1//===------ utils/obj2yaml.cpp - obj2yaml conversion tool -------*- C++ -*-===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9
10#include "obj2yaml.h"
11
12#include "llvm/Object/COFF.h"
13
14
15template <typename One, typename Two>
16struct pod_pair { // I'd much rather use std::pair, but it's not a POD
17  One first;
18  Two second;
19};
20
21#define STRING_PAIR(x)  {llvm::COFF::x, #x}
22static const pod_pair<llvm::COFF::MachineTypes, const char *>
23MachineTypePairs [] = {
24  STRING_PAIR(IMAGE_FILE_MACHINE_UNKNOWN),
25  STRING_PAIR(IMAGE_FILE_MACHINE_AM33),
26  STRING_PAIR(IMAGE_FILE_MACHINE_AMD64),
27  STRING_PAIR(IMAGE_FILE_MACHINE_ARM),
28  STRING_PAIR(IMAGE_FILE_MACHINE_ARMV7),
29  STRING_PAIR(IMAGE_FILE_MACHINE_EBC),
30  STRING_PAIR(IMAGE_FILE_MACHINE_I386),
31  STRING_PAIR(IMAGE_FILE_MACHINE_IA64),
32  STRING_PAIR(IMAGE_FILE_MACHINE_M32R),
33  STRING_PAIR(IMAGE_FILE_MACHINE_MIPS16),
34  STRING_PAIR(IMAGE_FILE_MACHINE_MIPSFPU),
35  STRING_PAIR(IMAGE_FILE_MACHINE_MIPSFPU16),
36  STRING_PAIR(IMAGE_FILE_MACHINE_POWERPC),
37  STRING_PAIR(IMAGE_FILE_MACHINE_POWERPCFP),
38  STRING_PAIR(IMAGE_FILE_MACHINE_R4000),
39  STRING_PAIR(IMAGE_FILE_MACHINE_SH3),
40  STRING_PAIR(IMAGE_FILE_MACHINE_SH3DSP),
41  STRING_PAIR(IMAGE_FILE_MACHINE_SH4),
42  STRING_PAIR(IMAGE_FILE_MACHINE_SH5),
43  STRING_PAIR(IMAGE_FILE_MACHINE_THUMB),
44  STRING_PAIR(IMAGE_FILE_MACHINE_WCEMIPSV2)
45};
46
47static const pod_pair<llvm::COFF::SectionCharacteristics, const char *>
48SectionCharacteristicsPairs1 [] = {
49  STRING_PAIR(IMAGE_SCN_TYPE_NO_PAD),
50  STRING_PAIR(IMAGE_SCN_CNT_CODE),
51  STRING_PAIR(IMAGE_SCN_CNT_INITIALIZED_DATA),
52  STRING_PAIR(IMAGE_SCN_CNT_UNINITIALIZED_DATA),
53  STRING_PAIR(IMAGE_SCN_LNK_OTHER),
54  STRING_PAIR(IMAGE_SCN_LNK_INFO),
55  STRING_PAIR(IMAGE_SCN_LNK_REMOVE),
56  STRING_PAIR(IMAGE_SCN_LNK_COMDAT),
57  STRING_PAIR(IMAGE_SCN_GPREL),
58  STRING_PAIR(IMAGE_SCN_MEM_PURGEABLE),
59  STRING_PAIR(IMAGE_SCN_MEM_16BIT),
60  STRING_PAIR(IMAGE_SCN_MEM_LOCKED),
61  STRING_PAIR(IMAGE_SCN_MEM_PRELOAD)
62};
63
64static const pod_pair<llvm::COFF::SectionCharacteristics, const char *>
65SectionCharacteristicsPairsAlignment [] = {
66  STRING_PAIR(IMAGE_SCN_ALIGN_1BYTES),
67  STRING_PAIR(IMAGE_SCN_ALIGN_2BYTES),
68  STRING_PAIR(IMAGE_SCN_ALIGN_4BYTES),
69  STRING_PAIR(IMAGE_SCN_ALIGN_8BYTES),
70  STRING_PAIR(IMAGE_SCN_ALIGN_16BYTES),
71  STRING_PAIR(IMAGE_SCN_ALIGN_32BYTES),
72  STRING_PAIR(IMAGE_SCN_ALIGN_64BYTES),
73  STRING_PAIR(IMAGE_SCN_ALIGN_128BYTES),
74  STRING_PAIR(IMAGE_SCN_ALIGN_256BYTES),
75  STRING_PAIR(IMAGE_SCN_ALIGN_512BYTES),
76  STRING_PAIR(IMAGE_SCN_ALIGN_1024BYTES),
77  STRING_PAIR(IMAGE_SCN_ALIGN_2048BYTES),
78  STRING_PAIR(IMAGE_SCN_ALIGN_4096BYTES),
79  STRING_PAIR(IMAGE_SCN_ALIGN_8192BYTES)
