1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
use crate::module::{
FuncRefIndex, Initializer, MemoryInitialization, MemoryInitializer, MemoryPlan, Module,
TablePlan, TableSegment, TableSegmentElements,
};
use crate::prelude::*;
use crate::{
ConstExpr, ConstOp, DataIndex, DefinedFuncIndex, ElemIndex, EngineOrModuleTypeIndex,
EntityIndex, EntityType, FuncIndex, GlobalIndex, IndexType, InitMemory, MemoryIndex,
ModuleInternedTypeIndex, ModuleTypesBuilder, PrimaryMap, SizeOverflow, StaticMemoryInitializer,
TableIndex, TableInitialValue, Tunables, TypeConvert, TypeIndex, Unsigned, WasmError,
WasmHeapTopType, WasmHeapType, WasmResult, WasmValType, WasmparserTypeConverter,
};
use anyhow::{bail, Result};
use cranelift_entity::packed_option::ReservedValue;
use std::borrow::Cow;
use std::collections::HashMap;
use std::mem;
use std::path::PathBuf;
use std::sync::Arc;
use wasmparser::{
types::Types, CustomSectionReader, DataKind, ElementItems, ElementKind, Encoding, ExternalKind,
FuncToValidate, FunctionBody, KnownCustom, NameSectionReader, Naming, Parser, Payload, TypeRef,
Validator, ValidatorResources,
};
/// Object containing the standalone environment information.
pub struct ModuleEnvironment<'a, 'data> {
/// The current module being translated
result: ModuleTranslation<'data>,
/// Intern'd types for this entire translation, shared by all modules.
types: &'a mut ModuleTypesBuilder,
// Various bits and pieces of configuration
validator: &'a mut Validator,
tunables: &'a Tunables,
}
/// The result of translating via `ModuleEnvironment`. Function bodies are not
/// yet translated, and data initializers have not yet been copied out of the
/// original buffer.
#[derive(Default)]
pub struct ModuleTranslation<'data> {
/// Module information.
pub module: Module,
/// The input wasm binary.
///
/// This can be useful, for example, when modules are parsed from a
/// component and the embedder wants access to the raw wasm modules
/// themselves.
pub wasm: &'data [u8],
/// References to the function bodies.
pub function_body_inputs: PrimaryMap<DefinedFuncIndex, FunctionBodyData<'data>>,
/// A list of type signatures which are considered exported from this
/// module, or those that can possibly be called. This list is sorted, and
/// trampolines for each of these signatures are required.
pub exported_signatures: Vec<ModuleInternedTypeIndex>,
/// DWARF debug information, if enabled, parsed from the module.
pub debuginfo: DebugInfoData<'data>,
/// Set if debuginfo was found but it was not parsed due to `Tunables`
/// configuration.
pub has_unparsed_debuginfo: bool,
/// List of data segments found in this module which should be concatenated
/// together for the final compiled artifact.
///
/// These data segments, when concatenated, are indexed by the
/// `MemoryInitializer` type.
pub data: Vec<Cow<'data, [u8]>>,
/// The desired alignment of `data` in the final data section of the object
/// file that we'll emit.
///
/// Note that this is 1 by default but `MemoryInitialization::Static` might
/// switch this to a higher alignment to facilitate mmap-ing data from
/// an object file into a linear memory.
pub data_align: Option<u64>,
/// Total size of all data pushed onto `data` so far.
total_data: u32,
/// List of passive element segments found in this module which will get
/// concatenated for the final artifact.
pub passive_data: Vec<&'data [u8]>,
/// Total size of all passive data pushed into `passive_data` so far.
total_passive_data: u32,
/// When we're parsing the code section this will be incremented so we know
/// which function is currently being defined.
code_index: u32,
/// The type information of the current module made available at the end of the
/// validation process.
types: Option<Types>,
}
impl<'data> ModuleTranslation<'data> {
/// Returns a reference to the type information of the current module.
pub fn get_types(&self) -> &Types {
self.types
.as_ref()
.expect("module type information to be available")
}
}
/// Contains function data: byte code and its offset in the module.
pub struct FunctionBodyData<'a> {
/// The body of the function, containing code and locals.
