1//===-- llvm/lib/CodeGen/AsmPrinter/WinCodeViewLineTables.cpp --*- 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// This file contains support for writing line tables info into COFF files.
11//
12//===----------------------------------------------------------------------===//
13
14#include "WinCodeViewLineTables.h"
15#include "llvm/MC/MCExpr.h"
16#include "llvm/MC/MCSymbol.h"
17#include "llvm/Support/COFF.h"
18
19namespace llvm {
20
21StringRef WinCodeViewLineTables::getFullFilepath(const MDNode *S) {
22  assert(S);
23  assert((isa<MDCompileUnit>(S) || isa<MDFile>(S) || isa<MDSubprogram>(S) ||
24          isa<MDLexicalBlockBase>(S)) &&
25         "Unexpected scope info");
26
27  auto *Scope = cast<MDScope>(S);
28  StringRef Dir = Scope->getDirectory(),
29            Filename = Scope->getFilename();
30  char *&Result = DirAndFilenameToFilepathMap[std::make_pair(Dir, Filename)];
31  if (Result)
32    return Result;
33
34  // Clang emits directory and relative filename info into the IR, but CodeView
35  // operates on full paths.  We could change Clang to emit full paths too, but
36  // that would increase the IR size and probably not needed for other users.
37  // For now, just concatenate and canonicalize the path here.
38  std::string Filepath;
39  if (Filename.find(':') == 1)
40    Filepath = Filename;
41  else
42    Filepath = (Dir + "\\" + Filename).str();
43
44  // Canonicalize the path.  We have to do it textually because we may no longer
45  // have access the file in the filesystem.
46  // First, replace all slashes with backslashes.
47  std::replace(Filepath.begin(), Filepath.end(), '/', '\\');
48
49  // Remove all "\.\" with "\".
50  size_t Cursor = 0;
51  while ((Cursor = Filepath.find("\\.\\", Cursor)) != std::string::npos)
52    Filepath.erase(Cursor, 2);
53
54  // Replace all "\XXX\..\" with "\".  Don't try too hard though as the original
55  // path should be well-formatted, e.g. start with a drive letter, etc.
56  Cursor = 0;
57  while ((Cursor = Filepath.find("\\..\\", Cursor)) != std::string::npos) {
58    // Something's wrong if the path starts with "\..\", abort.
59    if (Cursor == 0)
60      break;
61
62    size_t PrevSlash = Filepath.rfind('\\', Cursor - 1);
63    if (PrevSlash == std::string::npos)
64      // Something's wrong, abort.
65      break;
66
67    Filepath.erase(PrevSlash, Cursor + 3 - PrevSlash);
68    // The next ".." might be following the one we've just erased.
69    Cursor = PrevSlash;
70  }
71
72  // Remove all duplicate backslashes.
73  Cursor = 0;
74  while ((Cursor = Filepath.find("\\\\", Cursor)) != std::string::npos)
75    Filepath.erase(Cursor, 1);
76
77  Result = strdup(Filepath.c_str());
78  return StringRef(Result);
79}
80
81void WinCodeViewLineTables::maybeRecordLocation(DebugLoc DL,
82                                                const MachineFunction *MF) {
83  const MDNode *Scope = DL.getScope();
84  if (!Scope)
85    return;
86  StringRef Filename = getFullFilepath(Scope);
87
88  // Skip this instruction if it has the same file:line as the previous one.
89  assert(CurFn);
90  if (!CurFn->Instrs.empty()) {
91    const InstrInfoTy &LastInstr = InstrInfo[CurFn->Instrs.back()];
92    if (LastInstr.Filename == Filename && LastInstr.LineNumber == DL.getLine())
93      return;
94  }
95  FileNameRegistry.add(Filename);
96
97  MCSymbol *MCL = Asm->MMI->getContext().CreateTempSymbol();
98  Asm->OutStreamer.EmitLabel(MCL);
99  CurFn->Instrs.push_back(MCL);
100  InstrInfo[MCL] = InstrInfoTy(Filename, DL.getLine());
101}
102
103WinCodeViewLineTables::WinCodeViewLineTables(AsmPrinter *AP)
104    : Asm(nullptr), CurFn(nullptr) {
105  MachineModuleInfo *MMI = AP->MMI;
106
107  // If module doesn't have named metadata anchors or COFF debug section
108  // is not available, skip any debug info related stuff.
109  if (!MMI->getModule()->getNamedMetadata("llvm.dbg.cu") ||
110      !AP->getObjFileLowering().getCOFFDebugSymbolsSection())
111    return;
112
113  // Tell MMI that we have debug info.
