1//===- yaml2obj - Convert YAML to a binary object file --------------------===//
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// This program takes a YAML description of an object file and outputs the
11// binary equivalent.
12//
13// This is used for writing tests that require binary files.
14//
15//===----------------------------------------------------------------------===//
16
17#include "llvm/ADT/SmallString.h"
18#include "llvm/ADT/StringExtras.h"
19#include "llvm/ADT/StringMap.h"
20#include "llvm/ADT/StringSwitch.h"
21#include "llvm/Support/COFF.h"
22#include "llvm/Support/Casting.h"
23#include "llvm/Support/CommandLine.h"
24#include "llvm/Support/Endian.h"
25#include "llvm/Support/ManagedStatic.h"
26#include "llvm/Support/MemoryBuffer.h"
27#include "llvm/Support/PrettyStackTrace.h"
28#include "llvm/Support/raw_ostream.h"
29#include "llvm/Support/Signals.h"
30#include "llvm/Support/SourceMgr.h"
31#include "llvm/Support/system_error.h"
32#include "llvm/Support/YAMLParser.h"
33
34#include <vector>
35
36using namespace llvm;
37
38static cl::opt<std::string>
39  Input(cl::Positional, cl::desc("<input>"), cl::init("-"));
40
41template<class T>
42typename llvm::enable_if_c<std::numeric_limits<T>::is_integer, bool>::type
43getAs(const llvm::yaml::ScalarNode *SN, T &Result) {
44  SmallString<4> Storage;
45  StringRef Value = SN->getValue(Storage);
46  if (Value.getAsInteger(0, Result))
47    return false;
48  return true;
49}
50
51// Given a container with begin and end with ::value_type of a character type.
52// Iterate through pairs of characters in the the set of [a-fA-F0-9] ignoring
53// all other characters.
54struct hex_pair_iterator {
55  StringRef::const_iterator Current, End;
56  typedef SmallVector<char, 2> value_type;
57  value_type Pair;
58  bool IsDone;
59
60  hex_pair_iterator(StringRef C)
61    : Current(C.begin()), End(C.end()), IsDone(false) {
62    // Initalize Pair.
63    ++*this;
64  }
65
66  // End iterator.
67  hex_pair_iterator() : Current(), End(), IsDone(true) {}
68
69  value_type operator *() const {
70    return Pair;
71  }
72
73  hex_pair_iterator operator ++() {
74    // We're at the end of the input.
75    if (Current == End) {
76      IsDone = true;
77      return *this;
78    }
79    Pair = value_type();
80    for (; Current != End && Pair.size() != 2; ++Current) {
81      // Is a valid hex digit.
82      if ((*Current >= '0' && *Current <= '9') ||
83          (*Current >= 'a' && *Current <= 'f') ||
84          (*Current >= 'A' && *Current <= 'F'))
85        Pair.push_back(*Current);
86    }
87    // Hit the end without getting 2 hex digits. Pair is invalid.
88    if (Pair.size() != 2)
89      IsDone = true;
90    return *this;
91  }
92
93  bool operator ==(const hex_pair_iterator Other) {
94    return (IsDone == Other.IsDone) ||
95           (Current == Other.Current && End == Other.End);
96  }
97
98  bool operator !=(const hex_pair_iterator Other) {
99    return !(*this == Other);
100  }
101};
102
103template <class ContainerOut>
104static bool hexStringToByteArray(StringRef Str, ContainerOut &Out) {
105  for (hex_pair_iterator I(Str), E; I != E; ++I) {
106    typename hex_pair_iterator::value_type Pair = *I;
107    typename ContainerOut::value_type Byte;
108    if (StringRef(Pair.data(), 2).getAsInteger(16, Byte))
109      return false;
110    Out.push_back(Byte);
111  }
112  return true;
113}
114
115/// This parses a yaml stream that represents a COFF object file.
116/// See docs/yaml2obj for the yaml scheema.
117struct COFFParser {
118  COFFParser(yaml::Stream &Input) : YS(Input) {
119    std::memset(&Header, 0, sizeof(Header));
120    // A COFF string table always starts with a 4 byte size field. Offsets into
121    // it include this size, so allocate it now.
122    StringTable.append(4, 0);
123  }
124
125  bool parseHeader(yaml::Node *HeaderN) {
126    yaml::MappingNode *MN = dyn_cast<yaml::MappingNode>(HeaderN);
127    if (!MN) {
128      YS.printError(HeaderN, "header's value must be a mapping node");
129      return false;
130    }
131    for (yaml::MappingNode::iterator i = MN->begin(), e = MN->end();
132                                     i != e; ++i) {
133      yaml::ScalarNode *Key = dyn_cast<yaml::ScalarNode>(i->getKey());
134      if (!Key) {
135        YS.printError(i->getKey(), "Keys must be scalar values");
136        return false;
137      }
138      SmallString<32> Storage;
139      StringRef KeyValue = Key->getValue(Storage);
140      if (KeyValue == "Characteristics") {
141        if (!parseHeaderCharacteristics(i->getValue()))
142          return false;
143      } else {
144        yaml::ScalarNode *Value = dyn_cast<yaml::ScalarNode>(i->getValue());
145        if (!Value) {
146          YS.printError(Value,
147            Twine(KeyValue) + " must be a scalar value");
148          return false;
149        }
150        if (KeyValue == "Machine") {
151          uint16_t Machine = COFF::MT_Invalid;
152          if (!getAs(Value, Machine)) {
153            // It's not a raw number, try matching the string.
