tools/lto/LTOModule.cpp

   1 //===-LTOModule.cpp - LLVM Link Time Optimizer ----------------------------===//
   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 Link Time Optimization library. This library is
  11 // intended to be used by linker to optimize code at link time.
  12 //
  13 //===----------------------------------------------------------------------===//
  14
  15 #include "LTOModule.h"
  16
  17 #include "llvm/Module.h"
  18 #include "llvm/ModuleProvider.h"
  19 #include "llvm/ADT/OwningPtr.h"
  20 #include "llvm/Bitcode/ReaderWriter.h"
  21 #include "llvm/Support/SystemUtils.h"
  22 #include "llvm/Support/Mangler.h"
  23 #include "llvm/Support/MemoryBuffer.h"
  24 #include "llvm/Support/MathExtras.h"
  25 #include "llvm/System/Path.h"
  26 #include "llvm/System/Process.h"
  27 #include "llvm/Target/SubtargetFeature.h"
  28 #include "llvm/Target/TargetMachine.h"
  29 #include "llvm/Target/TargetMachineRegistry.h"
  30 #include "llvm/Target/TargetAsmInfo.h"
  31
  32 #include <fstream>
  33
  34 using namespace llvm;
  35
  36 bool LTOModule::isBitcodeFile(const void* mem, size_t length)
  37 {
  38     return ( llvm::sys::IdentifyFileType((char*)mem, length)
  39                                             == llvm::sys::Bitcode_FileType );
  40 }
  41
  42 bool LTOModule::isBitcodeFile(const char* path)
  43 {
  44     return llvm::sys::Path(path).isBitcodeFile();
  45 }
  46
  47 bool LTOModule::isBitcodeFileForTarget(const void* mem, size_t length,
  48                                        const char* triplePrefix)
  49 {
  50     MemoryBuffer* buffer = makeBuffer(mem, length);
  51     if ( buffer == NULL )
  52         return false;
  53     return isTargetMatch(buffer, triplePrefix);
  54 }
  55
  56
  57 bool LTOModule::isBitcodeFileForTarget(const char* path,
  58                                        const char* triplePrefix)
  59 {
  60     MemoryBuffer *buffer = MemoryBuffer::getFile(path);
  61     if (buffer == NULL)
  62         return false;
  63     return isTargetMatch(buffer, triplePrefix);
  64 }
  65
  66 // takes ownership of buffer
  67 bool LTOModule::isTargetMatch(MemoryBuffer* buffer, const char* triplePrefix)
  68 {
  69     OwningPtr<ModuleProvider> mp(getBitcodeModuleProvider(buffer));
  70     // on success, mp owns buffer and both are deleted at end of this method
  71     if ( !mp ) {
  72         delete buffer;
  73         return false;
  74     }
  75     std::string actualTarget = mp->getModule()->getTargetTriple();
  76     return ( strncmp(actualTarget.c_str(), triplePrefix,
  77                     strlen(triplePrefix)) == 0);
  78 }
  79
  80
  81 LTOModule::LTOModule(Module* m, TargetMachine* t)
  82  : _module(m), _target(t), _symbolsParsed(false)
  83 {
  84 }
  85
  86 LTOModule* LTOModule::makeLTOModule(const char* path, std::string& errMsg)
  87 {
  88     OwningPtr<MemoryBuffer> buffer(MemoryBuffer::getFile(path, &errMsg));
  89     if ( !buffer )
  90         return NULL;
  91     return makeLTOModule(buffer.get(), errMsg);
  92 }
  93
  94 /// makeBuffer - create a MemoryBuffer from a memory range.
  95 /// MemoryBuffer requires the byte past end of the buffer to be a zero.
  96 /// We might get lucky and already be that way, otherwise make a copy.
  97 /// Also if next byte is on a different page, don't assume it is readable.
  98 MemoryBuffer* LTOModule::makeBuffer(const void* mem, size_t length)
  99 {
 100     const char* startPtr = (char*)mem;
 101     const char* endPtr = startPtr+length;
 102     if ( (((uintptr_t)endPtr & (sys::Process::GetPageSize()-1)) == 0)
 103         || (*endPtr != 0) )
 104         return MemoryBuffer::getMemBufferCopy(startPtr, endPtr);
 105     else
 106         return MemoryBuffer::getMemBuffer(startPtr, endPtr);
 107 }
 108
 109
 110 LTOModule* LTOModule::makeLTOModule(const void* mem, size_t length,
 111                                                         std::string& errMsg)
 112 {
 113     OwningPtr<MemoryBuffer> buffer(makeBuffer(mem, length));
 114     if ( !buffer )
 115         return NULL;
 116     return makeLTOModule(buffer.get(), errMsg);
 117 }
 118
 119 /// getFeatureString - Return a string listing the features associated with the
 120 /// target triple.
 121 ///
 122 /// FIXME: This is an inelegant way of specifying the features of a
 123 /// subtarget. It would be better if we could encode this information into the
 124 /// IR. See <rdar://5972456>.
