Home | History | Annotate | Download | only in NVPTX 
      1 //===-- NVPTXAsmPrinter.cpp - NVPTX LLVM assembly writer ------------------===//
      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 a printer that converts from our internal representation
     11 // of machine-dependent LLVM code to NVPTX assembly language.
     12 //
     13 //===----------------------------------------------------------------------===//
     14 
     15 #include "NVPTXAsmPrinter.h"
     16 #include "InstPrinter/NVPTXInstPrinter.h"
     17 #include "MCTargetDesc/NVPTXMCAsmInfo.h"
     18 #include "NVPTX.h"
     19 #include "NVPTXInstrInfo.h"
     20 #include "NVPTXMCExpr.h"
     21 #include "NVPTXMachineFunctionInfo.h"
     22 #include "NVPTXRegisterInfo.h"
     23 #include "NVPTXTargetMachine.h"
     24 #include "NVPTXUtilities.h"
     25 #include "cl_common_defines.h"
     26 #include "llvm/ADT/StringExtras.h"
     27 #include "llvm/Analysis/ConstantFolding.h"
     28 #include "llvm/CodeGen/Analysis.h"
     29 #include "llvm/CodeGen/MachineFrameInfo.h"
     30 #include "llvm/CodeGen/MachineLoopInfo.h"
     31 #include "llvm/CodeGen/MachineModuleInfo.h"
     32 #include "llvm/CodeGen/MachineRegisterInfo.h"
     33 #include "llvm/IR/DebugInfo.h"
     34 #include "llvm/IR/DerivedTypes.h"
     35 #include "llvm/IR/Function.h"
     36 #include "llvm/IR/GlobalVariable.h"
     37 #include "llvm/IR/Mangler.h"
     38 #include "llvm/IR/Module.h"
     39 #include "llvm/IR/Operator.h"
     40 #include "llvm/MC/MCInst.h"
     41 #include "llvm/MC/MCStreamer.h"
     42 #include "llvm/MC/MCSymbol.h"
     43 #include "llvm/Support/CommandLine.h"
     44 #include "llvm/Support/ErrorHandling.h"
     45 #include "llvm/Support/FormattedStream.h"
     46 #include "llvm/Support/Path.h"
     47 #include "llvm/Support/TargetRegistry.h"
     48 #include "llvm/Support/TimeValue.h"
     49 #include "llvm/Target/TargetLoweringObjectFile.h"
     50 #include "llvm/Transforms/Utils/UnrollLoop.h"
     51 #include <sstream>
     52 using namespace llvm;
     53 
     54 #define DEPOTNAME "__local_depot"
     55 
     56 static cl::opt<bool>
     57 EmitLineNumbers("nvptx-emit-line-numbers", cl::Hidden,
     58                 cl::desc("NVPTX Specific: Emit Line numbers even without -G"),
     59                 cl::init(true));
     60 
     61 static cl::opt<bool>
     62 InterleaveSrc("nvptx-emit-src", cl::ZeroOrMore, cl::Hidden,
     63               cl::desc("NVPTX Specific: Emit source line in ptx file"),
     64               cl::init(false));
     65 
     66 namespace {
     67 /// DiscoverDependentGlobals - Return a set of GlobalVariables on which \p V
     68 /// depends.
     69 void DiscoverDependentGlobals(const Value *V,
     70                               DenseSet<const GlobalVariable *> &Globals) {
     71   if (const GlobalVariable *GV = dyn_cast<GlobalVariable>(V))
     72     Globals.insert(GV);
     73   else {
     74     if (const User *U = dyn_cast<User>(V)) {
     75       for (unsigned i = 0, e = U->getNumOperands(); i != e; ++i) {
     76         DiscoverDependentGlobals(U->getOperand(i), Globals);
     77       }
     78     }
     79   }
     80 }
     81 
     82 /// VisitGlobalVariableForEmission - Add \p GV to the list of GlobalVariable
     83 /// instances to be emitted, but only after any dependents have been added
     84 /// first.
     85 void VisitGlobalVariableForEmission(
     86     const GlobalVariable *GV, SmallVectorImpl<const GlobalVariable *> &Order,
     87     DenseSet<const GlobalVariable *> &Visited,
     88     DenseSet<const GlobalVariable *> &Visiting) {
     89   // Have we already visited this one?
     90   if (Visited.count(GV))
     91     return;
     92 
     93   // Do we have a circular dependency?
     94   if (!Visiting.insert(GV).second)
     95     report_fatal_error("Circular dependency found in global variable set");
     96 
     97   // Make sure we visit all dependents first
     98   DenseSet<const GlobalVariable *> Others;
     99   for (unsigned i = 0, e = GV->getNumOperands(); i != e; ++i)
    100     DiscoverDependentGlobals(GV->getOperand(i), Others);
    101 
    102   for (DenseSet<const GlobalVariable *>::iterator I = Others.begin(),
    103                                                   E = Others.end();
    104        I != E; ++I)
    105     VisitGlobalVariableForEmission(*I, Order, Visited, Visiting);
    106 
    107   // Now we can visit ourself
    108   Order.push_back(GV);
    109   Visited.insert(GV);
    110   Visiting.erase(GV);
    111 }
    112 }
    113 
    114 void NVPTXAsmPrinter::emitLineNumberAsDotLoc(const MachineInstr &MI) {
    115   if (!EmitLineNumbers)
    116     return;
    117   if (ignoreLoc(MI))
    118     return;
    119 
    120   DebugLoc curLoc = MI.getDebugLoc();
    121 
    122   if (!prevDebugLoc && !curLoc)
    123     return;
    124 
    125   if (prevDebugLoc == curLoc)
    126     return;
    127 
    128   prevDebugLoc = curLoc;
    129 
    130   if (!curLoc)
    131     return;
    132 
    133   auto *Scope = cast_or_null<DIScope>(curLoc.getScope());
    134   if (!Scope)
    135      return;
    136 
    137   StringRef fileName(Scope->getFilename());
    138   StringRef dirName(Scope->getDirectory());
    139   SmallString<128> FullPathName = dirName;
    140   if (!dirName.empty() && !sys::path::is_absolute(fileName)) {
    141     sys::path::append(FullPathName, fileName);
    142     fileName = FullPathName;
    143   }
    144 
    145   if (filenameMap.find(fileName) == filenameMap.end())
    146     return;
    147 
    148   // Emit the line from the source file.
    149   if (InterleaveSrc)
    150     this->emitSrcInText(fileName, curLoc.getLine());
    151 
    152   std::stringstream temp;
    153   temp << "\t.loc " << filenameMap[fileName] << " " << curLoc.getLine()
    154        << " " << curLoc.getCol();
    155   OutStreamer->EmitRawText(temp.str());
    156 }
    157 
    158 void NVPTXAsmPrinter::EmitInstruction(const MachineInstr *MI) {
    159   SmallString<128> Str;
    160   raw_svector_ostream OS(Str);
    161   if (static_cast<NVPTXTargetMachine &>(TM).getDrvInterface() == NVPTX::CUDA)
    162     emitLineNumberAsDotLoc(*MI);
    163 
    164   MCInst Inst;
    165   lowerToMCInst(MI, Inst);
    166   EmitToStreamer(*OutStreamer, Inst);
    167 }
    168 
    169 // Handle symbol backtracking for targets that do not support image handles
    170 bool NVPTXAsmPrinter::lowerImageHandleOperand(const MachineInstr *MI,
    171                                            unsigned OpNo, MCOperand &MCOp) {
    172   const MachineOperand &MO = MI->getOperand(OpNo);
    173   const MCInstrDesc &MCID = MI->getDesc();
    174 
    175   if (MCID.TSFlags & NVPTXII::IsTexFlag) {
    176     // This is a texture fetch, so operand 4 is a texref and operand 5 is
    177     // a samplerref
    178     if (OpNo == 4 && MO.isImm()) {
    179       lowerImageHandleSymbol(MO.getImm(), MCOp);
    180       return true;
    181     }
    182     if (OpNo == 5 && MO.isImm() && !(MCID.TSFlags & NVPTXII::IsTexModeUnifiedFlag)) {
    183       lowerImageHandleSymbol(MO.getImm(), MCOp);
    184       return true;
    185     }
    186 
    187     return false;
    188   } else if (MCID.TSFlags & NVPTXII::IsSuldMask) {
    189     unsigned VecSize =
    190       1 << (((MCID.TSFlags & NVPTXII::IsSuldMask) >> NVPTXII::IsSuldShift) - 1);
    191 
    192     // For a surface load of vector size N, the Nth operand will be the surfref
    193     if (OpNo == VecSize && MO.isImm()) {
    194       lowerImageHandleSymbol(MO.getImm(), MCOp);
    195       return true;
    196     }
    197 
    198     return false;
    199   } else if (MCID.TSFlags & NVPTXII::IsSustFlag) {
    200     // This is a surface store, so operand 0 is a surfref
    201     if (OpNo == 0 && MO.isImm()) {
    202       lowerImageHandleSymbol(MO.getImm(), MCOp);
    203       return true;
    204     }
    205 
    206     return false;
    207   } else if (MCID.TSFlags & NVPTXII::IsSurfTexQueryFlag) {
    208     // This is a query, so operand 1 is a surfref/texref
    209     if (OpNo == 1 && MO.isImm()) {
    210       lowerImageHandleSymbol(MO.getImm(), MCOp);
    211       return true;
    212     }
    213 
    214     return false;
    215   }
    216 
    217   return false;
    218 }
    219 
    220 void NVPTXAsmPrinter::lowerImageHandleSymbol(unsigned Index, MCOperand &MCOp) {
    221   // Ewwww
    222   TargetMachine &TM = const_cast<TargetMachine&>(MF->getTarget());
    223   NVPTXTargetMachine &nvTM = static_cast<NVPTXTargetMachine&>(TM);
    224   const NVPTXMachineFunctionInfo *MFI = MF->getInfo<NVPTXMachineFunctionInfo>();
    225   const char *Sym = MFI->getImageHandleSymbol(Index);
    226   std::string *SymNamePtr =
    227     nvTM.getManagedStrPool()->getManagedString(Sym);
    228   MCOp = GetSymbolRef(OutContext.getOrCreateSymbol(
    229     StringRef(SymNamePtr->c_str())));
    230 }
    231 
    232 void NVPTXAsmPrinter::lowerToMCInst(const MachineInstr *MI, MCInst &OutMI) {
    233   OutMI.setOpcode(MI->getOpcode());
    234   // Special: Do not mangle symbol operand of CALL_PROTOTYPE
    235   if (MI->getOpcode() == NVPTX::CALL_PROTOTYPE) {
    236     const MachineOperand &MO = MI->getOperand(0);
    237     OutMI.addOperand(GetSymbolRef(
    238       OutContext.getOrCreateSymbol(Twine(MO.getSymbolName()))));
    239     return;
    240   }
    241 
    242   for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
    243     const MachineOperand &MO = MI->getOperand(i);
    244 
    245     MCOperand MCOp;
    246     if (!nvptxSubtarget->hasImageHandles()) {
    247       if (lowerImageHandleOperand(MI, i, MCOp)) {
    248         OutMI.addOperand(MCOp);
    249         continue;
    250       }
    251     }
    252 
    253     if (lowerOperand(MO, MCOp))
    254       OutMI.addOperand(MCOp);
    255   }
    256 }
    257 
    258 bool NVPTXAsmPrinter::lowerOperand(const MachineOperand &MO,
    259                                    MCOperand &MCOp) {
    260   switch (MO.getType()) {
    261   default: llvm_unreachable("unknown operand type");
    262   case MachineOperand::MO_Register:
    263     MCOp = MCOperand::createReg(encodeVirtualRegister(MO.getReg()));
    264     break;
    265   case MachineOperand::MO_Immediate:
    266     MCOp = MCOperand::createImm(MO.getImm());
    267     break;
    268   case MachineOperand::MO_MachineBasicBlock:
    269     MCOp = MCOperand::createExpr(MCSymbolRefExpr::create(
    270         MO.getMBB()->getSymbol(), OutContext));
    271     break;
    272   case MachineOperand::MO_ExternalSymbol:
    273     MCOp = GetSymbolRef(GetExternalSymbolSymbol(MO.getSymbolName()));
    274     break;
    275   case MachineOperand::MO_GlobalAddress:
    276     MCOp = GetSymbolRef(getSymbol(MO.getGlobal()));
    277     break;
    278   case MachineOperand::MO_FPImmediate: {
    279     const ConstantFP *Cnt = MO.getFPImm();
    280     APFloat Val = Cnt->getValueAPF();
    281 
    282     switch (Cnt->getType()->getTypeID()) {
    283     default: report_fatal_error("Unsupported FP type"); break;
    284     case Type::FloatTyID:
    285       MCOp = MCOperand::createExpr(
    286         NVPTXFloatMCExpr::createConstantFPSingle(Val, OutContext));
    287       break;
    288     case Type::DoubleTyID:
    289       MCOp = MCOperand::createExpr(
    290         NVPTXFloatMCExpr::createConstantFPDouble(Val, OutContext));
    291       break;
    292     }
    293     break;
    294   }
    295   }
    296   return true;
    297 }
    298 
    299 unsigned NVPTXAsmPrinter::encodeVirtualRegister(unsigned Reg) {
    300   if (TargetRegisterInfo::isVirtualRegister(Reg)) {
    301     const TargetRegisterClass *RC = MRI->getRegClass(Reg);
    302 
    303     DenseMap<unsigned, unsigned> &RegMap = VRegMapping[RC];
    304     unsigned RegNum = RegMap[Reg];
    305 
    306     // Encode the register class in the upper 4 bits
    307     // Must be kept in sync with NVPTXInstPrinter::printRegName
    308     unsigned Ret = 0;
    309     if (RC == &NVPTX::Int1RegsRegClass) {
    310       Ret = (1 << 28);
    311     } else if (RC == &NVPTX::Int16RegsRegClass) {
    312       Ret = (2 << 28);
    313     } else if (RC == &NVPTX::Int32RegsRegClass) {
    314       Ret = (3 << 28);
    315     } else if (RC == &NVPTX::Int64RegsRegClass) {
    316       Ret = (4 << 28);
    317     } else if (RC == &NVPTX::Float32RegsRegClass) {
    318       Ret = (5 << 28);
    319     } else if (RC == &NVPTX::Float64RegsRegClass) {
    320       Ret = (6 << 28);
    321     } else {
    322       report_fatal_error("Bad register class");
    323     }
    324 
    325     // Insert the vreg number
    326     Ret |= (RegNum & 0x0FFFFFFF);
    327     return Ret;
    328   } else {
    329     // Some special-use registers are actually physical registers.