80};
81
82static const pod_pair<llvm::COFF::SectionCharacteristics, const char *>
83SectionCharacteristicsPairs2 [] = {
84  STRING_PAIR(IMAGE_SCN_LNK_NRELOC_OVFL),
85  STRING_PAIR(IMAGE_SCN_MEM_DISCARDABLE),
86  STRING_PAIR(IMAGE_SCN_MEM_NOT_CACHED),
87  STRING_PAIR(IMAGE_SCN_MEM_NOT_PAGED),
88  STRING_PAIR(IMAGE_SCN_MEM_SHARED),
89  STRING_PAIR(IMAGE_SCN_MEM_EXECUTE),
90  STRING_PAIR(IMAGE_SCN_MEM_READ),
91  STRING_PAIR(IMAGE_SCN_MEM_WRITE)
92};
93
94static const pod_pair<llvm::COFF::SymbolBaseType, const char *>
95SymbolBaseTypePairs [] = {
96  STRING_PAIR(IMAGE_SYM_TYPE_NULL),
97  STRING_PAIR(IMAGE_SYM_TYPE_VOID),
98  STRING_PAIR(IMAGE_SYM_TYPE_CHAR),
99  STRING_PAIR(IMAGE_SYM_TYPE_SHORT),
100  STRING_PAIR(IMAGE_SYM_TYPE_INT),
101  STRING_PAIR(IMAGE_SYM_TYPE_LONG),
102  STRING_PAIR(IMAGE_SYM_TYPE_FLOAT),
103  STRING_PAIR(IMAGE_SYM_TYPE_DOUBLE),
104  STRING_PAIR(IMAGE_SYM_TYPE_STRUCT),
105  STRING_PAIR(IMAGE_SYM_TYPE_UNION),
106  STRING_PAIR(IMAGE_SYM_TYPE_ENUM),
107  STRING_PAIR(IMAGE_SYM_TYPE_MOE),
108  STRING_PAIR(IMAGE_SYM_TYPE_BYTE),
109  STRING_PAIR(IMAGE_SYM_TYPE_WORD),
110  STRING_PAIR(IMAGE_SYM_TYPE_UINT),
111  STRING_PAIR(IMAGE_SYM_TYPE_DWORD)
112};
113
114static const pod_pair<llvm::COFF::SymbolComplexType, const char *>
115SymbolComplexTypePairs [] = {
116  STRING_PAIR(IMAGE_SYM_DTYPE_NULL),
117  STRING_PAIR(IMAGE_SYM_DTYPE_POINTER),
118  STRING_PAIR(IMAGE_SYM_DTYPE_FUNCTION),
119  STRING_PAIR(IMAGE_SYM_DTYPE_ARRAY),
120};
121
122static const pod_pair<llvm::COFF::SymbolStorageClass, const char *>
123SymbolStorageClassPairs [] = {
124  STRING_PAIR(IMAGE_SYM_CLASS_END_OF_FUNCTION),
125  STRING_PAIR(IMAGE_SYM_CLASS_NULL),
126  STRING_PAIR(IMAGE_SYM_CLASS_AUTOMATIC),
127  STRING_PAIR(IMAGE_SYM_CLASS_EXTERNAL),
128  STRING_PAIR(IMAGE_SYM_CLASS_STATIC),
129  STRING_PAIR(IMAGE_SYM_CLASS_REGISTER),
130  STRING_PAIR(IMAGE_SYM_CLASS_EXTERNAL_DEF),
131  STRING_PAIR(IMAGE_SYM_CLASS_LABEL),
132  STRING_PAIR(IMAGE_SYM_CLASS_UNDEFINED_LABEL),
133  STRING_PAIR(IMAGE_SYM_CLASS_MEMBER_OF_STRUCT),
134  STRING_PAIR(IMAGE_SYM_CLASS_ARGUMENT),
135  STRING_PAIR(IMAGE_SYM_CLASS_STRUCT_TAG),
136  STRING_PAIR(IMAGE_SYM_CLASS_MEMBER_OF_UNION),
137  STRING_PAIR(IMAGE_SYM_CLASS_UNION_TAG),
138  STRING_PAIR(IMAGE_SYM_CLASS_TYPE_DEFINITION),
139  STRING_PAIR(IMAGE_SYM_CLASS_UNDEFINED_STATIC),
140  STRING_PAIR(IMAGE_SYM_CLASS_ENUM_TAG),
141  STRING_PAIR(IMAGE_SYM_CLASS_MEMBER_OF_ENUM),
142  STRING_PAIR(IMAGE_SYM_CLASS_REGISTER_PARAM),
143  STRING_PAIR(IMAGE_SYM_CLASS_BIT_FIELD),
144  STRING_PAIR(IMAGE_SYM_CLASS_BLOCK),
145  STRING_PAIR(IMAGE_SYM_CLASS_FUNCTION),
146  STRING_PAIR(IMAGE_SYM_CLASS_END_OF_STRUCT),
147  STRING_PAIR(IMAGE_SYM_CLASS_FILE),
148  STRING_PAIR(IMAGE_SYM_CLASS_SECTION),
