NVPTXAsmPrinter.cpp revision 249423 
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 "MCTargetDesc/NVPTXMCAsmInfo.h"
17#include "NVPTX.h"
18#include "NVPTXInstrInfo.h"
19#include "NVPTXNumRegisters.h"
20#include "NVPTXRegisterInfo.h"
21#include "NVPTXTargetMachine.h"
22#include "NVPTXUtilities.h"
23#include "cl_common_defines.h"
24#include "llvm/ADT/StringExtras.h"
25#include "llvm/Analysis/ConstantFolding.h"
26#include "llvm/Assembly/Writer.h"
27#include "llvm/CodeGen/Analysis.h"
28#include "llvm/CodeGen/MachineFrameInfo.h"
29#include "llvm/CodeGen/MachineModuleInfo.h"
30#include "llvm/CodeGen/MachineRegisterInfo.h"
31#include "llvm/DebugInfo.h"
32#include "llvm/IR/DerivedTypes.h"
33#include "llvm/IR/Function.h"
34#include "llvm/IR/GlobalVariable.h"
35#include "llvm/IR/Module.h"
36#include "llvm/IR/Operator.h"
37#include "llvm/MC/MCStreamer.h"
38#include "llvm/MC/MCSymbol.h"
39#include "llvm/Support/CommandLine.h"
40#include "llvm/Support/ErrorHandling.h"
41#include "llvm/Support/FormattedStream.h"
42#include "llvm/Support/Path.h"
43#include "llvm/Support/TargetRegistry.h"
44#include "llvm/Support/TimeValue.h"
45#include "llvm/Target/Mangler.h"
46#include "llvm/Target/TargetLoweringObjectFile.h"
47#include <sstream>
48using namespace llvm;
49
50#include "NVPTXGenAsmWriter.inc"
51
52bool RegAllocNilUsed = true;
53
54#define DEPOTNAME "__local_depot"
55
56static cl::opt<bool>
57EmitLineNumbers("nvptx-emit-line-numbers",
58                cl::desc("NVPTX Specific: Emit Line numbers even without -G"),
59                cl::init(true));
60
61namespace llvm { bool InterleaveSrcInPtx = false; }
62
63static cl::opt<bool, true>
64InterleaveSrc("nvptx-emit-src", cl::ZeroOrMore,
65              cl::desc("NVPTX Specific: Emit source line in ptx file"),
66              cl::location(llvm::InterleaveSrcInPtx));
67
68namespace {
69/// DiscoverDependentGlobals - Return a set of GlobalVariables on which \p V
70/// depends.
71void DiscoverDependentGlobals(Value *V, DenseSet<GlobalVariable *> &Globals) {
72  if (GlobalVariable *GV = dyn_cast<GlobalVariable>(V))
73    Globals.insert(GV);
74  else {
75    if (User *U = dyn_cast<User>(V)) {
76      for (unsigned i = 0, e = U->getNumOperands(); i != e; ++i) {
77        DiscoverDependentGlobals(U->getOperand(i), Globals);
78      }
79    }
80  }
81}
82
83/// VisitGlobalVariableForEmission - Add \p GV to the list of GlobalVariable
84/// instances to be emitted, but only after any dependents have been added
85/// first.
86void VisitGlobalVariableForEmission(
87    GlobalVariable *GV, SmallVectorImpl<GlobalVariable *> &Order,
88    DenseSet<GlobalVariable *> &Visited, DenseSet<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.count(GV))
95    report_fatal_error("Circular dependency found in global variable set");
96
97  // Start visiting this global
98  Visiting.insert(GV);
99
100  // Make sure we visit all dependents first
101  DenseSet<GlobalVariable *> Others;
102  for (unsigned i = 0, e = GV->getNumOperands(); i != e; ++i)
103    DiscoverDependentGlobals(GV->getOperand(i), Others);
104
105  for (DenseSet<GlobalVariable *>::iterator I = Others.begin(),
106                                            E = Others.end();
107       I != E; ++I)
108    VisitGlobalVariableForEmission(*I, Order, Visited, Visiting);
109
110  // Now we can visit ourself
111  Order.push_back(GV);
112  Visited.insert(GV);
113  Visiting.erase(GV);
114}
115}
116
117// @TODO: This is a copy from AsmPrinter.cpp.  The function is static, so we
118// cannot just link to the existing version.
119/// LowerConstant - Lower the specified LLVM Constant to an MCExpr.
120///
121using namespace nvptx;
122const MCExpr *nvptx::LowerConstant(const Constant *CV, AsmPrinter &AP) {
123  MCContext &Ctx = AP.OutContext;
124
125  if (CV->isNullValue() || isa<UndefValue>(CV))
126    return MCConstantExpr::Create(0, Ctx);
127
128  if (const ConstantInt *CI = dyn_cast<ConstantInt>(CV))
129    return MCConstantExpr::Create(CI->getZExtValue(), Ctx);
130
131  if (const GlobalValue *GV = dyn_cast<GlobalValue>(CV))
132    return MCSymbolRefExpr::Create(AP.Mang->getSymbol(GV), Ctx);
133
134  if (const BlockAddress *BA = dyn_cast<BlockAddress>(CV))
135    return MCSymbolRefExpr::Create(AP.GetBlockAddressSymbol(BA), Ctx);
136
137  const ConstantExpr *CE = dyn_cast<ConstantExpr>(CV);
138  if (CE == 0)
139    llvm_unreachable("Unknown constant value to lower!");
140
141  switch (CE->getOpcode()) {
142  default:
143    // If the code isn't optimized, there may be outstanding folding
144    // opportunities. Attempt to fold the expression using DataLayout as a
145    // last resort before giving up.
146    if (Constant *C = ConstantFoldConstantExpression(CE, AP.TM.getDataLayout()))
147      if (C != CE)
148        return LowerConstant(C, AP);
149
150    // Otherwise report the problem to the user.
151    {
152      std::string S;
153      raw_string_ostream OS(S);
154      OS << "Unsupported expression in static initializer: ";
155      WriteAsOperand(OS, CE, /*PrintType=*/ false,
156                     !AP.MF ? 0 : AP.MF->getFunction()->getParent());
157      report_fatal_error(OS.str());
158    }
159  case Instruction::GetElementPtr: {
160    const DataLayout &TD = *AP.TM.getDataLayout();
161    // Generate a symbolic expression for the byte address
162    APInt OffsetAI(TD.getPointerSizeInBits(), 0);
163    cast<GEPOperator>(CE)->accumulateConstantOffset(TD, OffsetAI);
164
165    const MCExpr *Base = LowerConstant(CE->getOperand(0), AP);
166    if (!OffsetAI)
167      return Base;
168
169    int64_t Offset = OffsetAI.getSExtValue();
170    return MCBinaryExpr::CreateAdd(Base, MCConstantExpr::Create(Offset, Ctx),
171                                   Ctx);
172  }
173
174  case Instruction::Trunc:
175    // We emit the value and depend on the assembler to truncate the generated
176    // expression properly.  This is important for differences between
177    // blockaddress labels.  Since the two labels are in the same function, it
178    // is reasonable to treat their delta as a 32-bit value.
179  // FALL THROUGH.
180  case Instruction::BitCast:
181    return LowerConstant(CE->getOperand(0), AP);
182
183  case Instruction::IntToPtr: {
184    const DataLayout &TD = *AP.TM.getDataLayout();
185    // Handle casts to pointers by changing them into casts to the appropriate
186    // integer type.  This promotes constant folding and simplifies this code.
187    Constant *Op = CE->getOperand(0);
188    Op = ConstantExpr::getIntegerCast(Op, TD.getIntPtrType(CV->getContext()),
189                                      false /*ZExt*/);
190    return LowerConstant(Op, AP);
191  }
192
193  case Instruction::PtrToInt: {
194    const DataLayout &TD = *AP.TM.getDataLayout();
195    // Support only foldable casts to/from pointers that can be eliminated by
196    // changing the pointer to the appropriately sized integer type.
197    Constant *Op = CE->getOperand(0);
198    Type *Ty = CE->getType();
199
200    const MCExpr *OpExpr = LowerConstant(Op, AP);
201
202    // We can emit the pointer value into this slot if the slot is an
203    // integer slot equal to the size of the pointer.
204    if (TD.getTypeAllocSize(Ty) == TD.getTypeAllocSize(Op->getType()))
205      return OpExpr;
206
207    // Otherwise the pointer is smaller than the resultant integer, mask off
208    // the high bits so we are sure to get a proper truncation if the input is
209    // a constant expr.
210    unsigned InBits = TD.getTypeAllocSizeInBits(Op->getType());
211    const MCExpr *MaskExpr =
212        MCConstantExpr::Create(~0ULL >> (64 - InBits), Ctx);
213    return MCBinaryExpr::CreateAnd(OpExpr, MaskExpr, Ctx);
214  }
215
216    // The MC library also has a right-shift operator, but it isn't consistently
217  // signed or unsigned between different targets.