    330     // Encode this as the register class ID of 0 and the real register ID.
    331     return Reg & 0x0FFFFFFF;
    332   }
    333 }
    334 
    335 MCOperand NVPTXAsmPrinter::GetSymbolRef(const MCSymbol *Symbol) {
    336   const MCExpr *Expr;
    337   Expr = MCSymbolRefExpr::create(Symbol, MCSymbolRefExpr::VK_None,
    338                                  OutContext);
    339   return MCOperand::createExpr(Expr);
    340 }
    341 
    342 void NVPTXAsmPrinter::printReturnValStr(const Function *F, raw_ostream &O) {
    343   const DataLayout &DL = getDataLayout();
    344   const TargetLowering *TLI = nvptxSubtarget->getTargetLowering();
    345 
    346   Type *Ty = F->getReturnType();
    347 
    348   bool isABI = (nvptxSubtarget->getSmVersion() >= 20);
    349 
    350   if (Ty->getTypeID() == Type::VoidTyID)
    351     return;
    352 
    353   O << " (";
    354 
    355   if (isABI) {
    356     if (Ty->isFloatingPointTy() || Ty->isIntegerTy()) {
    357       unsigned size = 0;
    358       if (auto *ITy = dyn_cast<IntegerType>(Ty)) {
    359         size = ITy->getBitWidth();
    360         if (size < 32)
    361           size = 32;
    362       } else {
    363         assert(Ty->isFloatingPointTy() && "Floating point type expected here");
    364         size = Ty->getPrimitiveSizeInBits();
    365       }
    366 
    367       O << ".param .b" << size << " func_retval0";
    368     } else if (isa<PointerType>(Ty)) {
    369       O << ".param .b" << TLI->getPointerTy(DL).getSizeInBits()
    370         << " func_retval0";
    371     } else if ((Ty->getTypeID() == Type::StructTyID) || isa<VectorType>(Ty)) {
    372       unsigned totalsz = DL.getTypeAllocSize(Ty);
    373        unsigned retAlignment = 0;
    374        if (!llvm::getAlign(*F, 0, retAlignment))
    375          retAlignment = DL.getABITypeAlignment(Ty);
    376        O << ".param .align " << retAlignment << " .b8 func_retval0[" << totalsz
    377          << "]";
    378     } else
    379       llvm_unreachable("Unknown return type");
    380   } else {
    381     SmallVector<EVT, 16> vtparts;
    382     ComputeValueVTs(*TLI, DL, Ty, vtparts);
    383     unsigned idx = 0;
    384     for (unsigned i = 0, e = vtparts.size(); i != e; ++i) {
    385       unsigned elems = 1;
    386       EVT elemtype = vtparts[i];
    387       if (vtparts[i].isVector()) {
    388         elems = vtparts[i].getVectorNumElements();
    389         elemtype = vtparts[i].getVectorElementType();
    390       }
    391 
    392       for (unsigned j = 0, je = elems; j != je; ++j) {
    393         unsigned sz = elemtype.getSizeInBits();
    394         if (elemtype.isInteger() && (sz < 32))
    395           sz = 32;
    396         O << ".reg .b" << sz << " func_retval" << idx;
    397         if (j < je - 1)
    398           O << ", ";
    399         ++idx;
    400       }
    401       if (i < e - 1)
    402         O << ", ";
    403     }
    404   }
    405   O << ") ";
    406   return;
    407 }
    408 
    409 void NVPTXAsmPrinter::printReturnValStr(const MachineFunction &MF,
    410                                         raw_ostream &O) {
    411   const Function *F = MF.getFunction();
    412   printReturnValStr(F, O);
    413 }
    414 
    415 // Return true if MBB is the header of a loop marked with
    416 // llvm.loop.unroll.disable.
    417 // TODO: consider "#pragma unroll 1" which is equivalent to "#pragma nounroll".
    418 bool NVPTXAsmPrinter::isLoopHeaderOfNoUnroll(
    419     const MachineBasicBlock &MBB) const {
    420   MachineLoopInfo &LI = getAnalysis<MachineLoopInfo>();
    421   // We insert .pragma "nounroll" only to the loop header.
    422   if (!LI.isLoopHeader(&MBB))
    423     return false;
    424 
    425   // llvm.loop.unroll.disable is marked on the back edges of a loop. Therefore,
    426   // we iterate through each back edge of the loop with header MBB, and check
    427   // whether its metadata contains llvm.loop.unroll.disable.
    428   for (auto I = MBB.pred_begin(); I != MBB.pred_end(); ++I) {
    429     const MachineBasicBlock *PMBB = *I;
    430     if (LI.getLoopFor(PMBB) != LI.getLoopFor(&MBB)) {
    431       // Edges from other loops to MBB are not back edges.
    432       continue;
    433     }
    434     if (const BasicBlock *PBB = PMBB->getBasicBlock()) {
    435       if (MDNode *LoopID = PBB->getTerminator()->getMetadata("llvm.loop")) {
    436         if (GetUnrollMetadata(LoopID, "llvm.loop.unroll.disable"))
    437           return true;
    438       }
    439     }
    440   }
    441   return false;
    442 }
    443 
    444 void NVPTXAsmPrinter::EmitBasicBlockStart(const MachineBasicBlock &MBB) const {
    445   AsmPrinter::EmitBasicBlockStart(MBB);
    446   if (isLoopHeaderOfNoUnroll(MBB))
    447     OutStreamer->EmitRawText(StringRef("\t.pragma \"nounroll\";\n"));
    448 }
    449 
    450 void NVPTXAsmPrinter::EmitFunctionEntryLabel() {
    451   SmallString<128> Str;
    452   raw_svector_ostream O(Str);
    453 
    454   if (!GlobalsEmitted) {
    455     emitGlobals(*MF->getFunction()->getParent());
    456     GlobalsEmitted = true;
    457   }
    458 
    459   // Set up
    460   MRI = &MF->getRegInfo();
    461   F = MF->getFunction();
    462   emitLinkageDirective(F, O);
    463   if (llvm::isKernelFunction(*F))
    464     O << ".entry ";
    465   else {
    466     O << ".func ";
    467     printReturnValStr(*MF, O);
    468   }
    469 
    470   CurrentFnSym->print(O, MAI);
    471 
    472   emitFunctionParamList(*MF, O);
    473 
    474   if (llvm::isKernelFunction(*F))
    475     emitKernelFunctionDirectives(*F, O);
    476 
    477   OutStreamer->EmitRawText(O.str());
    478 
    479   prevDebugLoc = DebugLoc();
    480 }
    481 
    482 void NVPTXAsmPrinter::EmitFunctionBodyStart() {
    483   VRegMapping.clear();
    484   OutStreamer->EmitRawText(StringRef("{\n"));
    485   setAndEmitFunctionVirtualRegisters(*MF);
    486 
    487   SmallString<128> Str;
    488   raw_svector_ostream O(Str);
    489   emitDemotedVars(MF->getFunction(), O);
    490   OutStreamer->EmitRawText(O.str());
    491 }
    492 
    493 void NVPTXAsmPrinter::EmitFunctionBodyEnd() {
    494   OutStreamer->EmitRawText(StringRef("}\n"));
    495   VRegMapping.clear();
    496 }
    497 
    498 void NVPTXAsmPrinter::emitImplicitDef(const MachineInstr *MI) const {
    499   unsigned RegNo = MI->getOperand(0).getReg();
    500   if (TargetRegisterInfo::isVirtualRegister(RegNo)) {
    501     OutStreamer->AddComment(Twine("implicit-def: ") +
    502                             getVirtualRegisterName(RegNo));
    503   } else {
    504     OutStreamer->AddComment(Twine("implicit-def: ") +
    505                             nvptxSubtarget->getRegisterInfo()->getName(RegNo));
    506   }
    507   OutStreamer->AddBlankLine();
    508 }
    509 
    510 void NVPTXAsmPrinter::emitKernelFunctionDirectives(const Function &F,
    511                                                    raw_ostream &O) const {
    512   // If the NVVM IR has some of reqntid* specified, then output
    513   // the reqntid directive, and set the unspecified ones to 1.
    514   // If none of reqntid* is specified, don't output reqntid directive.
    515   unsigned reqntidx, reqntidy, reqntidz;
    516   bool specified = false;
    517   if (!llvm::getReqNTIDx(F, reqntidx))
    518     reqntidx = 1;
    519   else
    520     specified = true;
    521   if (!llvm::getReqNTIDy(F, reqntidy))
    522     reqntidy = 1;
    523   else
    524     specified = true;
    525   if (!llvm::getReqNTIDz(F, reqntidz))
    526     reqntidz = 1;
    527   else
    528     specified = true;
    529 
    530   if (specified)
    531     O << ".reqntid " << reqntidx << ", " << reqntidy << ", " << reqntidz
    532       << "\n";
    533 
    534   // If the NVVM IR has some of maxntid* specified, then output
    535   // the maxntid directive, and set the unspecified ones to 1.
    536   // If none of maxntid* is specified, don't output maxntid directive.