149  STRING_PAIR(IMAGE_SYM_CLASS_WEAK_EXTERNAL),
150  STRING_PAIR(IMAGE_SYM_CLASS_CLR_TOKEN),
151};
152
153static const pod_pair<llvm::COFF::RelocationTypeX86, const char *>
154RelocationTypeX86Pairs [] = {
155  STRING_PAIR(IMAGE_REL_I386_ABSOLUTE),
156  STRING_PAIR(IMAGE_REL_I386_DIR16),
157  STRING_PAIR(IMAGE_REL_I386_REL16),
158  STRING_PAIR(IMAGE_REL_I386_DIR32),
159  STRING_PAIR(IMAGE_REL_I386_DIR32NB),
160  STRING_PAIR(IMAGE_REL_I386_SEG12),
161  STRING_PAIR(IMAGE_REL_I386_SECTION),
162  STRING_PAIR(IMAGE_REL_I386_SECREL),
163  STRING_PAIR(IMAGE_REL_I386_TOKEN),
164  STRING_PAIR(IMAGE_REL_I386_SECREL7),
165  STRING_PAIR(IMAGE_REL_I386_REL32),
166  STRING_PAIR(IMAGE_REL_AMD64_ABSOLUTE),
167  STRING_PAIR(IMAGE_REL_AMD64_ADDR64),
168  STRING_PAIR(IMAGE_REL_AMD64_ADDR32),
169  STRING_PAIR(IMAGE_REL_AMD64_ADDR32NB),
170  STRING_PAIR(IMAGE_REL_AMD64_REL32),
171  STRING_PAIR(IMAGE_REL_AMD64_REL32_1),
172  STRING_PAIR(IMAGE_REL_AMD64_REL32_2),
173  STRING_PAIR(IMAGE_REL_AMD64_REL32_3),
174  STRING_PAIR(IMAGE_REL_AMD64_REL32_4),
175  STRING_PAIR(IMAGE_REL_AMD64_REL32_5),
176  STRING_PAIR(IMAGE_REL_AMD64_SECTION),
177  STRING_PAIR(IMAGE_REL_AMD64_SECREL),
178  STRING_PAIR(IMAGE_REL_AMD64_SECREL7),
179  STRING_PAIR(IMAGE_REL_AMD64_TOKEN),
180  STRING_PAIR(IMAGE_REL_AMD64_SREL32),
181  STRING_PAIR(IMAGE_REL_AMD64_PAIR),
182  STRING_PAIR(IMAGE_REL_AMD64_SSPAN32)
183};
184
185static const pod_pair<llvm::COFF::RelocationTypesARM, const char *>
186RelocationTypesARMPairs [] = {
187  STRING_PAIR(IMAGE_REL_ARM_ABSOLUTE),
188  STRING_PAIR(IMAGE_REL_ARM_ADDR32),
189  STRING_PAIR(IMAGE_REL_ARM_ADDR32NB),
190  STRING_PAIR(IMAGE_REL_ARM_BRANCH24),
191  STRING_PAIR(IMAGE_REL_ARM_BRANCH11),
192  STRING_PAIR(IMAGE_REL_ARM_TOKEN),
193  STRING_PAIR(IMAGE_REL_ARM_BLX24),
194  STRING_PAIR(IMAGE_REL_ARM_BLX11),
195  STRING_PAIR(IMAGE_REL_ARM_SECTION),
196  STRING_PAIR(IMAGE_REL_ARM_SECREL),
197  STRING_PAIR(IMAGE_REL_ARM_MOV32A),
198  STRING_PAIR(IMAGE_REL_ARM_MOV32T),
199  STRING_PAIR(IMAGE_REL_ARM_BRANCH20T),
200  STRING_PAIR(IMAGE_REL_ARM_BRANCH24T),
201  STRING_PAIR(IMAGE_REL_ARM_BLX23T)
202};
203#undef STRING_PAIR
204
205
206static const char endl = '\n';
207
208namespace yaml {  // COFF-specific yaml-writing specific routines
209
210static llvm::raw_ostream &writeName(llvm::raw_ostream &Out,
211                             const char *Name, std::size_t NameSize) {
212  for (std::size_t i = 0; i < NameSize; ++i) {
213    if (!Name[i]) break;
214    Out << Name[i];
215  }
216  return Out;
217}
218
219// Given an array of pod_pair<enum, const char *>, write all enums that match
220template <typename T, std::size_t N>
221static llvm::raw_ostream &writeBitMask(llvm::raw_ostream &Out,
222              const pod_pair<T, const char *> (&Arr)[N], unsigned long Val) {
