Hans Wennborg | 83e6e1e | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 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 | |
Rafael Espindola | 97de474 | 2014-07-03 02:01:39 +0000 | [diff] [blame] | 10 | #include "llvm/MC/StringTableBuilder.h" |
Rafael Espindola | 0169a45 | 2015-10-22 15:15:44 +0000 | [diff] [blame] | 11 | #include "llvm/ADT/STLExtras.h" |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 12 | #include "llvm/Support/COFF.h" |
| 13 | #include "llvm/Support/Endian.h" |
Hans Wennborg | 83e6e1e | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 14 | |
Zachary Turner | c55a504 | 2015-10-22 16:42:31 +0000 | [diff] [blame] | 15 | #include <vector> |
| 16 | |
Hans Wennborg | 83e6e1e | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 17 | using namespace llvm; |
| 18 | |
Rafael Espindola | 758de9c | 2016-02-19 14:13:52 +0000 | [diff] [blame] | 19 | StringTableBuilder::StringTableBuilder(Kind K, unsigned Alignment) |
| 20 | : K(K), Alignment(Alignment) { |
Reid Kleckner | 2214ed8 | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 21 | // Account for leading bytes in table so that offsets returned from add are |
| 22 | // correct. |
| 23 | switch (K) { |
| 24 | case RAW: |
| 25 | Size = 0; |
| 26 | break; |
| 27 | case MachO: |
| 28 | case ELF: |
| 29 | Size = 1; |
| 30 | break; |
| 31 | case WinCOFF: |
| 32 | Size = 4; |
| 33 | break; |
| 34 | } |
| 35 | } |
Rafael Espindola | 21956e4 | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 36 | |
Rui Ueyama | 02d71ad | 2016-05-06 00:51:58 +0000 | [diff] [blame] | 37 | typedef std::pair<CachedHash<StringRef>, size_t> StringPair; |
Rui Ueyama | df94852 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 38 | |
| 39 | // Returns the character at Pos from end of a string. |
| 40 | static int charTailAt(StringPair *P, size_t Pos) { |
Rui Ueyama | 02d71ad | 2016-05-06 00:51:58 +0000 | [diff] [blame] | 41 | StringRef S = P->first.Val; |
Rui Ueyama | df94852 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 42 | if (Pos >= S.size()) |
| 43 | return -1; |
| 44 | return (unsigned char)S[S.size() - Pos - 1]; |
| 45 | } |
| 46 | |
| 47 | // Three-way radix quicksort. This is much faster than std::sort with strcmp |
| 48 | // because it does not compare characters that we already know the same. |
Rui Ueyama | 5579e0b | 2015-10-27 16:57:50 +0000 | [diff] [blame] | 49 | static void multikey_qsort(StringPair **Begin, StringPair **End, int Pos) { |
Rui Ueyama | df94852 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 50 | tailcall: |
| 51 | if (End - Begin <= 1) |
| 52 | return; |
| 53 | |
| 54 | // Partition items. Items in [Begin, P) are greater than the pivot, |
| 55 | // [P, Q) are the same as the pivot, and [Q, End) are less than the pivot. |
| 56 | int Pivot = charTailAt(*Begin, Pos); |
| 57 | StringPair **P = Begin; |
| 58 | StringPair **Q = End; |
| 59 | for (StringPair **R = Begin + 1; R < Q;) { |
| 60 | int C = charTailAt(*R, Pos); |
| 61 | if (C > Pivot) |
| 62 | std::swap(*P++, *R++); |
| 63 | else if (C < Pivot) |
| 64 | std::swap(*--Q, *R); |
| 65 | else |
| 66 | R++; |
Akira Hatanaka | 8e77dbb | 2014-09-24 20:37:14 +0000 | [diff] [blame] | 67 | } |
Rui Ueyama | df94852 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 68 | |
Rui Ueyama | 5579e0b | 2015-10-27 16:57:50 +0000 | [diff] [blame] | 69 | multikey_qsort(Begin, P, Pos); |
| 70 | multikey_qsort(Q, End, Pos); |
Rui Ueyama | df94852 | 2015-10-26 19:58:29 +0000 | [diff] [blame] | 71 | if (Pivot != -1) { |
| 72 | // qsort(P, Q, Pos + 1), but with tail call optimization. |
| 73 | Begin = P; |
| 74 | End = Q; |
| 75 | ++Pos; |
| 76 | goto tailcall; |
| 77 | } |
Akira Hatanaka | 8e77dbb | 2014-09-24 20:37:14 +0000 | [diff] [blame] | 78 | } |
| 79 | |
Rafael Espindola | 21956e4 | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 80 | void StringTableBuilder::finalize() { |