    537   unsigned maxntidx, maxntidy, maxntidz;
    538   specified = false;
    539   if (!llvm::getMaxNTIDx(F, maxntidx))
    540     maxntidx = 1;
    541   else
    542     specified = true;
    543   if (!llvm::getMaxNTIDy(F, maxntidy))
    544     maxntidy = 1;
    545   else
    546     specified = true;
    547   if (!llvm::getMaxNTIDz(F, maxntidz))
    548     maxntidz = 1;
    549   else
    550     specified = true;
    551 
    552   if (specified)
    553     O << ".maxntid " << maxntidx << ", " << maxntidy << ", " << maxntidz
    554       << "\n";
    555 
    556   unsigned mincta;
    557   if (llvm::getMinCTASm(F, mincta))
    558     O << ".minnctapersm " << mincta << "\n";
    559 }
    560 
    561 std::string
    562 NVPTXAsmPrinter::getVirtualRegisterName(unsigned Reg) const {
    563   const TargetRegisterClass *RC = MRI->getRegClass(Reg);
    564 
    565   std::string Name;
    566   raw_string_ostream NameStr(Name);
    567 
    568   VRegRCMap::const_iterator I = VRegMapping.find(RC);
    569   assert(I != VRegMapping.end() && "Bad register class");
    570   const DenseMap<unsigned, unsigned> &RegMap = I->second;
    571 
    572   VRegMap::const_iterator VI = RegMap.find(Reg);
    573   assert(VI != RegMap.end() && "Bad virtual register");
    574   unsigned MappedVR = VI->second;
    575 
    576   NameStr << getNVPTXRegClassStr(RC) << MappedVR;
    577 
    578   NameStr.flush();
    579   return Name;
    580 }
    581 
    582 void NVPTXAsmPrinter::emitVirtualRegister(unsigned int vr,
    583                                           raw_ostream &O) {
    584   O << getVirtualRegisterName(vr);
    585 }
    586 
    587 void NVPTXAsmPrinter::printVecModifiedImmediate(
    588     const MachineOperand &MO, const char *Modifier, raw_ostream &O) {
    589   static const char vecelem[] = { '0', '1', '2', '3', '0', '1', '2', '3' };
    590   int Imm = (int) MO.getImm();
    591   if (0 == strcmp(Modifier, "vecelem"))
    592     O << "_" << vecelem[Imm];
    593   else if (0 == strcmp(Modifier, "vecv4comm1")) {
    594     if ((Imm < 0) || (Imm > 3))
    595       O << "//";
    596   } else if (0 == strcmp(Modifier, "vecv4comm2")) {
    597     if ((Imm < 4) || (Imm > 7))
    598       O << "//";
    599   } else if (0 == strcmp(Modifier, "vecv4pos")) {
    600     if (Imm < 0)
    601       Imm = 0;
    602     O << "_" << vecelem[Imm % 4];
    603   } else if (0 == strcmp(Modifier, "vecv2comm1")) {
    604     if ((Imm < 0) || (Imm > 1))
    605       O << "//";
    606   } else if (0 == strcmp(Modifier, "vecv2comm2")) {
    607     if ((Imm < 2) || (Imm > 3))
    608       O << "//";
    609   } else if (0 == strcmp(Modifier, "vecv2pos")) {
    610     if (Imm < 0)
    611       Imm = 0;
    612     O << "_" << vecelem[Imm % 2];
    613   } else
    614     llvm_unreachable("Unknown Modifier on immediate operand");
    615 }
    616 
    617 
    618 
    619 void NVPTXAsmPrinter::emitDeclaration(const Function *F, raw_ostream &O) {
    620 
    621   emitLinkageDirective(F, O);
    622   if (llvm::isKernelFunction(*F))
    623     O << ".entry ";
    624   else
    625     O << ".func ";
    626   printReturnValStr(F, O);
    627   getSymbol(F)->print(O, MAI);
    628   O << "\n";
    629   emitFunctionParamList(F, O);
    630   O << ";\n";
    631 }
    632 
    633 static bool usedInGlobalVarDef(const Constant *C) {
    634   if (!C)
    635     return false;
    636 
    637   if (const GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
    638     return GV->getName() != "llvm.used";
    639   }
    640 
    641   for (const User *U : C->users())
    642     if (const Constant *C = dyn_cast<Constant>(U))
    643       if (usedInGlobalVarDef(C))
    644         return true;
    645 
    646   return false;
    647 }
    648 
    649 static bool usedInOneFunc(const User *U, Function const *&oneFunc) {
    650   if (const GlobalVariable *othergv = dyn_cast<GlobalVariable>(U)) {
    651     if (othergv->getName() == "llvm.used")
    652       return true;
    653   }
    654 
    655   if (const Instruction *instr = dyn_cast<Instruction>(U)) {
    656     if (instr->getParent() && instr->getParent()->getParent()) {
    657       const Function *curFunc = instr->getParent()->getParent();
    658       if (oneFunc && (curFunc != oneFunc))
    659         return false;
    660       oneFunc = curFunc;
    661       return true;
    662     } else
    663       return false;
    664   }
    665 
    666   for (const User *UU : U->users())
    667     if (!usedInOneFunc(UU, oneFunc))
    668       return false;
    669 
    670   return true;
    671 }
    672 
    673 /* Find out if a global variable can be demoted to local scope.
    674  * Currently, this is valid for CUDA shared variables, which have local
    675  * scope and global lifetime. So the conditions to check are :
    676  * 1. Is the global variable in shared address space?
    677  * 2. Does it have internal linkage?
    678  * 3. Is the global variable referenced only in one function?
    679  */
    680 static bool canDemoteGlobalVar(const GlobalVariable *gv, Function const *&f) {
    681   if (!gv->hasInternalLinkage())
    682     return false;
    683   PointerType *Pty = gv->getType();
    684   if (Pty->getAddressSpace() != llvm::ADDRESS_SPACE_SHARED)
    685     return false;
    686 
    687   const Function *oneFunc = nullptr;
    688 
    689   bool flag = usedInOneFunc(gv, oneFunc);
    690   if (!flag)
    691     return false;
    692   if (!oneFunc)
    693     return false;
    694   f = oneFunc;
    695   return true;
    696 }
    697 
    698 static bool useFuncSeen(const Constant *C,
    699                         llvm::DenseMap<const Function *, bool> &seenMap) {
    700   for (const User *U : C->users()) {
    701     if (const Constant *cu = dyn_cast<Constant>(U)) {
    702       if (useFuncSeen(cu, seenMap))
    703         return true;
    704     } else if (const Instruction *I = dyn_cast<Instruction>(U)) {
    705       const BasicBlock *bb = I->getParent();
    706       if (!bb)
    707         continue;
    708       const Function *caller = bb->getParent();
    709       if (!caller)
    710         continue;
    711       if (seenMap.find(caller) != seenMap.end())
    712         return true;
    713     }
    714   }
    715   return false;
    716 }
    717 
    718 void NVPTXAsmPrinter::emitDeclarations(const Module &M, raw_ostream &O) {
    719   llvm::DenseMap<const Function *, bool> seenMap;
    720   for (Module::const_iterator FI = M.begin(), FE = M.end(); FI != FE; ++FI) {
    721     const Function *F = &*FI;
    722 
    723     if (F->isDeclaration()) {
    724       if (F->use_empty())
    725         continue;
    726       if (F->getIntrinsicID())
    727         continue;
    728       emitDeclaration(F, O);
    729       continue;
    730     }
    731     for (const User *U : F->users()) {
    732       if (const Constant *C = dyn_cast<Constant>(U)) {
    733         if (usedInGlobalVarDef(C)) {
    734           // The use is in the initialization of a global variable
    735           // that is a function pointer, so print a declaration
    736           // for the original function
    737           emitDeclaration(F, O);
    738           break;
    739         }
    740         // Emit a declaration of this function if the function that
    741         // uses this constant expr has already been seen.
    742         if (useFuncSeen(C, seenMap)) {
    743           emitDeclaration(F, O);
    744           break;
    745         }
    746       }
    747 
    748       if (!isa<Instruction>(U))
    749         continue;
    750       const Instruction *instr = cast<Instruction>(U);
    751       const BasicBlock *bb = instr->getParent();
    752       if (!bb)
    753         continue;
    754       const Function *caller = bb->getParent();
    755       if (!caller)
    756         continue;
    757 
    758       // If a caller has already been seen, then the caller is
    759       // appearing in the module before the callee. so print out
    760       // a declaration for the callee.
    761       if (seenMap.find(caller) != seenMap.end()) {
    762         emitDeclaration(F, O);
    763         break;
    764       }
    765     }
    766     seenMap[F] = true;
    767   }
    768 }
    769 
    770 void NVPTXAsmPrinter::recordAndEmitFilenames(Module &M) {
    771   DebugInfoFinder DbgFinder;
    772   DbgFinder.processModule(M);
    773 
    774   unsigned i = 1;
    775   for (const DICompileUnit *DIUnit : DbgFinder.compile_units()) {
    776     StringRef Filename = DIUnit->getFilename();
    777     StringRef Dirname = DIUnit->getDirectory();
    778     SmallString<128> FullPathName = Dirname;
    779     if (!Dirname.empty() && !sys::path::is_absolute(Filename)) {
    780       sys::path::append(FullPathName, Filename);
    781       Filename = FullPathName;
    782     }
    783     if (filenameMap.find(Filename) != filenameMap.end())
    784       continue;
    785     filenameMap[Filename] = i;
    786     OutStreamer->EmitDwarfFileDirective(i, "", Filename);
    787     ++i;
    788   }
    789 
    790   for (DISubprogram *SP : DbgFinder.subprograms()) {
    791     StringRef Filename = SP->getFilename();
    792     StringRef Dirname = SP->getDirectory();
    793     SmallString<128> FullPathName = Dirname;
    794     if (!Dirname.empty() && !sys::path::is_absolute(Filename)) {
    795       sys::path::append(FullPathName, Filename);
    796       Filename = FullPathName;
    797     }
    798     if (filenameMap.find(Filename) != filenameMap.end())
    799       continue;
    800     filenameMap[Filename] = i;
    801     ++i;
    802   }
    803 }
    804 
    805 bool NVPTXAsmPrinter::doInitialization(Module &M) {
    806   // Construct a default subtarget off of the TargetMachine defaults. The
    807   // rest of NVPTX isn't friendly to change subtargets per function and
    808   // so the default TargetMachine will have all of the options.
    809   const Triple &TT = TM.getTargetTriple();
    810   StringRef CPU = TM.getTargetCPU();
    811   StringRef FS = TM.getTargetFeatureString();
    812   const NVPTXTargetMachine &NTM = static_cast<const NVPTXTargetMachine &>(TM);
    813   const NVPTXSubtarget STI(TT, CPU, FS, NTM);
    814 
    815   SmallString<128> Str1;
    816   raw_svector_ostream OS1(Str1);
    817 
    818   MMI = getAnalysisIfAvailable<MachineModuleInfo>();
    819 
    820   // We need to call the parent's one explicitly.
    821   //bool Result = AsmPrinter::doInitialization(M);
    822 
    823   // Initialize TargetLoweringObjectFile.
    824   const_cast<TargetLoweringObjectFile &>(getObjFileLowering())
    825       .Initialize(OutContext, TM);
    826 
    827   Mang = new Mangler();
    828 
    829   // Emit header before any dwarf directives are emitted below.
    830   emitHeader(M, OS1, STI);
    831   OutStreamer->EmitRawText(OS1.str());
    832 
    833   // Already commented out
    834   //bool Result = AsmPrinter::doInitialization(M);
    835 
    836   // Emit module-level inline asm if it exists.
    837   if (!M.getModuleInlineAsm().empty()) {
    838     OutStreamer->AddComment("Start of file scope inline assembly");
    839     OutStreamer->AddBlankLine();
    840     OutStreamer->EmitRawText(StringRef(M.getModuleInlineAsm()));
    841     OutStreamer->AddBlankLine();
    842     OutStreamer->AddComment("End of file scope inline assembly");
    843     OutStreamer->AddBlankLine();
    844   }
    845 
    846   // If we're not NVCL we're CUDA, go ahead and emit filenames.
    847   if (TM.getTargetTriple().getOS() != Triple::NVCL)
    848     recordAndEmitFilenames(M);
    849 
    850   GlobalsEmitted = false;
    851 
    852   return false; // success
    853 }
    854 
    855 void NVPTXAsmPrinter::emitGlobals(const Module &M) {
    856   SmallString<128> Str2;
    857   raw_svector_ostream OS2(Str2);
    858 
    859   emitDeclarations(M, OS2);
    860 
    861   // As ptxas does not support forward references of globals, we need to first
    862   // sort the list of module-level globals in def-use order. We visit each
    863   // global variable in order, and ensure that we emit it *after* its dependent
    864   // globals. We use a little extra memory maintaining both a set and a list to
    865   // have fast searches while maintaining a strict ordering.
    866   SmallVector<const GlobalVariable *, 8> Globals;
    867   DenseSet<const GlobalVariable *> GVVisited;
    868   DenseSet<const GlobalVariable *> GVVisiting;
    869 
    870   // Visit each global variable, in order
    871   for (const GlobalVariable &I : M.globals())
    872     VisitGlobalVariableForEmission(&I, Globals, GVVisited, GVVisiting);
    873 
    874   assert(GVVisited.size() == M.getGlobalList().size() &&
    875          "Missed a global variable");
    876   assert(GVVisiting.size() == 0 && "Did not fully process a global variable");
    877 
    878   // Print out module-level global variables in proper order
    879   for (unsigned i = 0, e = Globals.size(); i != e; ++i)
    880     printModuleLevelGV(Globals[i], OS2);
    881 
    882   OS2 << '\n';
    883 
    884   OutStreamer->EmitRawText(OS2.str());
    885 }
    886 
    887 void NVPTXAsmPrinter::emitHeader(Module &M, raw_ostream &O,
    888                                  const NVPTXSubtarget &STI) {
    889   O << "//\n";
    890   O << "// Generated by LLVM NVPTX Back-End\n";
    891   O << "//\n";
    892   O << "\n";
    893 
    894   unsigned PTXVersion = STI.getPTXVersion();
    895   O << ".version " << (PTXVersion / 10) << "." << (PTXVersion % 10) << "\n";
    896 
    897   O << ".target ";
    898   O << STI.getTargetName();
    899 
    900   const NVPTXTargetMachine &NTM = static_cast<const NVPTXTargetMachine &>(TM);
    901   if (NTM.getDrvInterface() == NVPTX::NVCL)
    902     O << ", texmode_independent";
    903   else {
    904     if (!STI.hasDouble())
    905       O << ", map_f64_to_f32";
    906   }
    907 
    908   if (MAI->doesSupportDebugInformation())
    909     O << ", debug";
    910 
    911   O << "\n";
    912 
    913   O << ".address_size ";
    914   if (NTM.is64Bit())
    915     O << "64";
    916   else
    917     O << "32";
    918   O << "\n";
    919 
    920   O << "\n";
    921 }
    922 
    923 bool NVPTXAsmPrinter::doFinalization(Module &M) {
    924   // If we did not emit any functions, then the global declarations have not
    925   // yet been emitted.
