1 //===- DWARFDebugLine.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_DWARF_DWARFDEBUGLINE_H
10 #define LLVM_DEBUGINFO_DWARF_DWARFDEBUGLINE_H
11 
12 #include "llvm/ADT/Optional.h"
13 #include "llvm/ADT/StringRef.h"
14 #include "llvm/DebugInfo/DIContext.h"
15 #include "llvm/DebugInfo/DWARF/DWARFCompileUnit.h"
16 #include "llvm/DebugInfo/DWARF/DWARFDataExtractor.h"
17 #include "llvm/DebugInfo/DWARF/DWARFFormValue.h"
18 #include "llvm/DebugInfo/DWARF/DWARFRelocMap.h"
19 #include "llvm/DebugInfo/DWARF/DWARFTypeUnit.h"
20 #include "llvm/Support/MD5.h"
21 #include "llvm/Support/Path.h"
22 #include <cstdint>
23 #include <map>
24 #include <string>
25 #include <vector>
26 
27 namespace llvm {
28 
29 class DWARFUnit;
30 class raw_ostream;
31 
32 class DWARFDebugLine {
33 public:
34   struct FileNameEntry {
35     FileNameEntry() = default;
36 
37     DWARFFormValue Name;
38     uint64_t DirIdx = 0;
39     uint64_t ModTime = 0;
40     uint64_t Length = 0;
41     MD5::MD5Result Checksum;
42     DWARFFormValue Source;
43   };
44 
45   /// Tracks which optional content types are present in a DWARF file name
46   /// entry format.
47   struct ContentTypeTracker {
48     ContentTypeTracker() = default;
49 
50     /// Whether filename entries provide a modification timestamp.
51     bool HasModTime = false;
52     /// Whether filename entries provide a file size.
53     bool HasLength = false;
54     /// For v5, whether filename entries provide an MD5 checksum.
55     bool HasMD5 = false;
56     /// For v5, whether filename entries provide source text.
57     bool HasSource = false;
58 
59     /// Update tracked content types with \p ContentType.
60     void trackContentType(dwarf::LineNumberEntryFormat ContentType);
61   };
62 
63   struct Prologue {
64     Prologue();
65 
66     /// The size in bytes of the statement information for this compilation unit
67     /// (not including the total_length field itself).
68     uint64_t TotalLength;
69     /// Version, address size (starting in v5), and DWARF32/64 format; these
70     /// parameters affect interpretation of forms (used in the directory and
71     /// file tables starting with v5).
72     dwarf::FormParams FormParams;
73     /// The number of bytes following the prologue_length field to the beginning
74     /// of the first byte of the statement program itself.
75     uint64_t PrologueLength;
76     /// In v5, size in bytes of a segment selector.
77     uint8_t SegSelectorSize;
78     /// The size in bytes of the smallest target machine instruction. Statement
79     /// program opcodes that alter the address register first multiply their
80     /// operands by this value.
81     uint8_t MinInstLength;
82     /// The maximum number of individual operations that may be encoded in an
83     /// instruction.
84     uint8_t MaxOpsPerInst;
85     /// The initial value of theis_stmtregister.
86     uint8_t DefaultIsStmt;
87     /// This parameter affects the meaning of the special opcodes. See below.
88     int8_t LineBase;
89     /// This parameter affects the meaning of the special opcodes. See below.
90     uint8_t LineRange;
91     /// The number assigned to the first special opcode.
92     uint8_t OpcodeBase;
93     /// This tracks which optional file format content types are present.
94     ContentTypeTracker ContentTypes;
95     std::vector<uint8_t> StandardOpcodeLengths;
96     std::vector<DWARFFormValue> IncludeDirectories;
97     std::vector<FileNameEntry> FileNames;
98 
99     const dwarf::FormParams getFormParams() const { return FormParams; }
100     uint16_t getVersion() const { return FormParams.Version; }
101     uint8_t getAddressSize() const { return FormParams.AddrSize; }
102     bool isDWARF64() const { return FormParams.Format == dwarf::DWARF64; }
103 
104     uint32_t sizeofTotalLength() const { return isDWARF64() ? 12 : 4; }
105 
106     uint32_t sizeofPrologueLength() const { return isDWARF64() ? 8 : 4; }
107 
108     bool totalLengthIsValid() const;
109 
110     /// Length of the prologue in bytes.
