lib/Debugger/ProgramInfo.cpp

   1 //===-- ProgramInfo.cpp - Compute and cache info about a program ----------===//
   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 implements the ProgramInfo and related classes, by sorting through
  11 // the loaded Module.
  12 //
  13 //===----------------------------------------------------------------------===//
  14
  15 #include "llvm/Debugger/ProgramInfo.h"
  16 #include "llvm/Constants.h"
  17 #include "llvm/Analysis/ValueTracking.h"
  18 #include "llvm/DerivedTypes.h"
  19 #include "llvm/Intrinsics.h"
  20 #include "llvm/IntrinsicInst.h"
  21 #include "llvm/Instructions.h"
  22 #include "llvm/Module.h"
  23 #include "llvm/Debugger/SourceFile.h"
  24 #include "llvm/Debugger/SourceLanguage.h"
  25 #include "llvm/Support/raw_ostream.h"
  26 #include "llvm/Support/SlowOperationInformer.h"
  27 #include "llvm/ADT/STLExtras.h"
  28 using namespace llvm;
  29
  30 /// getGlobalVariablesUsing - Return all of the global variables which have the
  31 /// specified value in their initializer somewhere.
  32 static void getGlobalVariablesUsing(Value *V,
  33                                     std::vector<GlobalVariable*> &Found) {
  34   for (Value::use_iterator I = V->use_begin(), E = V->use_end(); I != E; ++I) {
  35     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(*I))
  36       Found.push_back(GV);
  37     else if (Constant *C = dyn_cast<Constant>(*I))
  38       getGlobalVariablesUsing(C, Found);
  39   }
  40 }
  41
  42 /// getNextStopPoint - Follow the def-use chains of the specified LLVM value,
  43 /// traversing the use chains until we get to a stoppoint.  When we do, return
  44 /// the source location of the stoppoint.  If we don't find a stoppoint, return
  45 /// null.
  46 static const GlobalVariable *getNextStopPoint(const Value *V, unsigned &LineNo,
  47                                               unsigned &ColNo) {
  48   // The use-def chains can fork.  As such, we pick the lowest numbered one we
  49   // find.
  50   const GlobalVariable *LastDesc = 0;
  51   unsigned LastLineNo = ~0;
  52   unsigned LastColNo = ~0;
  53
  54   for (Value::use_const_iterator UI = V->use_begin(), E = V->use_end();
  55        UI != E; ++UI) {
  56     bool ShouldRecurse = true;
  57     if (cast<Instruction>(*UI)->getOpcode() == Instruction::PHI) {
  58       // Infinite loops == bad, ignore PHI nodes.
  59       ShouldRecurse = false;
  60     } else if (const CallInst *CI = dyn_cast<CallInst>(*UI)) {
  61
  62       // If we found a stop point, check to see if it is earlier than what we
  63       // already have.  If so, remember it.
  64       if (CI->getCalledFunction())
  65         if (const DbgStopPointInst *SPI = dyn_cast<DbgStopPointInst>(CI)) {
  66           unsigned CurLineNo = SPI->getLine();
  67           unsigned CurColNo = SPI->getColumn();
  68           const GlobalVariable *CurDesc = 0;
  69           const Value *Op = SPI->getContext();
  70
  71           if ((CurDesc = dyn_cast<GlobalVariable>(Op)) &&
  72               (LineNo < LastLineNo ||
  73                (LineNo == LastLineNo && ColNo < LastColNo))) {
  74             LastDesc = CurDesc;
  75             LastLineNo = CurLineNo;
  76             LastColNo = CurColNo;
  77           }
  78           ShouldRecurse = false;
  79         }
  80     }
  81
  82     // If this is not a phi node or a stopping point, recursively scan the users
  83     // of this instruction to skip over region.begin's and the like.