pub body: FunctionBody<'a>,
/// Validator for the function body
pub validator: FuncToValidate<ValidatorResources>,
}
#[derive(Debug, Default)]
#[allow(missing_docs)]
pub struct DebugInfoData<'a> {
pub dwarf: Dwarf<'a>,
pub name_section: NameSection<'a>,
pub wasm_file: WasmFileInfo,
pub debug_loc: gimli::DebugLoc<Reader<'a>>,
pub debug_loclists: gimli::DebugLocLists<Reader<'a>>,
pub debug_ranges: gimli::DebugRanges<Reader<'a>>,
pub debug_rnglists: gimli::DebugRngLists<Reader<'a>>,
pub debug_cu_index: gimli::DebugCuIndex<Reader<'a>>,
pub debug_tu_index: gimli::DebugTuIndex<Reader<'a>>,
}
#[allow(missing_docs)]
pub type Dwarf<'input> = gimli::Dwarf<Reader<'input>>;
type Reader<'input> = gimli::EndianSlice<'input, gimli::LittleEndian>;
#[derive(Debug, Default)]
#[allow(missing_docs)]
pub struct NameSection<'a> {
pub module_name: Option<&'a str>,
pub func_names: HashMap<FuncIndex, &'a str>,
pub locals_names: HashMap<FuncIndex, HashMap<u32, &'a str>>,
}
#[derive(Debug, Default)]
#[allow(missing_docs)]
pub struct WasmFileInfo {
pub path: Option<PathBuf>,
pub code_section_offset: u64,
pub imported_func_count: u32,
pub funcs: Vec<FunctionMetadata>,
}
#[derive(Debug)]
#[allow(missing_docs)]
pub struct FunctionMetadata {
pub params: Box<[WasmValType]>,
pub locals: Box<[(u32, WasmValType)]>,
}
impl<'a, 'data> ModuleEnvironment<'a, 'data> {
/// Allocates the environment data structures.
pub fn new(
tunables: &'a Tunables,
validator: &'a mut Validator,
types: &'a mut ModuleTypesBuilder,
) -> Self {
Self {
result: ModuleTranslation::default(),
types,
tunables,
validator,
}
}
/// Translate a wasm module using this environment.
///
/// This function will translate the `data` provided with `parser`,
/// validating everything along the way with this environment's validator.
///
/// The result of translation, [`ModuleTranslation`], contains everything
/// necessary to compile functions afterwards as well as learn type
/// information about the module at runtime.
pub fn translate(
mut self,
parser: Parser,
data: &'data [u8],
) -> Result<ModuleTranslation<'data>> {
self.result.wasm = data;
for payload in parser.parse_all(data) {
self.translate_payload(payload?)?;
}
Ok(self.result)
}
fn translate_payload(&mut self, payload: Payload<'data>) -> Result<()> {
match payload {
Payload::Version {
num,
encoding,
range,
} => {
self.validator.version(num, encoding, &range)?;
match encoding {
Encoding::Module => {}
Encoding::Component => {
bail!("expected a WebAssembly module but was given a WebAssembly component")
}
}
}
Payload::End(offset) => {
self.result.types = Some(self.validator.end(offset)?);
// With the `escaped_funcs` set of functions finished
// we can calculate the set of signatures that are exported as
// the set of exported functions' signatures.
self.result.exported_signatures = self
.result
.module
.functions
.iter()
.filter_map(|(_, func)| {
if func.is_escaping() {
Some(func.signature)
} else {
None
}
})
.collect();
self.result.exported_signatures.sort_unstable();
self.result.exported_signatures.dedup();
}
Payload::TypeSection(types) => {
self.validator.type_section(&types)?;
let count = self.validator.types(0).unwrap().core_type_count();
log::trace!("interning {count} Wasm types");
let capacity = usize::try_from(count).unwrap();
self.result.module.types.reserve(capacity);
self.types.reserve_wasm_signatures(capacity);
// Iterate over each *rec group* -- not type -- defined in the
// types section. Rec groups are the unit of canonicalization
// and therefore the unit at which we need to process at a
// time. `wasmparser` has already done the hard work of
// de-duplicating and canonicalizing the rec groups within the
// module for us, we just need to translate them into our data
// structures. Note that, if the Wasm defines duplicate rec
// groups, we need copy the duplicates over (shallowly) as well,
// so that our types index space doesn't have holes.
let mut type_index = 0;
while type_index < count {
let validator_types = self.validator.types(0).unwrap();
// Get the rec group for the current type index, which is
// always the first type defined in a rec group.
log::trace!("looking up wasmparser type for index {type_index}");
let core_type_id = validator_types.core_type_at(type_index).unwrap_sub();
log::trace!(
" --> {core_type_id:?} = {:?}",
validator_types[core_type_id],
);
let rec_group_id = validator_types.rec_group_id_of(core_type_id);
debug_assert_eq!(
validator_types
.rec_group_elements(rec_group_id)
.position(|id| id == core_type_id),
Some(0)
);
// Intern the rec group and then fill in this module's types
// index space.
let interned = self.types.intern_rec_group(
&self.result.module,
validator_types,
rec_group_id,
)?;
let elems = self.types.rec_group_elements(interned);
let len = elems.len();
self.result.module.types.reserve(len);
for ty in elems {
self.result.module.types.push(ty);
}
// Advance `type_index` to the start of the next rec group.
type_index += u32::try_from(len).unwrap();
}
}
Payload::ImportSection(imports) => {
self.validator.import_section(&imports)?;
let cnt = usize::try_from(imports.count()).unwrap();
self.result.module.initializers.reserve(cnt);
for entry in imports {
let import = entry?;
let ty = match import.ty {
TypeRef::Func(index) => {
let index = TypeIndex::from_u32(index);
let interned_index = self.result.module.types[index];
self.result.module.num_imported_funcs += 1;
self.result.debuginfo.wasm_file.imported_func_count += 1;
EntityType::Function(EngineOrModuleTypeIndex::Module(interned_index))
}
TypeRef::Memory(ty) => {
self.result.module.num_imported_memories += 1;
EntityType::Memory(ty.into())
}
TypeRef::Global(ty) => {
self.result.module.num_imported_globals += 1;
EntityType::Global(self.convert_global_type(&ty))
}
TypeRef::Table(ty) => {
self.result.module.num_imported_tables += 1;
EntityType::Table(self.convert_table_type(&ty)?)
