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     /// Get DWARF-version aware access to the file name entry at the provided
114     /// index.
115     const llvm::DWARFDebugLine::FileNameEntry &
116     getFileNameEntry(uint64_t Index) const;
117 
118     bool hasFileAtIndex(uint64_t FileIndex) const;
119 
120     Optional<uint64_t> getLastValidFileIndex() const;
121 
122     bool
123     getFileNameByIndex(uint64_t FileIndex, StringRef CompDir,
124                        DILineInfoSpecifier::FileLineInfoKind Kind,
125                        std::string &Result,
126                        sys::path::Style Style = sys::path::Style::native) const;
127 
128     void clear();
129     void dump(raw_ostream &OS, DIDumpOptions DumpOptions) const;
130     Error parse(DWARFDataExtractor Data, uint64_t *OffsetPtr,
131                 function_ref<void(Error)> RecoverableErrorHandler,
132                 const DWARFContext &Ctx, const DWARFUnit *U = nullptr);
133   };
134 
135   /// Standard .debug_line state machine structure.
136   struct Row {
137     explicit Row(bool DefaultIsStmt = false);
138 
139     /// Called after a row is appended to the matrix.
140     void postAppend();
141     void reset(bool DefaultIsStmt);
142     void dump(raw_ostream &OS) const;
143 
144     static void dumpTableHeader(raw_ostream &OS, unsigned Indent);
145 
146     static bool orderByAddress(const Row &LHS, const Row &RHS) {
147       return std::tie(LHS.Address.SectionIndex, LHS.Address.Address) <
148              std::tie(RHS.Address.SectionIndex, RHS.Address.Address);
149     }
150 
151     /// The program-counter value corresponding to a machine instruction
152     /// generated by the compiler and section index pointing to the section
153     /// containg this PC. If relocation information is present then section
154     /// index is the index of the section which contains above address.
155     /// Otherwise this is object::SectionedAddress::Undef value.
156     object::SectionedAddress Address;
157     /// An unsigned integer indicating a source line number. Lines are numbered
158     /// beginning at 1. The compiler may emit the value 0 in cases where an
159     /// instruction cannot be attributed to any source line.
160     uint32_t Line;
161     /// An unsigned integer indicating a column number within a source line.
162     /// Columns are numbered beginning at 1. The value 0 is reserved to indicate
163     /// that a statement begins at the 'left edge' of the line.
164     uint16_t Column;
165     /// An unsigned integer indicating the identity of the source file
166     /// corresponding to a machine instruction.
167     uint16_t File;
168     /// An unsigned integer representing the DWARF path discriminator value
169     /// for this location.
170     uint32_t Discriminator;
171     /// An unsigned integer whose value encodes the applicable instruction set
172     /// architecture for the current instruction.
173     uint8_t Isa;
174     /// A boolean indicating that the current instruction is the beginning of a
175     /// statement.
176     uint8_t IsStmt : 1,
177         /// A boolean indicating that the current instruction is the
178         /// beginning of a basic block.
179         BasicBlock : 1,
180         /// A boolean indicating that the current address is that of the
181         /// first byte after the end of a sequence of target machine
182         /// instructions.
183         EndSequence : 1,
184         /// A boolean indicating that the current address is one (of possibly
185         /// many) where execution should be suspended for an entry breakpoint
186         /// of a function.
187         PrologueEnd : 1,
188         /// A boolean indicating that the current address is one (of possibly
189         /// many) where execution should be suspended for an exit breakpoint
190         /// of a function.
191         EpilogueBegin : 1;
192   };
193 
194   /// Represents a series of contiguous machine instructions. Line table for
195   /// each compilation unit may consist of multiple sequences, which are not
196   /// guaranteed to be in the order of ascending instruction address.
197   struct Sequence {
198     Sequence();
199 
200     /// Sequence describes instructions at address range [LowPC, HighPC)
201     /// and is described by line table rows [FirstRowIndex, LastRowIndex).
202     uint64_t LowPC;
203     uint64_t HighPC;
204     /// If relocation information is present then this is the index of the
205     /// section which contains above addresses. Otherwise this is
206     /// object::SectionedAddress::Undef value.