  84     if (ShouldRecurse) {
  85       unsigned CurLineNo, CurColNo;
  86       if (const GlobalVariable *GV = getNextStopPoint(*UI, CurLineNo,CurColNo)){
  87         if (LineNo < LastLineNo || (LineNo == LastLineNo && ColNo < LastColNo)){
  88           LastDesc = GV;
  89           LastLineNo = CurLineNo;
  90           LastColNo = CurColNo;
  91         }
  92       }
  93     }
  94   }
  95
  96   if (LastDesc) {
  97     LineNo = LastLineNo != ~0U ? LastLineNo : 0;
  98     ColNo  = LastColNo  != ~0U ? LastColNo : 0;
  99   }
 100   return LastDesc;
 101 }
 102
 103
 104 //===----------------------------------------------------------------------===//
 105 // SourceFileInfo implementation
 106 //
 107
 108 SourceFileInfo::SourceFileInfo(const GlobalVariable *Desc,
 109                                const SourceLanguage &Lang)
 110   : Language(&Lang), Descriptor(Desc) {
 111   Version = 0;
 112   SourceText = 0;
 113
 114   if (Desc && Desc->hasInitializer())
 115     if (ConstantStruct *CS = dyn_cast<ConstantStruct>(Desc->getInitializer()))
 116       if (CS->getNumOperands() > 4) {
 117         if (ConstantInt *CUI = dyn_cast<ConstantInt>(CS->getOperand(1)))
 118           Version = CUI->getZExtValue();
 119
 120         if (!GetConstantStringInfo(CS->getOperand(3), BaseName))
 121           BaseName = "";
 122         if (!GetConstantStringInfo(CS->getOperand(4), Directory))
 123           Directory = "";
 124       }
 125 }
 126
 127 SourceFileInfo::~SourceFileInfo() {
 128   delete SourceText;
 129 }
 130
 131 SourceFile &SourceFileInfo::getSourceText() const {
 132   // FIXME: this should take into account the source search directories!
 133   if (SourceText == 0) { // Read the file in if we haven't already.
 134     sys::Path tmpPath;
 135     if (!Directory.empty())
 136       tmpPath.set(Directory);
 137     tmpPath.appendComponent(BaseName);
 138     if (tmpPath.canRead())
 139       SourceText = new SourceFile(tmpPath.str(), Descriptor);
 140     else
 141       SourceText = new SourceFile(BaseName, Descriptor);
 142   }
 143   return *SourceText;
 144 }
 145
 146
 147 //===----------------------------------------------------------------------===//
 148 // SourceFunctionInfo implementation
 149 //
 150 SourceFunctionInfo::SourceFunctionInfo(ProgramInfo &PI,
 151                                        const GlobalVariable *Desc)
 152   : Descriptor(Desc) {
 153   LineNo = ColNo = 0;
 154   if (Desc && Desc->hasInitializer())
 155     if (ConstantStruct *CS = dyn_cast<ConstantStruct>(Desc->getInitializer()))
 156       if (CS->getNumOperands() > 2) {
 157         // Entry #1 is the file descriptor.
 158         if (const GlobalVariable *GV =
 159             dyn_cast<GlobalVariable>(CS->getOperand(1)))
 160           SourceFile = &PI.getSourceFile(GV);
 161
 162         // Entry #2 is the function name.
 163         if (!GetConstantStringInfo(CS->getOperand(2), Name))
 164           Name = "";
 165       }
 166 }
 167
 168 /// getSourceLocation - This method returns the location of the first stopping
 169 /// point in the function.
 170 void SourceFunctionInfo::getSourceLocation(unsigned &RetLineNo,
 171                                            unsigned &RetColNo) const {
 172   // If we haven't computed this yet...
 173   if (!LineNo) {
 174     // Look at all of the users of the function descriptor, looking for calls to
 175     // %llvm.dbg.func.start.
 176     for (Value::use_const_iterator UI = Descriptor->use_begin(),
 177            E = Descriptor->use_end(); UI != E; ++UI)
 178       if (const CallInst *CI = dyn_cast<CallInst>(*UI))
 179         if (const Function *F = CI->getCalledFunction())
 180           if (F->getIntrinsicID() == Intrinsic::dbg_func_start) {
 181             // We found the start of the function.  Check to see if there are
 182             // any stop points on the use-list of the function start.
 183             const GlobalVariable *SD = getNextStopPoint(CI, LineNo, ColNo);
 184             if (SD) {             // We found the first stop point!
 185               // This is just a sanity check.