114  MMI->setDebugInfoAvailability(true);
115  Asm = AP;
116}
117
118void WinCodeViewLineTables::endModule() {
119  if (FnDebugInfo.empty())
120    return;
121
122  assert(Asm != nullptr);
123  Asm->OutStreamer.SwitchSection(
124      Asm->getObjFileLowering().getCOFFDebugSymbolsSection());
125  Asm->EmitInt32(COFF::DEBUG_SECTION_MAGIC);
126
127  // The COFF .debug$S section consists of several subsections, each starting
128  // with a 4-byte control code (e.g. 0xF1, 0xF2, etc) and then a 4-byte length
129  // of the payload followed by the payload itself.  The subsections are 4-byte
130  // aligned.
131
132  // Emit per-function debug information.  This code is extracted into a
133  // separate function for readability.
134  for (size_t I = 0, E = VisitedFunctions.size(); I != E; ++I)
135    emitDebugInfoForFunction(VisitedFunctions[I]);
136
137  // This subsection holds a file index to offset in string table table.
138  Asm->OutStreamer.AddComment("File index to string table offset subsection");
139  Asm->EmitInt32(COFF::DEBUG_INDEX_SUBSECTION);
140  size_t NumFilenames = FileNameRegistry.Infos.size();
141  Asm->EmitInt32(8 * NumFilenames);
142  for (size_t I = 0, E = FileNameRegistry.Filenames.size(); I != E; ++I) {
143    StringRef Filename = FileNameRegistry.Filenames[I];
144    // For each unique filename, just write its offset in the string table.
145    Asm->EmitInt32(FileNameRegistry.Infos[Filename].StartOffset);
146    // The function name offset is not followed by any additional data.
147    Asm->EmitInt32(0);
148  }
149
150  // This subsection holds the string table.
151  Asm->OutStreamer.AddComment("String table");
152  Asm->EmitInt32(COFF::DEBUG_STRING_TABLE_SUBSECTION);
153  Asm->EmitInt32(FileNameRegistry.LastOffset);
154  // The payload starts with a null character.
155  Asm->EmitInt8(0);
156
157  for (size_t I = 0, E = FileNameRegistry.Filenames.size(); I != E; ++I) {
158    // Just emit unique filenames one by one, separated by a null character.
159    Asm->OutStreamer.EmitBytes(FileNameRegistry.Filenames[I]);
160    Asm->EmitInt8(0);
161  }
162
163  // No more subsections. Fill with zeros to align the end of the section by 4.
164  Asm->OutStreamer.EmitFill((-FileNameRegistry.LastOffset) % 4, 0);
165
166  clear();
167}
168
169static void EmitLabelDiff(MCStreamer &Streamer,
170                          const MCSymbol *From, const MCSymbol *To,
171                          unsigned int Size = 4) {
172  MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None;
173  MCContext &Context = Streamer.getContext();
174  const MCExpr *FromRef = MCSymbolRefExpr::Create(From, Variant, Context),
175               *ToRef   = MCSymbolRefExpr::Create(To, Variant, Context);
176  const MCExpr *AddrDelta =
177      MCBinaryExpr::Create(MCBinaryExpr::Sub, ToRef, FromRef, Context);
178  Streamer.EmitValue(AddrDelta, Size);
179}
180
181void WinCodeViewLineTables::emitDebugInfoForFunction(const Function *GV) {
182  // For each function there is a separate subsection
183  // which holds the PC to file:line table.
184  const MCSymbol *Fn = Asm->getSymbol(GV);
185  assert(Fn);
186
187  const FunctionInfo &FI = FnDebugInfo[GV];
188  if (FI.Instrs.empty())
189    return;
190  assert(FI.End && "Don't know where the function ends?");
191
192  StringRef GVName = GV->getName();
193  StringRef FuncName;
194  if (DISubprogram SP = getDISubprogram(GV))
195    FuncName = SP->getDisplayName();
196
197  // FIXME Clang currently sets DisplayName to "bar" for a C++
198  // "namespace_foo::bar" function, see PR21528.  Luckily, dbghelp.dll is trying
199  // to demangle display names anyways, so let's just put a mangled name into
200  // the symbols subsection until Clang gives us what we need.
201  if (GVName.startswith("\01?"))
202    FuncName = GVName.substr(1);
203  // Emit a symbol subsection, required by VS2012+ to find function boundaries.
