsMap::new(); 265 let mut pending_die_refs = PendingUnitRefs::new(); 266 let mut pending_di_refs = PendingDebugInfoRefs::new(); 267 let mut stack = Vec::new(); 268 269 let mut program_unit = unit; 270 let mut skeleton_die = None; 271 272 // Get entries in outer scope to avoid borrowing on short lived temporary. 273 let mut skeleton_entries = unit.entries(); 274 if let Some(unit) = split_unit { 275 program_unit = unit; 276 277 // From the spec, a skeleton unit has no children so we can assume the first, and only, entry is the DW_TAG_skeleton_unit (https://dwarfstd.org/doc/DWARF5.pdf). 278 if let Some(die_tuple) = skeleton_entries.next_dfs()? { 279 skeleton_die = Some(die_tuple.1); 280 } 281 } 282 283 let dwarf = &compilation.translations[module].debuginfo.dwarf; 284 let memory_offset = &compilation.module_memory_offsets[module]; 285 286 // Iterate over all of this compilation unit's entries. 287 let mut entries = program_unit.entries(); 288 let (mut comp_unit, unit_id, file_map, file_index_base, cu_low_pc, wp_die_id, vmctx_die_id) = 289 if let Some((depth_delta, entry)) = entries.next_dfs()? { 290 assert_eq!(depth_delta, 0); 291 let (out_line_program, debug_line_offset, file_map, file_index_base) = 292 clone_line_program( 293 split_dwarf.unwrap_or(dwarf), 294 dwarf, 295 program_unit, 296 entry, 297 skeleton_die, 298 addr_tr, 299 out_encoding, 300 context.debug_str, 301 context.debug_line, 302 out_strings, 303 )?; 304 305 if entry.tag() == gimli::DW_TAG_compile_unit { 306 let unit_id = out_units.add(write::Unit::new(out_encoding, out_line_program)); 307 let comp_unit = out_units.get_mut(unit_id); 308 309 let root_id = comp_unit.root(); 310 die_ref_map.insert(entry.offset(), root_id); 311 let cu_low_pc = unit.low_pc; 312 313 clone_die_attributes( 314 split_dwarf.unwrap_or(dwarf), 315 &program_unit, 316 entry, 317 context, 318 addr_tr, 319 None, 320 comp_unit, 321 root_id, 322 None, 323 None, 324 cu_low_pc, 325 out_strings, 326 &mut pending_die_refs, 327 &mut pending_di_refs, 328 FileAttributeContext::Root(Some(debug_line_offset)), 329 isa, 330 )?; 331 332 let (wp_die_id, vmctx_die_id) = 333 add_internal_types(comp_unit, root_id, out_strings, memory_offset); 334 335 stack.push(root_id); 336 ( 337 comp_unit, 338 unit_id, 339 file_map, 340 file_index_base, 341 cu_low_pc, 342 wp_die_id, 343 vmctx_die_id, 344 ) 345 } else { 346 // Can happen when the DWARF is split and we dont have the package/dwo files. 347 // This is a better user experience than errorring. 348 return Ok(None); // empty: 349 } 350 } else { 351 return Ok(None); // empty 352 }; 353 let mut skip_at_depth = None; 354 let mut current_frame_base = InheritedAttr::new(); 355 let mut current_value_range = InheritedAttr::new(); 356 let mut current_scope_ranges = InheritedAttr::new(); 357 while let Some((depth_delta, entry)) = entries.next_dfs()? { 358 // If `skip_at_depth` is `Some` then we previously decided to skip over 359 // a node and all it's children. Let A be the last node processed, B be 360 // the first node skipped, C be previous node, and D the current node. 361 // Then `cached` is the difference from A to B, `depth` is the difference 362 // from B to C, and `depth_delta` is the differenc from C to D. 363 let depth_delta = if let Some((depth, cached)) = skip_at_depth { 364 // `new_depth` = B to D 365 let new_depth = depth + depth_delta; 366 // if D is below B continue to skip 367 if new_depth > 0 { 368 skip_at_depth = Some((new_depth, cached)); 369 continue; 370 } 371 // otherwise process D with `depth_delta` being the difference from A to D 372 skip_at_depth = None; 373 new_depth + cached 374 } else { 375 depth_delta 376 }; 377 378 if !context 379 .reachable 380 .contains(&entry.offset().to_unit_section_offset(&unit)) 381 || is_dead_code(&entry) 382 { 383 // entry is not reachable: discarding all its info. 384 // Here B = C so `depth` is 0. A is the previous node so `cached` = 385 // `depth_delta`. 