lib/Debugger/ProgramInfo.cpp

   1 //===-- ProgramInfo.cpp - Compute and cache info about a program ----------===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file was developed by the LLVM research group and is distributed under
   6 // the University of Illinois Open Source 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/DerivedTypes.h"
  18 #include "llvm/Intrinsics.h"
  19 #include "llvm/IntrinsicInst.h"
  20 #include "llvm/Instructions.h"
  21 #include "llvm/Module.h"
  22 #include "llvm/Debugger/SourceFile.h"
  23 #include "llvm/Debugger/SourceLanguage.h"
  24 #include "llvm/Support/SlowOperationInformer.h"
  25 #include "llvm/ADT/STLExtras.h"
  26 #include <iostream>
  27
  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 (const Function *F = 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 (ConstantUInt *CUI = dyn_cast<ConstantUInt>(CS->getOperand(1)))
 118           Version = CUI->getValue();
 119
 120         BaseName  = CS->getOperand(3)->getStringValue();
 121         Directory = CS->getOperand(4)->getStringValue();
 122       }
 123 }
 124
 125 SourceFileInfo::~SourceFileInfo() {
 126   delete SourceText;
 127 }
 128
 129 SourceFile &SourceFileInfo::getSourceText() const {
 130   // FIXME: this should take into account the source search directories!
 131   if (SourceText == 0) { // Read the file in if we haven't already.
 132     sys::Path tmpPath;
 133     if (!Directory.empty())
 134       tmpPath.set(Directory);
 135     tmpPath.appendComponent(BaseName);
 136     if (tmpPath.canRead())
 137       SourceText = new SourceFile(tmpPath.toString(), Descriptor);
 138     else
 139       SourceText = new SourceFile(BaseName, Descriptor);
 140   }
 141   return *SourceText;
 142 }
 143
 144
 145 //===----------------------------------------------------------------------===//
 146 // SourceFunctionInfo implementation
 147 //
 148 SourceFunctionInfo::SourceFunctionInfo(ProgramInfo &PI,
 149                                        const GlobalVariable *Desc)
 150   : Descriptor(Desc) {
 151   LineNo = ColNo = 0;
 152   if (Desc && Desc->hasInitializer())
 153     if (ConstantStruct *CS = dyn_cast<ConstantStruct>(Desc->getInitializer()))
 154       if (CS->getNumOperands() > 2) {
 155         // Entry #1 is the file descriptor.
 156         if (const GlobalVariable *GV =
 157             dyn_cast<GlobalVariable>(CS->getOperand(1)))
 158           SourceFile = &PI.getSourceFile(GV);
 159
 160         // Entry #2 is the function name.
 161         Name = CS->getOperand(2)->getStringValue();
 162       }
 163 }
 164
 165 /// getSourceLocation - This method returns the location of the first stopping
 166 /// point in the function.
 167 void SourceFunctionInfo::getSourceLocation(unsigned &RetLineNo,
 168                                            unsigned &RetColNo) const {
 169   // If we haven't computed this yet...
 170   if (!LineNo) {
 171     // Look at all of the users of the function descriptor, looking for calls to
 172     // %llvm.dbg.func.start.
 173     for (Value::use_const_iterator UI = Descriptor->use_begin(),
 174            E = Descriptor->use_end(); UI != E; ++UI)
 175       if (const CallInst *CI = dyn_cast<CallInst>(*UI))
 176         if (const Function *F = CI->getCalledFunction())
 177           if (F->getIntrinsicID() == Intrinsic::dbg_func_start) {
 178             // We found the start of the function.  Check to see if there are
 179             // any stop points on the use-list of the function start.
 180             const GlobalVariable *SD = getNextStopPoint(CI, LineNo, ColNo);
 181             if (SD) {             // We found the first stop point!
 182               // This is just a sanity check.
