Wasm.h revision 360784 
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/master/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/StringMap.h"
21#include "llvm/ADT/StringRef.h"
22#include "llvm/BinaryFormat/Wasm.h"
23#include "llvm/Config/llvm-config.h"
24#include "llvm/MC/MCSymbolWasm.h"
25#include "llvm/Object/Binary.h"
26#include "llvm/Object/ObjectFile.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::WasmEventType *EventType,
41             const wasm::WasmSignature *Signature)
42      : Info(Info), GlobalType(GlobalType), EventType(EventType),
43        Signature(Signature) {}
44
45  const wasm::WasmSymbolInfo &Info;
46  const wasm::WasmGlobalType *GlobalType;
47  const wasm::WasmEventType *EventType;
48  const wasm::WasmSignature *Signature;
49
50  bool isTypeFunction() const {
51    return Info.Kind == wasm::WASM_SYMBOL_TYPE_FUNCTION;
52  }
53
54  bool isTypeData() const { return Info.Kind == wasm::WASM_SYMBOL_TYPE_DATA; }
55
56  bool isTypeGlobal() const {
57    return Info.Kind == wasm::WASM_SYMBOL_TYPE_GLOBAL;
58  }
59
60  bool isTypeSection() const {
61    return Info.Kind == wasm::WASM_SYMBOL_TYPE_SECTION;
62  }
63
64  bool isTypeEvent() const { return Info.Kind == wasm::WASM_SYMBOL_TYPE_EVENT; }
65
66  bool isDefined() const { return !isUndefined(); }
67
68  bool isUndefined() const {
69    return (Info.Flags & wasm::WASM_SYMBOL_UNDEFINED) != 0;
70  }
71
72  bool isBindingWeak() const {
73    return getBinding() == wasm::WASM_SYMBOL_BINDING_WEAK;
74  }
75
76  bool isBindingGlobal() const {
77    return getBinding() == wasm::WASM_SYMBOL_BINDING_GLOBAL;
78  }
79
80  bool isBindingLocal() const {
81    return getBinding() == wasm::WASM_SYMBOL_BINDING_LOCAL;
82  }
83
84  unsigned getBinding() const {
85    return Info.Flags & wasm::WASM_SYMBOL_BINDING_MASK;
86  }
87
88  bool isHidden() const {
89    return getVisibility() == wasm::WASM_SYMBOL_VISIBILITY_HIDDEN;
90  }
91
92  unsigned getVisibility() const {
93    return Info.Flags & wasm::WASM_SYMBOL_VISIBILITY_MASK;
94  }
95
96  void print(raw_ostream &Out) const;
97
98#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
99  LLVM_DUMP_METHOD void dump() const;
100#endif
101};
102
103struct WasmSection {
104  WasmSection() = default;
105
106  uint32_t Type = 0;         // Section type (See below)
107  uint32_t Offset = 0;       // Offset with in the file
108  StringRef Name;            // Section name (User-defined sections only)
109  ArrayRef<uint8_t> Content; // Section content
110  std::vector<wasm::WasmRelocation> Relocations; // Relocations for this section
111};
112
113struct WasmSegment {
114  uint32_t SectionOffset;
115  wasm::WasmDataSegment Data;
116};
117
118class WasmObjectFile : public ObjectFile {
119
120public:
121  WasmObjectFile(MemoryBufferRef Object, Error &Err);
122
123  const wasm::WasmObjectHeader &getHeader() const;
124  const WasmSymbol &getWasmSymbol(const DataRefImpl &Symb) const;
125  const WasmSymbol &getWasmSymbol(const SymbolRef &Symbol) const;
126  const WasmSection &getWasmSection(const SectionRef &Section) const;
127  const wasm::WasmRelocation &getWasmRelocation(const RelocationRef &Ref) const;
128
129  static bool classof(const Binary *v) { return v->isWasm(); }
130
131  const wasm::WasmDylinkInfo &dylinkInfo() const { return DylinkInfo; }
132  const wasm::WasmProducerInfo &getProducerInfo() const { return ProducerInfo; }