111     uint64_t getLength() const;
112 
113     int32_t getMaxLineIncrementForSpecialOpcode() const {
114       return LineBase + (int8_t)LineRange - 1;
115     }
116 
117     /// Get DWARF-version aware access to the file name entry at the provided
118     /// index.
119     const llvm::DWARFDebugLine::FileNameEntry &
120     getFileNameEntry(uint64_t Index) const;
121 
122     bool hasFileAtIndex(uint64_t FileIndex) const;
123 
124     Optional<uint64_t> getLastValidFileIndex() const;
125 
126     bool
127     getFileNameByIndex(uint64_t FileIndex, StringRef CompDir,
128                        DILineInfoSpecifier::FileLineInfoKind Kind,
129                        std::string &Result,
130                        sys::path::Style Style = sys::path::Style::native) const;
131 
132     void clear();
133     void dump(raw_ostream &OS, DIDumpOptions DumpOptions) const;
134     Error parse(DWARFDataExtractor Data, uint64_t *OffsetPtr,
135                 function_ref<void(Error)> RecoverableErrorHandler,
136                 const DWARFContext &Ctx, const DWARFUnit *U = nullptr);
137   };
138 
139   /// Standard .debug_line state machine structure.
140   struct Row {
141     explicit Row(bool DefaultIsStmt = false);
142 
143     /// Called after a row is appended to the matrix.
144     void postAppend();
145     void reset(bool DefaultIsStmt);
146     void dump(raw_ostream &OS) const;
147 
148     static void dumpTableHeader(raw_ostream &OS, unsigned Indent);
149 
150     static bool orderByAddress(const Row &LHS, const Row &RHS) {
151       return std::tie(LHS.Address.SectionIndex, LHS.Address.Address) <
152              std::tie(RHS.Address.SectionIndex, RHS.Address.Address);
153     }
154 
155     /// The program-counter value corresponding to a machine instruction
156     /// generated by the compiler and section index pointing to the section
157     /// containg this PC. If relocation information is present then section
158     /// index is the index of the section which contains above address.
159     /// Otherwise this is object::SectionedAddress::Undef value.
160     object::SectionedAddress Address;
161     /// An unsigned integer indicating a source line number. Lines are numbered
162     /// beginning at 1. The compiler may emit the value 0 in cases where an
163     /// instruction cannot be attributed to any source line.
164     uint32_t Line;
165     /// An unsigned integer indicating a column number within a source line.
166     /// Columns are numbered beginning at 1. The value 0 is reserved to indicate
167     /// that a statement begins at the 'left edge' of the line.
168     uint16_t Column;
169     /// An unsigned integer indicating the identity of the source file
170     /// corresponding to a machine instruction.
171     uint16_t File;
172     /// An unsigned integer representing the DWARF path discriminator value
173     /// for this location.
174     uint32_t Discriminator;
175     /// An unsigned integer whose value encodes the applicable instruction set
176     /// architecture for the current instruction.
177     uint8_t Isa;
178     /// A boolean indicating that the current instruction is the beginning of a
179     /// statement.
180     uint8_t IsStmt : 1,
181         /// A boolean indicating that the current instruction is the
182         /// beginning of a basic block.
183         BasicBlock : 1,
184         /// A boolean indicating that the current address is that of the
185         /// first byte after the end of a sequence of target machine
186         /// instructions.
187         EndSequence : 1,
188         /// A boolean indicating that the current address is one (of possibly
189         /// many) where execution should be suspended for an entry breakpoint
190         /// of a function.
191         PrologueEnd : 1,
192         /// A boolean indicating that the current address is one (of possibly
193         /// many) where execution should be suspended for an exit breakpoint
194         /// of a function.
195         EpilogueBegin : 1;
196   };
197 
198   /// Represents a series of contiguous machine instructions. Line table for
199   /// each compilation unit may consist of multiple sequences, which are not
200   /// guaranteed to be in the order of ascending instruction address.
201   struct Sequence {
202     Sequence();
203 
204     /// Sequence describes instructions at address range [LowPC, HighPC)
205     /// and is described by line table rows [FirstRowIndex, LastRowIndex).
206     uint64_t LowPC;
207     uint64_t HighPC;
208     /// If relocation information is present then this is the index of the
209     /// section which contains above addresses. Otherwise this is
210     /// object::SectionedAddress::Undef value.