 183               if (getSourceFile().getDescriptor() != SD)
 184                 std::cout << "WARNING: first line of function is not in the"
 185                   " file that the function descriptor claims it is in.\n";
 186               break;
 187             }
 188           }
 189   }
 190   RetLineNo = LineNo; RetColNo = ColNo;
 191 }
 192
 193 //===----------------------------------------------------------------------===//
 194 // ProgramInfo implementation
 195 //
 196
 197 ProgramInfo::ProgramInfo(Module *m) : M(m), ProgramTimeStamp(0,0) {
 198   assert(M && "Cannot create program information with a null module!");
 199   sys::Path modulePath(M->getModuleIdentifier());
 200   ProgramTimeStamp = modulePath.getTimestamp();
 201
 202   SourceFilesIsComplete = false;
 203   SourceFunctionsIsComplete = false;
 204 }
 205
 206 ProgramInfo::~ProgramInfo() {
 207   // Delete cached information about source program objects...
 208   for (std::map<const GlobalVariable*, SourceFileInfo*>::iterator
 209          I = SourceFiles.begin(), E = SourceFiles.end(); I != E; ++I)
 210     delete I->second;
 211   for (std::map<const GlobalVariable*, SourceFunctionInfo*>::iterator
 212          I = SourceFunctions.begin(), E = SourceFunctions.end(); I != E; ++I)
 213     delete I->second;
 214
 215   // Delete the source language caches.
 216   for (unsigned i = 0, e = LanguageCaches.size(); i != e; ++i)
 217     delete LanguageCaches[i].second;
 218 }
 219
 220
 221 //===----------------------------------------------------------------------===//
 222 // SourceFileInfo tracking...
 223 //
 224
 225 /// getSourceFile - Return source file information for the specified source file
 226 /// descriptor object, adding it to the collection as needed.  This method
 227 /// always succeeds (is unambiguous), and is always efficient.
 228 ///
 229 const SourceFileInfo &
 230 ProgramInfo::getSourceFile(const GlobalVariable *Desc) {
 231   SourceFileInfo *&Result = SourceFiles[Desc];
 232   if (Result) return *Result;
 233
 234   // Figure out what language this source file comes from...
 235   unsigned LangID = 0;   // Zero is unknown language
 236   if (Desc && Desc->hasInitializer())
 237     if (ConstantStruct *CS = dyn_cast<ConstantStruct>(Desc->getInitializer()))
 238       if (CS->getNumOperands() > 2)
 239         if (ConstantUInt *CUI = dyn_cast<ConstantUInt>(CS->getOperand(2)))
 240           LangID = CUI->getValue();
 241
 242   const SourceLanguage &Lang = SourceLanguage::get(LangID);
 243   SourceFileInfo *New = Lang.createSourceFileInfo(Desc, *this);
 244
 245   // FIXME: this should check to see if there is already a Filename/WorkingDir
 246   // pair that matches this one.  If so, we shouldn't create the duplicate!
 247   //
 248   SourceFileIndex.insert(std::make_pair(New->getBaseName(), New));
 249   return *(Result = New);
 250 }
 251
 252
 253 /// getSourceFiles - Index all of the source files in the program and return
 254 /// a mapping of it.  This information is lazily computed the first time
 255 /// that it is requested.  Since this information can take a long time to
 256 /// compute, the user is given a chance to cancel it.  If this occurs, an
 257 /// exception is thrown.
 258 const std::map<const GlobalVariable*, SourceFileInfo*> &
 259 ProgramInfo::getSourceFiles(bool RequiresCompleteMap) {
 260   // If we have a fully populated map, or if the client doesn't need one, just
 261   // return what we have.
 262   if (SourceFilesIsComplete || !RequiresCompleteMap)
 263     return SourceFiles;
 264
 265   // Ok, all of the source file descriptors (compile_unit in dwarf terms),
 266   // should be on the use list of the llvm.dbg.translation_units global.
 267   //
 268   GlobalVariable *Units =
 269     M->getGlobalVariable("llvm.dbg.translation_units",
 270                          StructType::get(std::vector<const Type*>()));
 271   if (Units == 0)
 272     throw "Program contains no debugging information!";
 273
 274   std::vector<GlobalVariable*> TranslationUnits;
 275   getGlobalVariablesUsing(Units, TranslationUnits);
 276
 277   SlowOperationInformer SOI("building source files index");
 278
 279   // Loop over all of the translation units found, building the SourceFiles
 280   // mapping.