    926   if (!GlobalsEmitted) {
    927     emitGlobals(M);
    928     GlobalsEmitted = true;
    929   }
    930 
    931   // XXX Temproarily remove global variables so that doFinalization() will not
    932   // emit them again (global variables are emitted at beginning).
    933 
    934   Module::GlobalListType &global_list = M.getGlobalList();
    935   int i, n = global_list.size();
    936   GlobalVariable **gv_array = new GlobalVariable *[n];
    937 
    938   // first, back-up GlobalVariable in gv_array
    939   i = 0;
    940   for (Module::global_iterator I = global_list.begin(), E = global_list.end();
    941        I != E; ++I)
    942     gv_array[i++] = &*I;
    943 
    944   // second, empty global_list
    945   while (!global_list.empty())
    946     global_list.remove(global_list.begin());
    947 
    948   // call doFinalization
    949   bool ret = AsmPrinter::doFinalization(M);
    950 
    951   // now we restore global variables
    952   for (i = 0; i < n; i++)
    953     global_list.insert(global_list.end(), gv_array[i]);
    954 
    955   clearAnnotationCache(&M);
    956 
    957   delete[] gv_array;
    958   return ret;
    959 
    960   //bool Result = AsmPrinter::doFinalization(M);
    961   // Instead of calling the parents doFinalization, we may
    962   // clone parents doFinalization and customize here.
    963   // Currently, we if NVISA out the EmitGlobals() in
    964   // parent's doFinalization, which is too intrusive.
    965   //
    966   // Same for the doInitialization.
    967   //return Result;
    968 }
    969 
    970 // This function emits appropriate linkage directives for
    971 // functions and global variables.
    972 //
    973 // extern function declaration            -> .extern
    974 // extern function definition             -> .visible
    975 // external global variable with init     -> .visible
    976 // external without init                  -> .extern
    977 // appending                              -> not allowed, assert.
    978 // for any linkage other than
    979 // internal, private, linker_private,
    980 // linker_private_weak, linker_private_weak_def_auto,
    981 // we emit                                -> .weak.
    982 
    983 void NVPTXAsmPrinter::emitLinkageDirective(const GlobalValue *V,
    984                                            raw_ostream &O) {
    985   if (static_cast<NVPTXTargetMachine &>(TM).getDrvInterface() == NVPTX::CUDA) {
    986     if (V->hasExternalLinkage()) {
    987       if (isa<GlobalVariable>(V)) {
    988         const GlobalVariable *GVar = cast<GlobalVariable>(V);
    989         if (GVar) {
    990           if (GVar->hasInitializer())
    991             O << ".visible ";
    992           else
    993             O << ".extern ";
    994         }
    995       } else if (V->isDeclaration())
    996         O << ".extern ";
    997       else
    998         O << ".visible ";
    999     } else if (V->hasAppendingLinkage()) {
   1000       std::string msg;
   1001       msg.append("Error: ");
   1002       msg.append("Symbol ");
   1003       if (V->hasName())
   1004         msg.append(V->getName());
   1005       msg.append("has unsupported appending linkage type");
   1006       llvm_unreachable(msg.c_str());
   1007     } else if (!V->hasInternalLinkage() &&
   1008                !V->hasPrivateLinkage()) {
   1009       O << ".weak ";
   1010     }
   1011   }
   1012 }
   1013 
   1014 void NVPTXAsmPrinter::printModuleLevelGV(const GlobalVariable *GVar,
   1015                                          raw_ostream &O,
   1016                                          bool processDemoted) {
   1017 
   1018   // Skip meta data
   1019   if (GVar->hasSection()) {
   1020     if (GVar->getSection() == StringRef("llvm.metadata"))
   1021       return;
   1022   }
   1023 
   1024   // Skip LLVM intrinsic global variables
   1025   if (GVar->getName().startswith("llvm.") ||
   1026       GVar->getName().startswith("nvvm."))
   1027     return;
   1028 
   1029   const DataLayout &DL = getDataLayout();
   1030 
   1031   // GlobalVariables are always constant pointers themselves.
   1032   PointerType *PTy = GVar->getType();
   1033   Type *ETy = PTy->getElementType();
   1034 
   1035   if (GVar->hasExternalLinkage()) {
   1036     if (GVar->hasInitializer())
   1037       O << ".visible ";
   1038     else
   1039       O << ".extern ";
   1040   } else if (GVar->hasLinkOnceLinkage() || GVar->hasWeakLinkage() ||
   1041              GVar->hasAvailableExternallyLinkage() ||
   1042              GVar->hasCommonLinkage()) {
   1043     O << ".weak ";
   1044   }
   1045 
   1046   if (llvm::isTexture(*GVar)) {
   1047     O << ".global .texref " << llvm::getTextureName(*GVar) << ";\n";
   1048     return;
   1049   }
   1050 
   1051   if (llvm::isSurface(*GVar)) {
   1052     O << ".global .surfref " << llvm::getSurfaceName(*GVar) << ";\n";
   1053     return;
   1054   }
   1055 
   1056   if (GVar->isDeclaration()) {
   1057     // (extern) declarations, no definition or initializer
   1058     // Currently the only known declaration is for an automatic __local
   1059     // (.shared) promoted to global.
   1060     emitPTXGlobalVariable(GVar, O);
   1061     O << ";\n";
   1062     return;
   1063   }
   1064 
   1065   if (llvm::isSampler(*GVar)) {
   1066     O << ".global .samplerref " << llvm::getSamplerName(*GVar);
   1067 
   1068     const Constant *Initializer = nullptr;
   1069     if (GVar->hasInitializer())
   1070       Initializer = GVar->getInitializer();
   1071     const ConstantInt *CI = nullptr;
   1072     if (Initializer)
   1073       CI = dyn_cast<ConstantInt>(Initializer);
   1074     if (CI) {
   1075       unsigned sample = CI->getZExtValue();
   1076 
   1077       O << " = { ";
   1078 
   1079       for (int i = 0,
   1080                addr = ((sample & __CLK_ADDRESS_MASK) >> __CLK_ADDRESS_BASE);
   1081            i < 3; i++) {
   1082         O << "addr_mode_" << i << " = ";
   1083         switch (addr) {
   1084         case 0:
   1085           O << "wrap";
   1086           break;
   1087         case 1:
   1088           O << "clamp_to_border";
   1089           break;
   1090         case 2:
   1091           O << "clamp_to_edge";
   1092           break;
   1093         case 3:
   1094           O << "wrap";
   1095           break;
   1096         case 4:
   1097           O << "mirror";
   1098           break;
   1099         }
   1100         O << ", ";
   1101       }
   1102       O << "filter_mode = ";
   1103       switch ((sample & __CLK_FILTER_MASK) >> __CLK_FILTER_BASE) {
   1104       case 0:
   1105         O << "nearest";
   1106         break;
   1107       case 1:
   1108         O << "linear";
   1109         break;
   1110       case 2:
   1111         llvm_unreachable("Anisotropic filtering is not supported");
   1112       default:
   1113         O << "nearest";
   1114         break;
   1115       }
   1116       if (!((sample & __CLK_NORMALIZED_MASK) >> __CLK_NORMALIZED_BASE)) {
   1117         O << ", force_unnormalized_coords = 1";
   1118       }
   1119       O << " }";
   1120     }
   1121 
   1122     O << ";\n";
   1123     return;
   1124   }
   1125 
   1126   if (GVar->hasPrivateLinkage()) {
   1127 
   1128     if (!strncmp(GVar->getName().data(), "unrollpragma", 12))
   1129       return;
   1130 
   1131     // FIXME - need better way (e.g. Metadata) to avoid generating this global
   1132     if (!strncmp(GVar->getName().data(), "filename", 8))
   1133       return;
   1134     if (GVar->use_empty())
   1135       return;
   1136   }
   1137 
   1138   const Function *demotedFunc = nullptr;
   1139   if (!processDemoted && canDemoteGlobalVar(GVar, demotedFunc)) {
   1140     O << "// " << GVar->getName() << " has been demoted\n";
   1141     if (localDecls.find(demotedFunc) != localDecls.end())
   1142       localDecls[demotedFunc].push_back(GVar);
   1143     else {
   1144       std::vector<const GlobalVariable *> temp;
   1145       temp.push_back(GVar);
   1146       localDecls[demotedFunc] = temp;
   1147     }
   1148     return;
   1149   }
   1150 
   1151   O << ".";
   1152   emitPTXAddressSpace(PTy->getAddressSpace(), O);
   1153 
   1154   if (isManaged(*GVar)) {
   1155     O << " .attribute(.managed)";
   1156   }
   1157 
   1158   if (GVar->getAlignment() == 0)
   1159     O << " .align " << (int)DL.getPrefTypeAlignment(ETy);
   1160   else
   1161     O << " .align " << GVar->getAlignment();
   1162 
   1163   if (ETy->isFloatingPointTy() || ETy->isIntegerTy() || ETy->isPointerTy()) {
   1164     O << " .";
   1165     // Special case: ABI requires that we use .u8 for predicates
   1166     if (ETy->isIntegerTy(1))
   1167       O << "u8";
   1168     else
   1169       O << getPTXFundamentalTypeStr(ETy, false);
   1170     O << " ";
   1171     getSymbol(GVar)->print(O, MAI);
   1172 
   1173     // Ptx allows variable initilization only for constant and global state
   1174     // spaces.
   1175     if (GVar->hasInitializer()) {
   1176       if ((PTy->getAddressSpace() == llvm::ADDRESS_SPACE_GLOBAL) ||
   1177           (PTy->getAddressSpace() == llvm::ADDRESS_SPACE_CONST)) {
   1178         const Constant *Initializer = GVar->getInitializer();
   1179         // 'undef' is treated as there is no value specified.
   1180         if (!Initializer->isNullValue() && !isa<UndefValue>(Initializer)) {
   1181           O << " = ";
   1182           printScalarConstant(Initializer, O);
   1183         }
   1184       } else {
   1185         // The frontend adds zero-initializer to device and constant variables
   1186         // that don't have an initial value, and UndefValue to shared
   1187         // variables, so skip warning for this case.
   1188         if (!GVar->getInitializer()->isNullValue() &&
   1189             !isa<UndefValue>(GVar->getInitializer())) {
   1190           report_fatal_error("initial value of '" + GVar->getName() +
   1191                              "' is not allowed in addrspace(" +
   1192                              Twine(PTy->getAddressSpace()) + ")");
   1193         }
   1194       }
   1195     }
   1196   } else {
   1197     unsigned int ElementSize = 0;
   1198 
   1199     // Although PTX has direct support for struct type and array type and
   1200     // LLVM IR is very similar to PTX, the LLVM CodeGen does not support for
   1201     // targets that support these high level field accesses. Structs, arrays
   1202     // and vectors are lowered into arrays of bytes.
   1203     switch (ETy->getTypeID()) {
   1204     case Type::StructTyID:
   1205     case Type::ArrayTyID:
   1206     case Type::VectorTyID:
   1207       ElementSize = DL.getTypeStoreSize(ETy);
   1208       // Ptx allows variable initilization only for constant and
   1209       // global state spaces.