218  case Instruction::Add:
219  case Instruction::Sub:
220  case Instruction::Mul:
221  case Instruction::SDiv:
222  case Instruction::SRem:
223  case Instruction::Shl:
224  case Instruction::And:
225  case Instruction::Or:
226  case Instruction::Xor: {
227    const MCExpr *LHS = LowerConstant(CE->getOperand(0), AP);
228    const MCExpr *RHS = LowerConstant(CE->getOperand(1), AP);
229    switch (CE->getOpcode()) {
230    default:
231      llvm_unreachable("Unknown binary operator constant cast expr");
232    case Instruction::Add:
233      return MCBinaryExpr::CreateAdd(LHS, RHS, Ctx);
234    case Instruction::Sub:
235      return MCBinaryExpr::CreateSub(LHS, RHS, Ctx);
236    case Instruction::Mul:
237      return MCBinaryExpr::CreateMul(LHS, RHS, Ctx);
238    case Instruction::SDiv:
239      return MCBinaryExpr::CreateDiv(LHS, RHS, Ctx);
240    case Instruction::SRem:
241      return MCBinaryExpr::CreateMod(LHS, RHS, Ctx);
242    case Instruction::Shl:
243      return MCBinaryExpr::CreateShl(LHS, RHS, Ctx);
244    case Instruction::And:
245      return MCBinaryExpr::CreateAnd(LHS, RHS, Ctx);
246    case Instruction::Or:
247      return MCBinaryExpr::CreateOr(LHS, RHS, Ctx);
248    case Instruction::Xor:
249      return MCBinaryExpr::CreateXor(LHS, RHS, Ctx);
250    }
251  }
252  }
253}
254
255void NVPTXAsmPrinter::emitLineNumberAsDotLoc(const MachineInstr &MI) {
256  if (!EmitLineNumbers)
257    return;
258  if (ignoreLoc(MI))
259    return;
260
261  DebugLoc curLoc = MI.getDebugLoc();
262
263  if (prevDebugLoc.isUnknown() && curLoc.isUnknown())
264    return;
265
266  if (prevDebugLoc == curLoc)
267    return;
268
269  prevDebugLoc = curLoc;
270
271  if (curLoc.isUnknown())
272    return;
273
274  const MachineFunction *MF = MI.getParent()->getParent();
275  //const TargetMachine &TM = MF->getTarget();
276
277  const LLVMContext &ctx = MF->getFunction()->getContext();
278  DIScope Scope(curLoc.getScope(ctx));
279
280  if (!Scope.Verify())
281    return;
282
283  StringRef fileName(Scope.getFilename());
284  StringRef dirName(Scope.getDirectory());
285  SmallString<128> FullPathName = dirName;
286  if (!dirName.empty() && !sys::path::is_absolute(fileName)) {
287    sys::path::append(FullPathName, fileName);
288    fileName = FullPathName.str();
289  }
290
291  if (filenameMap.find(fileName.str()) == filenameMap.end())
292    return;
293
294  // Emit the line from the source file.
295  if (llvm::InterleaveSrcInPtx)
296    this->emitSrcInText(fileName.str(), curLoc.getLine());
297
298  std::stringstream temp;
299  temp << "\t.loc " << filenameMap[fileName.str()] << " " << curLoc.getLine()
300       << " " << curLoc.getCol();
301  OutStreamer.EmitRawText(Twine(temp.str().c_str()));
302}
303
304void NVPTXAsmPrinter::EmitInstruction(const MachineInstr *MI) {
305  SmallString<128> Str;
306  raw_svector_ostream OS(Str);
307  if (nvptxSubtarget.getDrvInterface() == NVPTX::CUDA)
308    emitLineNumberAsDotLoc(*MI);
309  printInstruction(MI, OS);
310  OutStreamer.EmitRawText(OS.str());
311}
312
313void NVPTXAsmPrinter::printReturnValStr(const Function *F, raw_ostream &O) {
314  const DataLayout *TD = TM.getDataLayout();
315  const TargetLowering *TLI = TM.getTargetLowering();
316
317  Type *Ty = F->getReturnType();
318
319  bool isABI = (nvptxSubtarget.getSmVersion() >= 20);
320
321  if (Ty->getTypeID() == Type::VoidTyID)
322    return;
323
324  O << " (";
325
326  if (isABI) {
327    if (Ty->isPrimitiveType() || Ty->isIntegerTy()) {
328      unsigned size = 0;
329      if (const IntegerType *ITy = dyn_cast<IntegerType>(Ty)) {
330        size = ITy->getBitWidth();
331        if (size < 32)
332          size = 32;
333      } else {
334        assert(Ty->isFloatingPointTy() && "Floating point type expected here");
335        size = Ty->getPrimitiveSizeInBits();
336      }
337
338      O << ".param .b" << size << " func_retval0";
339    } else if (isa<PointerType>(Ty)) {
340      O << ".param .b" << TLI->getPointerTy().getSizeInBits()
341        << " func_retval0";
342    } else {
343      if ((Ty->getTypeID() == Type::StructTyID) || isa<VectorType>(Ty)) {
344        SmallVector<EVT, 16> vtparts;
345        ComputeValueVTs(*TLI, Ty, vtparts);
346        unsigned totalsz = 0;
347        for (unsigned i = 0, e = vtparts.size(); i != e; ++i) {
348          unsigned elems = 1;
349          EVT elemtype = vtparts[i];
350          if (vtparts[i].isVector()) {
351            elems = vtparts[i].getVectorNumElements();
352            elemtype = vtparts[i].getVectorElementType();
353          }
354          for (unsigned j = 0, je = elems; j != je; ++j) {
355            unsigned sz = elemtype.getSizeInBits();
356            if (elemtype.isInteger() && (sz < 8))
357              sz = 8;
358            totalsz += sz / 8;
359          }
360        }
361        unsigned retAlignment = 0;
362        if (!llvm::getAlign(*F, 0, retAlignment))
363          retAlignment = TD->getABITypeAlignment(Ty);
364        O << ".param .align " << retAlignment << " .b8 func_retval0[" << totalsz
365          << "]";
366      } else
367        assert(false && "Unknown return type");
368    }
369  } else {
370    SmallVector<EVT, 16> vtparts;
371    ComputeValueVTs(*TLI, Ty, vtparts);
372    unsigned idx = 0;
373    for (unsigned i = 0, e = vtparts.size(); i != e; ++i) {
374      unsigned elems = 1;
375      EVT elemtype = vtparts[i];
376      if (vtparts[i].isVector()) {
377        elems = vtparts[i].getVectorNumElements();
378        elemtype = vtparts[i].getVectorElementType();
379      }
380
381      for (unsigned j = 0, je = elems; j != je; ++j) {
382        unsigned sz = elemtype.getSizeInBits();
383        if (elemtype.isInteger() && (sz < 32))
384          sz = 32;
385        O << ".reg .b" << sz << " func_retval" << idx;
386        if (j < je - 1)
387          O << ", ";
388        ++idx;
389      }
390      if (i < e - 1)
391        O << ", ";
392    }
393  }
394  O << ") ";
395  return;
396}
397
398void NVPTXAsmPrinter::printReturnValStr(const MachineFunction &MF,
399                                        raw_ostream &O) {
400  const Function *F = MF.getFunction();
401  printReturnValStr(F, O);
402}
403
404void NVPTXAsmPrinter::EmitFunctionEntryLabel() {
405  SmallString<128> Str;
406  raw_svector_ostream O(Str);
407
408  // Set up
409  MRI = &MF->getRegInfo();
410  F = MF->getFunction();
411  emitLinkageDirective(F, O);
412  if (llvm::isKernelFunction(*F))
413    O << ".entry ";
414  else {
415    O << ".func ";
416    printReturnValStr(*MF, O);
417  }
418
419  O << *CurrentFnSym;
420
421  emitFunctionParamList(*MF, O);
422
423  if (llvm::isKernelFunction(*F))
424    emitKernelFunctionDirectives(*F, O);
425
426  OutStreamer.EmitRawText(O.str());
427
428  prevDebugLoc = DebugLoc();
429}
430
431void NVPTXAsmPrinter::EmitFunctionBodyStart() {
432  const TargetRegisterInfo &TRI = *TM.getRegisterInfo();
433  unsigned numRegClasses = TRI.getNumRegClasses();
434  VRidGlobal2LocalMap = new std::map<unsigned, unsigned>[numRegClasses + 1];
435  OutStreamer.EmitRawText(StringRef("{\n"));
436  setAndEmitFunctionVirtualRegisters(*MF);
437
438  SmallString<128> Str;
439  raw_svector_ostream O(Str);
440  emitDemotedVars(MF->getFunction(), O);
441  OutStreamer.EmitRawText(O.str());
442}
443
444void NVPTXAsmPrinter::EmitFunctionBodyEnd() {
445  OutStreamer.EmitRawText(StringRef("}\n"));
446  delete[] VRidGlobal2LocalMap;
447}
448
449void NVPTXAsmPrinter::emitKernelFunctionDirectives(const Function &F,
450                                                   raw_ostream &O) const {
451  // If the NVVM IR has some of reqntid* specified, then output
452  // the reqntid directive, and set the unspecified ones to 1.
453  // If none of reqntid* is specified, don't output reqntid directive.
454  unsigned reqntidx, reqntidy, reqntidz;
455  bool specified = false;
456  if (llvm::getReqNTIDx(F, reqntidx) == false)
457    reqntidx = 1;
458  else
459    specified = true;
460  if (llvm::getReqNTIDy(F, reqntidy) == false)
461    reqntidy = 1;
462  else
463    specified = true;
464  if (llvm::getReqNTIDz(F, reqntidz) == false)
465    reqntidz = 1;
466  else
467    specified = true;
468
469  if (specified)
470    O << ".reqntid " << reqntidx << ", " << reqntidy << ", " << reqntidz
471      << "\n";
472
473  // If the NVVM IR has some of maxntid* specified, then output
474  // the maxntid directive, and set the unspecified ones to 1.
475  // If none of maxntid* is specified, don't output maxntid directive.
