blob: de40a7728d3fcfb6775b30e91ecf828cb6b01fcf [file] [log] [blame]
Eugene Zelenkoa700a602017-02-11 00:27:28 +00001//===- StringTableBuilder.cpp - String table building utility -------------===//
Hans Wennborga8febf22014-04-30 16:25:02 +00002//
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 Carruthe3e43d92017-06-06 11:49:48 +000010#include "llvm/MC/StringTableBuilder.h"
Eugene Zelenkoa700a602017-02-11 00:27:28 +000011#include "llvm/ADT/CachedHashString.h"
Justin Lebarfd7d73f2016-10-17 22:24:36 +000012#include "llvm/ADT/SmallString.h"
Eugene Zelenkoa700a602017-02-11 00:27:28 +000013#include "llvm/ADT/StringRef.h"
Zachary Turner19ca2b02017-06-07 03:48:56 +000014#include "llvm/BinaryFormat/COFF.h"
Hans Wennborg4edcbae2014-09-29 22:43:20 +000015#include "llvm/Support/Endian.h"
Eugene Zelenkoa700a602017-02-11 00:27:28 +000016#include "llvm/Support/MathExtras.h"
Rafael Espindola2638e452016-10-04 22:43:25 +000017#include "llvm/Support/raw_ostream.h"
Eugene Zelenkoa700a602017-02-11 00:27:28 +000018#include <cassert>
19#include <cstddef>
20#include <cstdint>
21#include <cstring>
22#include <utility>
Zachary Turner30385e02015-10-22 16:42:31 +000023#include <vector>
24
Hans Wennborga8febf22014-04-30 16:25:02 +000025using namespace llvm;
26
Eugene Zelenkoa700a602017-02-11 00:27:28 +000027StringTableBuilder::~StringTableBuilder() = default;
Rafael Espindola2638e452016-10-04 22:43:25 +000028
29void StringTableBuilder::initSize() {
Reid Kleckner1689efb2016-01-29 00:49:42 +000030 // Account for leading bytes in table so that offsets returned from add are
31 // correct.
32 switch (K) {
33 case RAW:
Paul Robinson92a98cd2018-02-06 20:29:21 +000034 case DWARF:
Reid Kleckner1689efb2016-01-29 00:49:42 +000035 Size = 0;
36 break;
37 case MachO:
38 case ELF:
Rafael Espindola2638e452016-10-04 22:43:25 +000039 // Start the table with a NUL byte.
Reid Kleckner1689efb2016-01-29 00:49:42 +000040 Size = 1;
41 break;
42 case WinCOFF:
Rafael Espindola2638e452016-10-04 22:43:25 +000043 // Make room to write the table size later.
Reid Kleckner1689efb2016-01-29 00:49:42 +000044 Size = 4;
45 break;
46 }
47}
Rafael Espindola715bcca2015-10-23 21:48:05 +000048
Rafael Espindola2638e452016-10-04 22:43:25 +000049StringTableBuilder::StringTableBuilder(Kind K, unsigned Alignment)
50 : K(K), Alignment(Alignment) {
51 initSize();
52}
53
54void StringTableBuilder::write(raw_ostream &OS) const {
55 assert(isFinalized());
56 SmallString<0> Data;
57 Data.resize(getSize());
Peter Collingbournece637fe2017-04-06 00:10:17 +000058 write((uint8_t *)Data.data());
Rafael Espindola2638e452016-10-04 22:43:25 +000059 OS << Data;
60}
61
Eugene Zelenkof49f90e2017-04-26 22:31:39 +000062using StringPair = std::pair<CachedHashStringRef, size_t>;
Rafael Espindola2638e452016-10-04 22:43:25 +000063
64void StringTableBuilder::write(uint8_t *Buf) const {
65 assert(isFinalized());
66 for (const StringPair &P : StringIndexMap) {
67 StringRef Data = P.first.val();
Rafael Espindolab53fb3c2016-10-05 16:33:03 +000068 if (!Data.empty())
69 memcpy(Buf + P.second, Data.data(), Data.size());
Rafael Espindola2638e452016-10-04 22:43:25 +000070 }
71 if (K != WinCOFF)
72 return;
73 support::endian::write32le(Buf, Size);
74}
Rui Ueyamad2a6a142015-10-26 19:58:29 +000075
76// Returns the character at Pos from end of a string.
