1//===- Wasm.h - Wasm object file implementation -----------------*- C++ -*-===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8//
9// This file declares the WasmObjectFile class, which implements the ObjectFile
10// interface for Wasm files.
11//
12// See: https://github.com/WebAssembly/design/blob/main/BinaryEncoding.md
13//
14//===----------------------------------------------------------------------===//
15
16#ifndef LLVM_OBJECT_WASM_H
17#define LLVM_OBJECT_WASM_H
18
19#include "llvm/ADT/ArrayRef.h"
20#include "llvm/ADT/StringRef.h"
21#include "llvm/BinaryFormat/Wasm.h"
22#include "llvm/Config/llvm-config.h"
23#include "llvm/MC/MCSymbolWasm.h"
24#include "llvm/Object/Binary.h"
25#include "llvm/Object/ObjectFile.h"
26#include "llvm/Support/Compiler.h"
27#include "llvm/Support/Error.h"
28#include "llvm/Support/MemoryBuffer.h"
29#include <cstddef>
30#include <cstdint>
31#include <vector>
32
33namespace llvm {
34namespace object {
35
36class WasmSymbol {
37public:
38 WasmSymbol(const wasm::WasmSymbolInfo &Info,
39 const wasm::WasmGlobalType *GlobalType,
40 const wasm::WasmTableType *TableType,
41 const wasm::WasmSignature *Signature)
42 : Info(Info), GlobalType(GlobalType), TableType(TableType),
43 Signature(Signature) {
44 assert(!Signature || Signature->Kind != wasm::WasmSignature::Placeholder);
45 }
46
47 // Symbol info as represented in the symbol's 'syminfo' entry of an object
48 // file's symbol table.
49 wasm::WasmSymbolInfo Info;
50 const wasm::WasmGlobalType *GlobalType;
51 const wasm::WasmTableType *TableType;
52 const wasm::WasmSignature *Signature;
53
54 bool isTypeFunction() const {
55 return Info.Kind == wasm::WASM_SYMBOL_TYPE_FUNCTION;
56 }
57
58 bool isTypeTable() const { return Info.Kind == wasm::WASM_SYMBOL_TYPE_TABLE; }
59
60 bool isTypeData() const { return Info.Kind == wasm::WASM_SYMBOL_TYPE_DATA; }
61
62 bool isTypeGlobal() const {
63 return Info.Kind == wasm::WASM_SYMBOL_TYPE_GLOBAL;
64 }
65
66 bool isTypeSection() const {
67 return Info.Kind == wasm::WASM_SYMBOL_TYPE_SECTION;
68 }
69
70 bool isTypeTag() const { return Info.Kind == wasm::WASM_SYMBOL_TYPE_TAG; }
71
72 bool isDefined() const { return !isUndefined(); }
73
74 bool isUndefined() const {
75 return (Info.Flags & wasm::WASM_SYMBOL_UNDEFINED) != 0;
76 }
77
78 bool isBindingWeak() const {
79 return getBinding() == wasm::WASM_SYMBOL_BINDING_WEAK;
80 }
81
82 bool isBindingGlobal() const {
83 return getBinding() == wasm::WASM_SYMBOL_BINDING_GLOBAL;
84 }
85
86 bool isBindingLocal() const {
87 return getBinding() == wasm::WASM_SYMBOL_BINDING_LOCAL;
88 }
89
90 unsigned getBinding() const {
91 return Info.Flags & wasm::WASM_SYMBOL_BINDING_MASK;
92 }
93
94 bool isHidden() const {
95 return getVisibility() == wasm::WASM_SYMBOL_VISIBILITY_HIDDEN;
96 }
97
98 unsigned getVisibility() const {
99 return Info.Flags & wasm::WASM_SYMBOL_VISIBILITY_MASK;
100 }
101
102 LLVM_ABI void print(raw_ostream &Out) const;
103
104#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
105 LLVM_DUMP_METHOD void dump() const;
106#endif
107};
108
109struct WasmSection {
110 WasmSection() = default;
111
112 uint32_t Type = 0;
113 uint32_t Offset = 0; // Offset within the file
114 StringRef Name; // Section name (User-defined sections only)
115 uint32_t Comdat = UINT32_MAX; // From the "comdat info" section
116 ArrayRef<uint8_t> Content;
117 std::vector<wasm::WasmRelocation> Relocations;
118 // Length of the LEB encoding of the section header's size field
119 std::optional<uint8_t> HeaderSecSizeEncodingLen;
120};
121
122struct WasmSegment {
123 uint32_t SectionOffset;
124 wasm::WasmDataSegment Data;
125};
126
127class LLVM_ABI WasmObjectFile : public ObjectFile {
128
129public:
130 WasmObjectFile(MemoryBufferRef Object, Error &Err);
131
132 const wasm::WasmObjectHeader &getHeader() const;
133 const WasmSymbol &getWasmSymbol(const DataRefImpl &Symb) const;