476  unsigned maxntidx, maxntidy, maxntidz;
477  specified = false;
478  if (llvm::getMaxNTIDx(F, maxntidx) == false)
479    maxntidx = 1;
480  else
481    specified = true;
482  if (llvm::getMaxNTIDy(F, maxntidy) == false)
483    maxntidy = 1;
484  else
485    specified = true;
486  if (llvm::getMaxNTIDz(F, maxntidz) == false)
487    maxntidz = 1;
488  else
489    specified = true;
490
491  if (specified)
492    O << ".maxntid " << maxntidx << ", " << maxntidy << ", " << maxntidz
493      << "\n";
494
495  unsigned mincta;
496  if (llvm::getMinCTASm(F, mincta))
497    O << ".minnctapersm " << mincta << "\n";
498}
499
500void NVPTXAsmPrinter::getVirtualRegisterName(unsigned vr, bool isVec,
501                                             raw_ostream &O) {
502  const TargetRegisterClass *RC = MRI->getRegClass(vr);
503  unsigned id = RC->getID();
504
505  std::map<unsigned, unsigned> &regmap = VRidGlobal2LocalMap[id];
506  unsigned mapped_vr = regmap[vr];
507
508  if (!isVec) {
509    O << getNVPTXRegClassStr(RC) << mapped_vr;
510    return;
511  }
512  report_fatal_error("Bad register!");
513}
514
515void NVPTXAsmPrinter::emitVirtualRegister(unsigned int vr, bool isVec,
516                                          raw_ostream &O) {
517  getVirtualRegisterName(vr, isVec, O);
518}
519
520void NVPTXAsmPrinter::printVecModifiedImmediate(
521    const MachineOperand &MO, const char *Modifier, raw_ostream &O) {
522  static const char vecelem[] = { '0', '1', '2', '3', '0', '1', '2', '3' };
523  int Imm = (int) MO.getImm();
524  if (0 == strcmp(Modifier, "vecelem"))
525    O << "_" << vecelem[Imm];
526  else if (0 == strcmp(Modifier, "vecv4comm1")) {
527    if ((Imm < 0) || (Imm > 3))
528      O << "//";
529  } else if (0 == strcmp(Modifier, "vecv4comm2")) {
530    if ((Imm < 4) || (Imm > 7))
531      O << "//";
532  } else if (0 == strcmp(Modifier, "vecv4pos")) {
533    if (Imm < 0)
534      Imm = 0;
535    O << "_" << vecelem[Imm % 4];
536  } else if (0 == strcmp(Modifier, "vecv2comm1")) {
537    if ((Imm < 0) || (Imm > 1))
538      O << "//";
539  } else if (0 == strcmp(Modifier, "vecv2comm2")) {
540    if ((Imm < 2) || (Imm > 3))
541      O << "//";
542  } else if (0 == strcmp(Modifier, "vecv2pos")) {
543    if (Imm < 0)
544      Imm = 0;
545    O << "_" << vecelem[Imm % 2];
546  } else
547    llvm_unreachable("Unknown Modifier on immediate operand");
548}
549
550void NVPTXAsmPrinter::printOperand(const MachineInstr *MI, int opNum,
551                                   raw_ostream &O, const char *Modifier) {
552  const MachineOperand &MO = MI->getOperand(opNum);
553  switch (MO.getType()) {
554  case MachineOperand::MO_Register:
555    if (TargetRegisterInfo::isPhysicalRegister(MO.getReg())) {
556      if (MO.getReg() == NVPTX::VRDepot)
557        O << DEPOTNAME << getFunctionNumber();
558      else
559        O << getRegisterName(MO.getReg());
560    } else {
561      if (!Modifier)
562        emitVirtualRegister(MO.getReg(), false, O);
563      else {
564        if (strcmp(Modifier, "vecfull") == 0)
565          emitVirtualRegister(MO.getReg(), true, O);
566        else
567          llvm_unreachable(
568              "Don't know how to handle the modifier on virtual register.");
569      }
570    }
571    return;
572
573  case MachineOperand::MO_Immediate:
574    if (!Modifier)
575      O << MO.getImm();
576    else if (strstr(Modifier, "vec") == Modifier)
577      printVecModifiedImmediate(MO, Modifier, O);
578    else
579      llvm_unreachable(
580          "Don't know how to handle modifier on immediate operand");
581    return;
582
583  case MachineOperand::MO_FPImmediate:
584    printFPConstant(MO.getFPImm(), O);
585    break;
586
587  case MachineOperand::MO_GlobalAddress:
588    O << *Mang->getSymbol(MO.getGlobal());
589    break;
590
591  case MachineOperand::MO_ExternalSymbol: {
592    const char *symbname = MO.getSymbolName();
593    if (strstr(symbname, ".PARAM") == symbname) {
594      unsigned index;
595      sscanf(symbname + 6, "%u[];", &index);
596      printParamName(index, O);
597    } else if (strstr(symbname, ".HLPPARAM") == symbname) {
598      unsigned index;
599      sscanf(symbname + 9, "%u[];", &index);
600      O << *CurrentFnSym << "_param_" << index << "_offset";
601    } else
602      O << symbname;
603    break;
604  }
605
606  case MachineOperand::MO_MachineBasicBlock:
607    O << *MO.getMBB()->getSymbol();
608    return;
609
610  default:
611    llvm_unreachable("Operand type not supported.");
612  }
613}
614
615void NVPTXAsmPrinter::printImplicitDef(const MachineInstr *MI,
616                                       raw_ostream &O) const {
617#ifndef __OPTIMIZE__
618  O << "\t// Implicit def :";
619  //printOperand(MI, 0);
620  O << "\n";
621#endif
622}
623
624void NVPTXAsmPrinter::printMemOperand(const MachineInstr *MI, int opNum,
625                                      raw_ostream &O, const char *Modifier) {
626  printOperand(MI, opNum, O);
627
628  if (Modifier && !strcmp(Modifier, "add")) {
629    O << ", ";
630    printOperand(MI, opNum + 1, O);
631  } else {
632    if (MI->getOperand(opNum + 1).isImm() &&
633        MI->getOperand(opNum + 1).getImm() == 0)
634      return; // don't print ',0' or '+0'
635    O << "+";
636    printOperand(MI, opNum + 1, O);
637  }
638}
639
640void NVPTXAsmPrinter::printLdStCode(const MachineInstr *MI, int opNum,
641                                    raw_ostream &O, const char *Modifier) {
642  if (Modifier) {
643    const MachineOperand &MO = MI->getOperand(opNum);
644    int Imm = (int) MO.getImm();
645    if (!strcmp(Modifier, "volatile")) {
646      if (Imm)
647        O << ".volatile";
648    } else if (!strcmp(Modifier, "addsp")) {
649      switch (Imm) {
650      case NVPTX::PTXLdStInstCode::GLOBAL:
651        O << ".global";
652        break;
653      case NVPTX::PTXLdStInstCode::SHARED:
654        O << ".shared";
655        break;
656      case NVPTX::PTXLdStInstCode::LOCAL:
657        O << ".local";
658        break;
659      case NVPTX::PTXLdStInstCode::PARAM:
660        O << ".param";
661        break;
662      case NVPTX::PTXLdStInstCode::CONSTANT:
663        O << ".const";
664        break;
665      case NVPTX::PTXLdStInstCode::GENERIC:
666        if (!nvptxSubtarget.hasGenericLdSt())
667          O << ".global";
668        break;
669      default:
670        llvm_unreachable("Wrong Address Space");
671      }
672    } else if (!strcmp(Modifier, "sign")) {
673      if (Imm == NVPTX::PTXLdStInstCode::Signed)
674        O << "s";
675      else if (Imm == NVPTX::PTXLdStInstCode::Unsigned)
676        O << "u";
677      else
678        O << "f";
679    } else if (!strcmp(Modifier, "vec")) {
680      if (Imm == NVPTX::PTXLdStInstCode::V2)
681        O << ".v2";
682      else if (Imm == NVPTX::PTXLdStInstCode::V4)
683        O << ".v4";
684    } else
685      llvm_unreachable("Unknown Modifier");
686  } else
687    llvm_unreachable("Empty Modifier");
688}
689
690void NVPTXAsmPrinter::emitDeclaration(const Function *F, raw_ostream &O) {
691
692  emitLinkageDirective(F, O);
693  if (llvm::isKernelFunction(*F))
694    O << ".entry ";
695  else
696    O << ".func ";
697  printReturnValStr(F, O);
698  O << *CurrentFnSym << "\n";
699  emitFunctionParamList(F, O);
700  O << ";\n";
701}
702
703static bool usedInGlobalVarDef(const Constant *C) {
704  if (!C)
705    return false;
706
707  if (const GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
708    if (GV->getName().str() == "llvm.used")
709      return false;
710    return true;
711  }
712
713  for (Value::const_use_iterator ui = C->use_begin(), ue = C->use_end();
714       ui != ue; ++ui) {
715    const Constant *C = dyn_cast<Constant>(*ui);
716    if (usedInGlobalVarDef(C))
717      return true;
718  }
719  return false;
720}
721
722static bool usedInOneFunc(const User *U, Function const *&oneFunc) {
723  if (const GlobalVariable *othergv = dyn_cast<GlobalVariable>(U)) {
724    if (othergv->getName().str() == "llvm.used")
725      return true;
726  }
727
728  if (const Instruction *instr = dyn_cast<Instruction>(U)) {
729    if (instr->getParent() && instr->getParent()->getParent()) {
730      const Function *curFunc = instr->getParent()->getParent();
731      if (oneFunc && (curFunc != oneFunc))
732        return false;
733      oneFunc = curFunc;
734      return true;
735    } else
736      return false;
737  }
738
739  if (const MDNode *md = dyn_cast<MDNode>(U))
740    if (md->hasName() && ((md->getName().str() == "llvm.dbg.gv") ||
741                          (md->getName().str() == "llvm.dbg.sp")))
742      return true;
743
744  for (User::const_use_iterator ui = U->use_begin(), ue = U->use_end();
745       ui != ue; ++ui) {
746    if (usedInOneFunc(*ui, oneFunc) == false)
747      return false;
748  }
749  return true;
750}
751
752/* Find out if a global variable can be demoted to local scope.