211     uint64_t SectionIndex;
212     unsigned FirstRowIndex;
213     unsigned LastRowIndex;
214     bool Empty;
215 
216     void reset();
217 
218     static bool orderByHighPC(const Sequence &LHS, const Sequence &RHS) {
219       return std::tie(LHS.SectionIndex, LHS.HighPC) <
220              std::tie(RHS.SectionIndex, RHS.HighPC);
221     }
222 
223     bool isValid() const {
224       return !Empty && (LowPC < HighPC) && (FirstRowIndex < LastRowIndex);
225     }
226 
227     bool containsPC(object::SectionedAddress PC) const {
228       return SectionIndex == PC.SectionIndex &&
229              (LowPC <= PC.Address && PC.Address < HighPC);
230     }
231   };
232 
233   struct LineTable {
234     LineTable();
235 
236     /// Represents an invalid row
237     const uint32_t UnknownRowIndex = UINT32_MAX;
238 
239     void appendRow(const DWARFDebugLine::Row &R) { Rows.push_back(R); }
240 
241     void appendSequence(const DWARFDebugLine::Sequence &S) {
242       Sequences.push_back(S);
243     }
244 
245     /// Returns the index of the row with file/line info for a given address,
246     /// or UnknownRowIndex if there is no such row.
247     uint32_t lookupAddress(object::SectionedAddress Address) const;
248 
249     bool lookupAddressRange(object::SectionedAddress Address, uint64_t Size,
250                             std::vector<uint32_t> &Result) const;
251 
252     bool hasFileAtIndex(uint64_t FileIndex) const {
253       return Prologue.hasFileAtIndex(FileIndex);
254     }
255 
256     Optional<uint64_t> getLastValidFileIndex() const {
257       return Prologue.getLastValidFileIndex();
258     }
259 
260     /// Extracts filename by its index in filename table in prologue.
261     /// In Dwarf 4, the files are 1-indexed and the current compilation file
262     /// name is not represented in the list. In DWARF v5, the files are
263     /// 0-indexed and the primary source file has the index 0.
264     /// Returns true on success.
265     bool getFileNameByIndex(uint64_t FileIndex, StringRef CompDir,
266                             DILineInfoSpecifier::FileLineInfoKind Kind,
267                             std::string &Result) const {
268       return Prologue.getFileNameByIndex(FileIndex, CompDir, Kind, Result);
269     }
270 
271     /// Fills the Result argument with the file and line information
272     /// corresponding to Address. Returns true on success.
273     bool getFileLineInfoForAddress(object::SectionedAddress Address,
274                                    const char *CompDir,
275                                    DILineInfoSpecifier::FileLineInfoKind Kind,
276                                    DILineInfo &Result) const;
277 
278     void dump(raw_ostream &OS, DIDumpOptions DumpOptions) const;
279     void clear();
280 
281     /// Parse prologue and all rows.
282     Error parse(DWARFDataExtractor &DebugLineData, uint64_t *OffsetPtr,
283                 const DWARFContext &Ctx, const DWARFUnit *U,
284                 function_ref<void(Error)> RecoverableErrorHandler,
285                 raw_ostream *OS = nullptr, bool Verbose = false);
286 
287     using RowVector = std::vector<Row>;
288     using RowIter = RowVector::const_iterator;
289     using SequenceVector = std::vector<Sequence>;
290     using SequenceIter = SequenceVector::const_iterator;
291 
292     struct Prologue Prologue;
293     RowVector Rows;
294     SequenceVector Sequences;
295 
296   private:
297     uint32_t findRowInSeq(const DWARFDebugLine::Sequence &Seq,
298                           object::SectionedAddress Address) const;
299     Optional<StringRef>
300     getSourceByIndex(uint64_t FileIndex,
301                      DILineInfoSpecifier::FileLineInfoKind Kind) const;
302 
303     uint32_t lookupAddressImpl(object::SectionedAddress Address) const;
304 
305     bool lookupAddressRangeImpl(object::SectionedAddress Address, uint64_t Size,
306                                 std::vector<uint32_t> &Result) const;
307   };
308 
309   const LineTable *getLineTable(uint64_t Offset) const;
310   Expected<const LineTable *>
311   getOrParseLineTable(DWARFDataExtractor &DebugLineData, uint64_t Offset,
312                       const DWARFContext &Ctx, const DWARFUnit *U,
313                       function_ref<void(Error)> RecoverableErrorHandler);
314 
315   /// Helper to allow for parsing of an entire .debug_line section in sequence.