154            StringRef ValueValue = Value->getValue(Storage);
155            Machine = StringSwitch<COFF::MachineTypes>(ValueValue)
156              .Case( "IMAGE_FILE_MACHINE_UNKNOWN"
157                   , COFF::IMAGE_FILE_MACHINE_UNKNOWN)
158              .Case( "IMAGE_FILE_MACHINE_AM33"
159                   , COFF::IMAGE_FILE_MACHINE_AM33)
160              .Case( "IMAGE_FILE_MACHINE_AMD64"
161                   , COFF::IMAGE_FILE_MACHINE_AMD64)
162              .Case( "IMAGE_FILE_MACHINE_ARM"
163                   , COFF::IMAGE_FILE_MACHINE_ARM)
164              .Case( "IMAGE_FILE_MACHINE_ARMV7"
165                   , COFF::IMAGE_FILE_MACHINE_ARMV7)
166              .Case( "IMAGE_FILE_MACHINE_EBC"
167                   , COFF::IMAGE_FILE_MACHINE_EBC)
168              .Case( "IMAGE_FILE_MACHINE_I386"
169                   , COFF::IMAGE_FILE_MACHINE_I386)
170              .Case( "IMAGE_FILE_MACHINE_IA64"
171                   , COFF::IMAGE_FILE_MACHINE_IA64)
172              .Case( "IMAGE_FILE_MACHINE_M32R"
173                   , COFF::IMAGE_FILE_MACHINE_M32R)
174              .Case( "IMAGE_FILE_MACHINE_MIPS16"
175                   , COFF::IMAGE_FILE_MACHINE_MIPS16)
176              .Case( "IMAGE_FILE_MACHINE_MIPSFPU"
177                   , COFF::IMAGE_FILE_MACHINE_MIPSFPU)
178              .Case( "IMAGE_FILE_MACHINE_MIPSFPU16"
179                   , COFF::IMAGE_FILE_MACHINE_MIPSFPU16)
180              .Case( "IMAGE_FILE_MACHINE_POWERPC"
181                   , COFF::IMAGE_FILE_MACHINE_POWERPC)
182              .Case( "IMAGE_FILE_MACHINE_POWERPCFP"
183                   , COFF::IMAGE_FILE_MACHINE_POWERPCFP)
184              .Case( "IMAGE_FILE_MACHINE_R4000"
185                   , COFF::IMAGE_FILE_MACHINE_R4000)
186              .Case( "IMAGE_FILE_MACHINE_SH3"
187                   , COFF::IMAGE_FILE_MACHINE_SH3)
188              .Case( "IMAGE_FILE_MACHINE_SH3DSP"
189                   , COFF::IMAGE_FILE_MACHINE_SH3DSP)
190              .Case( "IMAGE_FILE_MACHINE_SH4"
191                   , COFF::IMAGE_FILE_MACHINE_SH4)
192              .Case( "IMAGE_FILE_MACHINE_SH5"
193                   , COFF::IMAGE_FILE_MACHINE_SH5)
194              .Case( "IMAGE_FILE_MACHINE_THUMB"
195                   , COFF::IMAGE_FILE_MACHINE_THUMB)
196              .Case( "IMAGE_FILE_MACHINE_WCEMIPSV2"
197                   , COFF::IMAGE_FILE_MACHINE_WCEMIPSV2)
198              .Default(COFF::MT_Invalid);
199            if (Machine == COFF::MT_Invalid) {
200              YS.printError(Value, "Invalid value for Machine");
201              return false;
202            }
203          }
204          Header.Machine = Machine;
205        } else if (KeyValue == "NumberOfSections") {
206          if (!getAs(Value, Header.NumberOfSections)) {
207              YS.printError(Value, "Invalid value for NumberOfSections");
208              return false;
209          }
210        } else if (KeyValue == "TimeDateStamp") {
211          if (!getAs(Value, Header.TimeDateStamp)) {
212              YS.printError(Value, "Invalid value for TimeDateStamp");
213              return false;
214          }
215        } else if (KeyValue == "PointerToSymbolTable") {
216          if (!getAs(Value, Header.PointerToSymbolTable)) {
217              YS.printError(Value, "Invalid value for PointerToSymbolTable");
218              return false;
219          }
220        } else if (KeyValue == "NumberOfSymbols") {
221          if (!getAs(Value, Header.NumberOfSymbols)) {
222              YS.printError(Value, "Invalid value for NumberOfSymbols");
223              return false;
224          }
225        } else if (KeyValue == "SizeOfOptionalHeader") {
226          if (!getAs(Value, Header.SizeOfOptionalHeader)) {
227              YS.printError(Value, "Invalid value for SizeOfOptionalHeader");
228              return false;
229          }
230        } else {
231          YS.printError(Key, "Unrecognized key in header");
232          return false;
233        }
234      }
235    }
236    return true;
237  }
238
239  bool parseHeaderCharacteristics(yaml::Node *Characteristics) {
240    yaml::ScalarNode *Value = dyn_cast<yaml::ScalarNode>(Characteristics);
241    yaml::SequenceNode *SeqValue
242      = dyn_cast<yaml::SequenceNode>(Characteristics);
243    if (!Value && !SeqValue) {