 281   for (unsigned i = 0, e = TranslationUnits.size(); i != e; ++i) {
 282     getSourceFile(TranslationUnits[i]);
 283     SOI.progress(i+1, e);
 284   }
 285
 286   // Ok, if we got this far, then we indexed the whole program.
 287   SourceFilesIsComplete = true;
 288   return SourceFiles;
 289 }
 290
 291 /// getSourceFile - Look up the file with the specified name.  If there is
 292 /// more than one match for the specified filename, prompt the user to pick
 293 /// one.  If there is no source file that matches the specified name, throw
 294 /// an exception indicating that we can't find the file.  Otherwise, return
 295 /// the file information for that file.
 296 const SourceFileInfo &ProgramInfo::getSourceFile(const std::string &Filename) {
 297   std::multimap<std::string, SourceFileInfo*>::const_iterator Start, End;
 298   getSourceFiles();
 299   tie(Start, End) = SourceFileIndex.equal_range(Filename);
 300
 301   if (Start == End) throw "Could not find source file '" + Filename + "'!";
 302   const SourceFileInfo &SFI = *Start->second;
 303   ++Start;
 304   if (Start == End) return SFI;
 305
 306   throw "FIXME: Multiple source files with the same name not implemented!";
 307 }
 308
 309
 310 //===----------------------------------------------------------------------===//
 311 // SourceFunctionInfo tracking...
 312 //
 313
 314
 315 /// getFunction - Return function information for the specified function
 316 /// descriptor object, adding it to the collection as needed.  This method
 317 /// always succeeds (is unambiguous), and is always efficient.
 318 ///
 319 const SourceFunctionInfo &
 320 ProgramInfo::getFunction(const GlobalVariable *Desc) {
 321   SourceFunctionInfo *&Result = SourceFunctions[Desc];
 322   if (Result) return *Result;
 323
 324   // Figure out what language this function comes from...
 325   const GlobalVariable *SourceFileDesc = 0;
 326   if (Desc && Desc->hasInitializer())
 327     if (ConstantStruct *CS = dyn_cast<ConstantStruct>(Desc->getInitializer()))
 328       if (CS->getNumOperands() > 0)
 329         if (const GlobalVariable *GV =
 330             dyn_cast<GlobalVariable>(CS->getOperand(1)))
 331           SourceFileDesc = GV;
 332
 333   const SourceLanguage &Lang = getSourceFile(SourceFileDesc).getLanguage();
 334   return *(Result = Lang.createSourceFunctionInfo(Desc, *this));
 335 }
 336
 337
 338 // getSourceFunctions - Index all of the functions in the program and return
 339 // them.  This information is lazily computed the first time that it is
 340 // requested.  Since this information can take a long time to compute, the user
 341 // is given a chance to cancel it.  If this occurs, an exception is thrown.
 342 const std::map<const GlobalVariable*, SourceFunctionInfo*> &
 343 ProgramInfo::getSourceFunctions(bool RequiresCompleteMap) {
 344   if (SourceFunctionsIsComplete || !RequiresCompleteMap)
 345     return SourceFunctions;
 346
 347   // Ok, all of the source function descriptors (subprogram in dwarf terms),
 348   // should be on the use list of the llvm.dbg.translation_units global.
 349   //
 350   GlobalVariable *Units =
 351     M->getGlobalVariable("llvm.dbg.globals",
 352                          StructType::get(std::vector<const Type*>()));
 353   if (Units == 0)
 354     throw "Program contains no debugging information!";
 355
 356   std::vector<GlobalVariable*> Functions;
 357   getGlobalVariablesUsing(Units, Functions);
 358
 359   SlowOperationInformer SOI("building functions index");
 360
 361   // Loop over all of the functions found, building the SourceFunctions mapping.
 362   for (unsigned i = 0, e = Functions.size(); i != e; ++i) {
 363     getFunction(Functions[i]);
 364     SOI.progress(i+1, e);
 365   }
 366
 367   // Ok, if we got this far, then we indexed the whole program.
 368   SourceFunctionsIsComplete = true;
 369   return SourceFunctions;
 370 }