133  ArrayRef<wasm::WasmFeatureEntry> getTargetFeatures() const {
134    return TargetFeatures;
135  }
136  ArrayRef<wasm::WasmSignature> types() const { return Signatures; }
137  ArrayRef<uint32_t> functionTypes() const { return FunctionTypes; }
138  ArrayRef<wasm::WasmImport> imports() const { return Imports; }
139  ArrayRef<wasm::WasmTable> tables() const { return Tables; }
140  ArrayRef<wasm::WasmLimits> memories() const { return Memories; }
141  ArrayRef<wasm::WasmGlobal> globals() const { return Globals; }
142  ArrayRef<wasm::WasmEvent> events() const { return Events; }
143  ArrayRef<wasm::WasmExport> exports() const { return Exports; }
144  ArrayRef<WasmSymbol> syms() const { return Symbols; }
145  const wasm::WasmLinkingData &linkingData() const { return LinkingData; }
146  uint32_t getNumberOfSymbols() const { return Symbols.size(); }
147  ArrayRef<wasm::WasmElemSegment> elements() const { return ElemSegments; }
148  ArrayRef<WasmSegment> dataSegments() const { return DataSegments; }
149  ArrayRef<wasm::WasmFunction> functions() const { return Functions; }
150  ArrayRef<wasm::WasmFunctionName> debugNames() const { return DebugNames; }
151  uint32_t startFunction() const { return StartFunction; }
152  uint32_t getNumImportedGlobals() const { return NumImportedGlobals; }
153  uint32_t getNumImportedFunctions() const { return NumImportedFunctions; }
154  uint32_t getNumImportedEvents() const { return NumImportedEvents; }
155  void moveSymbolNext(DataRefImpl &Symb) const override;
156
157  uint32_t getSymbolFlags(DataRefImpl Symb) const override;
158
159  basic_symbol_iterator symbol_begin() const override;
160
161  basic_symbol_iterator symbol_end() const override;
162  Expected<StringRef> getSymbolName(DataRefImpl Symb) const override;
163
164  Expected<uint64_t> getSymbolAddress(DataRefImpl Symb) const override;
165  uint64_t getWasmSymbolValue(const WasmSymbol &Sym) const;
166  uint64_t getSymbolValueImpl(DataRefImpl Symb) const override;
167  uint32_t getSymbolAlignment(DataRefImpl Symb) const override;
168  uint64_t getCommonSymbolSizeImpl(DataRefImpl Symb) const override;
169  Expected<SymbolRef::Type> getSymbolType(DataRefImpl Symb) const override;
170  Expected<section_iterator> getSymbolSection(DataRefImpl Symb) const override;
171
172  // Overrides from SectionRef.
173  void moveSectionNext(DataRefImpl &Sec) const override;
174  Expected<StringRef> getSectionName(DataRefImpl Sec) const override;
175  uint64_t getSectionAddress(DataRefImpl Sec) const override;
176  uint64_t getSectionIndex(DataRefImpl Sec) const override;
177  uint64_t getSectionSize(DataRefImpl Sec) const override;
178  Expected<ArrayRef<uint8_t>>
179  getSectionContents(DataRefImpl Sec) const override;
180  uint64_t getSectionAlignment(DataRefImpl Sec) const override;
181  bool isSectionCompressed(DataRefImpl Sec) const override;
182  bool isSectionText(DataRefImpl Sec) const override;
183  bool isSectionData(DataRefImpl Sec) const override;
184  bool isSectionBSS(DataRefImpl Sec) const override;
185  bool isSectionVirtual(DataRefImpl Sec) const override;
186  bool isSectionBitcode(DataRefImpl Sec) const override;
187  relocation_iterator section_rel_begin(DataRefImpl Sec) const override;
188  relocation_iterator section_rel_end(DataRefImpl Sec) const override;
189
190  // Overrides from RelocationRef.