   1210       if (((PTy->getAddressSpace() == llvm::ADDRESS_SPACE_GLOBAL) ||
   1211            (PTy->getAddressSpace() == llvm::ADDRESS_SPACE_CONST)) &&
   1212           GVar->hasInitializer()) {
   1213         const Constant *Initializer = GVar->getInitializer();
   1214         if (!isa<UndefValue>(Initializer) && !Initializer->isNullValue()) {
   1215           AggBuffer aggBuffer(ElementSize, O, *this);
   1216           bufferAggregateConstant(Initializer, &aggBuffer);
   1217           if (aggBuffer.numSymbols) {
   1218             if (static_cast<const NVPTXTargetMachine &>(TM).is64Bit()) {
   1219               O << " .u64 ";
   1220               getSymbol(GVar)->print(O, MAI);
   1221               O << "[";
   1222               O << ElementSize / 8;
   1223             } else {
   1224               O << " .u32 ";
   1225               getSymbol(GVar)->print(O, MAI);
   1226               O << "[";
   1227               O << ElementSize / 4;
   1228             }
   1229             O << "]";
   1230           } else {
   1231             O << " .b8 ";
   1232             getSymbol(GVar)->print(O, MAI);
   1233             O << "[";
   1234             O << ElementSize;
   1235             O << "]";
   1236           }
   1237           O << " = {";
   1238           aggBuffer.print();
   1239           O << "}";
   1240         } else {
   1241           O << " .b8 ";
   1242           getSymbol(GVar)->print(O, MAI);
   1243           if (ElementSize) {
   1244             O << "[";
   1245             O << ElementSize;
   1246             O << "]";
   1247           }
   1248         }
   1249       } else {
   1250         O << " .b8 ";
   1251         getSymbol(GVar)->print(O, MAI);
   1252         if (ElementSize) {
   1253           O << "[";
   1254           O << ElementSize;
   1255           O << "]";
   1256         }
   1257       }
   1258       break;
   1259     default:
   1260       llvm_unreachable("type not supported yet");
   1261     }
   1262 
   1263   }
   1264   O << ";\n";
   1265 }
   1266 
   1267 void NVPTXAsmPrinter::emitDemotedVars(const Function *f, raw_ostream &O) {
   1268   if (localDecls.find(f) == localDecls.end())
   1269     return;
   1270 
   1271   std::vector<const GlobalVariable *> &gvars = localDecls[f];
   1272 
   1273   for (unsigned i = 0, e = gvars.size(); i != e; ++i) {
   1274     O << "\t// demoted variable\n\t";
   1275     printModuleLevelGV(gvars[i], O, true);
   1276   }
   1277 }
   1278 
   1279 void NVPTXAsmPrinter::emitPTXAddressSpace(unsigned int AddressSpace,
   1280                                           raw_ostream &O) const {
   1281   switch (AddressSpace) {
   1282   case llvm::ADDRESS_SPACE_LOCAL:
   1283     O << "local";
   1284     break;
   1285   case llvm::ADDRESS_SPACE_GLOBAL:
   1286     O << "global";
   1287     break;
   1288   case llvm::ADDRESS_SPACE_CONST:
   1289     O << "const";
   1290     break;
   1291   case llvm::ADDRESS_SPACE_SHARED:
   1292     O << "shared";
   1293     break;
   1294   default:
   1295     report_fatal_error("Bad address space found while emitting PTX");
   1296     break;
   1297   }
   1298 }
   1299 
   1300 std::string
   1301 NVPTXAsmPrinter::getPTXFundamentalTypeStr(Type *Ty, bool useB4PTR) const {
   1302   switch (Ty->getTypeID()) {
   1303   default:
   1304     llvm_unreachable("unexpected type");
   1305     break;
   1306   case Type::IntegerTyID: {
   1307     unsigned NumBits = cast<IntegerType>(Ty)->getBitWidth();
   1308     if (NumBits == 1)
   1309       return "pred";
   1310     else if (NumBits <= 64) {
   1311       std::string name = "u";
   1312       return name + utostr(NumBits);
   1313     } else {
   1314       llvm_unreachable("Integer too large");
   1315       break;
   1316     }
   1317     break;
   1318   }
   1319   case Type::FloatTyID:
   1320     return "f32";
   1321   case Type::DoubleTyID:
   1322     return "f64";
   1323   case Type::PointerTyID:
   1324     if (static_cast<const NVPTXTargetMachine &>(TM).is64Bit())
   1325       if (useB4PTR)
   1326         return "b64";
   1327       else
   1328         return "u64";
   1329     else if (useB4PTR)
   1330       return "b32";
   1331     else
   1332       return "u32";
   1333   }
   1334   llvm_unreachable("unexpected type");
   1335   return nullptr;
   1336 }
   1337 
   1338 void NVPTXAsmPrinter::emitPTXGlobalVariable(const GlobalVariable *GVar,
   1339                                             raw_ostream &O) {
   1340 
   1341   const DataLayout &DL = getDataLayout();
   1342 
   1343   // GlobalVariables are always constant pointers themselves.
   1344   PointerType *PTy = GVar->getType();
   1345   Type *ETy = PTy->getElementType();
   1346 
   1347   O << ".";
   1348   emitPTXAddressSpace(PTy->getAddressSpace(), O);
   1349   if (GVar->getAlignment() == 0)
   1350     O << " .align " << (int)DL.getPrefTypeAlignment(ETy);
   1351   else
   1352     O << " .align " << GVar->getAlignment();
   1353 
   1354   if (ETy->isFloatingPointTy() || ETy->isIntegerTy() || ETy->isPointerTy()) {
   1355     O << " .";
   1356     O << getPTXFundamentalTypeStr(ETy);
   1357     O << " ";
   1358     getSymbol(GVar)->print(O, MAI);
   1359     return;
   1360   }
   1361 
   1362   int64_t ElementSize = 0;
   1363 
   1364   // Although PTX has direct support for struct type and array type and LLVM IR
   1365   // is very similar to PTX, the LLVM CodeGen does not support for targets that
   1366   // support these high level field accesses. Structs and arrays are lowered
   1367   // into arrays of bytes.
   1368   switch (ETy->getTypeID()) {
   1369   case Type::StructTyID:
   1370   case Type::ArrayTyID:
   1371   case Type::VectorTyID:
   1372     ElementSize = DL.getTypeStoreSize(ETy);
   1373     O << " .b8 ";
   1374     getSymbol(GVar)->print(O, MAI);
   1375     O << "[";
   1376     if (ElementSize) {
   1377       O << ElementSize;
   1378     }
   1379     O << "]";
   1380     break;
   1381   default:
   1382     llvm_unreachable("type not supported yet");
   1383   }
   1384   return;
   1385 }
   1386 
   1387 static unsigned int getOpenCLAlignment(const DataLayout &DL, Type *Ty) {
   1388   if (Ty->isSingleValueType())
   1389     return DL.getPrefTypeAlignment(Ty);
   1390 
   1391   auto *ATy = dyn_cast<ArrayType>(Ty);
   1392   if (ATy)
   1393     return getOpenCLAlignment(DL, ATy->getElementType());
   1394 
   1395   auto *STy = dyn_cast<StructType>(Ty);
   1396   if (STy) {
   1397     unsigned int alignStruct = 1;
   1398     // Go through each element of the struct and find the
   1399     // largest alignment.
   1400     for (unsigned i = 0, e = STy->getNumElements(); i != e; i++) {
   1401       Type *ETy = STy->getElementType(i);
   1402       unsigned int align = getOpenCLAlignment(DL, ETy);
   1403       if (align > alignStruct)
   1404         alignStruct = align;
   1405     }
   1406     return alignStruct;
   1407   }
   1408 
   1409   auto *FTy = dyn_cast<FunctionType>(Ty);
   1410   if (FTy)
   1411     return DL.getPointerPrefAlignment();
   1412   return DL.getPrefTypeAlignment(Ty);
   1413 }
   1414 
   1415 void NVPTXAsmPrinter::printParamName(Function::const_arg_iterator I,
   1416                                      int paramIndex, raw_ostream &O) {
   1417   getSymbol(I->getParent())->print(O, MAI);
   1418   O << "_param_" << paramIndex;
   1419 }
   1420 
   1421 void NVPTXAsmPrinter::emitFunctionParamList(const Function *F, raw_ostream &O) {
   1422   const DataLayout &DL = getDataLayout();
   1423   const AttributeSet &PAL = F->getAttributes();
   1424   const TargetLowering *TLI = nvptxSubtarget->getTargetLowering();
   1425   Function::const_arg_iterator I, E;
   1426   unsigned paramIndex = 0;
   1427   bool first = true;
   1428   bool isKernelFunc = llvm::isKernelFunction(*F);
   1429   bool isABI = (nvptxSubtarget->getSmVersion() >= 20);
   1430   MVT thePointerTy = TLI->getPointerTy(DL);
   1431 
   1432   O << "(\n";
   1433 
   1434   for (I = F->arg_begin(), E = F->arg_end(); I != E; ++I, paramIndex++) {
   1435     Type *Ty = I->getType();
   1436 
   1437     if (!first)
   1438       O << ",\n";
   1439 
   1440     first = false;
   1441 
   1442     // Handle image/sampler parameters
   1443     if (isKernelFunction(*F)) {
   1444       if (isSampler(*I) || isImage(*I)) {
   1445         if (isImage(*I)) {
   1446           std::string sname = I->getName();
   1447           if (isImageWriteOnly(*I) || isImageReadWrite(*I)) {
   1448             if (nvptxSubtarget->hasImageHandles())
   1449               O << "\t.param .u64 .ptr .surfref ";
   1450             else
   1451               O << "\t.param .surfref ";
   1452             CurrentFnSym->print(O, MAI);
   1453             O << "_param_" << paramIndex;
   1454           }
   1455           else { // Default image is read_only
   1456             if (nvptxSubtarget->hasImageHandles())
   1457               O << "\t.param .u64 .ptr .texref ";
   1458             else
   1459               O << "\t.param .texref ";
   1460             CurrentFnSym->print(O, MAI);
   1461             O << "_param_" << paramIndex;
   1462           }
   1463         } else {
   1464           if (nvptxSubtarget->hasImageHandles())
   1465             O << "\t.param .u64 .ptr .samplerref ";
   1466           else
   1467             O << "\t.param .samplerref ";
   1468           CurrentFnSym->print(O, MAI);
   1469           O << "_param_" << paramIndex;
   1470         }
   1471         continue;
   1472       }
   1473     }
   1474 
   1475     if (!PAL.hasAttribute(paramIndex + 1, Attribute::ByVal)) {
   1476       if (Ty->isAggregateType() || Ty->isVectorTy()) {
   1477         // Just print .param .align <a> .b8 .param[size];
   1478         // <a> = PAL.getparamalignment
   1479         // size = typeallocsize of element type
   1480         unsigned align = PAL.getParamAlignment(paramIndex + 1);
   1481         if (align == 0)
   1482           align = DL.getABITypeAlignment(Ty);
   1483 
   1484         unsigned sz = DL.getTypeAllocSize(Ty);
   1485         O << "\t.param .align " << align << " .b8 ";
   1486         printParamName(I, paramIndex, O);
   1487         O << "[" << sz << "]";
   1488 
   1489         continue;
   1490       }
   1491       // Just a scalar
   1492       auto *PTy = dyn_cast<PointerType>(Ty);
   1493       if (isKernelFunc) {
   1494         if (PTy) {
   1495           // Special handling for pointer arguments to kernel
   1496           O << "\t.param .u" << thePointerTy.getSizeInBits() << " ";
   1497 
   1498           if (static_cast<NVPTXTargetMachine &>(TM).getDrvInterface() !=
   1499               NVPTX::CUDA) {
   1500             Type *ETy = PTy->getElementType();
   1501             int addrSpace = PTy->getAddressSpace();
   1502             switch (addrSpace) {
   1503             default:
   1504               O << ".ptr ";
   1505               break;
   1506             case llvm::ADDRESS_SPACE_CONST:
   1507               O << ".ptr .const ";
   1508               break;
   1509             case llvm::ADDRESS_SPACE_SHARED:
   1510               O << ".ptr .shared ";
   1511               break;
   1512             case llvm::ADDRESS_SPACE_GLOBAL:
   1513               O << ".ptr .global ";
   1514               break;
   1515             }
   1516             O << ".align " << (int)getOpenCLAlignment(DL, ETy) << " ";
   1517           }
   1518           printParamName(I, paramIndex, O);
   1519           continue;
   1520         }
   1521 
   1522         // non-pointer scalar to kernel func
   1523         O << "\t.param .";
   1524         // Special case: predicate operands become .u8 types
   1525         if (Ty->isIntegerTy(1))
   1526           O << "u8";
   1527         else
   1528           O << getPTXFundamentalTypeStr(Ty);
   1529         O << " ";
   1530         printParamName(I, paramIndex, O);
   1531         continue;
   1532       }
   1533       // Non-kernel function, just print .param .b<size> for ABI
   1534       // and .reg .b<size> for non-ABI
   1535       unsigned sz = 0;
   1536       if (isa<IntegerType>(Ty)) {
   1537         sz = cast<IntegerType>(Ty)->getBitWidth();
   1538         if (sz < 32)
   1539           sz = 32;
   1540       } else if (isa<PointerType>(Ty))
   1541         sz = thePointerTy.getSizeInBits();
   1542       else
   1543         sz = Ty->getPrimitiveSizeInBits();
   1544       if (isABI)
   1545         O << "\t.param .b" << sz << " ";
   1546       else
   1547         O << "\t.reg .b" << sz << " ";
   1548       printParamName(I, paramIndex, O);
   1549       continue;
   1550     }
   1551 
   1552     // param has byVal attribute. So should be a pointer
   1553     auto *PTy = dyn_cast<PointerType>(Ty);
   1554     assert(PTy && "Param with byval attribute should be a pointer type");
   1555     Type *ETy = PTy->getElementType();
   1556 
   1557     if (isABI || isKernelFunc) {
   1558       // Just print .param .align <a> .b8 .param[size];
   1559       // <a> = PAL.getparamalignment
   1560       // size = typeallocsize of element type
   1561       unsigned align = PAL.getParamAlignment(paramIndex + 1);
   1562       if (align == 0)
   1563         align = DL.getABITypeAlignment(ETy);
   1564 
   1565       unsigned sz = DL.getTypeAllocSize(ETy);
   1566       O << "\t.param .align " << align << " .b8 ";
   1567       printParamName(I, paramIndex, O);
   1568       O << "[" << sz << "]";
   1569       continue;
   1570     } else {
   1571       // Split the ETy into constituent parts and
   1572       // print .param .b<size> <name> for each part.