207     uint64_t SectionIndex;
208     unsigned FirstRowIndex;
209     unsigned LastRowIndex;
210     bool Empty;
211 
212     void reset();
213 
214     static bool orderByHighPC(const Sequence &LHS, const Sequence &RHS) {
215       return std::tie(LHS.SectionIndex, LHS.HighPC) <
216              std::tie(RHS.SectionIndex, RHS.HighPC);
217     }
218 
219     bool isValid() const {
220       return !Empty && (LowPC < HighPC) && (FirstRowIndex < LastRowIndex);
221     }
222 
223     bool containsPC(object::SectionedAddress PC) const {
224       return SectionIndex == PC.SectionIndex &&
225              (LowPC <= PC.Address && PC.Address < HighPC);
226     }
227   };
228 
229   struct LineTable {
230     LineTable();
231 
232     /// Represents an invalid row
233     const uint32_t UnknownRowIndex = UINT32_MAX;
234 
235     void appendRow(const DWARFDebugLine::Row &R) { Rows.push_back(R); }
236 
237     void appendSequence(const DWARFDebugLine::Sequence &S) {
238       Sequences.push_back(S);
239     }
240 
241     /// Returns the index of the row with file/line info for a given address,
242     /// or UnknownRowIndex if there is no such row.
243     uint32_t lookupAddress(object::SectionedAddress Address) const;
244 
245     bool lookupAddressRange(object::SectionedAddress Address, uint64_t Size,
246                             std::vector<uint32_t> &Result) const;
247 
248     bool hasFileAtIndex(uint64_t FileIndex) const {
249       return Prologue.hasFileAtIndex(FileIndex);
250     }
251 
252     Optional<uint64_t> getLastValidFileIndex() const {
253       return Prologue.getLastValidFileIndex();
254     }
255 
256     /// Extracts filename by its index in filename table in prologue.
257     /// In Dwarf 4, the files are 1-indexed and the current compilation file
258     /// name is not represented in the list. In DWARF v5, the files are
259     /// 0-indexed and the primary source file has the index 0.
260     /// Returns true on success.
261     bool getFileNameByIndex(uint64_t FileIndex, StringRef CompDir,
262                             DILineInfoSpecifier::FileLineInfoKind Kind,
263                             std::string &Result) const {
264       return Prologue.getFileNameByIndex(FileIndex, CompDir, Kind, Result);
265     }
266 
267     /// Fills the Result argument with the file and line information
268     /// corresponding to Address. Returns true on success.
269     bool getFileLineInfoForAddress(object::SectionedAddress Address,
270                                    const char *CompDir,
271                                    DILineInfoSpecifier::FileLineInfoKind Kind,
272                                    DILineInfo &Result) const;
273 
274     void dump(raw_ostream &OS, DIDumpOptions DumpOptions) const;
275     void clear();
276 
277     /// Parse prologue and all rows.
278     Error parse(DWARFDataExtractor &DebugLineData, uint64_t *OffsetPtr,
279                 const DWARFContext &Ctx, const DWARFUnit *U,
280                 function_ref<void(Error)> RecoverableErrorHandler,
281                 raw_ostream *OS = nullptr, bool Verbose = false);
282 
283     using RowVector = std::vector<Row>;
284     using RowIter = RowVector::const_iterator;
285     using SequenceVector = std::vector<Sequence>;
286     using SequenceIter = SequenceVector::const_iterator;
287 
288     struct Prologue Prologue;
289     RowVector Rows;
290     SequenceVector Sequences;
291 
292   private:
293     uint32_t findRowInSeq(const DWARFDebugLine::Sequence &Seq,
294                           object::SectionedAddress Address) const;
295     Optional<StringRef>
296     getSourceByIndex(uint64_t FileIndex,
297                      DILineInfoSpecifier::FileLineInfoKind Kind) const;
298 
299     uint32_t lookupAddressImpl(object::SectionedAddress Address) const;
300 
301     bool lookupAddressRangeImpl(object::SectionedAddress Address, uint64_t Size,
302                                 std::vector<uint32_t> &Result) const;
303   };
304 
305   const LineTable *getLineTable(uint64_t Offset) const;
306   Expected<const LineTable *>
307   getOrParseLineTable(DWARFDataExtractor &DebugLineData, uint64_t Offset,
308                       const DWARFContext &Ctx, const DWARFUnit *U,
309                       function_ref<void(Error)> RecoverableErrorHandler);
310 
311   /// Helper to allow for parsing of an entire .debug_line section in sequence.