77static int charTailAt(StringPair *P, size_t Pos) {
Rafael Espindola2638e452016-10-04 22:43:25 +000078 StringRef S = P->first.val();
Rui Ueyamad2a6a142015-10-26 19:58:29 +000079 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 Ueyamaf3dc1162017-10-03 23:12:01 +000086static void multikeySort(MutableArrayRef<StringPair *> Vec, int Pos) {
Rui Ueyamad2a6a142015-10-26 19:58:29 +000087tailcall:
Rui Ueyamaf3dc1162017-10-03 23:12:01 +000088 if (Vec.size() <= 1)
Rui Ueyamad2a6a142015-10-26 19:58:29 +000089 return;
90
Rui Ueyamaf3dc1162017-10-03 23:12:01 +000091 // 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 Ueyamad2a6a142015-10-26 19:58:29 +000099 if (C > Pivot)
Rui Ueyamaf3dc1162017-10-03 23:12:01 +0000100 std::swap(Vec[I++], Vec[K++]);
Rui Ueyamad2a6a142015-10-26 19:58:29 +0000101 else if (C < Pivot)
Rui Ueyamaf3dc1162017-10-03 23:12:01 +0000102 std::swap(Vec[--J], Vec[K]);
Rui Ueyamad2a6a142015-10-26 19:58:29 +0000103 else
Rui Ueyamaf3dc1162017-10-03 23:12:01 +0000104 K++;
Akira Hatanakad3d620b2014-09-24 20:37:14 +0000105 }
Rui Ueyamad2a6a142015-10-26 19:58:29 +0000106
Rui Ueyamaf3dc1162017-10-03 23:12:01 +0000107 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 Ueyamad2a6a142015-10-26 19:58:29 +0000112 if (Pivot != -1) {
Rui Ueyamaf3dc1162017-10-03 23:12:01 +0000113 Vec = Vec.slice(I, J - I);
Rui Ueyamad2a6a142015-10-26 19:58:29 +0000114 ++Pos;
115 goto tailcall;
116 }
Akira Hatanakad3d620b2014-09-24 20:37:14 +0000117}
118
Rafael Espindola715bcca2015-10-23 21:48:05 +0000119void StringTableBuilder::finalize() {
Paul Robinson92a98cd2018-02-06 20:29:21 +0000120 assert(K != DWARF);
Reid Kleckner1689efb2016-01-29 00:49:42 +0000121 finalizeStringTable(/*Optimize=*/true);
122}
123
124void StringTableBuilder::finalizeInOrder() {
125 finalizeStringTable(/*Optimize=*/false);
126}
127
128void StringTableBuilder::finalizeStringTable(bool Optimize) {
Rafael Espindola2638e452016-10-04 22:43:25 +0000129 Finalized = true;
Hans Wennborga8febf22014-04-30 16:25:02 +0000130
Rafael Espindola2638e452016-10-04 22:43:25 +0000131 if (Optimize) {
132 std::vector<StringPair *> Strings;
133 Strings.reserve(StringIndexMap.size());
134 for (StringPair &P : StringIndexMap)
135 Strings.push_back(&P);
136
Rui Ueyamaf3dc1162017-10-03 23:12:01 +0000137 multikeySort(Strings, 0);
Rafael Espindola2638e452016-10-04 22:43:25 +0000138 initSize();
Hans Wennborga8febf22014-04-30 16:25:02 +0000139
Rafael Espindola2638e452016-10-04 22:43:25 +0000140 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 Espindola3e185f62016-02-19 14:13:52 +0000149 }
Hans Wennborga8febf22014-04-30 16:25:02 +0000150
Rafael Espindola2638e452016-10-04 22:43:25 +0000151 Size = alignTo(Size, Alignment);
152 P->second = Size;
Reid Kleckner1689efb2016-01-29 00:49:42 +0000153
Rafael Espindola2638e452016-10-04 22:43:25 +0000154 Size += S.size();
155 if (K != RAW)
156 ++Size;
157 Previous = S;
158 }
Hans Wennborga8febf22014-04-30 16:25:02 +0000159 }
Hans Wennborg4edcbae2014-09-29 22:43:20 +0000160
Rafael Espindola2638e452016-10-04 22:43:25 +0000161 if (K == MachO)
162 Size = alignTo(Size, 4); // Pad to multiple of 4.
Hans Wennborg4edcbae2014-09-29 22:43:20 +0000163}
164
165void StringTableBuilder::clear() {
Rafael Espindola2638e452016-10-04 22:43:25 +0000166 Finalized = false;
Hans Wennborg4edcbae2014-09-29 22:43:20 +0000167 StringIndexMap.clear();
Hans Wennborga8febf22014-04-30 16:25:02 +0000168}
Rafael Espindolad43c0012015-10-22 18:32:06 +0000169
Justin Lebarfd7d73f2016-10-17 22:24:36 +0000170size_t StringTableBuilder::getOffset(CachedHashStringRef S) const {
Rafael Espindolad43c0012015-10-22 18:32:06 +0000171 assert(isFinalized());
Rafael Espindola89cb1b72015-10-23 20:15:35 +0000172 auto I = StringIndexMap.find(S);
Rafael Espindolad43c0012015-10-22 18:32:06 +0000173 assert(I != StringIndexMap.end() && "String is not in table!");
174 return I->second;
175}
176
Justin Lebarfd7d73f2016-10-17 22:24:36 +0000177size_t StringTableBuilder::add(CachedHashStringRef S) {
Rafael Espindola2638e452016-10-04 22:43:25 +0000178 if (K == WinCOFF)
179 assert(S.size() > COFF::NameSize && "Short string in COFF string table!");
180
Rafael Espindolad43c0012015-10-22 18:32:06 +0000181 assert(!isFinalized());
Rafael Espindola7c52e332016-10-14 17:01:39 +0000182 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 Espindola3e185f62016-02-19 14:13:52 +0000186 Size = Start + S.size() + (K != RAW);
Rafael Espindola7c52e332016-10-14 17:01:39 +0000187 }
Rafael Espindola715bcca2015-10-23 21:48:05 +0000188 return P.first->second;
Rafael Espindolad43c0012015-10-22 18:32:06 +0000189}