1 //===- LineTable.h ----------------------------------------------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 
9 #ifndef LLVM_DEBUGINFO_GSYM_LINETABLE_H
10 #define LLVM_DEBUGINFO_GSYM_LINETABLE_H
11 
12 #include "llvm/DebugInfo/GSYM/LineEntry.h"
13 #include "llvm/Support/Error.h"
14 #include <cstdint>
15 #include <vector>
16 
17 namespace llvm {
18 namespace gsym {
19 
20 struct FunctionInfo;
21 class FileWriter;
22 
23 /// LineTable class contains deserialized versions of line tables for each
24 /// function's address ranges.
25 ///
26 /// When saved to disk, the line table is encoded using a modified version of
27 /// the DWARF line tables that only tracks address to source file and line.
28 ///
29 /// ENCODING
30 ///
31 /// The line table starts with a small prolog that contains the following
32 /// values:
33 ///
34 /// ENCODING NAME        DESCRIPTION
35 /// ======== =========== ====================================================
36 /// SLEB     MinDelta    The min line delta for special opcodes that  advance
37 ///                      the address and line number.
38 /// SLEB     MaxDelta    The max line delta for single byte opcodes that
39 ///                      advance the address and line number.
40 /// ULEB     FirstLine   The value of the first source line number to
41 ///                      initialize the LineEntry with.
42 ///
43 /// Once these prolog items are read, we initialize a LineEntry struct with
44 /// the start address of the function from the FunctionInfo's address range,
45 /// a default file index of 1, and the line number set to "FirstLine" from
46 /// the prolog above:
47 ///
48 ///   LineEntry Row(BaseAddr, 1, FirstLine);
49 ///
50 /// The line table state machine is now initialized and ready to be parsed.
51 /// The stream that follows this encodes the line entries in a compact
52 /// form. Some opcodes cause "Row" to be modified and some opcodes may also
53 /// push "Row" onto the end of the "LineTable.Lines" vector. The end result
54 /// is a vector of LineEntry structs that is sorted in ascending address
55 /// order.
56 ///
57 /// NORMAL OPCODES
58 ///
59 /// The opcodes 0 through 3 are normal in opcodes. Their encoding and
60 /// descriptions are listed below:
61 ///
62 /// ENCODING ENUMERATION       VALUE DESCRIPTION
63 /// ======== ================  ===== ========================================
64 ///          LTOC_EndSequence  0x00  Parsing is done.
65 /// ULEB     LTOC_SetFile      0x01  Row.File = ULEB
66 /// ULEB     LTOC_AdvancePC    0x02  Row.Addr += ULEB, push "Row".
67 /// SLEB     LTOC_AdvanceLine  0x03  Row.Line += SLEB
68 ///          LTOC_FirstSpecial 0x04  First special opcode (see SPECIAL
69 ///                                  OPCODES below).
70 ///
71 /// SPECIAL OPCODES
72 ///
73 /// Opcodes LTOC_FirstSpecial through 255 are special opcodes that always
74 /// increment both the Row.Addr and Row.Line and push "Row" onto the
75 /// LineEntry.Lines array. They do this by using some of the bits to
76 /// increment/decrement the source line number, and some of the bits to
77 /// increment the address. Line numbers can go up or down when making line
78 /// tables, where addresses always only increase since line tables are sorted
79 /// by address.
80 ///
81 /// In order to calculate the amount to increment the line and address for
82 /// these special opcodes, we calculate the number of values reserved for the
83 /// line increment/decrement using the "MinDelta" and "MaxDelta" from the
84 /// prolog:
85 ///
86 ///     const int64_t LineRange = MaxDelta - MinDelta + 1;
87 ///
88 /// Then we can adjust the opcode to not include any of the normal opcodes:
89 ///
90 ///     const uint8_t AdjustedOp = Opcode - LTOC_FirstSpecial;
91 ///
92 /// And we can calculate the line offset, and address offset:
93 ///
94 ///     const int64_t LineDelta = MinDelta + (AdjustedOp % LineRange);
95 ///     const uint64_t AddrDelta = (AdjustedOp / LineRange);
96 ///
97 /// And use these to modify our "Row":
98 ///
99 ///     Row.Line += LineDelta;
100 ///     Row.Addr += AddrDelta;
101 ///
102 /// And push a row onto the line table:
103 ///
104 ///     Lines.push_back(Row);
105 ///
106 /// This is verify similar to the way that DWARF encodes its line tables. The
107 /// only difference is the DWARF line tables have more normal opcodes and the
108 /// "Row" contains more members, like source column number, bools for end of
109 /// prologue, beginnging of epilogue, is statement and many others. There are
110 /// also more complex rules that happen for the extra normal opcodes. By
111 /// leaving these extra opcodes out, we leave more bits for the special
112 /// opcodes that allows us to encode line tables in fewer bytes than standard
113 /// DWARF encodings.