Reid Kleckner | 2214ed8 | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 81 | finalizeStringTable(/*Optimize=*/true); |
| 82 | } |
| 83 | |
| 84 | void StringTableBuilder::finalizeInOrder() { |
| 85 | finalizeStringTable(/*Optimize=*/false); |
| 86 | } |
| 87 | |
| 88 | void StringTableBuilder::finalizeStringTable(bool Optimize) { |
Rui Ueyama | 02d71ad | 2016-05-06 00:51:58 +0000 | [diff] [blame] | 89 | typedef std::pair<CachedHash<StringRef>, size_t> StringOffsetPair; |
Reid Kleckner | 2214ed8 | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 90 | std::vector<StringOffsetPair *> Strings; |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 91 | Strings.reserve(StringIndexMap.size()); |
Reid Kleckner | 2214ed8 | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 92 | for (StringOffsetPair &P : StringIndexMap) |
Rafael Espindola | e015f66 | 2015-10-22 15:26:35 +0000 | [diff] [blame] | 93 | Strings.push_back(&P); |
Hans Wennborg | 83e6e1e | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 94 | |
Reid Kleckner | 2214ed8 | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 95 | if (!Strings.empty()) { |
| 96 | // If we're optimizing, sort by name. If not, sort by previously assigned |
| 97 | // offset. |
| 98 | if (Optimize) { |
| 99 | multikey_qsort(&Strings[0], &Strings[0] + Strings.size(), 0); |
| 100 | } else { |
| 101 | std::sort(Strings.begin(), Strings.end(), |
| 102 | [](const StringOffsetPair *LHS, const StringOffsetPair *RHS) { |
| 103 | return LHS->second < RHS->second; |
| 104 | }); |
| 105 | } |
| 106 | } |
Hans Wennborg | 83e6e1e | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 107 | |
Rafael Espindola | a9b3944 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 108 | switch (K) { |
Rafael Espindola | 21956e4 | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 109 | case RAW: |
| 110 | break; |
Hans Wennborg | 1b1a399 | 2014-10-06 17:05:19 +0000 | [diff] [blame] | 111 | case ELF: |
| 112 | case MachO: |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 113 | // Start the table with a NUL byte. |
| 114 | StringTable += '\x00'; |
Hans Wennborg | 1b1a399 | 2014-10-06 17:05:19 +0000 | [diff] [blame] | 115 | break; |
| 116 | case WinCOFF: |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 117 | // Make room to write the table size later. |
| 118 | StringTable.append(4, '\x00'); |
Hans Wennborg | 1b1a399 | 2014-10-06 17:05:19 +0000 | [diff] [blame] | 119 | break; |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 120 | } |
Hans Wennborg | 83e6e1e | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 121 | |
| 122 | StringRef Previous; |
Reid Kleckner | 2214ed8 | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 123 | for (StringOffsetPair *P : Strings) { |
Rui Ueyama | 02d71ad | 2016-05-06 00:51:58 +0000 | [diff] [blame] | 124 | StringRef S = P->first.Val; |
Rafael Espindola | a9b3944 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 125 | if (K == WinCOFF) |
| 126 | assert(S.size() > COFF::NameSize && "Short string in COFF string table!"); |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 127 | |
Reid Kleckner | 2214ed8 | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 128 | if (Optimize && Previous.endswith(S)) { |
Rafael Espindola | 758de9c | 2016-02-19 14:13:52 +0000 | [diff] [blame] | 129 | size_t Pos = StringTable.size() - S.size() - (K != RAW); |
| 130 | if (!(Pos & (Alignment - 1))) { |
| 131 | P->second = Pos; |
| 132 | continue; |
| 133 | } |
Hans Wennborg | 83e6e1e | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 134 | } |
| 135 | |
Rafael Espindola | 758de9c | 2016-02-19 14:13:52 +0000 | [diff] [blame] | 136 | if (Optimize) { |
| 137 | size_t Start = alignTo(StringTable.size(), Alignment); |
| 138 | P->second = Start; |
| 139 | StringTable.append(Start - StringTable.size(), '\0'); |
| 140 | } else { |
Reid Kleckner | 2214ed8 | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 141 | assert(P->second == StringTable.size() && |
