13add3e9cSRafael Espindola //===- yaml2obj - Convert YAML to a binary object file --------------------===//
23add3e9cSRafael Espindola //
33add3e9cSRafael Espindola //                     The LLVM Compiler Infrastructure
43add3e9cSRafael Espindola //
53add3e9cSRafael Espindola // This file is distributed under the University of Illinois Open Source
63add3e9cSRafael Espindola // License. See LICENSE.TXT for details.
73add3e9cSRafael Espindola //
83add3e9cSRafael Espindola //===----------------------------------------------------------------------===//
93add3e9cSRafael Espindola //
103add3e9cSRafael Espindola // This program takes a YAML description of an object file and outputs the
113add3e9cSRafael Espindola // binary equivalent.
123add3e9cSRafael Espindola //
133add3e9cSRafael Espindola // This is used for writing tests that require binary files.
143add3e9cSRafael Espindola //
153add3e9cSRafael Espindola //===----------------------------------------------------------------------===//
163add3e9cSRafael Espindola 
173add3e9cSRafael Espindola #include "llvm/ADT/SmallString.h"
183add3e9cSRafael Espindola #include "llvm/ADT/StringExtras.h"
193add3e9cSRafael Espindola #include "llvm/ADT/StringMap.h"
203add3e9cSRafael Espindola #include "llvm/ADT/StringSwitch.h"
21*f5bb53f1SRafael Espindola #include "llvm/Object/COFFYaml.h"
223add3e9cSRafael Espindola #include "llvm/Support/Casting.h"
233add3e9cSRafael Espindola #include "llvm/Support/CommandLine.h"
243add3e9cSRafael Espindola #include "llvm/Support/Endian.h"
253add3e9cSRafael Espindola #include "llvm/Support/ManagedStatic.h"
263add3e9cSRafael Espindola #include "llvm/Support/MemoryBuffer.h"
273add3e9cSRafael Espindola #include "llvm/Support/PrettyStackTrace.h"
283add3e9cSRafael Espindola #include "llvm/Support/Signals.h"
293add3e9cSRafael Espindola #include "llvm/Support/SourceMgr.h"
303add3e9cSRafael Espindola #include "llvm/Support/raw_ostream.h"
313add3e9cSRafael Espindola #include "llvm/Support/system_error.h"
323add3e9cSRafael Espindola #include <vector>
333add3e9cSRafael Espindola 
343add3e9cSRafael Espindola using namespace llvm;
353add3e9cSRafael Espindola 
363add3e9cSRafael Espindola static cl::opt<std::string>
373add3e9cSRafael Espindola   Input(cl::Positional, cl::desc("<input>"), cl::init("-"));
383add3e9cSRafael Espindola 
393add3e9cSRafael Espindola /// This parses a yaml stream that represents a COFF object file.
403add3e9cSRafael Espindola /// See docs/yaml2obj for the yaml scheema.
413add3e9cSRafael Espindola struct COFFParser {
423add3e9cSRafael Espindola   COFFParser(COFFYAML::Object &Obj) : Obj(Obj) {
433add3e9cSRafael Espindola     // A COFF string table always starts with a 4 byte size field. Offsets into
443add3e9cSRafael Espindola     // it include this size, so allocate it now.
453add3e9cSRafael Espindola     StringTable.append(4, 0);
463add3e9cSRafael Espindola   }
473add3e9cSRafael Espindola 
483add3e9cSRafael Espindola   bool parseSections() {
493add3e9cSRafael Espindola     for (std::vector<COFFYAML::Section>::iterator i = Obj.Sections.begin(),
503add3e9cSRafael Espindola            e = Obj.Sections.end(); i != e; ++i) {
51ad3b9753SRafael Espindola       COFFYAML::Section &Sec = *i;
523add3e9cSRafael Espindola 
533add3e9cSRafael Espindola       // If the name is less than 8 bytes, store it in place, otherwise
543add3e9cSRafael Espindola       // store it in the string table.
