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