753 * Currently, this is valid for CUDA shared variables, which have local
754 * scope and global lifetime. So the conditions to check are :
755 * 1. Is the global variable in shared address space?
756 * 2. Does it have internal linkage?
757 * 3. Is the global variable referenced only in one function?
758 */
759static bool canDemoteGlobalVar(const GlobalVariable *gv, Function const *&f) {
760  if (gv->hasInternalLinkage() == false)
761    return false;
762  const PointerType *Pty = gv->getType();
763  if (Pty->getAddressSpace() != llvm::ADDRESS_SPACE_SHARED)
764    return false;
765
766  const Function *oneFunc = 0;
767
768  bool flag = usedInOneFunc(gv, oneFunc);
769  if (flag == false)
770    return false;
771  if (!oneFunc)
772    return false;
773  f = oneFunc;
774  return true;
775}
776
777static bool useFuncSeen(const Constant *C,
778                        llvm::DenseMap<const Function *, bool> &seenMap) {
779  for (Value::const_use_iterator ui = C->use_begin(), ue = C->use_end();
780       ui != ue; ++ui) {
781    if (const Constant *cu = dyn_cast<Constant>(*ui)) {
782      if (useFuncSeen(cu, seenMap))
783        return true;
784    } else if (const Instruction *I = dyn_cast<Instruction>(*ui)) {
785      const BasicBlock *bb = I->getParent();
786      if (!bb)
787        continue;
788      const Function *caller = bb->getParent();
789      if (!caller)
790        continue;
791      if (seenMap.find(caller) != seenMap.end())
792        return true;
793    }
794  }
795  return false;
796}
797
798void NVPTXAsmPrinter::emitDeclarations(Module &M, raw_ostream &O) {
799  llvm::DenseMap<const Function *, bool> seenMap;
800  for (Module::const_iterator FI = M.begin(), FE = M.end(); FI != FE; ++FI) {
801    const Function *F = FI;
802
803    if (F->isDeclaration()) {
804      if (F->use_empty())
805        continue;
806      if (F->getIntrinsicID())
807        continue;
808      CurrentFnSym = Mang->getSymbol(F);
809      emitDeclaration(F, O);
810      continue;
811    }
812    for (Value::const_use_iterator iter = F->use_begin(),
813                                   iterEnd = F->use_end();
814         iter != iterEnd; ++iter) {
815      if (const Constant *C = dyn_cast<Constant>(*iter)) {
816        if (usedInGlobalVarDef(C)) {
817          // The use is in the initialization of a global variable
818          // that is a function pointer, so print a declaration
819          // for the original function
820          CurrentFnSym = Mang->getSymbol(F);
821          emitDeclaration(F, O);
822          break;
823        }
824        // Emit a declaration of this function if the function that
825        // uses this constant expr has already been seen.
826        if (useFuncSeen(C, seenMap)) {
827          CurrentFnSym = Mang->getSymbol(F);
828          emitDeclaration(F, O);
829          break;
830        }
831      }
832
833      if (!isa<Instruction>(*iter))
834        continue;
835      const Instruction *instr = cast<Instruction>(*iter);
836      const BasicBlock *bb = instr->getParent();
837      if (!bb)
838        continue;
839      const Function *caller = bb->getParent();
840      if (!caller)
841        continue;
842
843      // If a caller has already been seen, then the caller is
844      // appearing in the module before the callee. so print out
845      // a declaration for the callee.
846      if (seenMap.find(caller) != seenMap.end()) {
847        CurrentFnSym = Mang->getSymbol(F);
848        emitDeclaration(F, O);
849        break;
850      }
851    }
852    seenMap[F] = true;
853  }
854}
855
856void NVPTXAsmPrinter::recordAndEmitFilenames(Module &M) {
857  DebugInfoFinder DbgFinder;
858  DbgFinder.processModule(M);
859
860  unsigned i = 1;
861  for (DebugInfoFinder::iterator I = DbgFinder.compile_unit_begin(),
862                                 E = DbgFinder.compile_unit_end();
863       I != E; ++I) {
864    DICompileUnit DIUnit(*I);
865    StringRef Filename(DIUnit.getFilename());
866    StringRef Dirname(DIUnit.getDirectory());
867    SmallString<128> FullPathName = Dirname;
868    if (!Dirname.empty() && !sys::path::is_absolute(Filename)) {
869      sys::path::append(FullPathName, Filename);
870      Filename = FullPathName.str();
871    }
872    if (filenameMap.find(Filename.str()) != filenameMap.end())
873      continue;
874    filenameMap[Filename.str()] = i;
875    OutStreamer.EmitDwarfFileDirective(i, "", Filename.str());
876    ++i;
877  }
878
879  for (DebugInfoFinder::iterator I = DbgFinder.subprogram_begin(),
880                                 E = DbgFinder.subprogram_end();
881       I != E; ++I) {
882    DISubprogram SP(*I);
883    StringRef Filename(SP.getFilename());
884    StringRef Dirname(SP.getDirectory());
885    SmallString<128> FullPathName = Dirname;
886    if (!Dirname.empty() && !sys::path::is_absolute(Filename)) {
887      sys::path::append(FullPathName, Filename);
888      Filename = FullPathName.str();
889    }
890    if (filenameMap.find(Filename.str()) != filenameMap.end())
891      continue;
892    filenameMap[Filename.str()] = i;
893    ++i;
894  }
895}
896
897bool NVPTXAsmPrinter::doInitialization(Module &M) {
898
899  SmallString<128> Str1;
900  raw_svector_ostream OS1(Str1);
901
902  MMI = getAnalysisIfAvailable<MachineModuleInfo>();
903  MMI->AnalyzeModule(M);
904
905  // We need to call the parent's one explicitly.
906  //bool Result = AsmPrinter::doInitialization(M);
907
908  // Initialize TargetLoweringObjectFile.
909  const_cast<TargetLoweringObjectFile &>(getObjFileLowering())
910      .Initialize(OutContext, TM);
911
912  Mang = new Mangler(OutContext, *TM.getDataLayout());
913
914  // Emit header before any dwarf directives are emitted below.
915  emitHeader(M, OS1);
916  OutStreamer.EmitRawText(OS1.str());
917
918  // Already commented out
919  //bool Result = AsmPrinter::doInitialization(M);
920
921  if (nvptxSubtarget.getDrvInterface() == NVPTX::CUDA)
922    recordAndEmitFilenames(M);
923
924  SmallString<128> Str2;
925  raw_svector_ostream OS2(Str2);
926
927  emitDeclarations(M, OS2);
928
929  // As ptxas does not support forward references of globals, we need to first
930  // sort the list of module-level globals in def-use order. We visit each
931  // global variable in order, and ensure that we emit it *after* its dependent
932  // globals. We use a little extra memory maintaining both a set and a list to
933  // have fast searches while maintaining a strict ordering.
934  SmallVector<GlobalVariable *, 8> Globals;
935  DenseSet<GlobalVariable *> GVVisited;
936  DenseSet<GlobalVariable *> GVVisiting;
937
938  // Visit each global variable, in order
939  for (Module::global_iterator I = M.global_begin(), E = M.global_end(); I != E;
940       ++I)
941    VisitGlobalVariableForEmission(I, Globals, GVVisited, GVVisiting);
942
943  assert(GVVisited.size() == M.getGlobalList().size() &&
944         "Missed a global variable");
945  assert(GVVisiting.size() == 0 && "Did not fully process a global variable");
946
947  // Print out module-level global variables in proper order
948  for (unsigned i = 0, e = Globals.size(); i != e; ++i)
949    printModuleLevelGV(Globals[i], OS2);
950
951  OS2 << '\n';
952
953  OutStreamer.EmitRawText(OS2.str());
954  return false; // success
955}
956
957void NVPTXAsmPrinter::emitHeader(Module &M, raw_ostream &O) {
958  O << "//\n";
959  O << "// Generated by LLVM NVPTX Back-End\n";
960  O << "//\n";
961  O << "\n";
962
963  unsigned PTXVersion = nvptxSubtarget.getPTXVersion();
964  O << ".version " << (PTXVersion / 10) << "." << (PTXVersion % 10) << "\n";
965
966  O << ".target ";
967  O << nvptxSubtarget.getTargetName();
968
969  if (nvptxSubtarget.getDrvInterface() == NVPTX::NVCL)
970    O << ", texmode_independent";
971  if (nvptxSubtarget.getDrvInterface() == NVPTX::CUDA) {
972    if (!nvptxSubtarget.hasDouble())
973      O << ", map_f64_to_f32";
974  }
975
976  if (MAI->doesSupportDebugInformation())
977    O << ", debug";
978
979  O << "\n";
980
981  O << ".address_size ";
982  if (nvptxSubtarget.is64Bit())
983    O << "64";
984  else
985    O << "32";
986  O << "\n";
987
988  O << "\n";
989}
990
991bool NVPTXAsmPrinter::doFinalization(Module &M) {
992  // XXX Temproarily remove global variables so that doFinalization() will not
993  // emit them again (global variables are emitted at beginning).
994
995  Module::GlobalListType &global_list = M.getGlobalList();
996  int i, n = global_list.size();
997  GlobalVariable **gv_array = new GlobalVariable *[n];
998
999  // first, back-up GlobalVariable in gv_array
1000  i = 0;
1001  for (Module::global_iterator I = global_list.begin(), E = global_list.end();
1002       I != E; ++I)
1003    gv_array[i++] = &*I;
1004
1005  // second, empty global_list
1006  while (!global_list.empty())
1007    global_list.remove(global_list.begin());
1008
1009  // call doFinalization
1010  bool ret = AsmPrinter::doFinalization(M);
1011
1012  // now we restore global variables
1013  for (i = 0; i < n; i++)
1014    global_list.insert(global_list.end(), gv_array[i]);
1015
1016  delete[] gv_array;
1017  return ret;
1018
1019  //bool Result = AsmPrinter::doFinalization(M);
1020  // Instead of calling the parents doFinalization, we may
1021  // clone parents doFinalization and customize here.