244      YS.printError(Characteristics,
245        "Characteristics must either be a number or sequence");
246      return false;
247    }
248    if (Value) {
249      if (!getAs(Value, Header.Characteristics)) {
250        YS.printError(Value, "Invalid value for Characteristics");
251        return false;
252      }
253    } else {
254      for (yaml::SequenceNode::iterator ci = SeqValue->begin(),
255                                        ce = SeqValue->end();
256                                        ci != ce; ++ci) {
257        yaml::ScalarNode *CharValue = dyn_cast<yaml::ScalarNode>(&*ci);
258        if (!CharValue) {
259          YS.printError(CharValue,
260            "Characteristics must be scalar values");
261          return false;
262        }
263        SmallString<32> Storage;
264        StringRef Char = CharValue->getValue(Storage);
265        uint16_t Characteristic = StringSwitch<COFF::Characteristics>(Char)
266          .Case( "IMAGE_FILE_RELOCS_STRIPPED"
267                , COFF::IMAGE_FILE_RELOCS_STRIPPED)
268          .Case( "IMAGE_FILE_EXECUTABLE_IMAGE"
269                , COFF::IMAGE_FILE_EXECUTABLE_IMAGE)
270          .Case( "IMAGE_FILE_LINE_NUMS_STRIPPED"
271                , COFF::IMAGE_FILE_LINE_NUMS_STRIPPED)
272          .Case( "IMAGE_FILE_LOCAL_SYMS_STRIPPED"
273                , COFF::IMAGE_FILE_LOCAL_SYMS_STRIPPED)
274          .Case( "IMAGE_FILE_AGGRESSIVE_WS_TRIM"
275                , COFF::IMAGE_FILE_AGGRESSIVE_WS_TRIM)
276          .Case( "IMAGE_FILE_LARGE_ADDRESS_AWARE"
277                , COFF::IMAGE_FILE_LARGE_ADDRESS_AWARE)
278          .Case( "IMAGE_FILE_BYTES_REVERSED_LO"
279                , COFF::IMAGE_FILE_BYTES_REVERSED_LO)
280          .Case( "IMAGE_FILE_32BIT_MACHINE"
281                , COFF::IMAGE_FILE_32BIT_MACHINE)
282          .Case( "IMAGE_FILE_DEBUG_STRIPPED"
283                , COFF::IMAGE_FILE_DEBUG_STRIPPED)
284          .Case( "IMAGE_FILE_REMOVABLE_RUN_FROM_SWAP"
285                , COFF::IMAGE_FILE_REMOVABLE_RUN_FROM_SWAP)
286          .Case( "IMAGE_FILE_SYSTEM"
287                , COFF::IMAGE_FILE_SYSTEM)
288          .Case( "IMAGE_FILE_DLL"
289                , COFF::IMAGE_FILE_DLL)
290          .Case( "IMAGE_FILE_UP_SYSTEM_ONLY"
291                , COFF::IMAGE_FILE_UP_SYSTEM_ONLY)
292          .Default(COFF::C_Invalid);
293        if (Characteristic == COFF::C_Invalid) {
294          // TODO: Typo-correct.
295          YS.printError(CharValue,
296            "Invalid value for Characteristic");
297          return false;
298        }
299        Header.Characteristics |= Characteristic;
300      }
301    }
302    return true;
303  }
304
305  bool parseSections(yaml::Node *SectionsN) {
306    yaml::SequenceNode *SN = dyn_cast<yaml::SequenceNode>(SectionsN);
307    if (!SN) {
308      YS.printError(SectionsN, "Sections must be a sequence");
309      return false;
310    }
311    for (yaml::SequenceNode::iterator i = SN->begin(), e = SN->end();
312                                      i != e; ++i) {
313      Section Sec;
314      std::memset(&Sec.Header, 0, sizeof(Sec.Header));
315      yaml::MappingNode *SecMap = dyn_cast<yaml::MappingNode>(&*i);
316      if (!SecMap) {
317        YS.printError(&*i, "Section entry must be a map");
318        return false;
319      }
320      for (yaml::MappingNode::iterator si = SecMap->begin(), se = SecMap->end();
321                                       si != se; ++si) {
322        yaml::ScalarNode *Key = dyn_cast<yaml::ScalarNode>(si->getKey());
323        if (!Key) {
324          YS.printError(si->getKey(), "Keys must be scalar values");
325          return false;
326        }
327        SmallString<32> Storage;
328        StringRef KeyValue = Key->getValue(Storage);
329
330        yaml::ScalarNode *Value = dyn_cast<yaml::ScalarNode>(si->getValue());
331        if (KeyValue == "Name") {
332          // If the name is less than 8 bytes, store it in place, otherwise
333          // store it in the string table.
334          StringRef Name = Value->getValue(Storage);
335          std::fill_n(Sec.Header.Name, unsigned(COFF::NameSize), 0);
336          if (Name.size() <= COFF::NameSize) {
337            std::copy(Name.begin(), Name.end(), Sec.Header.Name);
338          } else {
339            // Add string to the string table and format the index for output.