55ad3b9753SRafael Espindola       StringRef Name = Sec.Name;
56ad3b9753SRafael Espindola 
573add3e9cSRafael Espindola       if (Name.size() <= COFF::NameSize) {
583add3e9cSRafael Espindola         std::copy(Name.begin(), Name.end(), Sec.Header.Name);
593add3e9cSRafael Espindola       } else {
603add3e9cSRafael Espindola         // Add string to the string table and format the index for output.
613add3e9cSRafael Espindola         unsigned Index = getStringIndex(Name);
623add3e9cSRafael Espindola         std::string str = utostr(Index);
633add3e9cSRafael Espindola         if (str.size() > 7) {
643add3e9cSRafael Espindola           errs() << "String table got too large";
653add3e9cSRafael Espindola           return false;
663add3e9cSRafael Espindola         }
673add3e9cSRafael Espindola         Sec.Header.Name[0] = '/';
683add3e9cSRafael Espindola         std::copy(str.begin(), str.end(), Sec.Header.Name + 1);
693add3e9cSRafael Espindola       }
70fa4513a1SRafael Espindola 
71fa4513a1SRafael Espindola       Sec.Header.Characteristics |= (Log2_32(Sec.Alignment) + 1) << 20;
723add3e9cSRafael Espindola     }
733add3e9cSRafael Espindola     return true;
743add3e9cSRafael Espindola   }
753add3e9cSRafael Espindola 
763add3e9cSRafael Espindola   bool parseSymbols() {
773add3e9cSRafael Espindola     for (std::vector<COFFYAML::Symbol>::iterator i = Obj.Symbols.begin(),
783add3e9cSRafael Espindola            e = Obj.Symbols.end(); i != e; ++i) {
79ad3b9753SRafael Espindola       COFFYAML::Symbol &Sym = *i;
803add3e9cSRafael Espindola 
813add3e9cSRafael Espindola       // If the name is less than 8 bytes, store it in place, otherwise
823add3e9cSRafael Espindola       // store it in the string table.
83ad3b9753SRafael Espindola       StringRef Name = Sym.Name;
843add3e9cSRafael Espindola       if (Name.size() <= COFF::NameSize) {
853add3e9cSRafael Espindola         std::copy(Name.begin(), Name.end(), Sym.Header.Name);
863add3e9cSRafael Espindola       } else {
873add3e9cSRafael Espindola         // Add string to the string table and format the index for output.
883add3e9cSRafael Espindola         unsigned Index = getStringIndex(Name);
893add3e9cSRafael Espindola         *reinterpret_cast<support::aligned_ulittle32_t*>(
903add3e9cSRafael Espindola             Sym.Header.Name + 4) = Index;
913add3e9cSRafael Espindola       }
923add3e9cSRafael Espindola 
93ad3b9753SRafael Espindola       Sym.Header.Type = Sym.SimpleType;
94ad3b9753SRafael Espindola       Sym.Header.Type |= Sym.ComplexType << COFF::SCT_COMPLEX_TYPE_SHIFT;
953add3e9cSRafael Espindola     }
963add3e9cSRafael Espindola     return true;
973add3e9cSRafael Espindola   }
983add3e9cSRafael Espindola 
993add3e9cSRafael Espindola   bool parse() {
1003add3e9cSRafael Espindola     if (!parseSections())
1013add3e9cSRafael Espindola       return false;
1023add3e9cSRafael Espindola     if (!parseSymbols())
1033add3e9cSRafael Espindola       return false;
1043add3e9cSRafael Espindola     return true;
1053add3e9cSRafael Espindola   }
1063add3e9cSRafael Espindola 
1073add3e9cSRafael Espindola   unsigned getStringIndex(StringRef Str) {
1083add3e9cSRafael Espindola     StringMap<unsigned>::iterator i = StringTableMap.find(Str);
1093add3e9cSRafael Espindola     if (i == StringTableMap.end()) {
1103add3e9cSRafael Espindola       unsigned Index = StringTable.size();
1113add3e9cSRafael Espindola       StringTable.append(Str.begin(), Str.end());
1123add3e9cSRafael Espindola       StringTable.push_back(0);
1133add3e9cSRafael Espindola       StringTableMap[Str] = Index;
1143add3e9cSRafael Espindola       return Index;
1153add3e9cSRafael Espindola     }
1163add3e9cSRafael Espindola     return i->second;
1173add3e9cSRafael Espindola   }
1183add3e9cSRafael Espindola 
1193add3e9cSRafael Espindola   COFFYAML::Object &Obj;
1203add3e9cSRafael Espindola 
1213add3e9cSRafael Espindola   StringMap<unsigned> StringTableMap;
1223add3e9cSRafael Espindola   std::string StringTable;
1233add3e9cSRafael Espindola };
1243add3e9cSRafael Espindola 
1253add3e9cSRafael Espindola // Take a CP and assign addresses and sizes to everything. Returns false if the
1263add3e9cSRafael Espindola // layout is not valid to do.