1022  // Currently, we if NVISA out the EmitGlobals() in
1023  // parent's doFinalization, which is too intrusive.
1024  //
1025  // Same for the doInitialization.
1026  //return Result;
1027}
1028
1029// This function emits appropriate linkage directives for
1030// functions and global variables.
1031//
1032// extern function declaration            -> .extern
1033// extern function definition             -> .visible
1034// external global variable with init     -> .visible
1035// external without init                  -> .extern
1036// appending                              -> not allowed, assert.
1037
1038void NVPTXAsmPrinter::emitLinkageDirective(const GlobalValue *V,
1039                                           raw_ostream &O) {
1040  if (nvptxSubtarget.getDrvInterface() == NVPTX::CUDA) {
1041    if (V->hasExternalLinkage()) {
1042      if (isa<GlobalVariable>(V)) {
1043        const GlobalVariable *GVar = cast<GlobalVariable>(V);
1044        if (GVar) {
1045          if (GVar->hasInitializer())
1046            O << ".visible ";
1047          else
1048            O << ".extern ";
1049        }
1050      } else if (V->isDeclaration())
1051        O << ".extern ";
1052      else
1053        O << ".visible ";
1054    } else if (V->hasAppendingLinkage()) {
1055      std::string msg;
1056      msg.append("Error: ");
1057      msg.append("Symbol ");
1058      if (V->hasName())
1059        msg.append(V->getName().str());
1060      msg.append("has unsupported appending linkage type");
1061      llvm_unreachable(msg.c_str());
1062    }
1063  }
1064}
1065
1066void NVPTXAsmPrinter::printModuleLevelGV(GlobalVariable *GVar, raw_ostream &O,
1067                                         bool processDemoted) {
1068
1069  // Skip meta data
1070  if (GVar->hasSection()) {
1071    if (GVar->getSection() == "llvm.metadata")
1072      return;
1073  }
1074
1075  const DataLayout *TD = TM.getDataLayout();
1076
1077  // GlobalVariables are always constant pointers themselves.
1078  const PointerType *PTy = GVar->getType();
1079  Type *ETy = PTy->getElementType();
1080
1081  if (GVar->hasExternalLinkage()) {
1082    if (GVar->hasInitializer())
1083      O << ".visible ";
1084    else
1085      O << ".extern ";
1086  }
1087
1088  if (llvm::isTexture(*GVar)) {
1089    O << ".global .texref " << llvm::getTextureName(*GVar) << ";\n";
1090    return;
1091  }
1092
1093  if (llvm::isSurface(*GVar)) {
1094    O << ".global .surfref " << llvm::getSurfaceName(*GVar) << ";\n";
1095    return;
1096  }
1097
1098  if (GVar->isDeclaration()) {
1099    // (extern) declarations, no definition or initializer
1100    // Currently the only known declaration is for an automatic __local
1101    // (.shared) promoted to global.
1102    emitPTXGlobalVariable(GVar, O);
1103    O << ";\n";
1104    return;
1105  }
1106
1107  if (llvm::isSampler(*GVar)) {
1108    O << ".global .samplerref " << llvm::getSamplerName(*GVar);
1109
1110    Constant *Initializer = NULL;
1111    if (GVar->hasInitializer())
1112      Initializer = GVar->getInitializer();
1113    ConstantInt *CI = NULL;
1114    if (Initializer)
1115      CI = dyn_cast<ConstantInt>(Initializer);
1116    if (CI) {
1117      unsigned sample = CI->getZExtValue();
1118
1119      O << " = { ";
1120
1121      for (int i = 0,
1122               addr = ((sample & __CLK_ADDRESS_MASK) >> __CLK_ADDRESS_BASE);
1123           i < 3; i++) {
1124        O << "addr_mode_" << i << " = ";
1125        switch (addr) {
1126        case 0:
1127          O << "wrap";
1128          break;
1129        case 1:
1130          O << "clamp_to_border";
1131          break;
1132        case 2:
1133          O << "clamp_to_edge";
1134          break;
1135        case 3:
1136          O << "wrap";
1137          break;
1138        case 4:
1139          O << "mirror";
1140          break;
1141        }
1142        O << ", ";
1143      }
1144      O << "filter_mode = ";
1145      switch ((sample & __CLK_FILTER_MASK) >> __CLK_FILTER_BASE) {
1146      case 0:
1147        O << "nearest";
1148        break;
1149      case 1:
1150        O << "linear";
1151        break;
1152      case 2:
1153        assert(0 && "Anisotropic filtering is not supported");
1154      default:
1155        O << "nearest";
1156        break;
1157      }
1158      if (!((sample & __CLK_NORMALIZED_MASK) >> __CLK_NORMALIZED_BASE)) {
1159        O << ", force_unnormalized_coords = 1";
1160      }
1161      O << " }";
1162    }
1163
1164    O << ";\n";
1165    return;
1166  }
1167
1168  if (GVar->hasPrivateLinkage()) {
1169
1170    if (!strncmp(GVar->getName().data(), "unrollpragma", 12))
1171      return;
1172
1173    // FIXME - need better way (e.g. Metadata) to avoid generating this global
1174    if (!strncmp(GVar->getName().data(), "filename", 8))
1175      return;
1176    if (GVar->use_empty())
1177      return;
1178  }
1179
1180  const Function *demotedFunc = 0;
1181  if (!processDemoted && canDemoteGlobalVar(GVar, demotedFunc)) {
1182    O << "// " << GVar->getName().str() << " has been demoted\n";
1183    if (localDecls.find(demotedFunc) != localDecls.end())
1184      localDecls[demotedFunc].push_back(GVar);
1185    else {
1186      std::vector<GlobalVariable *> temp;
1187      temp.push_back(GVar);
1188      localDecls[demotedFunc] = temp;
1189    }
1190    return;
1191  }
1192
1193  O << ".";
1194  emitPTXAddressSpace(PTy->getAddressSpace(), O);
1195  if (GVar->getAlignment() == 0)
1196    O << " .align " << (int) TD->getPrefTypeAlignment(ETy);
1197  else
1198    O << " .align " << GVar->getAlignment();
1199
1200  if (ETy->isPrimitiveType() || ETy->isIntegerTy() || isa<PointerType>(ETy)) {
1201    O << " .";
1202    O << getPTXFundamentalTypeStr(ETy, false);
1203    O << " ";
1204    O << *Mang->getSymbol(GVar);
1205
1206    // Ptx allows variable initilization only for constant and global state
1207    // spaces.
1208    if (((PTy->getAddressSpace() == llvm::ADDRESS_SPACE_GLOBAL) ||
1209         (PTy->getAddressSpace() == llvm::ADDRESS_SPACE_CONST_NOT_GEN) ||
1210         (PTy->getAddressSpace() == llvm::ADDRESS_SPACE_CONST)) &&
1211        GVar->hasInitializer()) {
1212      Constant *Initializer = GVar->getInitializer();
1213      if (!Initializer->isNullValue()) {
1214        O << " = ";
1215        printScalarConstant(Initializer, O);
1216      }
1217    }
1218  } else {
1219    unsigned int ElementSize = 0;
1220
1221    // Although PTX has direct support for struct type and array type and
1222    // LLVM IR is very similar to PTX, the LLVM CodeGen does not support for
1223    // targets that support these high level field accesses. Structs, arrays
1224    // and vectors are lowered into arrays of bytes.
1225    switch (ETy->getTypeID()) {
1226    case Type::StructTyID:
1227    case Type::ArrayTyID:
1228    case Type::VectorTyID:
1229      ElementSize = TD->getTypeStoreSize(ETy);
1230      // Ptx allows variable initilization only for constant and
1231      // global state spaces.