134 const WasmSymbol &getWasmSymbol(const SymbolRef &Symbol) const;
135 const WasmSection &getWasmSection(const SectionRef &Section) const;
136 const wasm::WasmRelocation &getWasmRelocation(const RelocationRef &Ref) const;
137
138 static bool classof(const Binary *v) { return v->isWasm(); }
139
140 const wasm::WasmDylinkInfo &dylinkInfo() const { return DylinkInfo; }
141 const wasm::WasmProducerInfo &getProducerInfo() const { return ProducerInfo; }
142 ArrayRef<wasm::WasmFeatureEntry> getTargetFeatures() const {
143 return TargetFeatures;
144 }
145 ArrayRef<wasm::WasmSignature> types() const { return Signatures; }
146 ArrayRef<wasm::WasmImport> imports() const { return Imports; }
147 ArrayRef<wasm::WasmTable> tables() const { return Tables; }
148 ArrayRef<wasm::WasmLimits> memories() const { return Memories; }
149 ArrayRef<wasm::WasmGlobal> globals() const { return Globals; }
150 ArrayRef<wasm::WasmTag> tags() const { return Tags; }
151 ArrayRef<wasm::WasmExport> exports() const { return Exports; }
152 const wasm::WasmLinkingData &linkingData() const { return LinkingData; }
153 uint32_t getNumberOfSymbols() const { return Symbols.size(); }
154 ArrayRef<wasm::WasmElemSegment> elements() const { return ElemSegments; }
155 ArrayRef<WasmSegment> dataSegments() const { return DataSegments; }
156 ArrayRef<wasm::WasmFunction> functions() const { return Functions; }
157 ArrayRef<wasm::WasmDebugName> debugNames() const { return DebugNames; }
158 uint32_t startFunction() const { return StartFunction; }
159 uint32_t getNumImportedGlobals() const { return NumImportedGlobals; }
160 uint32_t getNumImportedTables() const { return NumImportedTables; }
161 uint32_t getNumImportedFunctions() const { return NumImportedFunctions; }
162 uint32_t getNumImportedTags() const { return NumImportedTags; }
163 uint32_t getNumSections() const { return Sections.size(); }
164 void moveSymbolNext(DataRefImpl &Symb) const override;
165
166 Expected<uint32_t> getSymbolFlags(DataRefImpl Symb) const override;
167
168 basic_symbol_iterator symbol_begin() const override;
169
170 basic_symbol_iterator symbol_end() const override;
171 Expected<StringRef> getSymbolName(DataRefImpl Symb) const override;
172
173 bool is64Bit() const override { return false; }
174
175 Expected<uint64_t> getSymbolAddress(DataRefImpl Symb) const override;
176 uint64_t getWasmSymbolValue(const WasmSymbol &Sym) const;
177 uint64_t getSymbolValueImpl(DataRefImpl Symb) const override;
178 uint32_t getSymbolAlignment(DataRefImpl Symb) const override;
179 uint64_t getCommonSymbolSizeImpl(DataRefImpl Symb) const override;
180 Expected<SymbolRef::Type> getSymbolType(DataRefImpl Symb) const override;
181 Expected<section_iterator> getSymbolSection(DataRefImpl Symb) const override;
182 uint32_t getSymbolSectionId(SymbolRef Sym) const;
183 uint32_t getSymbolSize(SymbolRef Sym) const;
184
185 // Overrides from SectionRef.
186 void moveSectionNext(DataRefImpl &Sec) const override;
187 Expected<StringRef> getSectionName(DataRefImpl Sec) const override;
188 uint64_t getSectionAddress(DataRefImpl Sec) const override;
189 uint64_t getSectionIndex(DataRefImpl Sec) const override;
190 uint64_t getSectionSize(DataRefImpl Sec) const override;
191 Expected<ArrayRef<uint8_t>>
192 getSectionContents(DataRefImpl Sec) const override;
193 uint64_t getSectionAlignment(DataRefImpl Sec) const override;
194 bool isSectionCompressed(DataRefImpl Sec) const override;
195 bool isSectionText(DataRefImpl Sec) const override;
196 bool isSectionData(DataRefImpl Sec) const override;
197 bool isSectionBSS(DataRefImpl Sec) const override;
198 bool isSectionVirtual(DataRefImpl Sec) const override;
199 relocation_iterator section_rel_begin(DataRefImpl Sec) const override;
200 relocation_iterator section_rel_end(DataRefImpl Sec) const override;
201
202 // Overrides from RelocationRef.