 125 std::string getFeatureString(const char *TargetTriple) {
 126   SubtargetFeatures Features;
 127
 128   if (strncmp(TargetTriple, "powerpc-apple-", 14) == 0) {
 129     Features.AddFeature("altivec", true);
 130   } else if (strncmp(TargetTriple, "powerpc64-apple-", 16) == 0) {
 131     Features.AddFeature("64bit", true);
 132     Features.AddFeature("altivec", true);
 133   }
 134
 135   return Features.getString();
 136 }
 137
 138 LTOModule* LTOModule::makeLTOModule(MemoryBuffer* buffer, std::string& errMsg)
 139 {
 140     // parse bitcode buffer
 141     OwningPtr<Module> m(ParseBitcodeFile(buffer, &errMsg));
 142     if ( !m )
 143         return NULL;
 144     // find machine architecture for this module
 145     const TargetMachineRegistry::entry* march =
 146             TargetMachineRegistry::getClosestStaticTargetForModule(*m, errMsg);
 147
 148     if ( march == NULL )
 149         return NULL;
 150
 151     // construct LTModule, hand over ownership of module and target
 152     std::string FeatureStr = getFeatureString(m->getTargetTriple().c_str());
 153     TargetMachine* target = march->CtorFn(*m, FeatureStr);
 154     return new LTOModule(m.take(), target);
 155 }
 156
 157
 158 const char* LTOModule::getTargetTriple()
 159 {
 160     return _module->getTargetTriple().c_str();
 161 }
 162
 163 void LTOModule::addDefinedFunctionSymbol(Function* f, Mangler &mangler)
 164 {
 165     // add to list of defined symbols
 166     addDefinedSymbol(f, mangler, true);
 167
 168     // add external symbols referenced by this function.
 169     for (Function::iterator b = f->begin(); b != f->end(); ++b) {
 170         for (BasicBlock::iterator i = b->begin(); i != b->end(); ++i) {
 171             for (unsigned count = 0, total = i->getNumOperands();
 172                                         count != total; ++count) {
 173                 findExternalRefs(i->getOperand(count), mangler);
 174             }
 175         }
 176     }
 177 }
 178
 179 void LTOModule::addDefinedDataSymbol(GlobalValue* v, Mangler &mangler)
 180 {
 181     // add to list of defined symbols
 182     addDefinedSymbol(v, mangler, false);
 183
 184     // add external symbols referenced by this data.
 185     for (unsigned count = 0, total = v->getNumOperands();
 186                                                 count != total; ++count) {
 187         findExternalRefs(v->getOperand(count), mangler);
 188     }
 189 }
 190
 191
 192 void LTOModule::addDefinedSymbol(GlobalValue* def, Mangler &mangler,
 193                                 bool isFunction)
 194 {
 195     // string is owned by _defines
 196     const char* symbolName = ::strdup(mangler.getValueName(def).c_str());
 197
 198     // set alignment part log2() can have rounding errors
 199     uint32_t align = def->getAlignment();
 200     uint32_t attr = align ? CountTrailingZeros_32(def->getAlignment()) : 0;
 201
 202     // set permissions part
 203     if ( isFunction )
 204         attr |= LTO_SYMBOL_PERMISSIONS_CODE;
 205     else {
 206         GlobalVariable* gv = dyn_cast<GlobalVariable>(def);
 207         if ( (gv != NULL) && gv->isConstant() )
 208             attr |= LTO_SYMBOL_PERMISSIONS_RODATA;
 209         else
 210             attr |= LTO_SYMBOL_PERMISSIONS_DATA;
 211     }
 212
 213     // set definition part
 214     if ( def->hasWeakLinkage() || def->hasLinkOnceLinkage() ) {
 215         attr |= LTO_SYMBOL_DEFINITION_WEAK;
 216     }
 217     else if ( def->hasCommonLinkage()) {
 218         attr |= LTO_SYMBOL_DEFINITION_TENTATIVE;
 219     }
 220     else {
 221         attr |= LTO_SYMBOL_DEFINITION_REGULAR;
 222     }
 223
 224     // set scope part
 225     if ( def->hasHiddenVisibility() )
 226         attr |= LTO_SYMBOL_SCOPE_HIDDEN;
 227     else if ( def->hasExternalLinkage() || def->hasWeakLinkage()
 228               || def->hasLinkOnceLinkage() || def->hasCommonLinkage() )
 229         attr |= LTO_SYMBOL_SCOPE_DEFAULT;
 230     else
 231         attr |= LTO_SYMBOL_SCOPE_INTERNAL;
 232
 233     // add to table of symbols
 234     NameAndAttributes info;
 235     info.name = symbolName;
 236     info.attributes = (lto_symbol_attributes)attr;
 237     _symbols.push_back(info);
 238     _defines[info.name] = 1;
 239 }
 240
 241 void LTOModule::addAsmGlobalSymbol(const char *name) {
 242   // string is owned by _defines
 243   const char *symbolName = ::strdup(name);
 244   uint32_t attr = LTO_SYMBOL_DEFINITION_REGULAR;
 245   attr |= LTO_SYMBOL_SCOPE_DEFAULT;
 246
 247   // add to table of symbols
 248   NameAndAttributes info;
 249   info.name = symbolName;
 250   info.attributes = (lto_symbol_attributes)attr;
 251   _symbols.push_back(info);
 252   _defines[info.name] = 1;
 253 }
 254
 255 void LTOModule::addPotentialUndefinedSymbol(GlobalValue* decl, Mangler &mangler)
 256 {
 257    const char* name = mangler.getValueName(decl).c_str();
 258     // ignore all llvm.* symbols
 259     if ( strncmp(name, "llvm.", 5) != 0 ) {
 260         _undefines[name] = 1;
 261     }
 262 }
 263
 264
 265
 266 // Find exeternal symbols referenced by VALUE. This is a recursive function.