340            unsigned Index = getStringIndex(Name);
341            std::string str = utostr(Index);
342            if (str.size() > 7) {
343              YS.printError(Value, "String table got too large");
344              return false;
345            }
346            Sec.Header.Name[0] = '/';
347            std::copy(str.begin(), str.end(), Sec.Header.Name + 1);
348          }
349        } else if (KeyValue == "VirtualSize") {
350          if (!getAs(Value, Sec.Header.VirtualSize)) {
351            YS.printError(Value, "Invalid value for VirtualSize");
352            return false;
353          }
354        } else if (KeyValue == "VirtualAddress") {
355          if (!getAs(Value, Sec.Header.VirtualAddress)) {
356            YS.printError(Value, "Invalid value for VirtualAddress");
357            return false;
358          }
359        } else if (KeyValue == "SizeOfRawData") {
360          if (!getAs(Value, Sec.Header.SizeOfRawData)) {
361            YS.printError(Value, "Invalid value for SizeOfRawData");
362            return false;
363          }
364        } else if (KeyValue == "PointerToRawData") {
365          if (!getAs(Value, Sec.Header.PointerToRawData)) {
366            YS.printError(Value, "Invalid value for PointerToRawData");
367            return false;
368          }
369        } else if (KeyValue == "PointerToRelocations") {
370          if (!getAs(Value, Sec.Header.PointerToRelocations)) {
371            YS.printError(Value, "Invalid value for PointerToRelocations");
372            return false;
373          }
374        } else if (KeyValue == "PointerToLineNumbers") {
375          if (!getAs(Value, Sec.Header.PointerToLineNumbers)) {
376            YS.printError(Value, "Invalid value for PointerToLineNumbers");
377            return false;
378          }
379        } else if (KeyValue == "NumberOfRelocations") {
380          if (!getAs(Value, Sec.Header.NumberOfRelocations)) {
381            YS.printError(Value, "Invalid value for NumberOfRelocations");
382            return false;
383          }
384        } else if (KeyValue == "NumberOfLineNumbers") {
385          if (!getAs(Value, Sec.Header.NumberOfLineNumbers)) {
386            YS.printError(Value, "Invalid value for NumberOfLineNumbers");
387            return false;
388          }
389        } else if (KeyValue == "Characteristics") {
390          yaml::SequenceNode *SeqValue
391            = dyn_cast<yaml::SequenceNode>(si->getValue());
392          if (!Value && !SeqValue) {
393            YS.printError(si->getValue(),
394              "Characteristics must either be a number or sequence");
395            return false;
396          }
397          if (Value) {
398            if (!getAs(Value, Sec.Header.Characteristics)) {
399              YS.printError(Value, "Invalid value for Characteristics");
400              return false;
401            }
402          } else {
403            for (yaml::SequenceNode::iterator ci = SeqValue->begin(),
404                                              ce = SeqValue->end();
405                                              ci != ce; ++ci) {
406              yaml::ScalarNode *CharValue = dyn_cast<yaml::ScalarNode>(&*ci);
407              if (!CharValue) {
408                YS.printError(CharValue, "Invalid value for Characteristics");
409                return false;
410              }
411              StringRef Char = CharValue->getValue(Storage);
412              uint32_t Characteristic =
413                StringSwitch<COFF::SectionCharacteristics>(Char)
414                .Case( "IMAGE_SCN_TYPE_NO_PAD"
415                     , COFF::IMAGE_SCN_TYPE_NO_PAD)
416                .Case( "IMAGE_SCN_CNT_CODE"
417                     , COFF::IMAGE_SCN_CNT_CODE)
418                .Case( "IMAGE_SCN_CNT_INITIALIZED_DATA"
419                     , COFF::IMAGE_SCN_CNT_INITIALIZED_DATA)
420                .Case( "IMAGE_SCN_CNT_UNINITIALIZED_DATA"
421                     , COFF::IMAGE_SCN_CNT_UNINITIALIZED_DATA)
422                .Case( "IMAGE_SCN_LNK_OTHER"
423                     , COFF::IMAGE_SCN_LNK_OTHER)
424                .Case( "IMAGE_SCN_LNK_INFO"
425                     , COFF::IMAGE_SCN_LNK_INFO)
426                .Case( "IMAGE_SCN_LNK_REMOVE"
427                     , COFF::IMAGE_SCN_LNK_REMOVE)
428                .Case( "IMAGE_SCN_LNK_COMDAT"
429                     , COFF::IMAGE_SCN_LNK_COMDAT)
430                .Case( "IMAGE_SCN_GPREL"
431                     , COFF::IMAGE_SCN_GPREL)
432                .Case( "IMAGE_SCN_MEM_PURGEABLE"
433                     , COFF::IMAGE_SCN_MEM_PURGEABLE)
434                .Case( "IMAGE_SCN_MEM_16BIT"
435                     , COFF::IMAGE_SCN_MEM_16BIT)
436                .Case( "IMAGE_SCN_MEM_LOCKED"
437                     , COFF::IMAGE_SCN_MEM_LOCKED)
438                .Case( "IMAGE_SCN_MEM_PRELOAD"
439                     , COFF::IMAGE_SCN_MEM_PRELOAD)
440                .Case( "IMAGE_SCN_ALIGN_1BYTES"
441                     , COFF::IMAGE_SCN_ALIGN_1BYTES)
