Eugene Zelenko | a700a60 | 2017-02-11 00:27:28 +0000 | [diff] [blame] | 1 | //===- StringTableBuilder.cpp - String table building utility -------------===// |
Hans Wennborg | a8febf2 | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 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 | |
Chandler Carruth | e3e43d9 | 2017-06-06 11:49:48 +0000 | [diff] [blame] | 10 | #include "llvm/MC/StringTableBuilder.h" |
Eugene Zelenko | a700a60 | 2017-02-11 00:27:28 +0000 | [diff] [blame] | 11 | #include "llvm/ADT/CachedHashString.h" |
Justin Lebar | fd7d73f | 2016-10-17 22:24:36 +0000 | [diff] [blame] | 12 | #include "llvm/ADT/SmallString.h" |
Eugene Zelenko | a700a60 | 2017-02-11 00:27:28 +0000 | [diff] [blame] | 13 | #include "llvm/ADT/StringRef.h" |
Zachary Turner | 19ca2b0 | 2017-06-07 03:48:56 +0000 | [diff] [blame] | 14 | #include "llvm/BinaryFormat/COFF.h" |
Hans Wennborg | 4edcbae | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 15 | #include "llvm/Support/Endian.h" |
Eugene Zelenko | a700a60 | 2017-02-11 00:27:28 +0000 | [diff] [blame] | 16 | #include "llvm/Support/MathExtras.h" |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 17 | #include "llvm/Support/raw_ostream.h" |
Eugene Zelenko | a700a60 | 2017-02-11 00:27:28 +0000 | [diff] [blame] | 18 | #include <cassert> |
| 19 | #include <cstddef> |
| 20 | #include <cstdint> |
| 21 | #include <cstring> |
| 22 | #include <utility> |
Zachary Turner | 30385e0 | 2015-10-22 16:42:31 +0000 | [diff] [blame] | 23 | #include <vector> |
| 24 | |
Hans Wennborg | a8febf2 | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 25 | using namespace llvm; |
| 26 | |
Eugene Zelenko | a700a60 | 2017-02-11 00:27:28 +0000 | [diff] [blame] | 27 | StringTableBuilder::~StringTableBuilder() = default; |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 28 | |
| 29 | void StringTableBuilder::initSize() { |
Reid Kleckner | 1689efb | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 30 | // Account for leading bytes in table so that offsets returned from add are |
| 31 | // correct. |
| 32 | switch (K) { |
| 33 | case RAW: |
Paul Robinson | 92a98cd | 2018-02-06 20:29:21 +0000 | [diff] [blame] | 34 | case DWARF: |
Reid Kleckner | 1689efb | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 35 | Size = 0; |
| 36 | break; |
| 37 | case MachO: |
| 38 | case ELF: |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 39 | // Start the table with a NUL byte. |
Reid Kleckner | 1689efb | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 40 | Size = 1; |
| 41 | break; |
| 42 | case WinCOFF: |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 43 | // Make room to write the table size later. |
Reid Kleckner | 1689efb | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 44 | Size = 4; |
| 45 | break; |
| 46 | } |
| 47 | } |
Rafael Espindola | 715bcca | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 48 | |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 49 | StringTableBuilder::StringTableBuilder(Kind K, unsigned Alignment) |
| 50 | : K(K), Alignment(Alignment) { |
| 51 | initSize(); |
| 52 | } |
| 53 | |
| 54 | void StringTableBuilder::write(raw_ostream &OS) const { |
| 55 | assert(isFinalized()); |
| 56 | SmallString<0> Data; |
| 57 | Data.resize(getSize()); |
Peter Collingbourne | ce637fe | 2017-04-06 00:10:17 +0000 | [diff] [blame] | 58 | write((uint8_t *)Data.data()); |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 59 | OS << Data; |
| 60 | } |
| 61 | |
Eugene Zelenko | f49f90e | 2017-04-26 22:31:39 +0000 | [diff] [blame] | 62 | using StringPair = std::pair<CachedHashStringRef, size_t>; |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 63 | |
| 64 | void StringTableBuilder::write(uint8_t *Buf) const { |
| 65 | assert(isFinalized()); |
| 66 | for (const StringPair &P : StringIndexMap) { |
| 67 | StringRef Data = P.first.val(); |