1273add3e9cSRafael Espindola static bool layoutCOFF(COFFParser &CP) {
1283add3e9cSRafael Espindola   uint32_t SectionTableStart = 0;
1293add3e9cSRafael Espindola   uint32_t SectionTableSize  = 0;
1303add3e9cSRafael Espindola 
1313add3e9cSRafael Espindola   // The section table starts immediately after the header, including the
1323add3e9cSRafael Espindola   // optional header.
133ad3b9753SRafael Espindola   SectionTableStart = sizeof(COFF::header) + CP.Obj.Header.SizeOfOptionalHeader;
134ad3b9753SRafael Espindola   SectionTableSize = sizeof(COFF::section) * CP.Obj.Sections.size();
1353add3e9cSRafael Espindola 
1363add3e9cSRafael Espindola   uint32_t CurrentSectionDataOffset = SectionTableStart + SectionTableSize;
1373add3e9cSRafael Espindola 
1383add3e9cSRafael Espindola   // Assign each section data address consecutively.
139ad3b9753SRafael Espindola   for (std::vector<COFFYAML::Section>::iterator i = CP.Obj.Sections.begin(),
140ad3b9753SRafael Espindola                                                 e = CP.Obj.Sections.end();
1413add3e9cSRafael Espindola                                                 i != e; ++i) {
142ad3b9753SRafael Espindola     if (!i->SectionData.empty()) {
143ad3b9753SRafael Espindola       i->Header.SizeOfRawData = i->SectionData.size()/2;
1443add3e9cSRafael Espindola       i->Header.PointerToRawData = CurrentSectionDataOffset;
1453add3e9cSRafael Espindola       CurrentSectionDataOffset += i->Header.SizeOfRawData;
146b716e622SRafael Espindola       if (!i->Relocations.empty()) {
147b716e622SRafael Espindola         i->Header.PointerToRelocations = CurrentSectionDataOffset;
148b716e622SRafael Espindola         i->Header.NumberOfRelocations = i->Relocations.size();
149b716e622SRafael Espindola         CurrentSectionDataOffset += i->Header.NumberOfRelocations *
150b716e622SRafael Espindola           COFF::RelocationSize;
151b716e622SRafael Espindola       }
1523add3e9cSRafael Espindola       // TODO: Handle alignment.
1533add3e9cSRafael Espindola     } else {
1543add3e9cSRafael Espindola       i->Header.SizeOfRawData = 0;
1553add3e9cSRafael Espindola       i->Header.PointerToRawData = 0;
1563add3e9cSRafael Espindola     }
1573add3e9cSRafael Espindola   }
1583add3e9cSRafael Espindola 
1593add3e9cSRafael Espindola   uint32_t SymbolTableStart = CurrentSectionDataOffset;
1603add3e9cSRafael Espindola 
1613add3e9cSRafael Espindola   // Calculate number of symbols.