442                .Case( "IMAGE_SCN_ALIGN_2BYTES"
443                     , COFF::IMAGE_SCN_ALIGN_2BYTES)
444                .Case( "IMAGE_SCN_ALIGN_4BYTES"
445                     , COFF::IMAGE_SCN_ALIGN_4BYTES)
446                .Case( "IMAGE_SCN_ALIGN_8BYTES"
447                     , COFF::IMAGE_SCN_ALIGN_8BYTES)
448                .Case( "IMAGE_SCN_ALIGN_16BYTES"
449                     , COFF::IMAGE_SCN_ALIGN_16BYTES)
450                .Case( "IMAGE_SCN_ALIGN_32BYTES"
451                     , COFF::IMAGE_SCN_ALIGN_32BYTES)
452                .Case( "IMAGE_SCN_ALIGN_64BYTES"
453                     , COFF::IMAGE_SCN_ALIGN_64BYTES)
454                .Case( "IMAGE_SCN_ALIGN_128BYTES"
455                     , COFF::IMAGE_SCN_ALIGN_128BYTES)
456                .Case( "IMAGE_SCN_ALIGN_256BYTES"
457                     , COFF::IMAGE_SCN_ALIGN_256BYTES)
458                .Case( "IMAGE_SCN_ALIGN_512BYTES"
459                     , COFF::IMAGE_SCN_ALIGN_512BYTES)
460                .Case( "IMAGE_SCN_ALIGN_1024BYTES"
461                     , COFF::IMAGE_SCN_ALIGN_1024BYTES)
462                .Case( "IMAGE_SCN_ALIGN_2048BYTES"
463                     , COFF::IMAGE_SCN_ALIGN_2048BYTES)
464                .Case( "IMAGE_SCN_ALIGN_4096BYTES"
465                     , COFF::IMAGE_SCN_ALIGN_4096BYTES)
466                .Case( "IMAGE_SCN_ALIGN_8192BYTES"
467                     , COFF::IMAGE_SCN_ALIGN_8192BYTES)
468                .Case( "IMAGE_SCN_LNK_NRELOC_OVFL"
469                     , COFF::IMAGE_SCN_LNK_NRELOC_OVFL)
470                .Case( "IMAGE_SCN_MEM_DISCARDABLE"
471                     , COFF::IMAGE_SCN_MEM_DISCARDABLE)
472                .Case( "IMAGE_SCN_MEM_NOT_CACHED"
473                     , COFF::IMAGE_SCN_MEM_NOT_CACHED)
474                .Case( "IMAGE_SCN_MEM_NOT_PAGED"
475                     , COFF::IMAGE_SCN_MEM_NOT_PAGED)
476                .Case( "IMAGE_SCN_MEM_SHARED"
477                     , COFF::IMAGE_SCN_MEM_SHARED)
478                .Case( "IMAGE_SCN_MEM_EXECUTE"
479                     , COFF::IMAGE_SCN_MEM_EXECUTE)
480                .Case( "IMAGE_SCN_MEM_READ"
481                     , COFF::IMAGE_SCN_MEM_READ)
482                .Case( "IMAGE_SCN_MEM_WRITE"
483                     , COFF::IMAGE_SCN_MEM_WRITE)
484                .Default(COFF::SC_Invalid);
485              if (Characteristic == COFF::SC_Invalid) {
486                YS.printError(CharValue, "Invalid value for Characteristic");
487                return false;
488              }
489              Sec.Header.Characteristics |= Characteristic;
490            }
491          }
492        } else if (KeyValue == "SectionData") {
493          yaml::ScalarNode *Value = dyn_cast<yaml::ScalarNode>(si->getValue());
494          SmallString<32> Storage;
495          StringRef Data = Value->getValue(Storage);
496          if (!hexStringToByteArray(Data, Sec.Data)) {
497            YS.printError(Value, "SectionData must be a collection of pairs of"
498                                 "hex bytes");
499            return false;
500          }
501        } else
502          si->skip();
503      }
504      Sections.push_back(Sec);
505    }
506    return true;
507  }
508
509  bool parseSymbols(yaml::Node *SymbolsN) {
510    yaml::SequenceNode *SN = dyn_cast<yaml::SequenceNode>(SymbolsN);
511    if (!SN) {
512      YS.printError(SymbolsN, "Symbols must be a sequence");
513      return false;
514    }
515    for (yaml::SequenceNode::iterator i = SN->begin(), e = SN->end();
516                                      i != e; ++i) {
517      Symbol Sym;
518      std::memset(&Sym.Header, 0, sizeof(Sym.Header));
519      yaml::MappingNode *SymMap = dyn_cast<yaml::MappingNode>(&*i);
520      if (!SymMap) {
521        YS.printError(&*i, "Symbol must be a map");
522        return false;
523      }
524      for (yaml::MappingNode::iterator si = SymMap->begin(), se = SymMap->end();
525                                       si != se; ++si) {
526        yaml::ScalarNode *Key = dyn_cast<yaml::ScalarNode>(si->getKey());
527        if (!Key) {
528          YS.printError(si->getKey(), "Keys must be scalar values");
529          return false;
530        }
531        SmallString<32> Storage;
532        StringRef KeyValue = Key->getValue(Storage);
533
534        yaml::ScalarNode *Value = dyn_cast<yaml::ScalarNode>(si->getValue());
535        if (!Value) {
536          YS.printError(si->getValue(), "Must be a scalar value");
537          return false;
538        }
539        if (KeyValue == "Name") {
540          // If the name is less than 8 bytes, store it in place, otherwise
541          // store it in the string table.
542          StringRef Name = Value->getValue(Storage);
543          std::fill_n(Sym.Header.Name, unsigned(COFF::NameSize), 0);
544          if (Name.size() <= COFF::NameSize) {
545            std::copy(Name.begin(), Name.end(), Sym.Header.Name);
546          } else {
547            // Add string to the string table and format the index for output.