312   class SectionParser {
313   public:
314     using LineToUnitMap = std::map<uint64_t, DWARFUnit *>;
315 
316     SectionParser(DWARFDataExtractor &Data, const DWARFContext &C,
317                   DWARFUnitVector::iterator_range Units);
318 
319     /// Get the next line table from the section. Report any issues via the
320     /// handlers.
321     ///
322     /// \param RecoverableErrorHandler - any issues that don't prevent further
323     /// parsing of the table will be reported through this handler.
324     /// \param UnrecoverableErrorHandler - any issues that prevent further
325     /// parsing of the table will be reported through this handler.
326     /// \param OS - if not null, the parser will print information about the
327     /// table as it parses it.
328     /// \param Verbose - if true, the parser will print verbose information when
329     /// printing to the output.
330     LineTable parseNext(function_ref<void(Error)> RecoverableErrorHandler,
331                         function_ref<void(Error)> UnrecoverableErrorHandler,
332                         raw_ostream *OS = nullptr, bool Verbose = false);
333 
334     /// Skip the current line table and go to the following line table (if
335     /// present) immediately.
336     ///
337     /// \param RecoverableErrorHandler - report any recoverable prologue
338     /// parsing issues via this handler.
339     /// \param UnrecoverableErrorHandler - report any unrecoverable prologue
340     /// parsing issues via this handler.
341     void skip(function_ref<void(Error)> RecoverableErrorHandler,
342               function_ref<void(Error)> UnrecoverableErrorHandler);
343 
344     /// Indicates if the parser has parsed as much as possible.
345     ///
346     /// \note Certain problems with the line table structure might mean that
347     /// parsing stops before the end of the section is reached.
348     bool done() const { return Done; }
349 
350     /// Get the offset the parser has reached.
351     uint64_t getOffset() const { return Offset; }
352 
353   private:
354     DWARFUnit *prepareToParse(uint64_t Offset);
355     void moveToNextTable(uint64_t OldOffset, const Prologue &P);
356 
357     LineToUnitMap LineToUnit;
358 
359     DWARFDataExtractor &DebugLineData;
360     const DWARFContext &Context;
361     uint64_t Offset = 0;
362     bool Done = false;
363   };
364 
365 private:
366   struct ParsingState {
367     ParsingState(struct LineTable *LT, uint64_t TableOffset,
368                  function_ref<void(Error)> ErrorHandler);
369 
370     void resetRowAndSequence();
371     void appendRowToMatrix();
372 
373     /// Advance the address by the \p OperationAdvance value. \returns the
374     /// amount advanced by.
375     uint64_t advanceAddr(uint64_t OperationAdvance, uint8_t Opcode,
376                          uint64_t OpcodeOffset);
377 
378     struct AddrAndAdjustedOpcode {
379       uint64_t AddrDelta;
380       uint8_t AdjustedOpcode;
381     };
382 
383     /// Advance the address as required by the specified \p Opcode.
384     /// \returns the amount advanced by and the calculated adjusted opcode.
385     AddrAndAdjustedOpcode advanceAddrForOpcode(uint8_t Opcode,
386                                                uint64_t OpcodeOffset);
387 
388     struct AddrAndLineDelta {
389       uint64_t Address;
390       int32_t Line;
391     };
392 
393     /// Advance the line and address as required by the specified special \p
394     /// Opcode. \returns the address and line delta.
395     AddrAndLineDelta handleSpecialOpcode(uint8_t Opcode, uint64_t OpcodeOffset);
396 
397     /// Line table we're currently parsing.
398     struct LineTable *LineTable;
399     struct Row Row;
400     struct Sequence Sequence;
401 
402   private:
403     uint64_t LineTableOffset;
404 
405     bool ReportAdvanceAddrProblem = true;
406     bool ReportBadLineRange = true;
407     function_ref<void(Error)> ErrorHandler;
408   };
409 
410   using LineTableMapTy = std::map<uint64_t, LineTable>;
411   using LineTableIter = LineTableMapTy::iterator;
412   using LineTableConstIter = LineTableMapTy::const_iterator;
413 
414   LineTableMapTy LineTableMap;
415 };
416 
417 } // end namespace llvm
418 
419 #endif // LLVM_DEBUGINFO_DWARF_DWARFDEBUGLINE_H
420