386 skip_at_depth = Some((0, depth_delta)); 387 continue; 388 } 389 390 let new_stack_len = stack.len().wrapping_add(depth_delta as usize); 391 current_frame_base.update(new_stack_len); 392 current_scope_ranges.update(new_stack_len); 393 current_value_range.update(new_stack_len); 394 let range_builder = if entry.tag() == gimli::DW_TAG_subprogram { 395 let range_builder = RangeInfoBuilder::from_subprogram_die( 396 dwarf, &unit, entry, context, addr_tr, cu_low_pc, 397 )?; 398 if let RangeInfoBuilder::Function(func) = range_builder { 399 let frame_info = compilation.function_frame_info(module, func); 400 current_value_range.push(new_stack_len, frame_info); 401 let (symbol, _) = compilation.function(module, func); 402 translated.insert(symbol); 403 current_scope_ranges.push(new_stack_len, range_builder.get_ranges(addr_tr)); 404 Some(range_builder) 405 } else { 406 // FIXME current_scope_ranges.push() 407 None 408 } 409 } else { 410 let high_pc = entry.attr_value(gimli::DW_AT_high_pc)?; 411 let ranges = entry.attr_value(gimli::DW_AT_ranges)?; 412 if high_pc.is_some() || ranges.is_some() { 413 let range_builder = 414 RangeInfoBuilder::from(dwarf, &unit, entry, context, cu_low_pc)?; 415 current_scope_ranges.push(new_stack_len, range_builder.get_ranges(addr_tr)); 416 Some(range_builder) 417 } else { 418 None 419 } 420 }; 421 422 if depth_delta <= 0 { 423 for _ in depth_delta..1 { 424 stack.pop(); 425 } 426 } else { 427 assert_eq!(depth_delta, 1); 428 } 429 430 if let Some(AttributeValue::Exprloc(expr)) = entry.attr_value(gimli::DW_AT_frame_base)? { 431 if let Some(expr) = compile_expression(&expr, unit.encoding(), None)? { 432 current_frame_base.push(new_stack_len, expr); 433 } 434 } 435 436 let parent = stack.last().unwrap(); 437 438 if entry.tag() == gimli::DW_TAG_pointer_type || entry.tag() == gimli::DW_TAG_reference_type 439 { 440 // Wrap pointer types. 441 let pointer_kind = match entry.tag() { 442 gimli::DW_TAG_pointer_type => WebAssemblyPtrKind::Pointer, 443 gimli::DW_TAG_reference_type => WebAssemblyPtrKind::Reference, 444 _ => panic!(), 445 }; 446 let die_id = replace_pointer_type( 447 *parent, 448 pointer_kind, 449 comp_unit, 450 wp_die_id, 451 entry, 452 &unit, 453 context, 454 out_strings, 455 &mut pending_die_refs, 456 )?; 457 stack.push(die_id); 458 assert_eq!(stack.len(), new_stack_len); 459 die_ref_map.insert(entry.offset(), die_id); 460 continue; 461 } 462 463 let die_id = comp_unit.add(*parent, entry.tag()); 464 465 stack.push(die_id); 466 assert_eq!(stack.len(), new_stack_len); 467 die_ref_map.insert(entry.offset(), die_id); 468 469 clone_die_attributes( 470 split_dwarf.unwrap_or(dwarf), 471 &unit, 472 entry, 473 context, 474 addr_tr, 475 current_value_range.top(), 476 &mut comp_unit, 477 die_id, 478 range_builder, 479 current_scope_ranges.top(), 480 cu_low_pc, 481 out_strings, 482 &mut pending_die_refs, 483 &mut pending_di_refs, 484 FileAttributeContext::Children { 485 file_map: &file_map, 486 file_index_base, 487 frame_base: current_frame_base.top(), 488 }, 489 isa, 490 )?; 491 492 // Data in WebAssembly memory always uses little-endian byte order. 493 // If the native architecture is big-endian, we need to mark all 494 // base types used to refer to WebAssembly memory as little-endian 495 // using the DW_AT_endianity attribute, so that the debugger will 496 // be able to correctly access them. 497 if entry.tag() == gimli::DW_TAG_base_type && isa.endianness() == Endianness::Big { 498 let current_scope = comp_unit.get_mut(die_id); 499 current_scope.set( 500 gimli::DW_AT_endianity, 501 write::AttributeValue::Endianity(gimli::DW_END_little), 502 ); 503 } 504 505 if entry.tag() == gimli::DW_TAG_subprogram && !current_scope_ranges.is_empty() { 506 append_vmctx_info( 507 comp_unit, 508 die_id, 509 vmctx_die_id, 510 addr_tr, 511 current_value_range.top(), 512 current_scope_ranges.top().context("range")?, 513 out_strings, 514 isa, 515 )?; 516 } 517 } 518 die_ref_map.patch(pending_die_refs, comp_unit); 519 Ok(Some((unit_id, die_ref_map, pending_di_refs))) 520 } 521