xref: /llvm-project/llvm/lib/DebugInfo/PDB/Native/PDBStringTableBuilder.cpp (revision a6fb536e5b5c98d3ca46fbd7ea21893f633d21d8)
1 //===- PDBStringTableBuilder.cpp - PDB String Table -------------*- C++ -*-===//
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/DebugInfo/PDB/Native/PDBStringTableBuilder.h"
11 
12 #include "llvm/ADT/ArrayRef.h"
13 #include "llvm/DebugInfo/PDB/Native/Hash.h"
14 #include "llvm/DebugInfo/PDB/Native/RawTypes.h"
15 #include "llvm/Support/BinaryStreamWriter.h"
16 #include "llvm/Support/Endian.h"
17 
18 #include <map>
19 
20 using namespace llvm;
21 using namespace llvm::msf;
22 using namespace llvm::support;
23 using namespace llvm::support::endian;
24 using namespace llvm::pdb;
25 
26 uint32_t PDBStringTableBuilder::insert(StringRef S) {
27   return Strings.insert(S);
28 }
29 
30 uint32_t PDBStringTableBuilder::getIdForString(StringRef S) const {
31   return Strings.getIdForString(S);
32 }
33 
34 StringRef PDBStringTableBuilder::getStringForId(uint32_t Id) const {
35   return Strings.getStringForId(Id);
36 }
37 
38 // This is a precomputed list of Buckets given the specified number of
39 // strings.  Matching the reference algorithm exactly is not strictly
40 // necessary for correctness, but it helps when comparing LLD's PDBs with
41 // Microsoft's PDBs so as to eliminate superfluous differences.
42 static std::map<uint32_t, uint32_t> StringsToBuckets = {
43     {1, 2},
44     {2, 4},
45     {4, 7},
46     {6, 11},
47     {9, 17},
48     {13, 26},
49     {20, 40},
50     {31, 61},
51     {46, 92},
52     {70, 139},
53     {105, 209},
54     {157, 314},
55     {236, 472},
56     {355, 709},
57     {532, 1064},
58     {799, 1597},
59     {1198, 2396},
60     {1798, 3595},
61     {2697, 5393},
62     {4045, 8090},
63     {6068, 12136},
64     {9103, 18205},
65     {13654, 27308},
66     {20482, 40963},
67     {30723, 61445},
68     {46084, 92168},
69     {69127, 138253},
70     {103690, 207380},
71     {155536, 311071},
72     {233304, 466607},
73     {349956, 699911},
74     {524934, 1049867},
75     {787401, 1574801},
76     {1181101, 2362202},
77     {1771652, 3543304},
78     {2657479, 5314957},
79     {3986218, 7972436},
80     {5979328, 11958655},
81     {8968992, 17937983},
82     {13453488, 26906975},
83     {20180232, 40360463},
84     {30270348, 60540695},
85     {45405522, 90811043},
86     {68108283, 136216565},
87     {102162424, 204324848},
88     {153243637, 306487273},
89     {229865455, 459730910},
90     {344798183, 689596366},
91     {517197275, 1034394550},
92     {775795913, 1551591826}};
93 
94 static uint32_t computeBucketCount(uint32_t NumStrings) {
95   auto Entry = StringsToBuckets.lower_bound(NumStrings);
96   assert(Entry != StringsToBuckets.end());
97   return Entry->second;
98 }
99 
100 uint32_t PDBStringTableBuilder::calculateHashTableSize() const {
101   uint32_t Size = sizeof(uint32_t); // Hash table begins with 4-byte size field.
102   Size += sizeof(uint32_t) * computeBucketCount(Strings.size());
103 
104   return Size;
105 }
106 
107 uint32_t PDBStringTableBuilder::calculateSerializedSize() const {
108   uint32_t Size = 0;
109   Size += sizeof(PDBStringTableHeader);
110   Size += Strings.calculateSerializedSize();
111   Size += calculateHashTableSize();
112   Size += sizeof(uint32_t); // The /names stream ends with the string count.
113   return Size;
114 }
115 
116 void PDBStringTableBuilder::setStrings(
117     const codeview::DebugStringTableSubsection &Strings) {
118   this->Strings = Strings;
119 }
120 
121 Error PDBStringTableBuilder::writeHeader(BinaryStreamWriter &Writer) const {
122   // Write a header
123   PDBStringTableHeader H;
124   H.Signature = PDBStringTableSignature;
125   H.HashVersion = 1;
126   H.ByteSize = Strings.calculateSerializedSize();
127   if (auto EC = Writer.writeObject(H))
128     return EC;
129   assert(Writer.bytesRemaining() == 0);
130   return Error::success();
131 }
132 
133 Error PDBStringTableBuilder::writeStrings(BinaryStreamWriter &Writer) const {
134   if (auto EC = Strings.commit(Writer))
135     return EC;
136 
137   assert(Writer.bytesRemaining() == 0);
138   return Error::success();
139 }
140 
141 Error PDBStringTableBuilder::writeHashTable(BinaryStreamWriter &Writer) const {
142   // Write a hash table.
143   uint32_t BucketCount = computeBucketCount(Strings.size());
144   if (auto EC = Writer.writeInteger(BucketCount))
145     return EC;
146   std::vector<ulittle32_t> Buckets(BucketCount);
147 
148   for (auto &Pair : Strings) {
149     StringRef S = Pair.getKey();
150     uint32_t Offset = Pair.getValue();
151     uint32_t Hash = hashStringV1(S);
152 
153     for (uint32_t I = 0; I != BucketCount; ++I) {
154       uint32_t Slot = (Hash + I) % BucketCount;
155       if (Buckets[Slot] != 0)
156         continue;
157       Buckets[Slot] = Offset;
158       break;
159     }
160   }
161 
162   if (auto EC = Writer.writeArray(ArrayRef<ulittle32_t>(Buckets)))
163     return EC;
164 
165   assert(Writer.bytesRemaining() == 0);
166   return Error::success();
167 }
168 
169 Error PDBStringTableBuilder::writeEpilogue(BinaryStreamWriter &Writer) const {
170   if (auto EC = Writer.writeInteger<uint32_t>(Strings.size()))
171     return EC;
172   assert(Writer.bytesRemaining() == 0);
173   return Error::success();
174 }
175 
176 Error PDBStringTableBuilder::commit(BinaryStreamWriter &Writer) const {
177   BinaryStreamWriter SectionWriter;
178 
179   std::tie(SectionWriter, Writer) = Writer.split(sizeof(PDBStringTableHeader));
180   if (auto EC = writeHeader(SectionWriter))
181     return EC;
182 
183   std::tie(SectionWriter, Writer) =
184       Writer.split(Strings.calculateSerializedSize());
185   if (auto EC = writeStrings(SectionWriter))
186     return EC;
187 
188   std::tie(SectionWriter, Writer) = Writer.split(calculateHashTableSize());
189   if (auto EC = writeHashTable(SectionWriter))
190     return EC;
191 
192   std::tie(SectionWriter, Writer) = Writer.split(sizeof(uint32_t));
193   if (auto EC = writeEpilogue(SectionWriter))
194     return EC;
195 
196   return Error::success();
197 }
198