}
// doesn't get past validation
TypeRef::Tag(_) => unreachable!(),
};
self.declare_import(import.module, import.name, ty);
}
}
Payload::FunctionSection(functions) => {
self.validator.function_section(&functions)?;
let cnt = usize::try_from(functions.count()).unwrap();
self.result.module.functions.reserve_exact(cnt);
for entry in functions {
let sigindex = entry?;
let ty = TypeIndex::from_u32(sigindex);
let interned_index = self.result.module.types[ty];
self.result.module.push_function(interned_index);
}
}
Payload::TableSection(tables) => {
self.validator.table_section(&tables)?;
let cnt = usize::try_from(tables.count()).unwrap();
self.result.module.table_plans.reserve_exact(cnt);
for entry in tables {
let wasmparser::Table { ty, init } = entry?;
let table = self.convert_table_type(&ty)?;
let plan = TablePlan::for_table(table, &self.tunables);
self.result.module.table_plans.push(plan);
let init = match init {
wasmparser::TableInit::RefNull => TableInitialValue::Null {
precomputed: Vec::new(),
},
wasmparser::TableInit::Expr(expr) => {
let (init, escaped) = ConstExpr::from_wasmparser(expr)?;
for f in escaped {
self.flag_func_escaped(f);
}
TableInitialValue::Expr(init)
}
};
self.result
.module
.table_initialization
.initial_values
.push(init);
}
}
Payload::MemorySection(memories) => {
self.validator.memory_section(&memories)?;
let cnt = usize::try_from(memories.count()).unwrap();
self.result.module.memory_plans.reserve_exact(cnt);
for entry in memories {
let memory = entry?;
let plan = MemoryPlan::for_memory(memory.into(), &self.tunables);
self.result.module.memory_plans.push(plan);
}
}
Payload::TagSection(tags) => {
self.validator.tag_section(&tags)?;
// This feature isn't enabled at this time, so we should
// never get here.
unreachable!();
}
Payload::GlobalSection(globals) => {
self.validator.global_section(&globals)?;
let cnt = usize::try_from(globals.count()).unwrap();
self.result.module.globals.reserve_exact(cnt);
for entry in globals {
let wasmparser::Global { ty, init_expr } = entry?;
let (initializer, escaped) = ConstExpr::from_wasmparser(init_expr)?;
for f in escaped {
self.flag_func_escaped(f);
}
let ty = self.convert_global_type(&ty);
self.result.module.globals.push(ty);
self.result.module.global_initializers.push(initializer);
}
}
Payload::ExportSection(exports) => {
self.validator.export_section(&exports)?;
let cnt = usize::try_from(exports.count()).unwrap();
self.result.module.exports.reserve(cnt);
for entry in exports {
let wasmparser::Export { name, kind, index } = entry?;
let entity = match kind {
ExternalKind::Func => {
let index = FuncIndex::from_u32(index);
self.flag_func_escaped(index);
EntityIndex::Function(index)
}
ExternalKind::Table => EntityIndex::Table(TableIndex::from_u32(index)),
ExternalKind::Memory => EntityIndex::Memory(MemoryIndex::from_u32(index)),
ExternalKind::Global => EntityIndex::Global(GlobalIndex::from_u32(index)),
// this never gets past validation
ExternalKind::Tag => unreachable!(),
};
self.result
.module
.exports
.insert(String::from(name), entity);
}
}
Payload::StartSection { func, range } => {
self.validator.start_section(func, &range)?;
let func_index = FuncIndex::from_u32(func);
self.flag_func_escaped(func_index);
debug_assert!(self.result.module.start_func.is_none());
self.result.module.start_func = Some(func_index);
}
Payload::ElementSection(elements) => {
self.validator.element_section(&elements)?;
for (index, entry) in elements.into_iter().enumerate() {
let wasmparser::Element {
kind,
items,
range: _,
} = entry?;
// Build up a list of `FuncIndex` corresponding to all the
// entries listed in this segment. Note that it's not
// possible to create anything other than a `ref.null
// extern` for externref segments, so those just get
// translated to the reserved value of `FuncIndex`.