316   class SectionParser {
317   public:
318     using LineToUnitMap = std::map<uint64_t, DWARFUnit *>;
319 
320     SectionParser(DWARFDataExtractor &Data, const DWARFContext &C,
321                   DWARFUnitVector::iterator_range Units);
322 
323     /// Get the next line table from the section. Report any issues via the
324     /// handlers.
325     ///
326     /// \param RecoverableErrorHandler - any issues that don't prevent further
327     /// parsing of the table will be reported through this handler.
328     /// \param UnrecoverableErrorHandler - any issues that prevent further
329     /// parsing of the table will be reported through this handler.
330     /// \param OS - if not null, the parser will print information about the
331     /// table as it parses it.
332     /// \param Verbose - if true, the parser will print verbose information when
333     /// printing to the output.
334     LineTable parseNext(function_ref<void(Error)> RecoverableErrorHandler,
335                         function_ref<void(Error)> UnrecoverableErrorHandler,
336                         raw_ostream *OS = nullptr, bool Verbose = false);
337 
338     /// Skip the current line table and go to the following line table (if
339     /// present) immediately.
340     ///
341     /// \param RecoverableErrorHandler - report any recoverable prologue
342     /// parsing issues via this handler.
343     /// \param UnrecoverableErrorHandler - report any unrecoverable prologue
344     /// parsing issues via this handler.
345     void skip(function_ref<void(Error)> RecoverableErrorHandler,
346               function_ref<void(Error)> UnrecoverableErrorHandler);
347 
348     /// Indicates if the parser has parsed as much as possible.
349     ///
350     /// \note Certain problems with the line table structure might mean that
351     /// parsing stops before the end of the section is reached.
352     bool done() const { return Done; }
353 
354     /// Get the offset the parser has reached.
355     uint64_t getOffset() const { return Offset; }
356 
357   private:
358     DWARFUnit *prepareToParse(uint64_t Offset);
359     void moveToNextTable(uint64_t OldOffset, const Prologue &P);
360 
361     LineToUnitMap LineToUnit;
362 
363     DWARFDataExtractor &DebugLineData;
364     const DWARFContext &Context;
365     uint64_t Offset = 0;
366     bool Done = false;
367   };
368 
369 private:
370   struct ParsingState {
371     ParsingState(struct LineTable *LT, uint64_t TableOffset,
372                  function_ref<void(Error)> ErrorHandler);
373 
374     void resetRowAndSequence();
375     void appendRowToMatrix();
376 
377     /// Advance the address by the \p OperationAdvance value. \returns the
378     /// amount advanced by.
379     uint64_t advanceAddr(uint64_t OperationAdvance, uint8_t Opcode,
380                          uint64_t OpcodeOffset);
381 
382     struct AddrAndAdjustedOpcode {
383       uint64_t AddrDelta;
384       uint8_t AdjustedOpcode;
385     };
386 
387     /// Advance the address as required by the specified \p Opcode.
388     /// \returns the amount advanced by and the calculated adjusted opcode.
389     AddrAndAdjustedOpcode advanceAddrForOpcode(uint8_t Opcode,
390                                                uint64_t OpcodeOffset);
391 
392     struct AddrAndLineDelta {
393       uint64_t Address;
394       int32_t Line;
395     };
396 
397     /// Advance the line and address as required by the specified special \p
398     /// Opcode. \returns the address and line delta.
399     AddrAndLineDelta handleSpecialOpcode(uint8_t Opcode, uint64_t OpcodeOffset);
400 
401     /// Line table we're currently parsing.
402     struct LineTable *LineTable;
403     struct Row Row;
404     struct Sequence Sequence;
405 
406   private:
407     uint64_t LineTableOffset;
408 
409     bool ReportAdvanceAddrProblem = true;
410     bool ReportBadLineRange = true;
411     function_ref<void(Error)> ErrorHandler;
412   };
413 
414   using LineTableMapTy = std::map<uint64_t, LineTable>;
415   using LineTableIter = LineTableMapTy::iterator;
416   using LineTableConstIter = LineTableMapTy::const_iterator;
417 
418   LineTableMapTy LineTableMap;
419 };
420 
421 } // end namespace llvm
422 
423 #endif // LLVM_DEBUGINFO_DWARF_DWARFDEBUGLINE_H
424