114 ///
115 /// Opcodes that will push "Row" onto the LineEntry.Lines include the
116 /// LTOC_AdvancePC opcode and all special opcodes. All other opcodes
117 /// only modify the current "Row", or cause the line table to end.
118 class LineTable {
119   typedef std::vector<gsym::LineEntry> Collection;
120   Collection Lines; ///< All line entries in the line table.
121 public:
122   /// Lookup a single address within a line table's data.
123   ///
124   /// Clients have the option to decode an entire line table using
125   /// LineTable::decode() or just find a single matching entry using this
126   /// function. The benefit of using this function is that parsed LineEntry
127   /// objects that do not match will not be stored in an array. This will avoid
128   /// memory allocation costs and parsing can stop once a match has been found.
129   ///
130   /// \param Data The binary stream to read the data from. This object must
131   /// have the data for the LineTable object starting at offset zero. The data
132   /// can contain more data than needed.
133   ///
134   /// \param BaseAddr The base address to use when decoding the line table.
135   /// This will be the FunctionInfo's start address and will be used to
136   /// initialize the line table row prior to parsing any opcodes.
137   ///
138   /// \returns An LineEntry object if a match is found, error otherwise.
139   static Expected<LineEntry> lookup(DataExtractor &Data, uint64_t BaseAddr,
140                                     uint64_t Addr);
141 
142   /// Decode an LineTable object from a binary data stream.
143   ///
144   /// \param Data The binary stream to read the data from. This object must
145   /// have the data for the LineTable object starting at offset zero. The data
146   /// can contain more data than needed.
147   ///
148   /// \param BaseAddr The base address to use when decoding the line table.
149   /// This will be the FunctionInfo's start address and will be used to
150   /// initialize the line table row prior to parsing any opcodes.
151   ///
152   /// \returns An LineTable or an error describing the issue that was
153   /// encountered during decoding.
154   static llvm::Expected<LineTable> decode(DataExtractor &Data,
155                                           uint64_t BaseAddr);
156   /// Encode this LineTable object into FileWriter stream.
157   ///
158   /// \param O The binary stream to write the data to at the current file
159   /// position.
160   ///
161   /// \param BaseAddr The base address to use when decoding the line table.
162   /// This will be the FunctionInfo's start address.
163   ///
164   /// \returns An error object that indicates success or failure or the
165   /// encoding process.
166   llvm::Error encode(FileWriter &O, uint64_t BaseAddr) const;
167   bool empty() const { return Lines.empty(); }
168   void clear() { Lines.clear(); }
169   void push(const LineEntry &LE) {
170     Lines.push_back(LE);
171   }
172   size_t isValid() const {
173     return !Lines.empty();
174   }
175   size_t size() const {
176     return Lines.size();
177   }
178   LineEntry &get(size_t i) {
179     assert(i < Lines.size());
180     return Lines[i];
181   }
182   const LineEntry &get(size_t i) const {
183     assert(i < Lines.size());
184     return Lines[i];
185   }
186   LineEntry &operator[](size_t i) {
187     return get(i);
188   }
189   const LineEntry &operator[](size_t i) const {
190     return get(i);
191   }
192   bool operator==(const LineTable &RHS) const {
193     return Lines == RHS.Lines;
194   }
195   bool operator!=(const LineTable &RHS) const {
196     return Lines != RHS.Lines;
197   }
198   bool operator<(const LineTable &RHS) const {
199     const auto LHSSize = Lines.size();
200     const auto RHSSize = RHS.Lines.size();
201     if (LHSSize == RHSSize)
202       return Lines < RHS.Lines;
203     return LHSSize < RHSSize;
204   }
205   Collection::const_iterator begin() const { return Lines.begin(); }
206   Collection::const_iterator end() const { return Lines.end(); }
207 
208 };
209 
210 raw_ostream &operator<<(raw_ostream &OS, const gsym::LineTable &LT);
211 
212 } // namespace gsym
213 } // namespace llvm
214 
215 #endif // #ifndef LLVM_DEBUGINFO_GSYM_LINETABLE_H
216