let elements = match items {
ElementItems::Functions(funcs) => {
let mut elems =
Vec::with_capacity(usize::try_from(funcs.count()).unwrap());
for func in funcs {
let func = FuncIndex::from_u32(func?);
self.flag_func_escaped(func);
elems.push(func);
}
TableSegmentElements::Functions(elems.into())
}
ElementItems::Expressions(_ty, items) => {
let mut exprs =
Vec::with_capacity(usize::try_from(items.count()).unwrap());
for expr in items {
let (expr, escaped) = ConstExpr::from_wasmparser(expr?)?;
exprs.push(expr);
for func in escaped {
self.flag_func_escaped(func);
}
}
TableSegmentElements::Expressions(exprs.into())
}
};
match kind {
ElementKind::Active {
table_index,
offset_expr,
} => {
let table_index = TableIndex::from_u32(table_index.unwrap_or(0));
let (offset, escaped) = ConstExpr::from_wasmparser(offset_expr)?;
debug_assert!(escaped.is_empty());
self.result
.module
.table_initialization
.segments
.push(TableSegment {
table_index,
offset,
elements: elements.into(),
});
}
ElementKind::Passive => {
let elem_index = ElemIndex::from_u32(index as u32);
let index = self.result.module.passive_elements.len();
self.result.module.passive_elements.push(elements.into());
self.result
.module
.passive_elements_map
.insert(elem_index, index);
}
ElementKind::Declared => {}
}
}
}
Payload::CodeSectionStart { count, range, .. } => {
self.validator.code_section_start(count, &range)?;
let cnt = usize::try_from(count).unwrap();
self.result.function_body_inputs.reserve_exact(cnt);
self.result.debuginfo.wasm_file.code_section_offset = range.start as u64;
}
Payload::CodeSectionEntry(body) => {
let validator = self.validator.code_section_entry(&body)?;
let func_index =
self.result.code_index + self.result.module.num_imported_funcs as u32;
let func_index = FuncIndex::from_u32(func_index);
if self.tunables.generate_native_debuginfo {
let sig_index = self.result.module.functions[func_index].signature;
let sig = self.types[sig_index].unwrap_func();
let mut locals = Vec::new();
for pair in body.get_locals_reader()? {
let (cnt, ty) = pair?;
let ty = self.convert_valtype(ty);
locals.push((cnt, ty));
}
self.result
.debuginfo
.wasm_file
.funcs
.push(FunctionMetadata {
locals: locals.into_boxed_slice(),
params: sig.params().into(),
});
}
self.result
.function_body_inputs
.push(FunctionBodyData { validator, body });
self.result.code_index += 1;
}
Payload::DataSection(data) => {
self.validator.data_section(&data)?;
let initializers = match &mut self.result.module.memory_initialization {
MemoryInitialization::Segmented(i) => i,
_ => unreachable!(),
};
let cnt = usize::try_from(data.count()).unwrap();
initializers.reserve_exact(cnt);
self.result.data.reserve_exact(cnt);
for (index, entry) in data.into_iter().enumerate() {
let wasmparser::Data {
kind,
data,
range: _,
} = entry?;
let mk_range = |total: &mut u32| -> Result<_, WasmError> {
let range = u32::try_from(data.len())
.ok()
.and_then(|size| {
let start = *total;
let end = start.checked_add(size)?;
Some(start..end)
})
.ok_or_else(|| {
WasmError::Unsupported(format!(
"more than 4 gigabytes of data in wasm module",
))
})?;
*total += range.end - range.start;
Ok(range)
};
match kind {
DataKind::Active {
memory_index,
offset_expr,
} => {
let range = mk_range(&mut self.result.total_data)?;
let memory_index = MemoryIndex::from_u32(memory_index);
let (offset, escaped) = ConstExpr::from_wasmparser(offset_expr)?;
debug_assert!(escaped.is_empty());
initializers.push(MemoryInitializer {
memory_index,
offset,
data: range,
});
self.result.data.push(data.into());
}
DataKind::Passive => {
let data_index = DataIndex::from_u32(index as u32);
let range = mk_range(&mut self.result.total_passive_data)?;
self.result.passive_data.push(data);
self.result
.module
.passive_data_map
.insert(data_index, range);
}
}
}
}
Payload::DataCountSection { count, range } => {
self.validator.data_count_section(count, &range)?;
// Note: the count passed in here is the *total* segment count
// There is no way to reserve for just the passive segments as
// they are discovered when iterating the data section entries
// Given that the total segment count might be much larger than
// the passive count, do not reserve anything here.
}
Payload::CustomSection(s)
if s.name() == "webidl-bindings" || s.name() == "wasm-interface-types" =>
{
bail!(
"\
Support for interface types has temporarily been removed from `wasmtime`.