203 void moveRelocationNext(DataRefImpl &Rel) const override;
204 uint64_t getRelocationOffset(DataRefImpl Rel) const override;
205 symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override;
206 uint64_t getRelocationType(DataRefImpl Rel) const override;
207 void getRelocationTypeName(DataRefImpl Rel,
208 SmallVectorImpl<char> &Result) const override;
209
210 section_iterator section_begin() const override;
211 section_iterator section_end() const override;
212 uint8_t getBytesInAddress() const override;
213 StringRef getFileFormatName() const override;
214 Triple::ArchType getArch() const override;
215 Expected<SubtargetFeatures> getFeatures() const override;
216 bool isRelocatableObject() const override;
217 bool isSharedObject() const;
218 bool hasUnmodeledTypes() const { return HasUnmodeledTypes; }
219
220 struct ReadContext {
221 const uint8_t *Start;
222 const uint8_t *Ptr;
223 const uint8_t *End;
224 };
225
226private:
227 bool isValidFunctionIndex(uint32_t Index) const;
228 bool isDefinedFunctionIndex(uint32_t Index) const;
229 bool isValidGlobalIndex(uint32_t Index) const;
230 bool isValidTableNumber(uint32_t Index) const;
231 bool isDefinedGlobalIndex(uint32_t Index) const;
232 bool isDefinedTableNumber(uint32_t Index) const;
233 bool isValidTagIndex(uint32_t Index) const;
234 bool isDefinedTagIndex(uint32_t Index) const;
235 bool isValidFunctionSymbol(uint32_t Index) const;
236 bool isValidTableSymbol(uint32_t Index) const;
237 bool isValidGlobalSymbol(uint32_t Index) const;
238 bool isValidTagSymbol(uint32_t Index) const;
239 bool isValidDataSymbol(uint32_t Index) const;
240 bool isValidSectionSymbol(uint32_t Index) const;
241 wasm::WasmFunction &getDefinedFunction(uint32_t Index);
242 const wasm::WasmFunction &getDefinedFunction(uint32_t Index) const;
243 const wasm::WasmGlobal &getDefinedGlobal(uint32_t Index) const;
244 wasm::WasmTag &getDefinedTag(uint32_t Index);
245
246 const WasmSection &getWasmSection(DataRefImpl Ref) const;
247 const wasm::WasmRelocation &getWasmRelocation(DataRefImpl Ref) const;
248 uint32_t getSymbolSectionIdImpl(const WasmSymbol &Symb) const;
249
250 Error parseSection(WasmSection &Sec);
251 Error parseCustomSection(WasmSection &Sec, ReadContext &Ctx);
252
253 // Standard section types
254 Error parseTypeSection(ReadContext &Ctx);
255 Error parseImportSection(ReadContext &Ctx);
256 Error parseFunctionSection(ReadContext &Ctx);
257 Error parseTableSection(ReadContext &Ctx);
258 Error parseMemorySection(ReadContext &Ctx);
259 Error parseTagSection(ReadContext &Ctx);
260 Error parseGlobalSection(ReadContext &Ctx);
261 Error parseExportSection(ReadContext &Ctx);
262 Error parseStartSection(ReadContext &Ctx);
263 Error parseElemSection(ReadContext &Ctx);
264 Error parseCodeSection(ReadContext &Ctx);
265 Error parseDataSection(ReadContext &Ctx);
266 Error parseDataCountSection(ReadContext &Ctx);
267
268 // Custom section types
269 Error parseDylinkSection(ReadContext &Ctx);
270 Error parseDylink0Section(ReadContext &Ctx);
271 Error parseNameSection(ReadContext &Ctx);
272 Error parseLinkingSection(ReadContext &Ctx);
273 Error parseLinkingSectionSymtab(ReadContext &Ctx);
274 Error parseLinkingSectionComdat(ReadContext &Ctx);
275 Error parseProducersSection(ReadContext &Ctx);
276 Error parseTargetFeaturesSection(ReadContext &Ctx);