Rafael Espindola | b53fb3c | 2016-10-05 16:33:03 +0000 | [diff] [blame] | 68 | if (!Data.empty()) |
| 69 | memcpy(Buf + P.second, Data.data(), Data.size()); |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 70 | } |
| 71 | if (K != WinCOFF) |
| 72 | return; |
| 73 | support::endian::write32le(Buf, Size); |
| 74 | } |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 75 | |
| 76 | // Returns the character at Pos from end of a string. |
| 77 | static int charTailAt(StringPair *P, size_t Pos) { |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 78 | StringRef S = P->first.val(); |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 79 | if (Pos >= S.size()) |
| 80 | return -1; |
| 81 | return (unsigned char)S[S.size() - Pos - 1]; |
| 82 | } |
| 83 | |
| 84 | // Three-way radix quicksort. This is much faster than std::sort with strcmp |
| 85 | // because it does not compare characters that we already know the same. |
Rui Ueyama | f3dc116 | 2017-10-03 23:12:01 +0000 | [diff] [blame] | 86 | static void multikeySort(MutableArrayRef<StringPair *> Vec, int Pos) { |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 87 | tailcall: |
Rui Ueyama | f3dc116 | 2017-10-03 23:12:01 +0000 | [diff] [blame] | 88 | if (Vec.size() <= 1) |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 89 | return; |
| 90 | |
Rui Ueyama | f3dc116 | 2017-10-03 23:12:01 +0000 | [diff] [blame] | 91 | // Partition items so that items in [0, I) are greater than the pivot, |
| 92 | // [I, J) are the same as the pivot, and [J, Vec.size()) are less than |
| 93 | // the pivot. |
| 94 | int Pivot = charTailAt(Vec[0], Pos); |
| 95 | size_t I = 0; |
| 96 | size_t J = Vec.size(); |
| 97 | for (size_t K = 1; K < J;) { |
| 98 | int C = charTailAt(Vec[K], Pos); |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 99 | if (C > Pivot) |
Rui Ueyama | f3dc116 | 2017-10-03 23:12:01 +0000 | [diff] [blame] | 100 | std::swap(Vec[I++], Vec[K++]); |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 101 | else if (C < Pivot) |
Rui Ueyama | f3dc116 | 2017-10-03 23:12:01 +0000 | [diff] [blame] | 102 | std::swap(Vec[--J], Vec[K]); |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 103 | else |
Rui Ueyama | f3dc116 | 2017-10-03 23:12:01 +0000 | [diff] [blame] | 104 | K++; |
Akira Hatanaka | d3d620b | 2014-09-24 20:37:14 +0000 | [diff] [blame] | 105 | } |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 106 | |
Rui Ueyama | f3dc116 | 2017-10-03 23:12:01 +0000 | [diff] [blame] | 107 | multikeySort(Vec.slice(0, I), Pos); |
| 108 | multikeySort(Vec.slice(J), Pos); |
| 109 | |
| 110 | // multikeySort(Vec.slice(I, J - I), Pos + 1), but with |
| 111 | // tail call optimization. |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 112 | if (Pivot != -1) { |
Rui Ueyama | f3dc116 | 2017-10-03 23:12:01 +0000 | [diff] [blame] | 113 | Vec = Vec.slice(I, J - I); |
Rui Ueyama | d2a6a14 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 114 | ++Pos; |
| 115 | goto tailcall; |
| 116 | } |
Akira Hatanaka | d3d620b | 2014-09-24 20:37:14 +0000 | [diff] [blame] | 117 | } |
| 118 | |
Rafael Espindola | 715bcca | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 119 | void StringTableBuilder::finalize() { |
Paul Robinson | 92a98cd | 2018-02-06 20:29:21 +0000 | [diff] [blame] | 120 | assert(K != DWARF); |
Reid Kleckner | 1689efb | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 121 | finalizeStringTable(/*Optimize=*/true); |
| 122 | } |
| 123 | |
| 124 | void StringTableBuilder::finalizeInOrder() { |
| 125 | finalizeStringTable(/*Optimize=*/false); |
| 126 | } |
| 127 | |
| 128 | void StringTableBuilder::finalizeStringTable(bool Optimize) { |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 129 | Finalized = true; |
Hans Wennborg | a8febf2 | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 130 | |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 131 | if (Optimize) { |