1232      if (((PTy->getAddressSpace() == llvm::ADDRESS_SPACE_GLOBAL) ||
1233           (PTy->getAddressSpace() == llvm::ADDRESS_SPACE_CONST_NOT_GEN) ||
1234           (PTy->getAddressSpace() == llvm::ADDRESS_SPACE_CONST)) &&
1235          GVar->hasInitializer()) {
1236        Constant *Initializer = GVar->getInitializer();
1237        if (!isa<UndefValue>(Initializer) && !Initializer->isNullValue()) {
1238          AggBuffer aggBuffer(ElementSize, O, *this);
1239          bufferAggregateConstant(Initializer, &aggBuffer);
1240          if (aggBuffer.numSymbols) {
1241            if (nvptxSubtarget.is64Bit()) {
1242              O << " .u64 " << *Mang->getSymbol(GVar) << "[";
1243              O << ElementSize / 8;
1244            } else {
1245              O << " .u32 " << *Mang->getSymbol(GVar) << "[";
1246              O << ElementSize / 4;
1247            }
1248            O << "]";
1249          } else {
1250            O << " .b8 " << *Mang->getSymbol(GVar) << "[";
1251            O << ElementSize;
1252            O << "]";
1253          }
1254          O << " = {";
1255          aggBuffer.print();
1256          O << "}";
1257        } else {
1258          O << " .b8 " << *Mang->getSymbol(GVar);
1259          if (ElementSize) {
1260            O << "[";
1261            O << ElementSize;
1262            O << "]";
1263          }
1264        }
1265      } else {
1266        O << " .b8 " << *Mang->getSymbol(GVar);
1267        if (ElementSize) {
1268          O << "[";
1269          O << ElementSize;
1270          O << "]";
1271        }
1272      }
1273      break;
1274    default:
1275      assert(0 && "type not supported yet");
1276    }
1277
1278  }
1279  O << ";\n";
1280}
1281
1282void NVPTXAsmPrinter::emitDemotedVars(const Function *f, raw_ostream &O) {
1283  if (localDecls.find(f) == localDecls.end())
1284    return;
1285
1286  std::vector<GlobalVariable *> &gvars = localDecls[f];
1287
1288  for (unsigned i = 0, e = gvars.size(); i != e; ++i) {
1289    O << "\t// demoted variable\n\t";
1290    printModuleLevelGV(gvars[i], O, true);
1291  }
1292}
1293
1294void NVPTXAsmPrinter::emitPTXAddressSpace(unsigned int AddressSpace,
1295                                          raw_ostream &O) const {
1296  switch (AddressSpace) {
1297  case llvm::ADDRESS_SPACE_LOCAL:
1298    O << "local";
1299    break;
1300  case llvm::ADDRESS_SPACE_GLOBAL:
1301    O << "global";
1302    break;
1303  case llvm::ADDRESS_SPACE_CONST:
1304    // This logic should be consistent with that in
1305    // getCodeAddrSpace() (NVPTXISelDATToDAT.cpp)
1306    if (nvptxSubtarget.hasGenericLdSt())
1307      O << "global";
1308    else
1309      O << "const";
1310    break;
1311  case llvm::ADDRESS_SPACE_CONST_NOT_GEN:
1312    O << "const";
1313    break;
1314  case llvm::ADDRESS_SPACE_SHARED:
1315    O << "shared";
1316    break;
1317  default:
1318    report_fatal_error("Bad address space found while emitting PTX");
1319    break;
1320  }
1321}
1322
1323std::string
1324NVPTXAsmPrinter::getPTXFundamentalTypeStr(const Type *Ty, bool useB4PTR) const {
1325  switch (Ty->getTypeID()) {
1326  default:
1327    llvm_unreachable("unexpected type");
1328    break;
1329  case Type::IntegerTyID: {
1330    unsigned NumBits = cast<IntegerType>(Ty)->getBitWidth();
1331    if (NumBits == 1)
1332      return "pred";
1333    else if (NumBits <= 64) {
1334      std::string name = "u";
1335      return name + utostr(NumBits);
1336    } else {
1337      llvm_unreachable("Integer too large");
1338      break;
1339    }
1340    break;
1341  }
1342  case Type::FloatTyID:
1343    return "f32";
1344  case Type::DoubleTyID:
1345    return "f64";
1346  case Type::PointerTyID:
1347    if (nvptxSubtarget.is64Bit())
1348      if (useB4PTR)
1349        return "b64";
1350      else
1351        return "u64";
1352    else if (useB4PTR)
1353      return "b32";
1354    else
1355      return "u32";
1356  }
1357  llvm_unreachable("unexpected type");
1358  return NULL;
1359}
1360
1361void NVPTXAsmPrinter::emitPTXGlobalVariable(const GlobalVariable *GVar,
1362                                            raw_ostream &O) {
1363
1364  const DataLayout *TD = TM.getDataLayout();
1365
1366  // GlobalVariables are always constant pointers themselves.
1367  const PointerType *PTy = GVar->getType();
1368  Type *ETy = PTy->getElementType();
1369
1370  O << ".";
1371  emitPTXAddressSpace(PTy->getAddressSpace(), O);
1372  if (GVar->getAlignment() == 0)
1373    O << " .align " << (int) TD->getPrefTypeAlignment(ETy);
1374  else
1375    O << " .align " << GVar->getAlignment();
1376
1377  if (ETy->isPrimitiveType() || ETy->isIntegerTy() || isa<PointerType>(ETy)) {
1378    O << " .";
1379    O << getPTXFundamentalTypeStr(ETy);
1380    O << " ";
1381    O << *Mang->getSymbol(GVar);
1382    return;
1383  }
1384
1385  int64_t ElementSize = 0;
1386
1387  // Although PTX has direct support for struct type and array type and LLVM IR
1388  // is very similar to PTX, the LLVM CodeGen does not support for targets that
1389  // support these high level field accesses. Structs and arrays are lowered
1390  // into arrays of bytes.
1391  switch (ETy->getTypeID()) {
1392  case Type::StructTyID:
1393  case Type::ArrayTyID:
1394  case Type::VectorTyID:
1395    ElementSize = TD->getTypeStoreSize(ETy);
1396    O << " .b8 " << *Mang->getSymbol(GVar) << "[";
1397    if (ElementSize) {
1398      O << itostr(ElementSize);
1399    }
1400    O << "]";
1401    break;
1402  default:
1403    assert(0 && "type not supported yet");
1404  }
1405  return;
1406}
1407
1408static unsigned int getOpenCLAlignment(const DataLayout *TD, Type *Ty) {
1409  if (Ty->isPrimitiveType() || Ty->isIntegerTy() || isa<PointerType>(Ty))
1410    return TD->getPrefTypeAlignment(Ty);
1411
1412  const ArrayType *ATy = dyn_cast<ArrayType>(Ty);
1413  if (ATy)
1414    return getOpenCLAlignment(TD, ATy->getElementType());
1415
1416  const VectorType *VTy = dyn_cast<VectorType>(Ty);
1417  if (VTy) {
1418    Type *ETy = VTy->getElementType();
1419    unsigned int numE = VTy->getNumElements();
1420    unsigned int alignE = TD->getPrefTypeAlignment(ETy);
1421    if (numE == 3)
1422      return 4 * alignE;
1423    else
1424      return numE * alignE;
1425  }
1426
1427  const StructType *STy = dyn_cast<StructType>(Ty);
1428  if (STy) {
1429    unsigned int alignStruct = 1;
1430    // Go through each element of the struct and find the
1431    // largest alignment.
1432    for (unsigned i = 0, e = STy->getNumElements(); i != e; i++) {
1433      Type *ETy = STy->getElementType(i);
1434      unsigned int align = getOpenCLAlignment(TD, ETy);
1435      if (align > alignStruct)
1436        alignStruct = align;
1437    }
1438    return alignStruct;
1439  }
1440
1441  const FunctionType *FTy = dyn_cast<FunctionType>(Ty);
1442  if (FTy)
1443    return TD->getPointerPrefAlignment();
1444  return TD->getPrefTypeAlignment(Ty);
1445}
1446
1447void NVPTXAsmPrinter::printParamName(Function::const_arg_iterator I,
1448                                     int paramIndex, raw_ostream &O) {
1449  if ((nvptxSubtarget.getDrvInterface() == NVPTX::NVCL) ||
1450      (nvptxSubtarget.getDrvInterface() == NVPTX::CUDA))
1451    O << *CurrentFnSym << "_param_" << paramIndex;
1452  else {
1453    std::string argName = I->getName();
1454    const char *p = argName.c_str();
1455    while (*p) {
1456      if (*p == '.')