For more information about this temporary change you can read on the issue online:
https://github.com/bytecodealliance/wasmtime/issues/1271
and for re-adding support for interface types you can see this issue:
https://github.com/bytecodealliance/wasmtime/issues/677
"
)
}
Payload::CustomSection(s) => {
self.register_custom_section(&s);
}
// It's expected that validation will probably reject other
// payloads such as `UnknownSection` or those related to the
// component model. If, however, something gets past validation then
// that's a bug in Wasmtime as we forgot to implement something.
other => {
self.validator.payload(&other)?;
panic!("unimplemented section in wasm file {other:?}");
}
}
Ok(())
}
fn register_custom_section(&mut self, section: &CustomSectionReader<'data>) {
match section.as_known() {
KnownCustom::Name(name) => {
let result = self.name_section(name);
if let Err(e) = result {
log::warn!("failed to parse name section {:?}", e);
}
}
_ => {
let name = section.name().trim_end_matches(".dwo");
if name.starts_with(".debug_") {
self.dwarf_section(name, section);
}
}
}
}
fn dwarf_section(&mut self, name: &str, section: &CustomSectionReader<'data>) {
if !self.tunables.generate_native_debuginfo && !self.tunables.parse_wasm_debuginfo {
self.result.has_unparsed_debuginfo = true;
return;
}
let info = &mut self.result.debuginfo;
let dwarf = &mut info.dwarf;
let endian = gimli::LittleEndian;
let data = section.data();
let slice = gimli::EndianSlice::new(data, endian);
match name {
// `gimli::Dwarf` fields.
".debug_abbrev" => dwarf.debug_abbrev = gimli::DebugAbbrev::new(data, endian),
".debug_addr" => dwarf.debug_addr = gimli::DebugAddr::from(slice),
".debug_info" => {
dwarf.debug_info = gimli::DebugInfo::new(data, endian);
}
".debug_line" => dwarf.debug_line = gimli::DebugLine::new(data, endian),
".debug_line_str" => dwarf.debug_line_str = gimli::DebugLineStr::from(slice),
".debug_str" => dwarf.debug_str = gimli::DebugStr::new(data, endian),
".debug_str_offsets" => dwarf.debug_str_offsets = gimli::DebugStrOffsets::from(slice),
".debug_str_sup" => {
let mut dwarf_sup: Dwarf<'data> = Default::default();
dwarf_sup.debug_str = gimli::DebugStr::from(slice);
dwarf.sup = Some(Arc::new(dwarf_sup));
}
".debug_types" => dwarf.debug_types = gimli::DebugTypes::from(slice),
// Additional fields.
".debug_loc" => info.debug_loc = gimli::DebugLoc::from(slice),
".debug_loclists" => info.debug_loclists = gimli::DebugLocLists::from(slice),
".debug_ranges" => info.debug_ranges = gimli::DebugRanges::new(data, endian),
".debug_rnglists" => info.debug_rnglists = gimli::DebugRngLists::new(data, endian),
// DWARF package fields
".debug_cu_index" => info.debug_cu_index = gimli::DebugCuIndex::new(data, endian),
".debug_tu_index" => info.debug_tu_index = gimli::DebugTuIndex::new(data, endian),
// We don't use these at the moment.
".debug_aranges" | ".debug_pubnames" | ".debug_pubtypes" => return,
other => {
log::warn!("unknown debug section `{}`", other);
return;
}
}
dwarf.ranges = gimli::RangeLists::new(info.debug_ranges, info.debug_rnglists);
dwarf.locations = gimli::LocationLists::new(info.debug_loc, info.debug_loclists);
}
/// Declares a new import with the `module` and `field` names, importing the
/// `ty` specified.
///
/// Note that this method is somewhat tricky due to the implementation of
/// the module linking proposal. In the module linking proposal two-level
/// imports are recast as single-level imports of instances. That recasting
/// happens here by recording an import of an instance for the first time
/// we see a two-level import.
///
/// When the module linking proposal is disabled, however, disregard this
/// logic and instead work directly with two-level imports since no
/// instances are defined.
fn declare_import(&mut self, module: &'data str, field: &'data str, ty: EntityType) {
let index = self.push_type(ty);
self.result.module.initializers.push(Initializer::Import {
name: module.to_owned(),
field: field.to_owned(),
index,
});
}
fn push_type(&mut self, ty: EntityType) -> EntityIndex {
match ty {
EntityType::Function(ty) => EntityIndex::Function({
let func_index = self
.result
.module
.push_function(ty.unwrap_module_type_index());
// Imported functions can escape; in fact, they've already done
// so to get here.
self.flag_func_escaped(func_index);
func_index
}),
EntityType::Table(ty) => {
let plan = TablePlan::for_table(ty, &self.tunables);
EntityIndex::Table(self.result.module.table_plans.push(plan))
}
EntityType::Memory(ty) => {
let plan = MemoryPlan::for_memory(ty, &self.tunables);
EntityIndex::Memory(self.result.module.memory_plans.push(plan))
}
EntityType::Global(ty) => EntityIndex::Global(self.result.module.globals.push(ty)),
EntityType::Tag(_) => unimplemented!(),
}
}
fn flag_func_escaped(&mut self, func: FuncIndex) {
let ty = &mut self.result.module.functions[func];
// If this was already assigned a funcref index no need to re-assign it.
if ty.is_escaping() {
return;
}
let index = self.result.module.num_escaped_funcs as u32;
ty.func_ref = FuncRefIndex::from_u32(index);
self.result.module.num_escaped_funcs += 1;
}
/// Parses the Name section of the wasm module.