204  MCSymbol *SymbolsBegin = Asm->MMI->getContext().CreateTempSymbol(),
205           *SymbolsEnd = Asm->MMI->getContext().CreateTempSymbol();
206  Asm->OutStreamer.AddComment("Symbol subsection for " + Twine(FuncName));
207  Asm->EmitInt32(COFF::DEBUG_SYMBOL_SUBSECTION);
208  EmitLabelDiff(Asm->OutStreamer, SymbolsBegin, SymbolsEnd);
209  Asm->OutStreamer.EmitLabel(SymbolsBegin);
210  {
211    MCSymbol *ProcSegmentBegin = Asm->MMI->getContext().CreateTempSymbol(),
212             *ProcSegmentEnd = Asm->MMI->getContext().CreateTempSymbol();
213    EmitLabelDiff(Asm->OutStreamer, ProcSegmentBegin, ProcSegmentEnd, 2);
214    Asm->OutStreamer.EmitLabel(ProcSegmentBegin);
215
216    Asm->EmitInt16(COFF::DEBUG_SYMBOL_TYPE_PROC_START);
217    // Some bytes of this segment don't seem to be required for basic debugging,
218    // so just fill them with zeroes.
219    Asm->OutStreamer.EmitFill(12, 0);
220    // This is the important bit that tells the debugger where the function
221    // code is located and what's its size:
222    EmitLabelDiff(Asm->OutStreamer, Fn, FI.End);
223    Asm->OutStreamer.EmitFill(12, 0);
224    Asm->OutStreamer.EmitCOFFSecRel32(Fn);
225    Asm->OutStreamer.EmitCOFFSectionIndex(Fn);
226    Asm->EmitInt8(0);
227    // Emit the function display name as a null-terminated string.
228    Asm->OutStreamer.EmitBytes(FuncName);
229    Asm->EmitInt8(0);
230    Asm->OutStreamer.EmitLabel(ProcSegmentEnd);
231
232    // We're done with this function.
233    Asm->EmitInt16(0x0002);
234    Asm->EmitInt16(COFF::DEBUG_SYMBOL_TYPE_PROC_END);
235  }
236  Asm->OutStreamer.EmitLabel(SymbolsEnd);
237  // Every subsection must be aligned to a 4-byte boundary.
238  Asm->OutStreamer.EmitFill((-FuncName.size()) % 4, 0);
239
240  // PCs/Instructions are grouped into segments sharing the same filename.
241  // Pre-calculate the lengths (in instructions) of these segments and store
242  // them in a map for convenience.  Each index in the map is the sequential
243  // number of the respective instruction that starts a new segment.
244  DenseMap<size_t, size_t> FilenameSegmentLengths;
245  size_t LastSegmentEnd = 0;
246  StringRef PrevFilename = InstrInfo[FI.Instrs[0]].Filename;
247  for (size_t J = 1, F = FI.Instrs.size(); J != F; ++J) {
248    if (PrevFilename == InstrInfo[FI.Instrs[J]].Filename)
249      continue;
250    FilenameSegmentLengths[LastSegmentEnd] = J - LastSegmentEnd;
251    LastSegmentEnd = J;
252    PrevFilename = InstrInfo[FI.Instrs[J]].Filename;
253  }
254  FilenameSegmentLengths[LastSegmentEnd] = FI.Instrs.size() - LastSegmentEnd;
255
256  // Emit a line table subsection, requred to do PC-to-file:line lookup.
257  Asm->OutStreamer.AddComment("Line table subsection for " + Twine(FuncName));
258  Asm->EmitInt32(COFF::DEBUG_LINE_TABLE_SUBSECTION);
259  MCSymbol *LineTableBegin = Asm->MMI->getContext().CreateTempSymbol(),
260           *LineTableEnd = Asm->MMI->getContext().CreateTempSymbol();
261  EmitLabelDiff(Asm->OutStreamer, LineTableBegin, LineTableEnd);
262  Asm->OutStreamer.EmitLabel(LineTableBegin);
263
264  // Identify the function this subsection is for.
265  Asm->OutStreamer.EmitCOFFSecRel32(Fn);
266  Asm->OutStreamer.EmitCOFFSectionIndex(Fn);
267  // Insert padding after a 16-bit section index.
268  Asm->EmitInt16(0);
269
270  // Length of the function's code, in bytes.
271  EmitLabelDiff(Asm->OutStreamer, Fn, FI.End);
272
273  // PC-to-linenumber lookup table:
274  MCSymbol *FileSegmentEnd = nullptr;
275  for (size_t J = 0, F = FI.Instrs.size(); J != F; ++J) {
276    MCSymbol *Instr = FI.Instrs[J];
277    assert(InstrInfo.count(Instr));
278
279    if (FilenameSegmentLengths.count(J)) {
280      // We came to a beginning of a new filename segment.