 267 void LTOModule::findExternalRefs(Value* value, Mangler &mangler) {
 268
 269     if (GlobalValue* gv = dyn_cast<GlobalValue>(value)) {
 270         if ( !gv->hasExternalLinkage() )
 271             addPotentialUndefinedSymbol(gv, mangler);
 272         // If this is a variable definition, do not recursively process
 273         // initializer.  It might contain a reference to this variable
 274         // and cause an infinite loop.  The initializer will be
 275         // processed in addDefinedDataSymbol().
 276         return;
 277     }
 278
 279     // GlobalValue, even with InternalLinkage type, may have operands with
 280     // ExternalLinkage type. Do not ignore these operands.
 281     if (Constant* c = dyn_cast<Constant>(value)) {
 282         // Handle ConstantExpr, ConstantStruct, ConstantArry etc..
 283         for (unsigned i = 0, e = c->getNumOperands(); i != e; ++i)
 284             findExternalRefs(c->getOperand(i), mangler);
 285     }
 286 }
 287
 288 void LTOModule::lazyParseSymbols()
 289 {
 290     if ( !_symbolsParsed ) {
 291         _symbolsParsed = true;
 292
 293         // Use mangler to add GlobalPrefix to names to match linker names.
 294         Mangler mangler(*_module, _target->getTargetAsmInfo()->getGlobalPrefix());
 295
 296         // add functions
 297         for (Module::iterator f = _module->begin(); f != _module->end(); ++f) {
 298             if ( f->isDeclaration() )
 299                 addPotentialUndefinedSymbol(f, mangler);
 300             else
 301                 addDefinedFunctionSymbol(f, mangler);
 302         }
 303
 304         // add data
 305         for (Module::global_iterator v = _module->global_begin(),
 306                                     e = _module->global_end(); v !=  e; ++v) {
 307             if ( v->isDeclaration() )
 308                 addPotentialUndefinedSymbol(v, mangler);
 309             else
 310                 addDefinedDataSymbol(v, mangler);
 311         }
 312
 313         // add asm globals
 314         const std::string &inlineAsm = _module->getModuleInlineAsm();
 315         const std::string glbl = ".globl";
 316         std::string asmSymbolName;
 317         std::string::size_type pos = inlineAsm.find(glbl, 0);
 318         while (pos != std::string::npos) {
 319           // eat .globl
 320           pos = pos + 6;
 321
 322           // skip white space between .globl and symbol name
 323           std::string::size_type pbegin = inlineAsm.find_first_not_of(' ', pos);
 324           if (pbegin == std::string::npos)
 325             break;
 326
 327           // find end-of-line
 328           std::string::size_type pend = inlineAsm.find_first_of('\n', pbegin);
 329           if (pend == std::string::npos)
 330             break;
 331
 332           asmSymbolName.assign(inlineAsm, pbegin, pend - pbegin);
 333           addAsmGlobalSymbol(asmSymbolName.c_str());
 334
 335           // search next .globl
 336           pos = inlineAsm.find(glbl, pend);
 337         }
 338
 339         // make symbols for all undefines
 340         for (StringSet::iterator it=_undefines.begin();
 341                                                 it != _undefines.end(); ++it) {
 342             // if this symbol also has a definition, then don't make an undefine
 343             // because it is a tentative definition
 344             if ( _defines.count(it->getKeyData(), it->getKeyData()+
 345                                                   it->getKeyLength()) == 0 ) {
 346                 NameAndAttributes info;
 347                 info.name = it->getKeyData();
 348                 info.attributes = LTO_SYMBOL_DEFINITION_UNDEFINED;
 349                 _symbols.push_back(info);
 350             }
 351         }
 352     }
 353 }
 354
 355
 356 uint32_t LTOModule::getSymbolCount()
 357 {
 358     lazyParseSymbols();
 359     return _symbols.size();
 360 }
 361
 362
 363 lto_symbol_attributes LTOModule::getSymbolAttributes(uint32_t index)
 364 {
 365     lazyParseSymbols();
 366     if ( index < _symbols.size() )
 367         return _symbols[index].attributes;
 368     else
 369         return lto_symbol_attributes(0);
 370 }
 371
 372 const char* LTOModule::getSymbolName(uint32_t index)
 373 {
 374     lazyParseSymbols();
 375     if ( index < _symbols.size() )
 376         return _symbols[index].name;
 377     else
 378         return NULL;
 379 }
 380