1623add3e9cSRafael Espindola   uint32_t NumberOfSymbols = 0;
163ad3b9753SRafael Espindola   for (std::vector<COFFYAML::Symbol>::iterator i = CP.Obj.Symbols.begin(),
164ad3b9753SRafael Espindola                                                e = CP.Obj.Symbols.end();
1653add3e9cSRafael Espindola                                                i != e; ++i) {
1667f08d1b9SRafael Espindola     unsigned AuxBytes = i->AuxiliaryData.size() / 2;
167ad3b9753SRafael Espindola     if (AuxBytes % COFF::SymbolSize != 0) {
1687f08d1b9SRafael Espindola       errs() << "AuxiliaryData size not a multiple of symbol size!\n";
1693add3e9cSRafael Espindola       return false;
1703add3e9cSRafael Espindola     }
171ad3b9753SRafael Espindola     i->Header.NumberOfAuxSymbols = AuxBytes / COFF::SymbolSize;
1723add3e9cSRafael Espindola     NumberOfSymbols += 1 + i->Header.NumberOfAuxSymbols;
1733add3e9cSRafael Espindola   }
1743add3e9cSRafael Espindola 
1753add3e9cSRafael Espindola   // Store all the allocated start addresses in the header.
176ad3b9753SRafael Espindola   CP.Obj.Header.NumberOfSections = CP.Obj.Sections.size();
177ad3b9753SRafael Espindola   CP.Obj.Header.NumberOfSymbols = NumberOfSymbols;
178ad3b9753SRafael Espindola   CP.Obj.Header.PointerToSymbolTable = SymbolTableStart;
1793add3e9cSRafael Espindola 
1803add3e9cSRafael Espindola   *reinterpret_cast<support::ulittle32_t *>(&CP.StringTable[0])
1813add3e9cSRafael Espindola     = CP.StringTable.size();
1823add3e9cSRafael Espindola 
1833add3e9cSRafael Espindola   return true;
1843add3e9cSRafael Espindola }
1853add3e9cSRafael Espindola 
1863add3e9cSRafael Espindola template <typename value_type>
1873add3e9cSRafael Espindola struct binary_le_impl {
1883add3e9cSRafael Espindola   value_type Value;
1893add3e9cSRafael Espindola   binary_le_impl(value_type V) : Value(V) {}
1903add3e9cSRafael Espindola };
1913add3e9cSRafael Espindola 
1923add3e9cSRafael Espindola template <typename value_type>
1933add3e9cSRafael Espindola raw_ostream &operator <<( raw_ostream &OS
1943add3e9cSRafael Espindola                         , const binary_le_impl<value_type> &BLE) {
1953add3e9cSRafael Espindola   char Buffer[sizeof(BLE.Value)];
1963add3e9cSRafael Espindola   support::endian::write<value_type, support::little, support::unaligned>(
1973add3e9cSRafael Espindola     Buffer, BLE.Value);
1983add3e9cSRafael Espindola   OS.write(Buffer, sizeof(BLE.Value));
1993add3e9cSRafael Espindola   return OS;
2003add3e9cSRafael Espindola }
2013add3e9cSRafael Espindola 
2023add3e9cSRafael Espindola template <typename value_type>
2033add3e9cSRafael Espindola binary_le_impl<value_type> binary_le(value_type V) {
2043add3e9cSRafael Espindola   return binary_le_impl<value_type>(V);
2053add3e9cSRafael Espindola }
2063add3e9cSRafael Espindola 
207ba5899c5SRafael Espindola static bool writeHexData(StringRef Data, raw_ostream &OS) {
208ba5899c5SRafael Espindola   unsigned Size = Data.size();
209ba5899c5SRafael Espindola   if (Size % 2)
210ba5899c5SRafael Espindola     return false;
211ba5899c5SRafael Espindola 
212ba5899c5SRafael Espindola   for (unsigned I = 0; I != Size; I += 2) {
213ba5899c5SRafael Espindola     uint8_t Byte;
214ba5899c5SRafael Espindola     if (Data.substr(I,  2).getAsInteger(16, Byte))
215ba5899c5SRafael Espindola       return false;
216ba5899c5SRafael Espindola     OS.write(Byte);
217ba5899c5SRafael Espindola   }
218ba5899c5SRafael Espindola 
219ba5899c5SRafael Espindola   return true;
220ba5899c5SRafael Espindola }
221ba5899c5SRafael Espindola 
222ad3b9753SRafael Espindola bool writeCOFF(COFFParser &CP, raw_ostream &OS) {
223ad3b9753SRafael Espindola   OS << binary_le(CP.Obj.Header.Machine)
224ad3b9753SRafael Espindola      << binary_le(CP.Obj.Header.NumberOfSections)
225ad3b9753SRafael Espindola      << binary_le(CP.Obj.Header.TimeDateStamp)
226ad3b9753SRafael Espindola      << binary_le(CP.Obj.Header.PointerToSymbolTable)
227ad3b9753SRafael Espindola      << binary_le(CP.Obj.Header.NumberOfSymbols)
228ad3b9753SRafael Espindola      << binary_le(CP.Obj.Header.SizeOfOptionalHeader)
229ad3b9753SRafael Espindola      << binary_le(CP.Obj.Header.Characteristics);
2303add3e9cSRafael Espindola 
2313add3e9cSRafael Espindola   // Output section table.