281      if (FileSegmentEnd)
282        Asm->OutStreamer.EmitLabel(FileSegmentEnd);
283      StringRef CurFilename = InstrInfo[FI.Instrs[J]].Filename;
284      assert(FileNameRegistry.Infos.count(CurFilename));
285      size_t IndexInStringTable =
286          FileNameRegistry.Infos[CurFilename].FilenameID;
287      // Each segment starts with the offset of the filename
288      // in the string table.
289      Asm->OutStreamer.AddComment(
290          "Segment for file '" + Twine(CurFilename) + "' begins");
291      MCSymbol *FileSegmentBegin = Asm->MMI->getContext().CreateTempSymbol();
292      Asm->OutStreamer.EmitLabel(FileSegmentBegin);
293      Asm->EmitInt32(8 * IndexInStringTable);
294
295      // Number of PC records in the lookup table.
296      size_t SegmentLength = FilenameSegmentLengths[J];
297      Asm->EmitInt32(SegmentLength);
298
299      // Full size of the segment for this filename, including the prev two
300      // records.
301      FileSegmentEnd = Asm->MMI->getContext().CreateTempSymbol();
302      EmitLabelDiff(Asm->OutStreamer, FileSegmentBegin, FileSegmentEnd);
303    }
304
305    // The first PC with the given linenumber and the linenumber itself.
306    EmitLabelDiff(Asm->OutStreamer, Fn, Instr);
307    Asm->EmitInt32(InstrInfo[Instr].LineNumber);
308  }
309
310  if (FileSegmentEnd)
311    Asm->OutStreamer.EmitLabel(FileSegmentEnd);
312  Asm->OutStreamer.EmitLabel(LineTableEnd);
313}
314
315void WinCodeViewLineTables::beginFunction(const MachineFunction *MF) {
316  assert(!CurFn && "Can't process two functions at once!");
317
318  if (!Asm || !Asm->MMI->hasDebugInfo())
319    return;
320
321  const Function *GV = MF->getFunction();
322  assert(FnDebugInfo.count(GV) == false);
323  VisitedFunctions.push_back(GV);
324  CurFn = &FnDebugInfo[GV];
325
326  // Find the end of the function prolog.
327  // FIXME: is there a simpler a way to do this? Can we just search
328  // for the first instruction of the function, not the last of the prolog?
329  DebugLoc PrologEndLoc;
330  bool EmptyPrologue = true;
331  for (const auto &MBB : *MF) {
332    if (PrologEndLoc)
333      break;
334    for (const auto &MI : MBB) {
335      if (MI.isDebugValue())
336        continue;
337
338      // First known non-DBG_VALUE and non-frame setup location marks
339      // the beginning of the function body.
340      // FIXME: do we need the first subcondition?
341      if (!MI.getFlag(MachineInstr::FrameSetup) && MI.getDebugLoc()) {
342        PrologEndLoc = MI.getDebugLoc();
343        break;
344      }
345      EmptyPrologue = false;
346    }
347  }
348  // Record beginning of function if we have a non-empty prologue.
349  if (PrologEndLoc && !EmptyPrologue) {
350    DebugLoc FnStartDL = PrologEndLoc.getFnDebugLoc();
351    maybeRecordLocation(FnStartDL, MF);
352  }
353}
354
355void WinCodeViewLineTables::endFunction(const MachineFunction *MF) {
356  if (!Asm || !CurFn)  // We haven't created any debug info for this function.
357    return;
358
359  const Function *GV = MF->getFunction();
360  assert(FnDebugInfo.count(GV));
361  assert(CurFn == &FnDebugInfo[GV]);
362
363  if (CurFn->Instrs.empty()) {
364    FnDebugInfo.erase(GV);
365    VisitedFunctions.pop_back();
366  } else {
367    CurFn->End = Asm->getFunctionEnd();
368  }
369  CurFn = nullptr;
370}
371
372void WinCodeViewLineTables::beginInstruction(const MachineInstr *MI) {
373  // Ignore DBG_VALUE locations and function prologue.
374  if (!Asm || MI->isDebugValue() || MI->getFlag(MachineInstr::FrameSetup))
375    return;
376  DebugLoc DL = MI->getDebugLoc();
377  if (DL == PrevInstLoc || !DL)
378    return;
379  maybeRecordLocation(DL, Asm->MF);
380}
381}
382