   1573       // Further, if a part is vector, print the above for
   1574       // each vector element.
   1575       SmallVector<EVT, 16> vtparts;
   1576       ComputeValueVTs(*TLI, DL, ETy, vtparts);
   1577       for (unsigned i = 0, e = vtparts.size(); i != e; ++i) {
   1578         unsigned elems = 1;
   1579         EVT elemtype = vtparts[i];
   1580         if (vtparts[i].isVector()) {
   1581           elems = vtparts[i].getVectorNumElements();
   1582           elemtype = vtparts[i].getVectorElementType();
   1583         }
   1584 
   1585         for (unsigned j = 0, je = elems; j != je; ++j) {
   1586           unsigned sz = elemtype.getSizeInBits();
   1587           if (elemtype.isInteger() && (sz < 32))
   1588             sz = 32;
   1589           O << "\t.reg .b" << sz << " ";
   1590           printParamName(I, paramIndex, O);
   1591           if (j < je - 1)
   1592             O << ",\n";
   1593           ++paramIndex;
   1594         }
   1595         if (i < e - 1)
   1596           O << ",\n";
   1597       }
   1598       --paramIndex;
   1599       continue;
   1600     }
   1601   }
   1602 
   1603   O << "\n)\n";
   1604 }
   1605 
   1606 void NVPTXAsmPrinter::emitFunctionParamList(const MachineFunction &MF,
   1607                                             raw_ostream &O) {
   1608   const Function *F = MF.getFunction();
   1609   emitFunctionParamList(F, O);
   1610 }
   1611 
   1612 void NVPTXAsmPrinter::setAndEmitFunctionVirtualRegisters(
   1613     const MachineFunction &MF) {
   1614   SmallString<128> Str;
   1615   raw_svector_ostream O(Str);
   1616 
   1617   // Map the global virtual register number to a register class specific
   1618   // virtual register number starting from 1 with that class.
   1619   const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo();
   1620   //unsigned numRegClasses = TRI->getNumRegClasses();
   1621 
   1622   // Emit the Fake Stack Object
   1623   const MachineFrameInfo *MFI = MF.getFrameInfo();
   1624   int NumBytes = (int) MFI->getStackSize();
   1625   if (NumBytes) {
   1626     O << "\t.local .align " << MFI->getMaxAlignment() << " .b8 \t" << DEPOTNAME
   1627       << getFunctionNumber() << "[" << NumBytes << "];\n";
   1628     if (static_cast<const NVPTXTargetMachine &>(MF.getTarget()).is64Bit()) {
   1629       O << "\t.reg .b64 \t%SP;\n";
   1630       O << "\t.reg .b64 \t%SPL;\n";
   1631     } else {
   1632       O << "\t.reg .b32 \t%SP;\n";
   1633       O << "\t.reg .b32 \t%SPL;\n";
   1634     }
   1635   }
   1636 
   1637   // Go through all virtual registers to establish the mapping between the
   1638   // global virtual
   1639   // register number and the per class virtual register number.
   1640   // We use the per class virtual register number in the ptx output.
   1641   unsigned int numVRs = MRI->getNumVirtRegs();
   1642   for (unsigned i = 0; i < numVRs; i++) {
   1643     unsigned int vr = TRI->index2VirtReg(i);
   1644     const TargetRegisterClass *RC = MRI->getRegClass(vr);
   1645     DenseMap<unsigned, unsigned> &regmap = VRegMapping[RC];
   1646     int n = regmap.size();
   1647     regmap.insert(std::make_pair(vr, n + 1));
   1648   }
   1649 
   1650   // Emit register declarations
   1651   // @TODO: Extract out the real register usage
   1652   // O << "\t.reg .pred %p<" << NVPTXNumRegisters << ">;\n";
   1653   // O << "\t.reg .s16 %rc<" << NVPTXNumRegisters << ">;\n";
   1654   // O << "\t.reg .s16 %rs<" << NVPTXNumRegisters << ">;\n";
   1655   // O << "\t.reg .s32 %r<" << NVPTXNumRegisters << ">;\n";
   1656   // O << "\t.reg .s64 %rd<" << NVPTXNumRegisters << ">;\n";
   1657   // O << "\t.reg .f32 %f<" << NVPTXNumRegisters << ">;\n";
   1658   // O << "\t.reg .f64 %fd<" << NVPTXNumRegisters << ">;\n";
   1659 
   1660   // Emit declaration of the virtual registers or 'physical' registers for
   1661   // each register class
   1662   for (unsigned i=0; i< TRI->getNumRegClasses(); i++) {
   1663     const TargetRegisterClass *RC = TRI->getRegClass(i);
   1664     DenseMap<unsigned, unsigned> &regmap = VRegMapping[RC];
   1665     std::string rcname = getNVPTXRegClassName(RC);
   1666     std::string rcStr = getNVPTXRegClassStr(RC);
   1667     int n = regmap.size();
   1668 
   1669     // Only declare those registers that may be used.
   1670     if (n) {
   1671        O << "\t.reg " << rcname << " \t" << rcStr << "<" << (n+1)
   1672          << ">;\n";
   1673     }
   1674   }
   1675 
   1676   OutStreamer->EmitRawText(O.str());
   1677 }
   1678 
   1679 void NVPTXAsmPrinter::printFPConstant(const ConstantFP *Fp, raw_ostream &O) {
   1680   APFloat APF = APFloat(Fp->getValueAPF()); // make a copy
   1681   bool ignored;
   1682   unsigned int numHex;
   1683   const char *lead;
   1684 
   1685   if (Fp->getType()->getTypeID() == Type::FloatTyID) {
   1686     numHex = 8;
   1687     lead = "0f";
   1688     APF.convert(APFloat::IEEEsingle, APFloat::rmNearestTiesToEven, &ignored);
   1689   } else if (Fp->getType()->getTypeID() == Type::DoubleTyID) {
   1690     numHex = 16;
   1691     lead = "0d";
   1692     APF.convert(APFloat::IEEEdouble, APFloat::rmNearestTiesToEven, &ignored);
   1693   } else
   1694     llvm_unreachable("unsupported fp type");
   1695 
   1696   APInt API = APF.bitcastToAPInt();
   1697   std::string hexstr(utohexstr(API.getZExtValue()));
   1698   O << lead;
   1699   if (hexstr.length() < numHex)
   1700     O << std::string(numHex - hexstr.length(), '0');
   1701   O << utohexstr(API.getZExtValue());
   1702 }
   1703 
   1704 void NVPTXAsmPrinter::printScalarConstant(const Constant *CPV, raw_ostream &O) {
   1705   if (const ConstantInt *CI = dyn_cast<ConstantInt>(CPV)) {
   1706     O << CI->getValue();
   1707     return;
   1708   }
   1709   if (const ConstantFP *CFP = dyn_cast<ConstantFP>(CPV)) {
   1710     printFPConstant(CFP, O);
   1711     return;
   1712   }
   1713   if (isa<ConstantPointerNull>(CPV)) {
   1714     O << "0";
   1715     return;
   1716   }
   1717   if (const GlobalValue *GVar = dyn_cast<GlobalValue>(CPV)) {
   1718     PointerType *PTy = dyn_cast<PointerType>(GVar->getType());
   1719     bool IsNonGenericPointer = false;
   1720     if (PTy && PTy->getAddressSpace() != 0) {
   1721       IsNonGenericPointer = true;
   1722     }
   1723     if (EmitGeneric && !isa<Function>(CPV) && !IsNonGenericPointer) {
   1724       O << "generic(";
   1725       getSymbol(GVar)->print(O, MAI);
   1726       O << ")";
   1727     } else {
   1728       getSymbol(GVar)->print(O, MAI);
   1729     }
   1730     return;
   1731   }
   1732   if (const ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
   1733     const Value *v = Cexpr->stripPointerCasts();
   1734     PointerType *PTy = dyn_cast<PointerType>(Cexpr->getType());
   1735     bool IsNonGenericPointer = false;
   1736     if (PTy && PTy->getAddressSpace() != 0) {
   1737       IsNonGenericPointer = true;
   1738     }
   1739     if (const GlobalValue *GVar = dyn_cast<GlobalValue>(v)) {
   1740       if (EmitGeneric && !isa<Function>(v) && !IsNonGenericPointer) {
   1741         O << "generic(";
   1742         getSymbol(GVar)->print(O, MAI);
   1743         O << ")";
   1744       } else {
   1745         getSymbol(GVar)->print(O, MAI);
   1746       }
   1747       return;
   1748     } else {
   1749       lowerConstant(CPV)->print(O, MAI);
   1750       return;
   1751     }
   1752   }
   1753   llvm_unreachable("Not scalar type found in printScalarConstant()");
   1754 }
   1755 
   1756 // These utility functions assure we get the right sequence of bytes for a given
   1757 // type even for big-endian machines
   1758 template <typename T> static void ConvertIntToBytes(unsigned char *p, T val) {
   1759   int64_t vp = (int64_t)val;
   1760   for (unsigned i = 0; i < sizeof(T); ++i) {
   1761     p[i] = (unsigned char)vp;
   1762     vp >>= 8;
   1763   }
   1764 }
   1765 static void ConvertFloatToBytes(unsigned char *p, float val) {
   1766   int32_t *vp = (int32_t *)&val;
   1767   for (unsigned i = 0; i < sizeof(int32_t); ++i) {
   1768     p[i] = (unsigned char)*vp;
   1769     *vp >>= 8;
   1770   }
   1771 }
   1772 static void ConvertDoubleToBytes(unsigned char *p, double val) {
   1773   int64_t *vp = (int64_t *)&val;
   1774   for (unsigned i = 0; i < sizeof(int64_t); ++i) {
   1775     p[i] = (unsigned char)*vp;
   1776     *vp >>= 8;
   1777   }
   1778 }
   1779 
   1780 void NVPTXAsmPrinter::bufferLEByte(const Constant *CPV, int Bytes,
   1781                                    AggBuffer *aggBuffer) {
   1782 
   1783   const DataLayout &DL = getDataLayout();
   1784 
   1785   if (isa<UndefValue>(CPV) || CPV->isNullValue()) {
   1786     int s = DL.getTypeAllocSize(CPV->getType());
   1787     if (s < Bytes)
   1788       s = Bytes;
   1789     aggBuffer->addZeros(s);
   1790     return;
   1791   }
   1792 
   1793   unsigned char ptr[8];
   1794   switch (CPV->getType()->getTypeID()) {
   1795 
   1796   case Type::IntegerTyID: {
   1797     Type *ETy = CPV->getType();
   1798     if (ETy == Type::getInt8Ty(CPV->getContext())) {
   1799       unsigned char c = (unsigned char)cast<ConstantInt>(CPV)->getZExtValue();
   1800       ConvertIntToBytes<>(ptr, c);
   1801       aggBuffer->addBytes(ptr, 1, Bytes);
   1802     } else if (ETy == Type::getInt16Ty(CPV->getContext())) {
   1803       short int16 = (short)cast<ConstantInt>(CPV)->getZExtValue();
   1804       ConvertIntToBytes<>(ptr, int16);
   1805       aggBuffer->addBytes(ptr, 2, Bytes);
   1806     } else if (ETy == Type::getInt32Ty(CPV->getContext())) {
   1807       if (const ConstantInt *constInt = dyn_cast<ConstantInt>(CPV)) {
   1808         int int32 = (int)(constInt->getZExtValue());
   1809         ConvertIntToBytes<>(ptr, int32);
   1810         aggBuffer->addBytes(ptr, 4, Bytes);