232ad3b9753SRafael Espindola   for (std::vector<COFFYAML::Section>::iterator i = CP.Obj.Sections.begin(),
233ad3b9753SRafael Espindola                                                 e = CP.Obj.Sections.end();
2343add3e9cSRafael Espindola                                                 i != e; ++i) {
2353add3e9cSRafael Espindola     OS.write(i->Header.Name, COFF::NameSize);
2363add3e9cSRafael Espindola     OS << binary_le(i->Header.VirtualSize)
2373add3e9cSRafael Espindola        << binary_le(i->Header.VirtualAddress)
2383add3e9cSRafael Espindola        << binary_le(i->Header.SizeOfRawData)
2393add3e9cSRafael Espindola        << binary_le(i->Header.PointerToRawData)
2403add3e9cSRafael Espindola        << binary_le(i->Header.PointerToRelocations)
2413add3e9cSRafael Espindola        << binary_le(i->Header.PointerToLineNumbers)
2423add3e9cSRafael Espindola        << binary_le(i->Header.NumberOfRelocations)
2433add3e9cSRafael Espindola        << binary_le(i->Header.NumberOfLineNumbers)
2443add3e9cSRafael Espindola        << binary_le(i->Header.Characteristics);
2453add3e9cSRafael Espindola   }
2463add3e9cSRafael Espindola 
2473add3e9cSRafael Espindola   // Output section data.
248ad3b9753SRafael Espindola   for (std::vector<COFFYAML::Section>::iterator i = CP.Obj.Sections.begin(),
249ad3b9753SRafael Espindola                                                 e = CP.Obj.Sections.end();
2503add3e9cSRafael Espindola                                                 i != e; ++i) {
251ad3b9753SRafael Espindola     if (!i->SectionData.empty()) {
252ba5899c5SRafael Espindola       if (!writeHexData(i->SectionData, OS)) {
253ad3b9753SRafael Espindola         errs() << "SectionData must be a collection of pairs of hex bytes";
254ad3b9753SRafael Espindola         return false;
255ad3b9753SRafael Espindola       }
256ad3b9753SRafael Espindola     }
257b716e622SRafael Espindola     for (unsigned I2 = 0, E2 = i->Relocations.size(); I2 != E2; ++I2) {
258b716e622SRafael Espindola       const COFF::relocation &R = i->Relocations[I2];
259b716e622SRafael Espindola       OS << binary_le(R.VirtualAddress)
260b716e622SRafael Espindola          << binary_le(R.SymbolTableIndex)
261b716e622SRafael Espindola          << binary_le(R.Type);
262b716e622SRafael Espindola     }
2633add3e9cSRafael Espindola   }
2643add3e9cSRafael Espindola 
2653add3e9cSRafael Espindola   // Output symbol table.