fn name_section(&mut self, names: NameSectionReader<'data>) -> WasmResult<()> {
for subsection in names {
match subsection? {
wasmparser::Name::Function(names) => {
for name in names {
let Naming { index, name } = name?;
// Skip this naming if it's naming a function that
// doesn't actually exist.
if (index as usize) >= self.result.module.functions.len() {
continue;
}
// Store the name unconditionally, regardless of
// whether we're parsing debuginfo, since function
// names are almost always present in the
// final compilation artifact.
let index = FuncIndex::from_u32(index);
self.result
.debuginfo
.name_section
.func_names
.insert(index, name);
}
}
wasmparser::Name::Module { name, .. } => {
self.result.module.name = Some(name.to_string());
if self.tunables.generate_native_debuginfo {
self.result.debuginfo.name_section.module_name = Some(name);
}
}
wasmparser::Name::Local(reader) => {
if !self.tunables.generate_native_debuginfo {
continue;
}
for f in reader {
let f = f?;
// Skip this naming if it's naming a function that
// doesn't actually exist.
if (f.index as usize) >= self.result.module.functions.len() {
continue;
}
for name in f.names {
let Naming { index, name } = name?;
self.result
.debuginfo
.name_section
.locals_names
.entry(FuncIndex::from_u32(f.index))
.or_insert(HashMap::new())
.insert(index, name);
}
}
}
wasmparser::Name::Label(_)
| wasmparser::Name::Type(_)
| wasmparser::Name::Table(_)
| wasmparser::Name::Global(_)
| wasmparser::Name::Memory(_)
| wasmparser::Name::Element(_)
| wasmparser::Name::Data(_)
| wasmparser::Name::Tag(_)
| wasmparser::Name::Field(_)
| wasmparser::Name::Unknown { .. } => {}
}
}
Ok(())
}
}
impl TypeConvert for ModuleEnvironment<'_, '_> {
fn lookup_heap_type(&self, index: wasmparser::UnpackedIndex) -> WasmHeapType {
WasmparserTypeConverter::new(&self.types, &self.result.module).lookup_heap_type(index)
}
fn lookup_type_index(&self, index: wasmparser::UnpackedIndex) -> EngineOrModuleTypeIndex {
WasmparserTypeConverter::new(&self.types, &self.result.module).lookup_type_index(index)
}
}
impl ModuleTranslation<'_> {
/// Attempts to convert segmented memory initialization into static
/// initialization for the module that this translation represents.
///
/// If this module's memory initialization is not compatible with paged
/// initialization then this won't change anything. Otherwise if it is
/// compatible then the `memory_initialization` field will be updated.
///
/// Takes a `page_size` argument in order to ensure that all
/// initialization is page-aligned for mmap-ability, and
/// `max_image_size_always_allowed` to control how we decide
/// whether to use static init.
///
/// We will try to avoid generating very sparse images, which are
/// possible if e.g. a module has an initializer at offset 0 and a
/// very high offset (say, 1 GiB). To avoid this, we use a dual
/// condition: we always allow images less than
/// `max_image_size_always_allowed`, and the embedder of Wasmtime
/// can set this if desired to ensure that static init should
/// always be done if the size of the module or its heaps is
/// otherwise bounded by the system. We also allow images with
/// static init data bigger than that, but only if it is "dense",
/// defined as having at least half (50%) of its pages with some
/// data.
///
/// We could do something slightly better by building a dense part
/// and keeping a sparse list of outlier/leftover segments (see
/// issue #3820). This would also allow mostly-static init of
/// modules that have some dynamically-placed data segments. But,
/// for now, this is sufficient to allow a system that "knows what
/// it's doing" to always get static init.
pub fn try_static_init(&mut self, page_size: u64, max_image_size_always_allowed: u64) {
// This method only attempts to transform a `Segmented` memory init
// into a `Static` one, no other state.
if !self.module.memory_initialization.is_segmented() {
return;
}
// First a dry run of memory initialization is performed. This
// collects information about the extent of memory initialized for each
// memory as well as the size of all data segments being copied in.
struct Memory {
data_size: u64,
min_addr: u64,
max_addr: u64,
// The `usize` here is a pointer into `self.data` which is the list
// of data segments corresponding to what was found in the original
// wasm module.