223  for (std::size_t i = 0; i < N; ++i)
224    if (Val & Arr[i].first)
225      Out << Arr[i].second << ", ";
226  return Out;
227}
228
229} // end of yaml namespace
230
231// Given an array of pod_pair<enum, const char *>, look up a value
232template <typename T, std::size_t N>
233const char *nameLookup(const pod_pair<T, const char *> (&Arr)[N],
234                           unsigned long Val, const char *NotFound = NULL) {
235  T n = static_cast<T>(Val);
236  for (std::size_t i = 0; i < N; ++i)
237    if (n == Arr[i].first)
238      return Arr[i].second;
239  return NotFound;
240}
241
242
243static llvm::raw_ostream &yamlCOFFHeader(
244          const llvm::object::coff_file_header *Header,llvm::raw_ostream &Out) {
245
246  Out << "header: !Header" << endl;
247  Out << "  Machine: ";
248  Out << nameLookup(MachineTypePairs, Header->Machine, "# Unknown_MachineTypes")
249      << " # (";
250  return yaml::writeHexNumber(Out, Header->Machine) << ")" << endl << endl;
251}
252
253
254static llvm::raw_ostream &yamlCOFFSections(llvm::object::COFFObjectFile &Obj,
255                            std::size_t NumSections, llvm::raw_ostream &Out) {
256  llvm::error_code ec;
257  Out << "sections:" << endl;
258  for (llvm::object::section_iterator iter = Obj.begin_sections();
259                             iter != Obj.end_sections(); iter.increment(ec)) {
260    const llvm::object::coff_section *sect = Obj.getCOFFSection(iter);
261
262    Out << "  - !Section" << endl;
263    Out << "    Name: ";
264    yaml::writeName(Out, sect->Name, sizeof(sect->Name)) << endl;
265
266    Out << "    Characteristics: [";
267    yaml::writeBitMask(Out, SectionCharacteristicsPairs1, sect->Characteristics);
268    Out << nameLookup(SectionCharacteristicsPairsAlignment,
269        sect->Characteristics & 0x00F00000, "# Unrecognized_IMAGE_SCN_ALIGN")
270        << ", ";
271    yaml::writeBitMask(Out, SectionCharacteristicsPairs2, sect->Characteristics);
272    Out << "] # ";
273    yaml::writeHexNumber(Out, sect->Characteristics) << endl;
274
275    llvm::ArrayRef<uint8_t> sectionData;
276    Obj.getSectionContents(sect, sectionData);
277    Out << "    SectionData: ";
278    yaml::writeHexStream(Out, sectionData) << endl;
279    if (iter->begin_relocations() != iter->end_relocations())
280      Out << "    Relocations:\n";
281    for (llvm::object::relocation_iterator rIter = iter->begin_relocations();
282                       rIter != iter->end_relocations(); rIter.increment(ec)) {
283      const llvm::object::coff_relocation *reloc = Obj.getCOFFRelocation(rIter);
284
285        Out << "      - !Relocation" << endl;
286        Out << "        VirtualAddress: " ;