2663add3e9cSRafael Espindola 
267ad3b9753SRafael Espindola   for (std::vector<COFFYAML::Symbol>::const_iterator i = CP.Obj.Symbols.begin(),
268ad3b9753SRafael Espindola                                                      e = CP.Obj.Symbols.end();
2693add3e9cSRafael Espindola                                                      i != e; ++i) {
2703add3e9cSRafael Espindola     OS.write(i->Header.Name, COFF::NameSize);
2713add3e9cSRafael Espindola     OS << binary_le(i->Header.Value)
2723add3e9cSRafael Espindola        << binary_le(i->Header.SectionNumber)
2733add3e9cSRafael Espindola        << binary_le(i->Header.Type)
2743add3e9cSRafael Espindola        << binary_le(i->Header.StorageClass)
2753add3e9cSRafael Espindola        << binary_le(i->Header.NumberOfAuxSymbols);
2767f08d1b9SRafael Espindola     if (!i->AuxiliaryData.empty()) {
277ba5899c5SRafael Espindola       if (!writeHexData(i->AuxiliaryData, OS)) {
2787f08d1b9SRafael Espindola         errs() << "AuxiliaryData must be a collection of pairs of hex bytes";
279ad3b9753SRafael Espindola         return false;
280ad3b9753SRafael Espindola       }
281ad3b9753SRafael Espindola     }
2823add3e9cSRafael Espindola   }
2833add3e9cSRafael Espindola 
2843add3e9cSRafael Espindola   // Output string table.
2853add3e9cSRafael Espindola   OS.write(&CP.StringTable[0], CP.StringTable.size());
286ad3b9753SRafael Espindola   return true;
2873add3e9cSRafael Espindola }
2883add3e9cSRafael Espindola 
2893add3e9cSRafael Espindola int main(int argc, char **argv) {
2903add3e9cSRafael Espindola   cl::ParseCommandLineOptions(argc, argv);
2913add3e9cSRafael Espindola   sys::PrintStackTraceOnErrorSignal();
2923add3e9cSRafael Espindola   PrettyStackTraceProgram X(argc, argv);
2933add3e9cSRafael Espindola   llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
2943add3e9cSRafael Espindola 
2953add3e9cSRafael Espindola   OwningPtr<MemoryBuffer> Buf;
2963add3e9cSRafael Espindola   if (MemoryBuffer::getFileOrSTDIN(Input, Buf))
2973add3e9cSRafael Espindola     return 1;
2983add3e9cSRafael Espindola 
2993add3e9cSRafael Espindola   yaml::Input YIn(Buf->getBuffer());
3003add3e9cSRafael Espindola   COFFYAML::Object Doc;
3013add3e9cSRafael Espindola   YIn >> Doc;
3023add3e9cSRafael Espindola   if (YIn.error()) {
3033add3e9cSRafael Espindola     errs() << "yaml2obj: Failed to parse YAML file!\n";
3043add3e9cSRafael Espindola     return 1;
3053add3e9cSRafael Espindola   }
3063add3e9cSRafael Espindola 
3073add3e9cSRafael Espindola   COFFParser CP(Doc);
3083add3e9cSRafael Espindola   if (!CP.parse()) {
3093add3e9cSRafael Espindola     errs() << "yaml2obj: Failed to parse YAML file!\n";
3103add3e9cSRafael Espindola     return 1;
3113add3e9cSRafael Espindola   }
3123add3e9cSRafael Espindola 
3133add3e9cSRafael Espindola   if (!layoutCOFF(CP)) {
3143add3e9cSRafael Espindola     errs() << "yaml2obj: Failed to layout COFF file!\n";
3153add3e9cSRafael Espindola     return 1;
3163add3e9cSRafael Espindola   }
317ad3b9753SRafael Espindola   if (!writeCOFF(CP, outs())) {
318ad3b9753SRafael Espindola     errs() << "yaml2obj: Failed to write COFF file!\n";
319ad3b9753SRafael Espindola     return 1;
320ad3b9753SRafael Espindola   }
3213add3e9cSRafael Espindola }
322