 186               if (getSourceFile().getDescriptor() != SD)
 187                 outs() << "WARNING: first line of function is not in the"
 188                      << " file that the function descriptor claims it is in.\n";
 189               break;
 190             }
 191           }
 192   }
 193   RetLineNo = LineNo; RetColNo = ColNo;
 194 }
 195
 196 //===----------------------------------------------------------------------===//
 197 // ProgramInfo implementation
 198 //
 199
 200 ProgramInfo::ProgramInfo(Module *m) : M(m), ProgramTimeStamp(0,0) {
 201   assert(M && "Cannot create program information with a null module!");
 202   sys::PathWithStatus ModPath(M->getModuleIdentifier());
 203   const sys::FileStatus *Stat = ModPath.getFileStatus();
 204   if (Stat)
 205     ProgramTimeStamp = Stat->getTimestamp();
 206
 207   SourceFilesIsComplete = false;
 208   SourceFunctionsIsComplete = false;
 209 }
 210
 211 ProgramInfo::~ProgramInfo() {
 212   // Delete cached information about source program objects...
 213   for (std::map<const GlobalVariable*, SourceFileInfo*>::iterator
 214          I = SourceFiles.begin(), E = SourceFiles.end(); I != E; ++I)
 215     delete I->second;
 216   for (std::map<const GlobalVariable*, SourceFunctionInfo*>::iterator
 217          I = SourceFunctions.begin(), E = SourceFunctions.end(); I != E; ++I)
 218     delete I->second;
 219
 220   // Delete the source language caches.
 221   for (unsigned i = 0, e = LanguageCaches.size(); i != e; ++i)
 222     delete LanguageCaches[i].second;
 223 }
 224
 225
 226 //===----------------------------------------------------------------------===//
 227 // SourceFileInfo tracking...
 228 //
 229
 230 /// getSourceFile - Return source file information for the specified source file
 231 /// descriptor object, adding it to the collection as needed.  This method
 232 /// always succeeds (is unambiguous), and is always efficient.
 233 ///
 234 const SourceFileInfo &
 235 ProgramInfo::getSourceFile(const GlobalVariable *Desc) {
 236   SourceFileInfo *&Result = SourceFiles[Desc];
 237   if (Result) return *Result;
 238
 239   // Figure out what language this source file comes from...
 240   unsigned LangID = 0;   // Zero is unknown language
 241   if (Desc && Desc->hasInitializer())
 242     if (ConstantStruct *CS = dyn_cast<ConstantStruct>(Desc->getInitializer()))
 243       if (CS->getNumOperands() > 2)
 244         if (ConstantInt *CUI = dyn_cast<ConstantInt>(CS->getOperand(2)))
 245           LangID = CUI->getZExtValue();
 246
 247   const SourceLanguage &Lang = SourceLanguage::get(LangID);
 248   SourceFileInfo *New = Lang.createSourceFileInfo(Desc, *this);
 249
 250   // FIXME: this should check to see if there is already a Filename/WorkingDir
 251   // pair that matches this one.  If so, we shouldn't create the duplicate!
 252   //
 253   SourceFileIndex.insert(std::make_pair(New->getBaseName(), New));
 254   return *(Result = New);
 255 }
 256
 257
 258 /// getSourceFiles - Index all of the source files in the program and return
 259 /// a mapping of it.  This information is lazily computed the first time
 260 /// that it is requested.  Since this information can take a long time to
 261 /// compute, the user is given a chance to cancel it.  If this occurs, an
 262 /// exception is thrown.
 263 const std::map<const GlobalVariable*, SourceFileInfo*> &
 264 ProgramInfo::getSourceFiles(bool RequiresCompleteMap) {
 265   // If we have a fully populated map, or if the client doesn't need one, just
 266   // return what we have.
 267   if (SourceFilesIsComplete || !RequiresCompleteMap)
 268     return SourceFiles;
 269
 270   // Ok, all of the source file descriptors (compile_unit in dwarf terms),
 271   // should be on the use list of the llvm.dbg.translation_units global.
 272   //
 273   GlobalVariable *Units =
 274     M->getGlobalVariable("llvm.dbg.translation_units",
 275                          StructType::get(M->getContext()));
 276   if (Units == 0)
 277     throw "Program contains no debugging information!";
 278
 279   std::vector<GlobalVariable*> TranslationUnits;
 280   getGlobalVariablesUsing(Units, TranslationUnits);
 281
 282   SlowOperationInformer SOI("building source files index");
 283
 284   // Loop over all of the translation units found, building the SourceFiles
 285   // mapping.