548            unsigned Index = getStringIndex(Name);
549            *reinterpret_cast<support::aligned_ulittle32_t*>(
550              Sym.Header.Name + 4) = Index;
551          }
552        } else if (KeyValue == "Value") {
553          if (!getAs(Value, Sym.Header.Value)) {
554            YS.printError(Value, "Invalid value for Value");
555            return false;
556          }
557        } else if (KeyValue == "SimpleType") {
558          Sym.Header.Type |= StringSwitch<COFF::SymbolBaseType>(
559            Value->getValue(Storage))
560            .Case("IMAGE_SYM_TYPE_NULL", COFF::IMAGE_SYM_TYPE_NULL)
561            .Case("IMAGE_SYM_TYPE_VOID", COFF::IMAGE_SYM_TYPE_VOID)
562            .Case("IMAGE_SYM_TYPE_CHAR", COFF::IMAGE_SYM_TYPE_CHAR)
563            .Case("IMAGE_SYM_TYPE_SHORT", COFF::IMAGE_SYM_TYPE_SHORT)
564            .Case("IMAGE_SYM_TYPE_INT", COFF::IMAGE_SYM_TYPE_INT)
565            .Case("IMAGE_SYM_TYPE_LONG", COFF::IMAGE_SYM_TYPE_LONG)
566            .Case("IMAGE_SYM_TYPE_FLOAT", COFF::IMAGE_SYM_TYPE_FLOAT)
567            .Case("IMAGE_SYM_TYPE_DOUBLE", COFF::IMAGE_SYM_TYPE_DOUBLE)
568            .Case("IMAGE_SYM_TYPE_STRUCT", COFF::IMAGE_SYM_TYPE_STRUCT)
569            .Case("IMAGE_SYM_TYPE_UNION", COFF::IMAGE_SYM_TYPE_UNION)
570            .Case("IMAGE_SYM_TYPE_ENUM", COFF::IMAGE_SYM_TYPE_ENUM)
571            .Case("IMAGE_SYM_TYPE_MOE", COFF::IMAGE_SYM_TYPE_MOE)
572            .Case("IMAGE_SYM_TYPE_BYTE", COFF::IMAGE_SYM_TYPE_BYTE)
573            .Case("IMAGE_SYM_TYPE_WORD", COFF::IMAGE_SYM_TYPE_WORD)
574            .Case("IMAGE_SYM_TYPE_UINT", COFF::IMAGE_SYM_TYPE_UINT)
575            .Case("IMAGE_SYM_TYPE_DWORD", COFF::IMAGE_SYM_TYPE_DWORD)
576            .Default(COFF::IMAGE_SYM_TYPE_NULL);
577        } else if (KeyValue == "ComplexType") {
578          Sym.Header.Type |= StringSwitch<COFF::SymbolComplexType>(
579            Value->getValue(Storage))
580            .Case("IMAGE_SYM_DTYPE_NULL", COFF::IMAGE_SYM_DTYPE_NULL)
581            .Case("IMAGE_SYM_DTYPE_POINTER", COFF::IMAGE_SYM_DTYPE_POINTER)
582            .Case("IMAGE_SYM_DTYPE_FUNCTION", COFF::IMAGE_SYM_DTYPE_FUNCTION)
583            .Case("IMAGE_SYM_DTYPE_ARRAY", COFF::IMAGE_SYM_DTYPE_ARRAY)
584            .Default(COFF::IMAGE_SYM_DTYPE_NULL)
585            << COFF::SCT_COMPLEX_TYPE_SHIFT;
586        } else if (KeyValue == "StorageClass") {
587          Sym.Header.StorageClass = StringSwitch<COFF::SymbolStorageClass>(
588            Value->getValue(Storage))
589            .Case( "IMAGE_SYM_CLASS_END_OF_FUNCTION"
590                 , COFF::IMAGE_SYM_CLASS_END_OF_FUNCTION)
591            .Case( "IMAGE_SYM_CLASS_NULL"
592                 , COFF::IMAGE_SYM_CLASS_NULL)
593            .Case( "IMAGE_SYM_CLASS_AUTOMATIC"
594                 , COFF::IMAGE_SYM_CLASS_AUTOMATIC)
595            .Case( "IMAGE_SYM_CLASS_EXTERNAL"
596                 , COFF::IMAGE_SYM_CLASS_EXTERNAL)
597            .Case( "IMAGE_SYM_CLASS_STATIC"
598                 , COFF::IMAGE_SYM_CLASS_STATIC)
599            .Case( "IMAGE_SYM_CLASS_REGISTER"
600                 , COFF::IMAGE_SYM_CLASS_REGISTER)
601            .Case( "IMAGE_SYM_CLASS_EXTERNAL_DEF"
602                 , COFF::IMAGE_SYM_CLASS_EXTERNAL_DEF)
603            .Case( "IMAGE_SYM_CLASS_LABEL"
604                 , COFF::IMAGE_SYM_CLASS_LABEL)
605            .Case( "IMAGE_SYM_CLASS_UNDEFINED_LABEL"
606                 , COFF::IMAGE_SYM_CLASS_UNDEFINED_LABEL)
607            .Case( "IMAGE_SYM_CLASS_MEMBER_OF_STRUCT"
608                 , COFF::IMAGE_SYM_CLASS_MEMBER_OF_STRUCT)
609            .Case( "IMAGE_SYM_CLASS_ARGUMENT"
610                 , COFF::IMAGE_SYM_CLASS_ARGUMENT)
611            .Case( "IMAGE_SYM_CLASS_STRUCT_TAG"
612                 , COFF::IMAGE_SYM_CLASS_STRUCT_TAG)
613            .Case( "IMAGE_SYM_CLASS_MEMBER_OF_UNION"
614                 , COFF::IMAGE_SYM_CLASS_MEMBER_OF_UNION)
615            .Case( "IMAGE_SYM_CLASS_UNION_TAG"
616                 , COFF::IMAGE_SYM_CLASS_UNION_TAG)
617            .Case( "IMAGE_SYM_CLASS_TYPE_DEFINITION"
618                 , COFF::IMAGE_SYM_CLASS_TYPE_DEFINITION)
619            .Case( "IMAGE_SYM_CLASS_UNDEFINED_STATIC"
620                 , COFF::IMAGE_SYM_CLASS_UNDEFINED_STATIC)
621            .Case( "IMAGE_SYM_CLASS_ENUM_TAG"
622                 , COFF::IMAGE_SYM_CLASS_ENUM_TAG)
623            .Case( "IMAGE_SYM_CLASS_MEMBER_OF_ENUM"
624                 , COFF::IMAGE_SYM_CLASS_MEMBER_OF_ENUM)