1457        O << "_";
1458      else
1459        O << *p;
1460      p++;
1461    }
1462  }
1463}
1464
1465void NVPTXAsmPrinter::printParamName(int paramIndex, raw_ostream &O) {
1466  Function::const_arg_iterator I, E;
1467  int i = 0;
1468
1469  if ((nvptxSubtarget.getDrvInterface() == NVPTX::NVCL) ||
1470      (nvptxSubtarget.getDrvInterface() == NVPTX::CUDA)) {
1471    O << *CurrentFnSym << "_param_" << paramIndex;
1472    return;
1473  }
1474
1475  for (I = F->arg_begin(), E = F->arg_end(); I != E; ++I, i++) {
1476    if (i == paramIndex) {
1477      printParamName(I, paramIndex, O);
1478      return;
1479    }
1480  }
1481  llvm_unreachable("paramIndex out of bound");
1482}
1483
1484void NVPTXAsmPrinter::emitFunctionParamList(const Function *F, raw_ostream &O) {
1485  const DataLayout *TD = TM.getDataLayout();
1486  const AttributeSet &PAL = F->getAttributes();
1487  const TargetLowering *TLI = TM.getTargetLowering();
1488  Function::const_arg_iterator I, E;
1489  unsigned paramIndex = 0;
1490  bool first = true;
1491  bool isKernelFunc = llvm::isKernelFunction(*F);
1492  bool isABI = (nvptxSubtarget.getSmVersion() >= 20);
1493  MVT thePointerTy = TLI->getPointerTy();
1494
1495  O << "(\n";
1496
1497  for (I = F->arg_begin(), E = F->arg_end(); I != E; ++I, paramIndex++) {
1498    Type *Ty = I->getType();
1499
1500    if (!first)
1501      O << ",\n";
1502
1503    first = false;
1504
1505    // Handle image/sampler parameters
1506    if (llvm::isSampler(*I) || llvm::isImage(*I)) {
1507      if (llvm::isImage(*I)) {
1508        std::string sname = I->getName();
1509        if (llvm::isImageWriteOnly(*I))
1510          O << "\t.param .surfref " << *CurrentFnSym << "_param_" << paramIndex;
1511        else // Default image is read_only
1512          O << "\t.param .texref " << *CurrentFnSym << "_param_" << paramIndex;
1513      } else // Should be llvm::isSampler(*I)
1514        O << "\t.param .samplerref " << *CurrentFnSym << "_param_"
1515          << paramIndex;
1516      continue;
1517    }
1518
1519    if (PAL.hasAttribute(paramIndex + 1, Attribute::ByVal) == false) {
1520      if (Ty->isVectorTy()) {
1521        // Just print .param .b8 .align <a> .param[size];
1522        // <a> = PAL.getparamalignment
1523        // size = typeallocsize of element type
1524        unsigned align = PAL.getParamAlignment(paramIndex + 1);
1525        if (align == 0)
1526          align = TD->getABITypeAlignment(Ty);
1527
1528        unsigned sz = TD->getTypeAllocSize(Ty);
1529        O << "\t.param .align " << align << " .b8 ";
1530        printParamName(I, paramIndex, O);
1531        O << "[" << sz << "]";
1532
1533        continue;
1534      }
1535      // Just a scalar
1536      const PointerType *PTy = dyn_cast<PointerType>(Ty);
1537      if (isKernelFunc) {
1538        if (PTy) {
1539          // Special handling for pointer arguments to kernel
1540          O << "\t.param .u" << thePointerTy.getSizeInBits() << " ";
1541
1542          if (nvptxSubtarget.getDrvInterface() != NVPTX::CUDA) {
1543            Type *ETy = PTy->getElementType();
1544            int addrSpace = PTy->getAddressSpace();
1545            switch (addrSpace) {
1546            default:
1547              O << ".ptr ";
1548              break;
1549            case llvm::ADDRESS_SPACE_CONST_NOT_GEN:
1550              O << ".ptr .const ";
1551              break;
1552            case llvm::ADDRESS_SPACE_SHARED:
1553              O << ".ptr .shared ";
1554              break;
1555            case llvm::ADDRESS_SPACE_GLOBAL:
1556            case llvm::ADDRESS_SPACE_CONST:
1557              O << ".ptr .global ";
1558              break;
1559            }
1560            O << ".align " << (int) getOpenCLAlignment(TD, ETy) << " ";
1561          }
1562          printParamName(I, paramIndex, O);
1563          continue;
1564        }
1565
1566        // non-pointer scalar to kernel func
1567        O << "\t.param ." << getPTXFundamentalTypeStr(Ty) << " ";
1568        printParamName(I, paramIndex, O);
1569        continue;
1570      }
1571      // Non-kernel function, just print .param .b<size> for ABI
1572      // and .reg .b<size> for non ABY
1573      unsigned sz = 0;
1574      if (isa<IntegerType>(Ty)) {
1575        sz = cast<IntegerType>(Ty)->getBitWidth();
1576        if (sz < 32)
1577          sz = 32;
1578      } else if (isa<PointerType>(Ty))
1579        sz = thePointerTy.getSizeInBits();
1580      else
1581        sz = Ty->getPrimitiveSizeInBits();
1582      if (isABI)
1583        O << "\t.param .b" << sz << " ";
1584      else
1585        O << "\t.reg .b" << sz << " ";
1586      printParamName(I, paramIndex, O);
1587      continue;
1588    }
1589
1590    // param has byVal attribute. So should be a pointer
1591    const PointerType *PTy = dyn_cast<PointerType>(Ty);
1592    assert(PTy && "Param with byval attribute should be a pointer type");
1593    Type *ETy = PTy->getElementType();
1594
1595    if (isABI || isKernelFunc) {
1596      // Just print .param .b8 .align <a> .param[size];
1597      // <a> = PAL.getparamalignment
1598      // size = typeallocsize of element type
1599      unsigned align = PAL.getParamAlignment(paramIndex + 1);
1600      if (align == 0)
1601        align = TD->getABITypeAlignment(ETy);
1602
1603      unsigned sz = TD->getTypeAllocSize(ETy);
1604      O << "\t.param .align " << align << " .b8 ";
1605      printParamName(I, paramIndex, O);
1606      O << "[" << sz << "]";
1607      continue;
1608    } else {
1609      // Split the ETy into constituent parts and
1610      // print .param .b<size> <name> for each part.
1611      // Further, if a part is vector, print the above for
1612      // each vector element.
1613      SmallVector<EVT, 16> vtparts;
1614      ComputeValueVTs(*TLI, ETy, vtparts);
1615      for (unsigned i = 0, e = vtparts.size(); i != e; ++i) {
1616        unsigned elems = 1;
1617        EVT elemtype = vtparts[i];
1618        if (vtparts[i].isVector()) {
1619          elems = vtparts[i].getVectorNumElements();
1620          elemtype = vtparts[i].getVectorElementType();
1621        }
1622
1623        for (unsigned j = 0, je = elems; j != je; ++j) {
1624          unsigned sz = elemtype.getSizeInBits();
1625          if (elemtype.isInteger() && (sz < 32))
1626            sz = 32;
1627          O << "\t.reg .b" << sz << " ";
1628          printParamName(I, paramIndex, O);
1629          if (j < je - 1)
1630            O << ",\n";
1631          ++paramIndex;
1632        }
1633        if (i < e - 1)
1634          O << ",\n";
1635      }
1636      --paramIndex;
1637      continue;
1638    }
1639  }
1640
1641  O << "\n)\n";
1642}
1643
1644void NVPTXAsmPrinter::emitFunctionParamList(const MachineFunction &MF,
1645                                            raw_ostream &O) {
1646  const Function *F = MF.getFunction();
1647  emitFunctionParamList(F, O);
1648}
1649
1650void NVPTXAsmPrinter::setAndEmitFunctionVirtualRegisters(
1651    const MachineFunction &MF) {
1652  SmallString<128> Str;
1653  raw_svector_ostream O(Str);
1654
1655  // Map the global virtual register number to a register class specific
1656  // virtual register number starting from 1 with that class.
1657  const TargetRegisterInfo *TRI = MF.getTarget().getRegisterInfo();
1658  //unsigned numRegClasses = TRI->getNumRegClasses();
1659
1660  // Emit the Fake Stack Object
1661  const MachineFrameInfo *MFI = MF.getFrameInfo();
1662  int NumBytes = (int) MFI->getStackSize();
1663  if (NumBytes) {
1664    O << "\t.local .align " << MFI->getMaxAlignment() << " .b8 \t" << DEPOTNAME
1665      << getFunctionNumber() << "[" << NumBytes << "];\n";
1666    if (nvptxSubtarget.is64Bit()) {
1667      O << "\t.reg .b64 \t%SP;\n";
1668      O << "\t.reg .b64 \t%SPL;\n";
1669    } else {
1670      O << "\t.reg .b32 \t%SP;\n";
1671      O << "\t.reg .b32 \t%SPL;\n";
1672    }
1673  }
1674
1675  // Go through all virtual registers to establish the mapping between the
1676  // global virtual
1677  // register number and the per class virtual register number.
1678  // We use the per class virtual register number in the ptx output.
1679  unsigned int numVRs = MRI->getNumVirtRegs();
1680  for (unsigned i = 0; i < numVRs; i++) {
1681    unsigned int vr = TRI->index2VirtReg(i);
1682    const TargetRegisterClass *RC = MRI->getRegClass(vr);
1683    std::map<unsigned, unsigned> &regmap = VRidGlobal2LocalMap[RC->getID()];
1684    int n = regmap.size();
1685    regmap.insert(std::make_pair(vr, n + 1));
1686  }
1687
1688  // Emit register declarations
1689  // @TODO: Extract out the real register usage
1690  O << "\t.reg .pred %p<" << NVPTXNumRegisters << ">;\n";
1691  O << "\t.reg .s16 %rc<" << NVPTXNumRegisters << ">;\n";
1692  O << "\t.reg .s16 %rs<" << NVPTXNumRegisters << ">;\n";
1693  O << "\t.reg .s32 %r<" << NVPTXNumRegisters << ">;\n";
1694  O << "\t.reg .s64 %rl<" << NVPTXNumRegisters << ">;\n";
1695  O << "\t.reg .f32 %f<" << NVPTXNumRegisters << ">;\n";
1696  O << "\t.reg .f64 %fl<" << NVPTXNumRegisters << ">;\n";
1697
1698  // Emit declaration of the virtual registers or 'physical' registers for
1699  // each register class
1700  //for (unsigned i=0; i< numRegClasses; i++) {
1701  //    std::map<unsigned, unsigned> &regmap = VRidGlobal2LocalMap[i];
1702  //    const TargetRegisterClass *RC = TRI->getRegClass(i);
1703  //    std::string rcname = getNVPTXRegClassName(RC);
1704  //    std::string rcStr = getNVPTXRegClassStr(RC);
1705  //    //int n = regmap.size();
1706  //    if (!isNVPTXVectorRegClass(RC)) {
1707  //      O << "\t.reg " << rcname << " \t" << rcStr << "<"
1708  //        << NVPTXNumRegisters << ">;\n";
1709  //    }
1710
1711  // Only declare those registers that may be used. And do not emit vector
1712  // registers as
1713  // they are all elementized to scalar registers.