segments: Vec<(usize, StaticMemoryInitializer)>,
}
let mut info = PrimaryMap::with_capacity(self.module.memory_plans.len());
for _ in 0..self.module.memory_plans.len() {
info.push(Memory {
data_size: 0,
min_addr: u64::MAX,
max_addr: 0,
segments: Vec::new(),
});
}
struct InitMemoryAtCompileTime<'a> {
module: &'a Module,
info: &'a mut PrimaryMap<MemoryIndex, Memory>,
idx: usize,
}
impl InitMemory for InitMemoryAtCompileTime<'_> {
fn memory_size_in_bytes(
&mut self,
memory_index: MemoryIndex,
) -> Result<u64, SizeOverflow> {
self.module.memory_plans[memory_index]
.memory
.minimum_byte_size()
}
fn eval_offset(&mut self, memory_index: MemoryIndex, expr: &ConstExpr) -> Option<u64> {
match (
expr.ops(),
self.module.memory_plans[memory_index].memory.idx_type,
) {
(&[ConstOp::I32Const(offset)], IndexType::I32) => {
Some(offset.unsigned().into())
}
(&[ConstOp::I64Const(offset)], IndexType::I64) => Some(offset.unsigned()),
_ => None,
}
}
fn write(&mut self, memory: MemoryIndex, init: &StaticMemoryInitializer) -> bool {
// Currently `Static` only applies to locally-defined memories,
// so if a data segment references an imported memory then
// transitioning to a `Static` memory initializer is not
// possible.
if self.module.defined_memory_index(memory).is_none() {
return false;
};
let info = &mut self.info[memory];
let data_len = u64::from(init.data.end - init.data.start);
if data_len > 0 {
info.data_size += data_len;
info.min_addr = info.min_addr.min(init.offset);
info.max_addr = info.max_addr.max(init.offset + data_len);
info.segments.push((self.idx, init.clone()));
}
self.idx += 1;
true
}
}
let ok = self
.module
.memory_initialization
.init_memory(&mut InitMemoryAtCompileTime {
idx: 0,
module: &self.module,
info: &mut info,
});
if !ok {
return;
}
// Validate that the memory information collected is indeed valid for
// static memory initialization.
for (i, info) in info.iter().filter(|(_, info)| info.data_size > 0) {
let image_size = info.max_addr - info.min_addr;
// Simplify things for now by bailing out entirely if any memory has
// a page size smaller than the host's page size. This fixes a case
// where currently initializers are created in host-page-size units
// of length which means that a larger-than-the-entire-memory
// initializer can be created. This can be handled technically but
// would require some more changes to help fix the assert elsewhere
// that this protects against.
if self.module.memory_plans[i].memory.page_size() < page_size {
return;
}
// If the range of memory being initialized is less than twice the
// total size of the data itself then it's assumed that static
// initialization is ok. This means we'll at most double memory
// consumption during the memory image creation process, which is
// currently assumed to "probably be ok" but this will likely need
// tweaks over time.
if image_size < info.data_size.saturating_mul(2) {
continue;
}
// If the memory initialization image is larger than the size of all
// data, then we still allow memory initialization if the image will
// be of a relatively modest size, such as 1MB here.
if image_size < max_image_size_always_allowed {
continue;
}
// At this point memory initialization is concluded to be too
// expensive to do at compile time so it's entirely deferred to
// happen at runtime.
return;
}
// Here's where we've now committed to changing to static memory. The
// memory initialization image is built here from the page data and then
// it's converted to a single initializer.
let data = mem::replace(&mut self.data, Vec::new());
let mut map = PrimaryMap::with_capacity(info.len());
let mut module_data_size = 0u32;
for (memory, info) in info.iter() {
// Create the in-memory `image` which is the initialized contents of
// this linear memory.
let extent = if info.segments.len() > 0 {
(info.max_addr - info.min_addr) as usize
} else {
0
};
let mut image = Vec::with_capacity(extent);
for (idx, init) in info.segments.iter() {
let data = &data[*idx];
assert_eq!(data.len(), init.data.len());
let offset = usize::try_from(init.offset - info.min_addr).unwrap();
if image.len() < offset {
image.resize(offset, 0u8);
image.extend_from_slice(data);
} else {
image.splice(
offset..(offset + data.len()).min(image.len()),
data.iter().copied(),
);
}
}
assert_eq!(image.len(), extent);
assert_eq!(image.capacity(), extent);
let mut offset = if info.segments.len() > 0 {
info.min_addr
} else {
0
};
// Chop off trailing zeros from the image as memory is already
// zero-initialized. Note that `i` is the position of a nonzero
// entry here, so to not lose it we truncate to `i + 1`.
if let Some(i) = image.iter().rposition(|i| *i != 0) {
image.truncate(i + 1);
}
// Also chop off leading zeros, if any.
if let Some(i) = image.iter().position(|i| *i != 0) {
offset += i as u64;
image.drain(..i);
}
let mut len = u64::try_from(image.len()).unwrap();
// The goal is to enable mapping this image directly into memory, so
// the offset into linear memory must be a multiple of the page
// size. If that's not already the case then the image is padded at
// the front and back with extra zeros as necessary
if offset % page_size != 0 {
let zero_padding = offset % page_size;
self.data.push(vec![0; zero_padding as usize].into());
offset -= zero_padding;
len += zero_padding;
}
self.data.push(image.into());
if len % page_size != 0 {
let zero_padding = page_size - (len % page_size);
self.data.push(vec![0; zero_padding as usize].into());
len += zero_padding;
}
// Offset/length should now always be page-aligned.