191  void moveRelocationNext(DataRefImpl &Rel) const override;
192  uint64_t getRelocationOffset(DataRefImpl Rel) const override;
193  symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override;
194  uint64_t getRelocationType(DataRefImpl Rel) const override;
195  void getRelocationTypeName(DataRefImpl Rel,
196                             SmallVectorImpl<char> &Result) const override;
197
198  section_iterator section_begin() const override;
199  section_iterator section_end() const override;
200  uint8_t getBytesInAddress() const override;
201  StringRef getFileFormatName() const override;
202  Triple::ArchType getArch() const override;
203  SubtargetFeatures getFeatures() const override;
204  bool isRelocatableObject() const override;
205  bool isSharedObject() const;
206
207  struct ReadContext {
208    const uint8_t *Start;
209    const uint8_t *Ptr;
210    const uint8_t *End;
211  };
212
213private:
214  bool isValidFunctionIndex(uint32_t Index) const;
215  bool isDefinedFunctionIndex(uint32_t Index) const;
216  bool isValidGlobalIndex(uint32_t Index) const;
217  bool isDefinedGlobalIndex(uint32_t Index) const;
218  bool isValidEventIndex(uint32_t Index) const;
219  bool isDefinedEventIndex(uint32_t Index) const;
220  bool isValidFunctionSymbol(uint32_t Index) const;
221  bool isValidGlobalSymbol(uint32_t Index) const;
222  bool isValidEventSymbol(uint32_t Index) const;
223  bool isValidDataSymbol(uint32_t Index) const;
224  bool isValidSectionSymbol(uint32_t Index) const;
225  wasm::WasmFunction &getDefinedFunction(uint32_t Index);
226  const wasm::WasmFunction &getDefinedFunction(uint32_t Index) const;
227  wasm::WasmGlobal &getDefinedGlobal(uint32_t Index);
228  wasm::WasmEvent &getDefinedEvent(uint32_t Index);
229
230  const WasmSection &getWasmSection(DataRefImpl Ref) const;
231  const wasm::WasmRelocation &getWasmRelocation(DataRefImpl Ref) const;
232
233  Error parseSection(WasmSection &Sec);
234  Error parseCustomSection(WasmSection &Sec, ReadContext &Ctx);
235
236  // Standard section types
237  Error parseTypeSection(ReadContext &Ctx);
238  Error parseImportSection(ReadContext &Ctx);
239  Error parseFunctionSection(ReadContext &Ctx);
240  Error parseTableSection(ReadContext &Ctx);
241  Error parseMemorySection(ReadContext &Ctx);
242  Error parseGlobalSection(ReadContext &Ctx);
243  Error parseEventSection(ReadContext &Ctx);
244  Error parseExportSection(ReadContext &Ctx);
245  Error parseStartSection(ReadContext &Ctx);
246  Error parseElemSection(ReadContext &Ctx);
247  Error parseCodeSection(ReadContext &Ctx);
248  Error parseDataSection(ReadContext &Ctx);
249  Error parseDataCountSection(ReadContext &Ctx);
250
251  // Custom section types
252  Error parseDylinkSection(ReadContext &Ctx);
253  Error parseNameSection(ReadContext &Ctx);
254  Error parseLinkingSection(ReadContext &Ctx);
255  Error parseLinkingSectionSymtab(ReadContext &Ctx);
256  Error parseLinkingSectionComdat(ReadContext &Ctx);
257  Error parseProducersSection(ReadContext &Ctx);
258  Error parseTargetFeaturesSection(ReadContext &Ctx);
259  Error parseRelocSection(StringRef Name, ReadContext &Ctx);
260
261  wasm::WasmObjectHeader Header;
262  std::vector<WasmSection> Sections;
263  wasm::WasmDylinkInfo DylinkInfo;