| 132 | std::vector<StringPair *> Strings; |
| 133 | Strings.reserve(StringIndexMap.size()); |
| 134 | for (StringPair &P : StringIndexMap) |
| 135 | Strings.push_back(&P); |
| 136 | |
Rui Ueyama | f3dc116 | 2017-10-03 23:12:01 +0000 | [diff] [blame] | 137 | multikeySort(Strings, 0); |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 138 | initSize(); |
Hans Wennborg | a8febf2 | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 139 | |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 140 | StringRef Previous; |
| 141 | for (StringPair *P : Strings) { |
| 142 | StringRef S = P->first.val(); |
| 143 | if (Previous.endswith(S)) { |
| 144 | size_t Pos = Size - S.size() - (K != RAW); |
| 145 | if (!(Pos & (Alignment - 1))) { |
| 146 | P->second = Pos; |
| 147 | continue; |
| 148 | } |
Rafael Espindola | 3e185f6 | 2016-02-19 14:13:52 +0000 | [diff] [blame] | 149 | } |
Hans Wennborg | a8febf2 | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 150 | |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 151 | Size = alignTo(Size, Alignment); |
| 152 | P->second = Size; |
Reid Kleckner | 1689efb | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 153 | |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 154 | Size += S.size(); |
| 155 | if (K != RAW) |
| 156 | ++Size; |
| 157 | Previous = S; |
| 158 | } |
Hans Wennborg | a8febf2 | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 159 | } |
Hans Wennborg | 4edcbae | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 160 | |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 161 | if (K == MachO) |
| 162 | Size = alignTo(Size, 4); // Pad to multiple of 4. |
Hans Wennborg | 4edcbae | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 163 | } |
| 164 | |
| 165 | void StringTableBuilder::clear() { |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 166 | Finalized = false; |
Hans Wennborg | 4edcbae | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 167 | StringIndexMap.clear(); |
Hans Wennborg | a8febf2 | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 168 | } |
Rafael Espindola | d43c001 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 169 | |
Justin Lebar | fd7d73f | 2016-10-17 22:24:36 +0000 | [diff] [blame] | 170 | size_t StringTableBuilder::getOffset(CachedHashStringRef S) const { |
Rafael Espindola | d43c001 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 171 | assert(isFinalized()); |
Rafael Espindola | 89cb1b7 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 172 | auto I = StringIndexMap.find(S); |
Rafael Espindola | d43c001 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 173 | assert(I != StringIndexMap.end() && "String is not in table!"); |
| 174 | return I->second; |
| 175 | } |
| 176 | |
Justin Lebar | fd7d73f | 2016-10-17 22:24:36 +0000 | [diff] [blame] | 177 | size_t StringTableBuilder::add(CachedHashStringRef S) { |
Rafael Espindola | 2638e45 | 2016-10-04 22:43:25 +0000 | [diff] [blame] | 178 | if (K == WinCOFF) |
| 179 | assert(S.size() > COFF::NameSize && "Short string in COFF string table!"); |
| 180 | |
Rafael Espindola | d43c001 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 181 | assert(!isFinalized()); |
Rafael Espindola | 7c52e33 | 2016-10-14 17:01:39 +0000 | [diff] [blame] | 182 | auto P = StringIndexMap.insert(std::make_pair(S, 0)); |
| 183 | if (P.second) { |
| 184 | size_t Start = alignTo(Size, Alignment); |
| 185 | P.first->second = Start; |
Rafael Espindola | 3e185f6 | 2016-02-19 14:13:52 +0000 | [diff] [blame] | 186 | Size = Start + S.size() + (K != RAW); |
Rafael Espindola | 7c52e33 | 2016-10-14 17:01:39 +0000 | [diff] [blame] | 187 | } |
Rafael Espindola | 715bcca | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 188 | return P.first->second; |
Rafael Espindola | d43c001 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 189 | } |