625            .Case( "IMAGE_SYM_CLASS_REGISTER_PARAM"
626                 , COFF::IMAGE_SYM_CLASS_REGISTER_PARAM)
627            .Case( "IMAGE_SYM_CLASS_BIT_FIELD"
628                 , COFF::IMAGE_SYM_CLASS_BIT_FIELD)
629            .Case( "IMAGE_SYM_CLASS_BLOCK"
630                 , COFF::IMAGE_SYM_CLASS_BLOCK)
631            .Case( "IMAGE_SYM_CLASS_FUNCTION"
632                 , COFF::IMAGE_SYM_CLASS_FUNCTION)
633            .Case( "IMAGE_SYM_CLASS_END_OF_STRUCT"
634                 , COFF::IMAGE_SYM_CLASS_END_OF_STRUCT)
635            .Case( "IMAGE_SYM_CLASS_FILE"
636                 , COFF::IMAGE_SYM_CLASS_FILE)
637            .Case( "IMAGE_SYM_CLASS_SECTION"
638                 , COFF::IMAGE_SYM_CLASS_SECTION)
639            .Case( "IMAGE_SYM_CLASS_WEAK_EXTERNAL"
640                 , COFF::IMAGE_SYM_CLASS_WEAK_EXTERNAL)
641            .Case( "IMAGE_SYM_CLASS_CLR_TOKEN"
642                 , COFF::IMAGE_SYM_CLASS_CLR_TOKEN)
643            .Default(COFF::SSC_Invalid);
644          if (Sym.Header.StorageClass == COFF::SSC_Invalid) {
645            YS.printError(Value, "Invalid value for StorageClass");
646            return false;
647          }
648        } else if (KeyValue == "SectionNumber") {
649          if (!getAs(Value, Sym.Header.SectionNumber)) {
650              YS.printError(Value, "Invalid value for SectionNumber");
651              return false;
652          }
653        } else if (KeyValue == "AuxillaryData") {
654          StringRef Data = Value->getValue(Storage);
655          if (!hexStringToByteArray(Data, Sym.AuxSymbols)) {
656            YS.printError(Value, "AuxillaryData must be a collection of pairs"
657                                 "of hex bytes");
658            return false;
659          }
660        } else
661          si->skip();
662      }
663      Symbols.push_back(Sym);
664    }
665    return true;
666  }
667
668  bool parse() {
669    yaml::Document &D = *YS.begin();
670    yaml::MappingNode *Root = dyn_cast<yaml::MappingNode>(D.getRoot());
671    if (!Root) {
672      YS.printError(D.getRoot(), "Root node must be a map");
673      return false;
674    }
675    for (yaml::MappingNode::iterator i = Root->begin(), e = Root->end();
676                                     i != e; ++i) {
677      yaml::ScalarNode *Key = dyn_cast<yaml::ScalarNode>(i->getKey());
678      if (!Key) {
679        YS.printError(i->getKey(), "Keys must be scalar values");
680        return false;
681      }
682      SmallString<32> Storage;
683      StringRef KeyValue = Key->getValue(Storage);
684      if (KeyValue == "header") {
685        if (!parseHeader(i->getValue()))
686          return false;
687      } else if (KeyValue == "sections") {
688        if (!parseSections(i->getValue()))
689          return false;
690      } else if (KeyValue == "symbols") {
691        if (!parseSymbols(i->getValue()))
692          return false;
693      }
694    }
695    return !YS.failed();
696  }
697
698  unsigned getStringIndex(StringRef Str) {
699    StringMap<unsigned>::iterator i = StringTableMap.find(Str);
700    if (i == StringTableMap.end()) {
701      unsigned Index = StringTable.size();
702      StringTable.append(Str.begin(), Str.end());
703      StringTable.push_back(0);
704      StringTableMap[Str] = Index;
705      return Index;
706    }
707    return i->second;
708  }
709
710  yaml::Stream &YS;
711  COFF::header Header;
712
713  struct Section {
714    COFF::section Header;
715    std::vector<uint8_t> Data;
716    std::vector<COFF::relocation> Relocations;
717  };
718
719  struct Symbol {
720    COFF::symbol Header;
721    std::vector<uint8_t> AuxSymbols;
722  };
723
724  std::vector<Section> Sections;
725  std::vector<Symbol> Symbols;
726  StringMap<unsigned> StringTableMap;
727  std::string StringTable;
728};
729
730// Take a CP and assign addresses and sizes to everything. Returns false if the
731// layout is not valid to do.
732static bool layoutCOFF(COFFParser &CP) {
733  uint32_t SectionTableStart = 0;
734  uint32_t SectionTableSize  = 0;
735
736  // The section table starts immediately after the header, including the
737  // optional header.
738  SectionTableStart = sizeof(COFF::header) + CP.Header.SizeOfOptionalHeader;
739  SectionTableSize = sizeof(COFF::section) * CP.Sections.size();
740
741  uint32_t CurrentSectionDataOffset = SectionTableStart + SectionTableSize;
742
743  // Assign each section data address consecutively.