287        yaml::writeHexNumber(Out, reloc->VirtualAddress) << endl;
288        Out << "        SymbolTableIndex: " << reloc->SymbolTableIndex << endl;
289        Out << "        Type: "
290            << nameLookup(RelocationTypeX86Pairs, reloc->Type) << endl;
291    // TODO: Use the correct reloc type for the machine.
292        Out << endl;
293      }
294
295  }
296  return Out;
297}
298
299static llvm::raw_ostream& yamlCOFFSymbols(llvm::object::COFFObjectFile &Obj,
300                              std::size_t NumSymbols, llvm::raw_ostream &Out) {
301  llvm::error_code ec;
302  Out << "symbols:" << endl;
303  for (llvm::object::symbol_iterator iter = Obj.begin_symbols();
304                             iter != Obj.end_symbols(); iter.increment(ec)) {
305 // Gather all the info that we need
306    llvm::StringRef str;
307    const llvm::object::coff_symbol *symbol = Obj.getCOFFSymbol(iter);
308    Obj.getSymbolName(symbol, str);
309    std::size_t  simpleType  = symbol->getBaseType();
310    std::size_t complexType  = symbol->getComplexType();
311    std::size_t storageClass = symbol->StorageClass;
312
313    Out << "  - !Symbol" << endl;
314    Out << "    Name: " << str << endl;
315
316    Out << "    Value: "         << symbol->Value << endl;
317    Out << "    SectionNumber: " << symbol->SectionNumber << endl;
318
319    Out << "    SimpleType: "
320        << nameLookup(SymbolBaseTypePairs, simpleType,
321            "# Unknown_SymbolBaseType")
322        << " # (" << simpleType << ")" << endl;
323
324    Out << "    ComplexType: "
325        << nameLookup(SymbolComplexTypePairs, complexType,
326                "# Unknown_SymbolComplexType")
327        << " # (" << complexType << ")" << endl;
328
329    Out << "    StorageClass: "
330        << nameLookup(SymbolStorageClassPairs, storageClass,
331              "# Unknown_StorageClass")
332        << " # (" << (int) storageClass << ")" << endl;
333
334    if (symbol->NumberOfAuxSymbols > 0) {
335      llvm::ArrayRef<uint8_t> aux = Obj.getSymbolAuxData(symbol);
336      Out << "    NumberOfAuxSymbols: "
337          << (int) symbol->NumberOfAuxSymbols << endl;
338      Out << "    AuxillaryData: ";
339      yaml::writeHexStream(Out, aux);
340    }
341
342    Out << endl;
343  }
344
345  return Out;
346}
347
348
349llvm::error_code coff2yaml(llvm::raw_ostream &Out, llvm::MemoryBuffer *TheObj) {
350  llvm::error_code ec;
351  llvm::object::COFFObjectFile obj(TheObj, ec);
352  if (!ec) {
353    const llvm::object::coff_file_header *hd;
354    ec = obj.getHeader(hd);
355    if (!ec) {
356      yamlCOFFHeader(hd, Out);
357      yamlCOFFSections(obj, hd->NumberOfSections, Out);
358      yamlCOFFSymbols(obj, hd->NumberOfSymbols, Out);
359    }
360  }
361  return ec;
362}
363