 286   for (unsigned i = 0, e = TranslationUnits.size(); i != e; ++i) {
 287     getSourceFile(TranslationUnits[i]);
 288     if (SOI.progress(i+1, e))
 289       throw "While building source files index, operation cancelled.";
 290   }
 291
 292   // Ok, if we got this far, then we indexed the whole program.
 293   SourceFilesIsComplete = true;
 294   return SourceFiles;
 295 }
 296
 297 /// getSourceFile - Look up the file with the specified name.  If there is
 298 /// more than one match for the specified filename, prompt the user to pick
 299 /// one.  If there is no source file that matches the specified name, throw
 300 /// an exception indicating that we can't find the file.  Otherwise, return
 301 /// the file information for that file.
 302 const SourceFileInfo &ProgramInfo::getSourceFile(const std::string &Filename) {
 303   std::multimap<std::string, SourceFileInfo*>::const_iterator Start, End;
 304   getSourceFiles();
 305   tie(Start, End) = SourceFileIndex.equal_range(Filename);
 306
 307   if (Start == End) throw "Could not find source file '" + Filename + "'!";
 308   const SourceFileInfo &SFI = *Start->second;
 309   ++Start;
 310   if (Start == End) return SFI;
 311
 312   throw "FIXME: Multiple source files with the same name not implemented!";
 313 }
 314
 315
 316 //===----------------------------------------------------------------------===//
 317 // SourceFunctionInfo tracking...
 318 //
 319
 320
 321 /// getFunction - Return function information for the specified function
 322 /// descriptor object, adding it to the collection as needed.  This method
 323 /// always succeeds (is unambiguous), and is always efficient.
 324 ///
 325 const SourceFunctionInfo &
 326 ProgramInfo::getFunction(const GlobalVariable *Desc) {
 327   SourceFunctionInfo *&Result = SourceFunctions[Desc];
 328   if (Result) return *Result;
 329
 330   // Figure out what language this function comes from...
 331   const GlobalVariable *SourceFileDesc = 0;
 332   if (Desc && Desc->hasInitializer())
 333     if (ConstantStruct *CS = dyn_cast<ConstantStruct>(Desc->getInitializer()))
 334       if (CS->getNumOperands() > 0)
 335         if (const GlobalVariable *GV =
 336             dyn_cast<GlobalVariable>(CS->getOperand(1)))
 337           SourceFileDesc = GV;
 338
 339   const SourceLanguage &Lang = getSourceFile(SourceFileDesc).getLanguage();
 340   return *(Result = Lang.createSourceFunctionInfo(Desc, *this));
 341 }
 342
 343
 344 // getSourceFunctions - Index all of the functions in the program and return
 345 // them.  This information is lazily computed the first time that it is
 346 // requested.  Since this information can take a long time to compute, the user
 347 // is given a chance to cancel it.  If this occurs, an exception is thrown.
 348 const std::map<const GlobalVariable*, SourceFunctionInfo*> &
 349 ProgramInfo::getSourceFunctions(bool RequiresCompleteMap) {
 350   if (SourceFunctionsIsComplete || !RequiresCompleteMap)
 351     return SourceFunctions;
 352
 353   // Ok, all of the source function descriptors (subprogram in dwarf terms),
 354   // should be on the use list of the llvm.dbg.translation_units global.
 355   //
 356   GlobalVariable *Units =
 357     M->getGlobalVariable("llvm.dbg.globals", StructType::get(M->getContext()));
 358   if (Units == 0)
 359     throw "Program contains no debugging information!";
 360
 361   std::vector<GlobalVariable*> Functions;
 362   getGlobalVariablesUsing(Units, Functions);
 363
 364   SlowOperationInformer SOI("building functions index");
 365
 366   // Loop over all of the functions found, building the SourceFunctions mapping.
 367   for (unsigned i = 0, e = Functions.size(); i != e; ++i) {
 368     getFunction(Functions[i]);
 369     if (SOI.progress(i+1, e))
 370       throw "While functions index, operation cancelled.";
 371   }
 372
 373   // Ok, if we got this far, then we indexed the whole program.
 374   SourceFunctionsIsComplete = true;
 375   return SourceFunctions;
 376 }