   1811         break;
   1812       } else if (const ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
   1813         if (const ConstantInt *constInt = dyn_cast<ConstantInt>(
   1814                 ConstantFoldConstantExpression(Cexpr, DL))) {
   1815           int int32 = (int)(constInt->getZExtValue());
   1816           ConvertIntToBytes<>(ptr, int32);
   1817           aggBuffer->addBytes(ptr, 4, Bytes);
   1818           break;
   1819         }
   1820         if (Cexpr->getOpcode() == Instruction::PtrToInt) {
   1821           Value *v = Cexpr->getOperand(0)->stripPointerCasts();
   1822           aggBuffer->addSymbol(v, Cexpr->getOperand(0));
   1823           aggBuffer->addZeros(4);
   1824           break;
   1825         }
   1826       }
   1827       llvm_unreachable("unsupported integer const type");
   1828     } else if (ETy == Type::getInt64Ty(CPV->getContext())) {
   1829       if (const ConstantInt *constInt = dyn_cast<ConstantInt>(CPV)) {
   1830         long long int64 = (long long)(constInt->getZExtValue());
   1831         ConvertIntToBytes<>(ptr, int64);
   1832         aggBuffer->addBytes(ptr, 8, Bytes);
   1833         break;
   1834       } else if (const ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
   1835         if (const ConstantInt *constInt = dyn_cast<ConstantInt>(
   1836                 ConstantFoldConstantExpression(Cexpr, DL))) {
   1837           long long int64 = (long long)(constInt->getZExtValue());
   1838           ConvertIntToBytes<>(ptr, int64);
   1839           aggBuffer->addBytes(ptr, 8, Bytes);
   1840           break;
   1841         }
   1842         if (Cexpr->getOpcode() == Instruction::PtrToInt) {
   1843           Value *v = Cexpr->getOperand(0)->stripPointerCasts();
   1844           aggBuffer->addSymbol(v, Cexpr->getOperand(0));
   1845           aggBuffer->addZeros(8);
   1846           break;
   1847         }
   1848       }
   1849       llvm_unreachable("unsupported integer const type");
   1850     } else
   1851       llvm_unreachable("unsupported integer const type");
   1852     break;
   1853   }
   1854   case Type::FloatTyID:
   1855   case Type::DoubleTyID: {
   1856     const ConstantFP *CFP = dyn_cast<ConstantFP>(CPV);
   1857     Type *Ty = CFP->getType();
   1858     if (Ty == Type::getFloatTy(CPV->getContext())) {
   1859       float float32 = (float) CFP->getValueAPF().convertToFloat();
   1860       ConvertFloatToBytes(ptr, float32);
   1861       aggBuffer->addBytes(ptr, 4, Bytes);
   1862     } else if (Ty == Type::getDoubleTy(CPV->getContext())) {
   1863       double float64 = CFP->getValueAPF().convertToDouble();
   1864       ConvertDoubleToBytes(ptr, float64);
   1865       aggBuffer->addBytes(ptr, 8, Bytes);
   1866     } else {
   1867       llvm_unreachable("unsupported fp const type");
   1868     }
   1869     break;
   1870   }
   1871   case Type::PointerTyID: {
   1872     if (const GlobalValue *GVar = dyn_cast<GlobalValue>(CPV)) {
   1873       aggBuffer->addSymbol(GVar, GVar);
   1874     } else if (const ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
   1875       const Value *v = Cexpr->stripPointerCasts();
   1876       aggBuffer->addSymbol(v, Cexpr);
   1877     }
   1878     unsigned int s = DL.getTypeAllocSize(CPV->getType());
   1879     aggBuffer->addZeros(s);
   1880     break;
   1881   }
   1882 
   1883   case Type::ArrayTyID:
   1884   case Type::VectorTyID:
   1885   case Type::StructTyID: {
   1886     if (isa<ConstantArray>(CPV) || isa<ConstantVector>(CPV) ||
   1887         isa<ConstantStruct>(CPV) || isa<ConstantDataSequential>(CPV)) {
   1888       int ElementSize = DL.getTypeAllocSize(CPV->getType());
   1889       bufferAggregateConstant(CPV, aggBuffer);
   1890       if (Bytes > ElementSize)
   1891         aggBuffer->addZeros(Bytes - ElementSize);
   1892     } else if (isa<ConstantAggregateZero>(CPV))
   1893       aggBuffer->addZeros(Bytes);
   1894     else
   1895       llvm_unreachable("Unexpected Constant type");
   1896     break;
   1897   }
   1898 
   1899   default:
   1900     llvm_unreachable("unsupported type");
   1901   }
   1902 }
   1903 
   1904 void NVPTXAsmPrinter::bufferAggregateConstant(const Constant *CPV,
   1905                                               AggBuffer *aggBuffer) {
   1906   const DataLayout &DL = getDataLayout();
   1907   int Bytes;
   1908 
   1909   // Old constants
   1910   if (isa<ConstantArray>(CPV) || isa<ConstantVector>(CPV)) {
   1911     if (CPV->getNumOperands())
   1912       for (unsigned i = 0, e = CPV->getNumOperands(); i != e; ++i)
   1913         bufferLEByte(cast<Constant>(CPV->getOperand(i)), 0, aggBuffer);
   1914     return;
   1915   }
   1916 
   1917   if (const ConstantDataSequential *CDS =
   1918           dyn_cast<ConstantDataSequential>(CPV)) {
   1919     if (CDS->getNumElements())
   1920       for (unsigned i = 0; i < CDS->getNumElements(); ++i)
   1921         bufferLEByte(cast<Constant>(CDS->getElementAsConstant(i)), 0,
   1922                      aggBuffer);
   1923     return;
   1924   }
   1925 
   1926   if (isa<ConstantStruct>(CPV)) {
   1927     if (CPV->getNumOperands()) {
   1928       StructType *ST = cast<StructType>(CPV->getType());
   1929       for (unsigned i = 0, e = CPV->getNumOperands(); i != e; ++i) {
   1930         if (i == (e - 1))
   1931           Bytes = DL.getStructLayout(ST)->getElementOffset(0) +
   1932                   DL.getTypeAllocSize(ST) -
   1933                   DL.getStructLayout(ST)->getElementOffset(i);
   1934         else
   1935           Bytes = DL.getStructLayout(ST)->getElementOffset(i + 1) -
   1936                   DL.getStructLayout(ST)->getElementOffset(i);
   1937         bufferLEByte(cast<Constant>(CPV->getOperand(i)), Bytes, aggBuffer);
   1938       }
   1939     }
   1940     return;
   1941   }
   1942   llvm_unreachable("unsupported constant type in printAggregateConstant()");
   1943 }
   1944 
   1945 // buildTypeNameMap - Run through symbol table looking for type names.
   1946 //
   1947 
   1948 
   1949 bool NVPTXAsmPrinter::ignoreLoc(const MachineInstr &MI) {
   1950   switch (MI.getOpcode()) {
   1951   default:
   1952     return false;
   1953   case NVPTX::CallArgBeginInst:
   1954   case NVPTX::CallArgEndInst0:
   1955   case NVPTX::CallArgEndInst1:
   1956   case NVPTX::CallArgF32:
   1957   case NVPTX::CallArgF64:
   1958   case NVPTX::CallArgI16:
   1959   case NVPTX::CallArgI32:
   1960   case NVPTX::CallArgI32imm:
   1961   case NVPTX::CallArgI64:
   1962   case NVPTX::CallArgParam:
   1963   case NVPTX::CallVoidInst:
   1964   case NVPTX::CallVoidInstReg:
   1965   case NVPTX::Callseq_End:
   1966   case NVPTX::CallVoidInstReg64:
   1967   case NVPTX::DeclareParamInst:
   1968   case NVPTX::DeclareRetMemInst:
   1969   case NVPTX::DeclareRetRegInst:
   1970   case NVPTX::DeclareRetScalarInst:
   1971   case NVPTX::DeclareScalarParamInst:
   1972   case NVPTX::DeclareScalarRegInst:
   1973   case NVPTX::StoreParamF32:
   1974   case NVPTX::StoreParamF64:
   1975   case NVPTX::StoreParamI16:
   1976   case NVPTX::StoreParamI32:
   1977   case NVPTX::StoreParamI64:
   1978   case NVPTX::StoreParamI8:
   1979   case NVPTX::StoreRetvalF32:
   1980   case NVPTX::StoreRetvalF64:
   1981   case NVPTX::StoreRetvalI16:
   1982   case NVPTX::StoreRetvalI32:
   1983   case NVPTX::StoreRetvalI64:
   1984   case NVPTX::StoreRetvalI8:
   1985   case NVPTX::LastCallArgF32:
   1986   case NVPTX::LastCallArgF64:
   1987   case NVPTX::LastCallArgI16:
   1988   case NVPTX::LastCallArgI32:
   1989   case NVPTX::LastCallArgI32imm:
   1990   case NVPTX::LastCallArgI64:
   1991   case NVPTX::LastCallArgParam:
   1992   case NVPTX::LoadParamMemF32:
   1993   case NVPTX::LoadParamMemF64:
   1994   case NVPTX::LoadParamMemI16:
   1995   case NVPTX::LoadParamMemI32:
   1996   case NVPTX::LoadParamMemI64:
   1997   case NVPTX::LoadParamMemI8:
   1998   case NVPTX::PrototypeInst:
   1999   case NVPTX::DBG_VALUE:
   2000     return true;
   2001   }
   2002   return false;
   2003 }
   2004 
   2005 /// lowerConstantForGV - Return an MCExpr for the given Constant.  This is mostly
   2006 /// a copy from AsmPrinter::lowerConstant, except customized to only handle
   2007 /// expressions that are representable in PTX and create
   2008 /// NVPTXGenericMCSymbolRefExpr nodes for addrspacecast instructions.
   2009 const MCExpr *
   2010 NVPTXAsmPrinter::lowerConstantForGV(const Constant *CV, bool ProcessingGeneric) {
   2011   MCContext &Ctx = OutContext;
   2012 
   2013   if (CV->isNullValue() || isa<UndefValue>(CV))
   2014     return MCConstantExpr::create(0, Ctx);
   2015 
   2016   if (const ConstantInt *CI = dyn_cast<ConstantInt>(CV))
   2017     return MCConstantExpr::create(CI->getZExtValue(), Ctx);
   2018 
   2019   if (const GlobalValue *GV = dyn_cast<GlobalValue>(CV)) {
   2020     const MCSymbolRefExpr *Expr =
   2021       MCSymbolRefExpr::create(getSymbol(GV), Ctx);
   2022     if (ProcessingGeneric) {
   2023       return NVPTXGenericMCSymbolRefExpr::create(Expr, Ctx);
   2024     } else {
   2025       return Expr;
   2026     }
   2027   }
   2028 
   2029   const ConstantExpr *CE = dyn_cast<ConstantExpr>(CV);
   2030   if (!CE) {
   2031     llvm_unreachable("Unknown constant value to lower!");
   2032   }
   2033 
   2034   switch (CE->getOpcode()) {
   2035   default:
   2036     // If the code isn't optimized, there may be outstanding folding
   2037     // opportunities. Attempt to fold the expression using DataLayout as a
   2038     // last resort before giving up.
   2039     if (Constant *C = ConstantFoldConstantExpression(CE, getDataLayout()))
   2040       if (C != CE)
   2041         return lowerConstantForGV(C, ProcessingGeneric);
   2042 
   2043     // Otherwise report the problem to the user.