264  wasm::WasmProducerInfo ProducerInfo;
265  std::vector<wasm::WasmFeatureEntry> TargetFeatures;
266  std::vector<wasm::WasmSignature> Signatures;
267  std::vector<uint32_t> FunctionTypes;
268  std::vector<wasm::WasmTable> Tables;
269  std::vector<wasm::WasmLimits> Memories;
270  std::vector<wasm::WasmGlobal> Globals;
271  std::vector<wasm::WasmEvent> Events;
272  std::vector<wasm::WasmImport> Imports;
273  std::vector<wasm::WasmExport> Exports;
274  std::vector<wasm::WasmElemSegment> ElemSegments;
275  std::vector<WasmSegment> DataSegments;
276  llvm::Optional<size_t> DataCount;
277  std::vector<wasm::WasmFunction> Functions;
278  std::vector<WasmSymbol> Symbols;
279  std::vector<wasm::WasmFunctionName> DebugNames;
280  uint32_t StartFunction = -1;
281  bool HasLinkingSection = false;
282  bool HasDylinkSection = false;
283  bool SeenCodeSection = false;
284  wasm::WasmLinkingData LinkingData;
285  uint32_t NumImportedGlobals = 0;
286  uint32_t NumImportedFunctions = 0;
287  uint32_t NumImportedEvents = 0;
288  uint32_t CodeSection = 0;
289  uint32_t DataSection = 0;
290  uint32_t GlobalSection = 0;
291  uint32_t EventSection = 0;
292};
293
294class WasmSectionOrderChecker {
295public:
296  // We define orders for all core wasm sections and known custom sections.
297  enum : int {
298    // Sentinel, must be zero
299    WASM_SEC_ORDER_NONE = 0,
300
301    // Core sections
302    WASM_SEC_ORDER_TYPE,
303    WASM_SEC_ORDER_IMPORT,
304    WASM_SEC_ORDER_FUNCTION,
305    WASM_SEC_ORDER_TABLE,
306    WASM_SEC_ORDER_MEMORY,
307    WASM_SEC_ORDER_GLOBAL,
308    WASM_SEC_ORDER_EVENT,
309    WASM_SEC_ORDER_EXPORT,
310    WASM_SEC_ORDER_START,
311    WASM_SEC_ORDER_ELEM,
312    WASM_SEC_ORDER_DATACOUNT,
313    WASM_SEC_ORDER_CODE,
314    WASM_SEC_ORDER_DATA,
315
316    // Custom sections
317    // "dylink" should be the very first section in the module
318    WASM_SEC_ORDER_DYLINK,
319    // "linking" section requires DATA section in order to validate data symbols
320    WASM_SEC_ORDER_LINKING,
321    // Must come after "linking" section in order to validate reloc indexes.
322    WASM_SEC_ORDER_RELOC,
323    // "name" section must appear after DATA. Comes after "linking" to allow
324    // symbol table to set default function name.
325    WASM_SEC_ORDER_NAME,
326    // "producers" section must appear after "name" section.
327    WASM_SEC_ORDER_PRODUCERS,
328    // "target_features" section must appear after producers section
329    WASM_SEC_ORDER_TARGET_FEATURES,
330
331    // Must be last
332    WASM_NUM_SEC_ORDERS
333
334  };
335
336  // Sections that may or may not be present, but cannot be predecessors
337  static int DisallowedPredecessors[WASM_NUM_SEC_ORDERS][WASM_NUM_SEC_ORDERS];
338
339  bool isValidSectionOrder(unsigned ID, StringRef CustomSectionName = "");
340
341private:
342  bool Seen[WASM_NUM_SEC_ORDERS] = {}; // Sections that have been seen already
343
344  // Returns -1 for unknown sections.
345  int getSectionOrder(unsigned ID, StringRef CustomSectionName = "");
346};
347
348} // end namespace object
349
350inline raw_ostream &operator<<(raw_ostream &OS, const object::WasmSymbol &Sym) {
351  Sym.print(OS);
352  return OS;
353}
354
355} // end namespace llvm
356
357#endif // LLVM_OBJECT_WASM_H
358