277 Error parseRelocSection(StringRef Name, ReadContext &Ctx);
278
279 wasm::WasmObjectHeader Header;
280 std::vector<WasmSection> Sections;
281 wasm::WasmDylinkInfo DylinkInfo;
282 wasm::WasmProducerInfo ProducerInfo;
283 std::vector<wasm::WasmFeatureEntry> TargetFeatures;
284 std::vector<wasm::WasmSignature> Signatures;
285 std::vector<wasm::WasmTable> Tables;
286 std::vector<wasm::WasmLimits> Memories;
287 std::vector<wasm::WasmGlobal> Globals;
288 std::vector<wasm::WasmTag> Tags;
289 std::vector<wasm::WasmImport> Imports;
290 std::vector<wasm::WasmExport> Exports;
291 std::vector<wasm::WasmElemSegment> ElemSegments;
292 std::vector<WasmSegment> DataSegments;
293 std::optional<size_t> DataCount;
294 std::vector<wasm::WasmFunction> Functions;
295 std::vector<WasmSymbol> Symbols;
296 std::vector<wasm::WasmDebugName> DebugNames;
297 uint32_t StartFunction = -1;
298 bool HasLinkingSection = false;
299 bool HasDylinkSection = false;
300 bool HasMemory64 = false;
301 bool HasUnmodeledTypes = false;
302 wasm::WasmLinkingData LinkingData;
303 uint32_t NumImportedGlobals = 0;
304 uint32_t NumImportedTables = 0;
305 uint32_t NumImportedFunctions = 0;
306 uint32_t NumImportedTags = 0;
307 uint32_t CodeSection = 0;
308 uint32_t DataSection = 0;
309 uint32_t TagSection = 0;
310 uint32_t GlobalSection = 0;
311 uint32_t TableSection = 0;
312};
313
314class WasmSectionOrderChecker {
315public:
316 // We define orders for all core wasm sections and known custom sections.
317 enum : int {
318 // Sentinel, must be zero
319 WASM_SEC_ORDER_NONE = 0,
320
321 // Core sections
322 WASM_SEC_ORDER_TYPE,
323 WASM_SEC_ORDER_IMPORT,
324 WASM_SEC_ORDER_FUNCTION,
325 WASM_SEC_ORDER_TABLE,
326 WASM_SEC_ORDER_MEMORY,
327 WASM_SEC_ORDER_TAG,
328 WASM_SEC_ORDER_GLOBAL,
329 WASM_SEC_ORDER_EXPORT,
330 WASM_SEC_ORDER_START,
331 WASM_SEC_ORDER_ELEM,
332 WASM_SEC_ORDER_DATACOUNT,
333 WASM_SEC_ORDER_CODE,
334 WASM_SEC_ORDER_DATA,
335
336 // Custom sections
337 // "dylink" should be the very first section in the module
338 WASM_SEC_ORDER_DYLINK,
339 // "linking" section requires DATA section in order to validate data symbols
340 WASM_SEC_ORDER_LINKING,
341 // Must come after "linking" section in order to validate reloc indexes.
342 WASM_SEC_ORDER_RELOC,
343 // "name" section must appear after DATA. Comes after "linking" to allow
344 // symbol table to set default function name.
345 WASM_SEC_ORDER_NAME,
346 // "producers" section must appear after "name" section.
347 WASM_SEC_ORDER_PRODUCERS,
348 // "target_features" section must appear after producers section
349 WASM_SEC_ORDER_TARGET_FEATURES,
350
351 // Must be last
352 WASM_NUM_SEC_ORDERS
353
354 };
355
356 // Sections that may or may not be present, but cannot be predecessors
357 LLVM_ABI static int DisallowedPredecessors[WASM_NUM_SEC_ORDERS]
358 [WASM_NUM_SEC_ORDERS];
359
360 LLVM_ABI bool isValidSectionOrder(unsigned ID,
361 StringRef CustomSectionName = "");
362
363private:
364 bool Seen[WASM_NUM_SEC_ORDERS] = {}; // Sections that have been seen already
365
366 // Returns -1 for unknown sections.
367 int getSectionOrder(unsigned ID, StringRef CustomSectionName = "");
368};
369
370} // end namespace object
371
372inline raw_ostream &operator<<(raw_ostream &OS, const object::WasmSymbol &Sym) {
373 Sym.print(Out&: OS);
374 return OS;
375}
376
377} // end namespace llvm
378
379#endif // LLVM_OBJECT_WASM_H
380