   2044     {
   2045       std::string S;
   2046       raw_string_ostream OS(S);
   2047       OS << "Unsupported expression in static initializer: ";
   2048       CE->printAsOperand(OS, /*PrintType=*/false,
   2049                      !MF ? nullptr : MF->getFunction()->getParent());
   2050       report_fatal_error(OS.str());
   2051     }
   2052 
   2053   case Instruction::AddrSpaceCast: {
   2054     // Strip the addrspacecast and pass along the operand
   2055     PointerType *DstTy = cast<PointerType>(CE->getType());
   2056     if (DstTy->getAddressSpace() == 0) {
   2057       return lowerConstantForGV(cast<const Constant>(CE->getOperand(0)), true);
   2058     }
   2059     std::string S;
   2060     raw_string_ostream OS(S);
   2061     OS << "Unsupported expression in static initializer: ";
   2062     CE->printAsOperand(OS, /*PrintType=*/ false,
   2063                        !MF ? 0 : MF->getFunction()->getParent());
   2064     report_fatal_error(OS.str());
   2065   }
   2066 
   2067   case Instruction::GetElementPtr: {
   2068     const DataLayout &DL = getDataLayout();
   2069 
   2070     // Generate a symbolic expression for the byte address
   2071     APInt OffsetAI(DL.getPointerTypeSizeInBits(CE->getType()), 0);
   2072     cast<GEPOperator>(CE)->accumulateConstantOffset(DL, OffsetAI);
   2073 
   2074     const MCExpr *Base = lowerConstantForGV(CE->getOperand(0),
   2075                                             ProcessingGeneric);
   2076     if (!OffsetAI)
   2077       return Base;
   2078 
   2079     int64_t Offset = OffsetAI.getSExtValue();
   2080     return MCBinaryExpr::createAdd(Base, MCConstantExpr::create(Offset, Ctx),
   2081                                    Ctx);
   2082   }
   2083 
   2084   case Instruction::Trunc:
   2085     // We emit the value and depend on the assembler to truncate the generated
   2086     // expression properly.  This is important for differences between
   2087     // blockaddress labels.  Since the two labels are in the same function, it
   2088     // is reasonable to treat their delta as a 32-bit value.
   2089     // FALL THROUGH.
   2090   case Instruction::BitCast:
   2091     return lowerConstantForGV(CE->getOperand(0), ProcessingGeneric);
   2092 
   2093   case Instruction::IntToPtr: {
   2094     const DataLayout &DL = getDataLayout();
   2095 
   2096     // Handle casts to pointers by changing them into casts to the appropriate
   2097     // integer type.  This promotes constant folding and simplifies this code.
   2098     Constant *Op = CE->getOperand(0);
   2099     Op = ConstantExpr::getIntegerCast(Op, DL.getIntPtrType(CV->getType()),
   2100                                       false/*ZExt*/);
   2101     return lowerConstantForGV(Op, ProcessingGeneric);
   2102   }
   2103 
   2104   case Instruction::PtrToInt: {
   2105     const DataLayout &DL = getDataLayout();
   2106 
   2107     // Support only foldable casts to/from pointers that can be eliminated by
   2108     // changing the pointer to the appropriately sized integer type.
   2109     Constant *Op = CE->getOperand(0);
   2110     Type *Ty = CE->getType();
   2111 
   2112     const MCExpr *OpExpr = lowerConstantForGV(Op, ProcessingGeneric);
   2113 
   2114     // We can emit the pointer value into this slot if the slot is an
   2115     // integer slot equal to the size of the pointer.
   2116     if (DL.getTypeAllocSize(Ty) == DL.getTypeAllocSize(Op->getType()))
   2117       return OpExpr;
   2118 
   2119     // Otherwise the pointer is smaller than the resultant integer, mask off
   2120     // the high bits so we are sure to get a proper truncation if the input is
   2121     // a constant expr.
   2122     unsigned InBits = DL.getTypeAllocSizeInBits(Op->getType());
   2123     const MCExpr *MaskExpr = MCConstantExpr::create(~0ULL >> (64-InBits), Ctx);
   2124     return MCBinaryExpr::createAnd(OpExpr, MaskExpr, Ctx);
   2125   }
   2126 
   2127   // The MC library also has a right-shift operator, but it isn't consistently
   2128   // signed or unsigned between different targets.
   2129   case Instruction::Add: {
   2130     const MCExpr *LHS = lowerConstantForGV(CE->getOperand(0), ProcessingGeneric);
   2131     const MCExpr *RHS = lowerConstantForGV(CE->getOperand(1), ProcessingGeneric);
   2132     switch (CE->getOpcode()) {
   2133     default: llvm_unreachable("Unknown binary operator constant cast expr");
   2134     case Instruction::Add: return MCBinaryExpr::createAdd(LHS, RHS, Ctx);
   2135     }
   2136   }
   2137   }
   2138 }
   2139 
   2140 // Copy of MCExpr::print customized for NVPTX
   2141 void NVPTXAsmPrinter::printMCExpr(const MCExpr &Expr, raw_ostream &OS) {
   2142   switch (Expr.getKind()) {
   2143   case MCExpr::Target:
   2144     return cast<MCTargetExpr>(&Expr)->printImpl(OS, MAI);
   2145   case MCExpr::Constant:
   2146     OS << cast<MCConstantExpr>(Expr).getValue();
   2147     return;
   2148 
   2149   case MCExpr::SymbolRef: {
   2150     const MCSymbolRefExpr &SRE = cast<MCSymbolRefExpr>(Expr);
   2151     const MCSymbol &Sym = SRE.getSymbol();
   2152     Sym.print(OS, MAI);
   2153     return;
   2154   }
   2155 
   2156   case MCExpr::Unary: {
   2157     const MCUnaryExpr &UE = cast<MCUnaryExpr>(Expr);
   2158     switch (UE.getOpcode()) {
   2159     case MCUnaryExpr::LNot:  OS << '!'; break;
   2160     case MCUnaryExpr::Minus: OS << '-'; break;
   2161     case MCUnaryExpr::Not:   OS << '~'; break;
   2162     case MCUnaryExpr::Plus:  OS << '+'; break;
   2163     }
   2164     printMCExpr(*UE.getSubExpr(), OS);
   2165     return;
   2166   }
   2167 
   2168   case MCExpr::Binary: {
   2169     const MCBinaryExpr &BE = cast<MCBinaryExpr>(Expr);
   2170 
   2171     // Only print parens around the LHS if it is non-trivial.
   2172     if (isa<MCConstantExpr>(BE.getLHS()) || isa<MCSymbolRefExpr>(BE.getLHS()) ||
   2173         isa<NVPTXGenericMCSymbolRefExpr>(BE.getLHS())) {
   2174       printMCExpr(*BE.getLHS(), OS);
   2175     } else {
   2176       OS << '(';
   2177       printMCExpr(*BE.getLHS(), OS);
   2178       OS<< ')';
   2179     }
   2180 
   2181     switch (BE.getOpcode()) {
   2182     case MCBinaryExpr::Add:
   2183       // Print "X-42" instead of "X+-42".
   2184       if (const MCConstantExpr *RHSC = dyn_cast<MCConstantExpr>(BE.getRHS())) {
   2185         if (RHSC->getValue() < 0) {
   2186           OS << RHSC->getValue();
   2187           return;
   2188         }
   2189       }
   2190 
   2191       OS <<  '+';
   2192       break;
   2193     default: llvm_unreachable("Unhandled binary operator");
   2194     }
   2195 
   2196     // Only print parens around the LHS if it is non-trivial.
   2197     if (isa<MCConstantExpr>(BE.getRHS()) || isa<MCSymbolRefExpr>(BE.getRHS())) {
   2198       printMCExpr(*BE.getRHS(), OS);
   2199     } else {
   2200       OS << '(';
   2201       printMCExpr(*BE.getRHS(), OS);
   2202       OS << ')';
   2203     }
   2204     return;
   2205   }
   2206   }
   2207 
   2208   llvm_unreachable("Invalid expression kind!");
   2209 }
   2210 
   2211 /// PrintAsmOperand - Print out an operand for an inline asm expression.
   2212 ///
   2213 bool NVPTXAsmPrinter::PrintAsmOperand(const MachineInstr *MI, unsigned OpNo,
   2214                                       unsigned AsmVariant,
   2215                                       const char *ExtraCode, raw_ostream &O) {
   2216   if (ExtraCode && ExtraCode[0]) {
   2217     if (ExtraCode[1] != 0)
   2218       return true; // Unknown modifier.
   2219 
   2220     switch (ExtraCode[0]) {
   2221     default:
   2222       // See if this is a generic print operand
   2223       return AsmPrinter::PrintAsmOperand(MI, OpNo, AsmVariant, ExtraCode, O);
   2224     case 'r':
   2225       break;
   2226     }
   2227   }
   2228 
   2229   printOperand(MI, OpNo, O);
   2230 
   2231   return false;
   2232 }
   2233 
   2234 bool NVPTXAsmPrinter::PrintAsmMemoryOperand(
   2235     const MachineInstr *MI, unsigned OpNo, unsigned AsmVariant,
   2236     const char *ExtraCode, raw_ostream &O) {
   2237   if (ExtraCode && ExtraCode[0])
   2238     return true; // Unknown modifier
   2239 
   2240   O << '[';
   2241   printMemOperand(MI, OpNo, O);
   2242   O << ']';
   2243 
   2244   return false;
   2245 }
   2246 
   2247 void NVPTXAsmPrinter::printOperand(const MachineInstr *MI, int opNum,
   2248                                    raw_ostream &O, const char *Modifier) {
   2249   const MachineOperand &MO = MI->getOperand(opNum);
   2250   switch (MO.getType()) {
   2251   case MachineOperand::MO_Register:
   2252     if (TargetRegisterInfo::isPhysicalRegister(MO.getReg())) {
   2253       if (MO.getReg() == NVPTX::VRDepot)
   2254         O << DEPOTNAME << getFunctionNumber();
   2255       else
   2256         O << NVPTXInstPrinter::getRegisterName(MO.getReg());
   2257     } else {
   2258       emitVirtualRegister(MO.getReg(), O);
   2259     }
   2260     return;
   2261 
   2262   case MachineOperand::MO_Immediate:
   2263     if (!Modifier)
   2264       O << MO.getImm();
   2265     else if (strstr(Modifier, "vec") == Modifier)
   2266       printVecModifiedImmediate(MO, Modifier, O);
   2267     else
   2268       llvm_unreachable(
   2269           "Don't know how to handle modifier on immediate operand");
   2270     return;
   2271 
   2272   case MachineOperand::MO_FPImmediate:
   2273     printFPConstant(MO.getFPImm(), O);
   2274     break;
   2275 
   2276   case MachineOperand::MO_GlobalAddress:
   2277     getSymbol(MO.getGlobal())->print(O, MAI);
   2278     break;
   2279 
   2280   case MachineOperand::MO_MachineBasicBlock:
   2281     MO.getMBB()->getSymbol()->print(O, MAI);
   2282     return;
   2283 
   2284   default:
   2285     llvm_unreachable("Operand type not supported.");
   2286   }
   2287 }
   2288 
   2289 void NVPTXAsmPrinter::printMemOperand(const MachineInstr *MI, int opNum,
   2290                                       raw_ostream &O, const char *Modifier) {
   2291   printOperand(MI, opNum, O);
   2292 
   2293   if (Modifier && !strcmp(Modifier, "add")) {
   2294     O << ", ";
   2295     printOperand(MI, opNum + 1, O);
   2296   } else {
   2297     if (MI->getOperand(opNum + 1).isImm() &&
   2298         MI->getOperand(opNum + 1).getImm() == 0)
   2299       return; // don't print ',0' or '+0'
   2300     O << "+";
   2301     printOperand(MI, opNum + 1, O);
   2302   }
   2303 }
   2304 
   2305 void NVPTXAsmPrinter::emitSrcInText(StringRef filename, unsigned line) {
   2306   std::stringstream temp;
   2307   LineReader *reader = this->getReader(filename);
   2308   temp << "\n//";
   2309   temp << filename.str();
   2310   temp << ":";
   2311   temp << line;
   2312   temp << " ";
   2313   temp << reader->readLine(line);
   2314   temp << "\n";
   2315   this->OutStreamer->EmitRawText(temp.str());
   2316 }
   2317 
   2318 LineReader *NVPTXAsmPrinter::getReader(std::string filename) {
   2319   if (!reader) {
   2320     reader = new LineReader(filename);
   2321   }
   2322 
   2323   if (reader->fileName() != filename) {
   2324     delete reader;
   2325     reader = new LineReader(filename);
   2326   }
   2327 
   2328   return reader;
   2329 }
   2330 
   2331 std::string LineReader::readLine(unsigned lineNum) {
   2332   if (lineNum < theCurLine) {
   2333     theCurLine = 0;
   2334     fstr.seekg(0, std::ios::beg);
   2335   }
   2336   while (theCurLine < lineNum) {
   2337     fstr.getline(buff, 500);
   2338     theCurLine++;
   2339   }
   2340   return buff;
   2341 }
   2342 
   2343 // Force static initialization.
   2344 extern "C" void LLVMInitializeNVPTXAsmPrinter() {
   2345   RegisterAsmPrinter<NVPTXAsmPrinter> X(TheNVPTXTarget32);
   2346   RegisterAsmPrinter<NVPTXAsmPrinter> Y(TheNVPTXTarget64);
   2347 }
   2348