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