1 //===- StringTableBuilder.cpp - String table building utility -------------===// 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 "llvm/MC/StringTableBuilder.h" 11 #include "llvm/ADT/CachedHashString.h" 12 #include "llvm/ADT/SmallString.h" 13 #include "llvm/ADT/StringRef.h" 14 #include "llvm/BinaryFormat/COFF.h" 15 #include "llvm/Support/Endian.h" 16 #include "llvm/Support/MathExtras.h" 17 #include "llvm/Support/raw_ostream.h" 18 #include <cassert> 19 #include <cstddef> 20 #include <cstdint> 21 #include <cstring> 22 #include <utility> 23 #include <vector> 24 25 using namespace llvm; 26 27 StringTableBuilder::~StringTableBuilder() = default; 28 29 void StringTableBuilder::initSize() { 30 // Account for leading bytes in table so that offsets returned from add are 31 // correct. 32 switch (K) { 33 case RAW: 34 Size = 0; 35 break; 36 case MachO: 37 case ELF: 38 // Start the table with a NUL byte. 39 Size = 1; 40 break; 41 case WinCOFF: 42 // Make room to write the table size later. 43 Size = 4; 44 break; 45 } 46 } 47 48 StringTableBuilder::StringTableBuilder(Kind K, unsigned Alignment) 49 : K(K), Alignment(Alignment) { 50 initSize(); 51 } 52 53 void StringTableBuilder::write(raw_ostream &OS) const { 54 assert(isFinalized()); 55 SmallString<0> Data; 56 Data.resize(getSize()); 57 write((uint8_t *)Data.data()); 58 OS << Data; 59 } 60 61 using StringPair = std::pair<CachedHashStringRef, size_t>; 62 63 void StringTableBuilder::write(uint8_t *Buf) const { 64 assert(isFinalized()); 65 for (const StringPair &P : StringIndexMap) { 66 StringRef Data = P.first.val(); 67 if (!Data.empty()) 68 memcpy(Buf + P.second, Data.data(), Data.size()); 69 } 70 if (K != WinCOFF) 71 return; 72 support::endian::write32le(Buf, Size); 73 } 74 75 // Returns the character at Pos from end of a string. 76 static int charTailAt(StringPair *P, size_t Pos) { 77 StringRef S = P->first.val(); 78 if (Pos >= S.size()) 79 return -1; 80 return (unsigned char)S[S.size() - Pos - 1]; 81 } 82 83 // Three-way radix quicksort. This is much faster than std::sort with strcmp 84 // because it does not compare characters that we already know the same. 85 static void multikeySort(MutableArrayRef<StringPair *> Vec, int Pos) { 86 tailcall: 87 if (Vec.size() <= 1) 88 return; 89 90 // Partition items so that items in [0, I) are greater than the pivot, 91 // [I, J) are the same as the pivot, and [J, Vec.size()) are less than 92 // the pivot. 93 int Pivot = charTailAt(Vec[0], Pos); 94 size_t I = 0; 95 size_t J = Vec.size(); 96 for (size_t K = 1; K < J;) { 97 int C = charTailAt(Vec[K], Pos); 98 if (C > Pivot) 99 std::swap(Vec[I++], Vec[K++]); 100 else if (C < Pivot) 101 std::swap(Vec[--J], Vec[K]); 102 else 103 K++; 104 } 105 106 multikeySort(Vec.slice(0, I), Pos); 107 multikeySort(Vec.slice(J), Pos); 108 109 // multikeySort(Vec.slice(I, J - I), Pos + 1), but with 110 // tail call optimization. 111 if (Pivot != -1) { 112 Vec = Vec.slice(I, J - I); 113 ++Pos; 114 goto tailcall; 115 } 116 } 117 118 void StringTableBuilder::finalize() { 119 finalizeStringTable(/*Optimize=*/true); 120 } 121 122 void StringTableBuilder::finalizeInOrder() { 123 finalizeStringTable(/*Optimize=*/false); 124 } 125 126 void StringTableBuilder::finalizeStringTable(bool Optimize) { 127 Finalized = true; 128 129 if (Optimize) { 130 std::vector<StringPair *> Strings; 131 Strings.reserve(StringIndexMap.size()); 132 for (StringPair &P : StringIndexMap) 133 Strings.push_back(&P); 134 135 multikeySort(Strings, 0); 136 initSize(); 137 138 StringRef Previous; 139 for (StringPair *P : Strings) { 140 StringRef S = P->first.val(); 141 if (Previous.endswith(S)) { 142 size_t Pos = Size - S.size() - (K != RAW); 143 if (!(Pos & (Alignment - 1))) { 144 P->second = Pos; 145 continue; 146 } 147 } 148 149 Size = alignTo(Size, Alignment); 150 P->second = Size; 151 152 Size += S.size(); 153 if (K != RAW) 154 ++Size; 155 Previous = S; 156 } 157 } 158 159 if (K == MachO) 160 Size = alignTo(Size, 4); // Pad to multiple of 4. 161 } 162 163 void StringTableBuilder::clear() { 164 Finalized = false; 165 StringIndexMap.clear(); 166 } 167 168 size_t StringTableBuilder::getOffset(CachedHashStringRef S) const { 169 assert(isFinalized()); 170 auto I = StringIndexMap.find(S); 171 assert(I != StringIndexMap.end() && "String is not in table!"); 172 return I->second; 173 } 174 175 size_t StringTableBuilder::add(CachedHashStringRef S) { 176 if (K == WinCOFF) 177 assert(S.size() > COFF::NameSize && "Short string in COFF string table!"); 178 179 assert(!isFinalized()); 180 auto P = StringIndexMap.insert(std::make_pair(S, 0)); 181 if (P.second) { 182 size_t Start = alignTo(Size, Alignment); 183 P.first->second = Start; 184 Size = Start + S.size() + (K != RAW); 185 } 186 return P.first->second; 187 } 188