744  for (std::vector<COFFParser::Section>::iterator i = CP.Sections.begin(),
745                                                  e = CP.Sections.end();
746                                                  i != e; ++i) {
747    if (!i->Data.empty()) {
748      i->Header.SizeOfRawData = i->Data.size();
749      i->Header.PointerToRawData = CurrentSectionDataOffset;
750      CurrentSectionDataOffset += i->Header.SizeOfRawData;
751      // TODO: Handle alignment.
752    } else {
753      i->Header.SizeOfRawData = 0;
754      i->Header.PointerToRawData = 0;
755    }
756  }
757
758  uint32_t SymbolTableStart = CurrentSectionDataOffset;
759
760  // Calculate number of symbols.
761  uint32_t NumberOfSymbols = 0;
762  for (std::vector<COFFParser::Symbol>::iterator i = CP.Symbols.begin(),
763                                                 e = CP.Symbols.end();
764                                                 i != e; ++i) {
765    if (i->AuxSymbols.size() % COFF::SymbolSize != 0) {
766      errs() << "AuxillaryData size not a multiple of symbol size!\n";
767      return false;
768    }
769    i->Header.NumberOfAuxSymbols = i->AuxSymbols.size() / COFF::SymbolSize;
770    NumberOfSymbols += 1 + i->Header.NumberOfAuxSymbols;
771  }
772
773  // Store all the allocated start addresses in the header.
774  CP.Header.NumberOfSections = CP.Sections.size();
775  CP.Header.NumberOfSymbols = NumberOfSymbols;
776  CP.Header.PointerToSymbolTable = SymbolTableStart;
777
778  *reinterpret_cast<support::ulittle32_t *>(&CP.StringTable[0])
779    = CP.StringTable.size();
780
781  return true;
782}
783
784template <typename value_type>
785struct binary_le_impl {
786  value_type Value;
787  binary_le_impl(value_type V) : Value(V) {}
788};
789
790template <typename value_type>
791raw_ostream &operator <<( raw_ostream &OS
792                        , const binary_le_impl<value_type> &BLE) {
793  char Buffer[sizeof(BLE.Value)];
794  support::endian::write_le<value_type, support::unaligned>(Buffer, BLE.Value);
795  OS.write(Buffer, sizeof(BLE.Value));
796  return OS;
797}
798
799template <typename value_type>
800binary_le_impl<value_type> binary_le(value_type V) {
801  return binary_le_impl<value_type>(V);
802}
803
804void writeCOFF(COFFParser &CP, raw_ostream &OS) {
805  OS << binary_le(CP.Header.Machine)
806     << binary_le(CP.Header.NumberOfSections)
807     << binary_le(CP.Header.TimeDateStamp)
808     << binary_le(CP.Header.PointerToSymbolTable)
809     << binary_le(CP.Header.NumberOfSymbols)
810     << binary_le(CP.Header.SizeOfOptionalHeader)
811     << binary_le(CP.Header.Characteristics);
812
813  // Output section table.
814  for (std::vector<COFFParser::Section>::const_iterator i = CP.Sections.begin(),
815                                                        e = CP.Sections.end();
816                                                        i != e; ++i) {
817    OS.write(i->Header.Name, COFF::NameSize);
818    OS << binary_le(i->Header.VirtualSize)
819       << binary_le(i->Header.VirtualAddress)
820       << binary_le(i->Header.SizeOfRawData)
821       << binary_le(i->Header.PointerToRawData)
822       << binary_le(i->Header.PointerToRelocations)
823       << binary_le(i->Header.PointerToLineNumbers)
824       << binary_le(i->Header.NumberOfRelocations)
825       << binary_le(i->Header.NumberOfLineNumbers)
826       << binary_le(i->Header.Characteristics);
827  }
828
829  // Output section data.
830  for (std::vector<COFFParser::Section>::const_iterator i = CP.Sections.begin(),
831                                                        e = CP.Sections.end();
832                                                        i != e; ++i) {
833    if (!i->Data.empty())
834      OS.write(reinterpret_cast<const char*>(&i->Data[0]), i->Data.size());
835  }
836
837  // Output symbol table.
838
839  for (std::vector<COFFParser::Symbol>::const_iterator i = CP.Symbols.begin(),
840                                                       e = CP.Symbols.end();
841                                                       i != e; ++i) {
842    OS.write(i->Header.Name, COFF::NameSize);
843    OS << binary_le(i->Header.Value)
844       << binary_le(i->Header.SectionNumber)
845       << binary_le(i->Header.Type)
846       << binary_le(i->Header.StorageClass)
847       << binary_le(i->Header.NumberOfAuxSymbols);
848    if (!i->AuxSymbols.empty())
849      OS.write( reinterpret_cast<const char*>(&i->AuxSymbols[0])
850              , i->AuxSymbols.size());
851  }
852
853  // Output string table.
854  OS.write(&CP.StringTable[0], CP.StringTable.size());
855}
856
857int main(int argc, char **argv) {
858  cl::ParseCommandLineOptions(argc, argv);
859  sys::PrintStackTraceOnErrorSignal();
860  PrettyStackTraceProgram X(argc, argv);
861  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
862
863  OwningPtr<MemoryBuffer> Buf;
864  if (MemoryBuffer::getFileOrSTDIN(Input, Buf))
865    return 1;
866
867  SourceMgr SM;
868  yaml::Stream S(Buf->getBuffer(), SM);
869  COFFParser CP(S);
870  if (!CP.parse()) {
871    errs() << "yaml2obj: Failed to parse YAML file!\n";
872    return 1;
873  }
874  if (!layoutCOFF(CP)) {
875    errs() << "yaml2obj: Failed to layout COFF file!\n";
876    return 1;
877  }
878  writeCOFF(CP, outs());
879}
880