assert!(offset % page_size == 0);
assert!(len % page_size == 0);
// Create the `StaticMemoryInitializer` which describes this image,
// only needed if the image is actually present and has a nonzero
// length. The `offset` has been calculates above, originally
// sourced from `info.min_addr`. The `data` field is the extent
// within the final data segment we'll emit to an ELF image, which
// is the concatenation of `self.data`, so here it's the size of
// the section-so-far plus the current segment we're appending.
let len = u32::try_from(len).unwrap();
let init = if len > 0 {
Some(StaticMemoryInitializer {
offset,
data: module_data_size..module_data_size + len,
})
} else {
None
};
let idx = map.push(init);
assert_eq!(idx, memory);
module_data_size += len;
}
self.data_align = Some(page_size);
self.module.memory_initialization = MemoryInitialization::Static { map };
}
/// Attempts to convert the module's table initializers to
/// FuncTable form where possible. This enables lazy table
/// initialization later by providing a one-to-one map of initial
/// table values, without having to parse all segments.
pub fn try_func_table_init(&mut self) {
// This should be large enough to support very large Wasm
// modules with huge funcref tables, but small enough to avoid
// OOMs or DoS on truly sparse tables.
const MAX_FUNC_TABLE_SIZE: u64 = 1024 * 1024;
// First convert any element-initialized tables to images of just that
// single function if the minimum size of the table allows doing so.
for ((_, init), (_, plan)) in self
.module
.table_initialization
.initial_values
.iter_mut()
.zip(
self.module
.table_plans
.iter()
.skip(self.module.num_imported_tables),
)
{
let table_size = plan.table.limits.min;
if table_size > MAX_FUNC_TABLE_SIZE {
continue;
}
if let TableInitialValue::Expr(expr) = init {
if let [ConstOp::RefFunc(f)] = expr.ops() {
*init = TableInitialValue::Null {
precomputed: vec![*f; table_size as usize],
};
}
}
}
let mut segments = mem::take(&mut self.module.table_initialization.segments)
.into_iter()
.peekable();
// The goal of this loop is to interpret a table segment and apply it
// "statically" to a local table. This will iterate over segments and
// apply them one-by-one to each table.
//
// If any segment can't be applied, however, then this loop exits and
// all remaining segments are placed back into the segment list. This is
// because segments are supposed to be initialized one-at-a-time which
// means that intermediate state is visible with respect to traps. If
// anything isn't statically known to not trap it's pessimistically
// assumed to trap meaning all further segment initializers must be
// applied manually at instantiation time.
while let Some(segment) = segments.peek() {
let defined_index = match self.module.defined_table_index(segment.table_index) {
Some(index) => index,
// Skip imported tables: we can't provide a preconstructed
// table for them, because their values depend on the
// imported table overlaid with whatever segments we have.
None => break,
};
// If the base of this segment is dynamic, then we can't
// include it in the statically-built array of initial
// contents.
let offset = match segment.offset.ops() {
&[ConstOp::I32Const(offset)] => u64::from(offset.unsigned()),
&[ConstOp::I64Const(offset)] => offset.unsigned(),
_ => break,
};
// Get the end of this segment. If out-of-bounds, or too
// large for our dense table representation, then skip the
// segment.
let top = match offset.checked_add(segment.elements.len()) {
Some(top) => top,
None => break,
};
let table_size = self.module.table_plans[segment.table_index]
.table
.limits
.min;
if top > table_size || top > MAX_FUNC_TABLE_SIZE {
break;
}
match self.module.table_plans[segment.table_index]
.table
.ref_type
.heap_type
.top()
{
WasmHeapTopType::Func => {}
// If this is not a funcref table, then we can't support a
// pre-computed table of function indices. Technically this
// initializer won't trap so we could continue processing
// segments, but that's left as a future optimization if
// necessary.
WasmHeapTopType::Any | WasmHeapTopType::Extern => break,
}
// Function indices can be optimized here, but fully general
// expressions are deferred to get evaluated at runtime.
let function_elements = match &segment.elements {
TableSegmentElements::Functions(indices) => indices,
TableSegmentElements::Expressions(_) => break,
};
let precomputed =
match &mut self.module.table_initialization.initial_values[defined_index] {
TableInitialValue::Null { precomputed } => precomputed,
// If this table is still listed as an initial value here
// then that means the initial size of the table doesn't
// support a precomputed function list, so skip this.
// Technically this won't trap so it's possible to process
// further initializers, but that's left as a future
// optimization.
TableInitialValue::Expr(_) => break,
};
// At this point we're committing to pre-initializing the table
// with the `segment` that's being iterated over. This segment is
// applied to the `precomputed` list for the table by ensuring
// it's large enough to hold the segment and then copying the
// segment into the precomputed list.
if precomputed.len() < top as usize {
precomputed.resize(top as usize, FuncIndex::reserved_value());
}
let dst = &mut precomputed[offset as usize..top as usize];
dst.copy_from_slice(&function_elements);
// advance the iterator to see the next segment
let _ = segments.next();
}
self.module.table_initialization.segments = segments.collect();
}
}