| 142 | "different strtab offset after finalization"); |
Rafael Espindola | 758de9c | 2016-02-19 14:13:52 +0000 | [diff] [blame] | 143 | } |
Reid Kleckner | 2214ed8 | 2016-01-29 00:49:42 +0000 | [diff] [blame] | 144 | |
Rafael Espindola | a9b3944 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 145 | StringTable += S; |
Rafael Espindola | 21956e4 | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 146 | if (K != RAW) |
| 147 | StringTable += '\x00'; |
Rafael Espindola | a9b3944 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 148 | Previous = S; |
Hans Wennborg | 83e6e1e | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 149 | } |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 150 | |
Rafael Espindola | a9b3944 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 151 | switch (K) { |
Rafael Espindola | 21956e4 | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 152 | case RAW: |
Hans Wennborg | 1b1a399 | 2014-10-06 17:05:19 +0000 | [diff] [blame] | 153 | case ELF: |
| 154 | break; |
| 155 | case MachO: |
| 156 | // Pad to multiple of 4. |
| 157 | while (StringTable.size() % 4) |
| 158 | StringTable += '\x00'; |
| 159 | break; |
| 160 | case WinCOFF: |
| 161 | // Write the table size in the first word. |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 162 | assert(StringTable.size() <= std::numeric_limits<uint32_t>::max()); |
Rafael Espindola | a9b3944 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 163 | uint32_t Size = static_cast<uint32_t>(StringTable.size()); |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 164 | support::endian::write<uint32_t, support::little, support::unaligned>( |
Rafael Espindola | a9b3944 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 165 | StringTable.data(), Size); |
Hans Wennborg | 1b1a399 | 2014-10-06 17:05:19 +0000 | [diff] [blame] | 166 | break; |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 167 | } |
Rafael Espindola | 21956e4 | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 168 | |
| 169 | Size = StringTable.size(); |
Hans Wennborg | f26bfc1 | 2014-09-29 22:43:20 +0000 | [diff] [blame] | 170 | } |
| 171 | |
| 172 | void StringTableBuilder::clear() { |
| 173 | StringTable.clear(); |
| 174 | StringIndexMap.clear(); |
Hans Wennborg | 83e6e1e | 2014-04-30 16:25:02 +0000 | [diff] [blame] | 175 | } |
Rafael Espindola | fc063e8 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 176 | |
Rafael Espindola | a9b3944 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 177 | size_t StringTableBuilder::getOffset(StringRef S) const { |
Rafael Espindola | fc063e8 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 178 | assert(isFinalized()); |
Rafael Espindola | a9b3944 | 2015-10-23 20:15:35 +0000 | [diff] [blame] | 179 | auto I = StringIndexMap.find(S); |
Rafael Espindola | fc063e8 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 180 | assert(I != StringIndexMap.end() && "String is not in table!"); |
| 181 | return I->second; |
| 182 | } |
| 183 | |
Rafael Espindola | 21956e4 | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 184 | size_t StringTableBuilder::add(StringRef S) { |
Rafael Espindola | fc063e8 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 185 | assert(!isFinalized()); |
Rafael Espindola | 758de9c | 2016-02-19 14:13:52 +0000 | [diff] [blame] | 186 | size_t Start = alignTo(Size, Alignment); |
| 187 | auto P = StringIndexMap.insert(std::make_pair(S, Start)); |
Rafael Espindola | 21956e4 | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 188 | if (P.second) |
Rafael Espindola | 758de9c | 2016-02-19 14:13:52 +0000 | [diff] [blame] | 189 | Size = Start + S.size() + (K != RAW); |
Rafael Espindola | 21956e4 | 2015-10-23 21:48:05 +0000 | [diff] [blame] | 190 | return P.first->second; |
Rafael Espindola | fc063e8 | 2015-10-22 18:32:06 +0000 | [diff] [blame] | 191 | } |