1714  //if (n && !isNVPTXVectorRegClass(RC)) {
1715  //    if (RegAllocNilUsed) {
1716  //        O << "\t.reg " << rcname << " \t" << rcStr << "<" << (n+1)
1717  //          << ">;\n";
1718  //    }
1719  //    else {
1720  //        O << "\t.reg " << rcname << " \t" << StrToUpper(rcStr)
1721  //          << "<" << 32 << ">;\n";
1722  //    }
1723  //}
1724  //}
1725
1726  OutStreamer.EmitRawText(O.str());
1727}
1728
1729void NVPTXAsmPrinter::printFPConstant(const ConstantFP *Fp, raw_ostream &O) {
1730  APFloat APF = APFloat(Fp->getValueAPF()); // make a copy
1731  bool ignored;
1732  unsigned int numHex;
1733  const char *lead;
1734
1735  if (Fp->getType()->getTypeID() == Type::FloatTyID) {
1736    numHex = 8;
1737    lead = "0f";
1738    APF.convert(APFloat::IEEEsingle, APFloat::rmNearestTiesToEven, &ignored);
1739  } else if (Fp->getType()->getTypeID() == Type::DoubleTyID) {
1740    numHex = 16;
1741    lead = "0d";
1742    APF.convert(APFloat::IEEEdouble, APFloat::rmNearestTiesToEven, &ignored);
1743  } else
1744    llvm_unreachable("unsupported fp type");
1745
1746  APInt API = APF.bitcastToAPInt();
1747  std::string hexstr(utohexstr(API.getZExtValue()));
1748  O << lead;
1749  if (hexstr.length() < numHex)
1750    O << std::string(numHex - hexstr.length(), '0');
1751  O << utohexstr(API.getZExtValue());
1752}
1753
1754void NVPTXAsmPrinter::printScalarConstant(Constant *CPV, raw_ostream &O) {
1755  if (ConstantInt *CI = dyn_cast<ConstantInt>(CPV)) {
1756    O << CI->getValue();
1757    return;
1758  }
1759  if (ConstantFP *CFP = dyn_cast<ConstantFP>(CPV)) {
1760    printFPConstant(CFP, O);
1761    return;
1762  }
1763  if (isa<ConstantPointerNull>(CPV)) {
1764    O << "0";
1765    return;
1766  }
1767  if (GlobalValue *GVar = dyn_cast<GlobalValue>(CPV)) {
1768    O << *Mang->getSymbol(GVar);
1769    return;
1770  }
1771  if (ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
1772    Value *v = Cexpr->stripPointerCasts();
1773    if (GlobalValue *GVar = dyn_cast<GlobalValue>(v)) {
1774      O << *Mang->getSymbol(GVar);
1775      return;
1776    } else {
1777      O << *LowerConstant(CPV, *this);
1778      return;
1779    }
1780  }
1781  llvm_unreachable("Not scalar type found in printScalarConstant()");
1782}
1783
1784void NVPTXAsmPrinter::bufferLEByte(Constant *CPV, int Bytes,
1785                                   AggBuffer *aggBuffer) {
1786
1787  const DataLayout *TD = TM.getDataLayout();
1788
1789  if (isa<UndefValue>(CPV) || CPV->isNullValue()) {
1790    int s = TD->getTypeAllocSize(CPV->getType());
1791    if (s < Bytes)
1792      s = Bytes;
1793    aggBuffer->addZeros(s);
1794    return;
1795  }
1796
1797  unsigned char *ptr;
1798  switch (CPV->getType()->getTypeID()) {
1799
1800  case Type::IntegerTyID: {
1801    const Type *ETy = CPV->getType();
1802    if (ETy == Type::getInt8Ty(CPV->getContext())) {
1803      unsigned char c =
1804          (unsigned char)(dyn_cast<ConstantInt>(CPV))->getZExtValue();
1805      ptr = &c;
1806      aggBuffer->addBytes(ptr, 1, Bytes);
1807    } else if (ETy == Type::getInt16Ty(CPV->getContext())) {
1808      short int16 = (short)(dyn_cast<ConstantInt>(CPV))->getZExtValue();
1809      ptr = (unsigned char *)&int16;
1810      aggBuffer->addBytes(ptr, 2, Bytes);
1811    } else if (ETy == Type::getInt32Ty(CPV->getContext())) {
1812      if (ConstantInt *constInt = dyn_cast<ConstantInt>(CPV)) {
1813        int int32 = (int)(constInt->getZExtValue());
1814        ptr = (unsigned char *)&int32;
1815        aggBuffer->addBytes(ptr, 4, Bytes);
1816        break;
1817      } else if (ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
1818        if (ConstantInt *constInt = dyn_cast<ConstantInt>(
1819                ConstantFoldConstantExpression(Cexpr, TD))) {
1820          int int32 = (int)(constInt->getZExtValue());
1821          ptr = (unsigned char *)&int32;
1822          aggBuffer->addBytes(ptr, 4, Bytes);
1823          break;
1824        }
1825        if (Cexpr->getOpcode() == Instruction::PtrToInt) {
1826          Value *v = Cexpr->getOperand(0)->stripPointerCasts();
1827          aggBuffer->addSymbol(v);
1828          aggBuffer->addZeros(4);
1829          break;
1830        }
1831      }
1832      llvm_unreachable("unsupported integer const type");
1833    } else if (ETy == Type::getInt64Ty(CPV->getContext())) {
1834      if (ConstantInt *constInt = dyn_cast<ConstantInt>(CPV)) {
1835        long long int64 = (long long)(constInt->getZExtValue());
1836        ptr = (unsigned char *)&int64;
1837        aggBuffer->addBytes(ptr, 8, Bytes);
1838        break;
1839      } else if (ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
1840        if (ConstantInt *constInt = dyn_cast<ConstantInt>(
1841                ConstantFoldConstantExpression(Cexpr, TD))) {
1842          long long int64 = (long long)(constInt->getZExtValue());
1843          ptr = (unsigned char *)&int64;
1844          aggBuffer->addBytes(ptr, 8, Bytes);
1845          break;
1846        }
1847        if (Cexpr->getOpcode() == Instruction::PtrToInt) {
1848          Value *v = Cexpr->getOperand(0)->stripPointerCasts();
1849          aggBuffer->addSymbol(v);
1850          aggBuffer->addZeros(8);
1851          break;
1852        }
1853      }
1854      llvm_unreachable("unsupported integer const type");
1855    } else
1856      llvm_unreachable("unsupported integer const type");
1857    break;
1858  }
1859  case Type::FloatTyID:
1860  case Type::DoubleTyID: {
1861    ConstantFP *CFP = dyn_cast<ConstantFP>(CPV);
1862    const Type *Ty = CFP->getType();
1863    if (Ty == Type::getFloatTy(CPV->getContext())) {
1864      float float32 = (float) CFP->getValueAPF().convertToFloat();
1865      ptr = (unsigned char *)&float32;
1866      aggBuffer->addBytes(ptr, 4, Bytes);
1867    } else if (Ty == Type::getDoubleTy(CPV->getContext())) {
1868      double float64 = CFP->getValueAPF().convertToDouble();
1869      ptr = (unsigned char *)&float64;
1870      aggBuffer->addBytes(ptr, 8, Bytes);
1871    } else {
1872      llvm_unreachable("unsupported fp const type");
1873    }
1874    break;
1875  }
1876  case Type::PointerTyID: {
1877    if (GlobalValue *GVar = dyn_cast<GlobalValue>(CPV)) {
1878      aggBuffer->addSymbol(GVar);
1879    } else if (ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
1880      Value *v = Cexpr->stripPointerCasts();
1881      aggBuffer->addSymbol(v);
1882    }
1883    unsigned int s = TD->getTypeAllocSize(CPV->getType());
1884    aggBuffer->addZeros(s);
1885    break;
1886  }
1887
1888  case Type::ArrayTyID:
1889  case Type::VectorTyID:
1890  case Type::StructTyID: {
1891    if (isa<ConstantArray>(CPV) || isa<ConstantVector>(CPV) ||
1892        isa<ConstantStruct>(CPV)) {
1893      int ElementSize = TD->getTypeAllocSize(CPV->getType());
1894      bufferAggregateConstant(CPV, aggBuffer);
1895      if (Bytes > ElementSize)
1896        aggBuffer->addZeros(Bytes - ElementSize);
1897    } else if (isa<ConstantAggregateZero>(CPV))
1898      aggBuffer->addZeros(Bytes);
1899    else
1900      llvm_unreachable("Unexpected Constant type");
1901    break;
1902  }
1903
1904  default:
1905    llvm_unreachable("unsupported type");
1906  }
1907}
1908
1909void NVPTXAsmPrinter::bufferAggregateConstant(Constant *CPV,
1910                                              AggBuffer *aggBuffer) {
1911  const DataLayout *TD = TM.getDataLayout();
1912  int Bytes;
1913
1914  // Old constants
1915  if (isa<ConstantArray>(CPV) || isa<ConstantVector>(CPV)) {
1916    if (CPV->getNumOperands())
1917      for (unsigned i = 0, e = CPV->getNumOperands(); i != e; ++i)
1918        bufferLEByte(cast<Constant>(CPV->getOperand(i)), 0, aggBuffer);
1919    return;
1920  }
1921
1922  if (const ConstantDataSequential *CDS =
1923          dyn_cast<ConstantDataSequential>(CPV)) {
1924    if (CDS->getNumElements())
1925      for (unsigned i = 0; i < CDS->getNumElements(); ++i)
1926        bufferLEByte(cast<Constant>(CDS->getElementAsConstant(i)), 0,
1927                     aggBuffer);
1928    return;
1929  }
1930
1931  if (isa<ConstantStruct>(CPV)) {
1932    if (CPV->getNumOperands()) {
1933      StructType *ST = cast<StructType>(CPV->getType());
1934      for (unsigned i = 0, e = CPV->getNumOperands(); i != e; ++i) {
1935        if (i == (e - 1))
1936          Bytes = TD->getStructLayout(ST)->getElementOffset(0) +
1937                  TD->getTypeAllocSize(ST) -
1938                  TD->getStructLayout(ST)->getElementOffset(i);
1939        else
1940          Bytes = TD->getStructLayout(ST)->getElementOffset(i + 1) -
1941                  TD->getStructLayout(ST)->getElementOffset(i);
1942        bufferLEByte(cast<Constant>(CPV->getOperand(i)), Bytes, aggBuffer);
1943      }
1944    }
1945    return;
1946  }
1947  llvm_unreachable("unsupported constant type in printAggregateConstant()");
1948}
1949
1950// buildTypeNameMap - Run through symbol table looking for type names.
1951//
1952
1953bool NVPTXAsmPrinter::isImageType(const Type *Ty) {
1954
1955  std::map<const Type *, std::string>::iterator PI = TypeNameMap.find(Ty);
1956
1957  if (PI != TypeNameMap.end() && (!PI->second.compare("struct._image1d_t") ||
1958                                  !PI->second.compare("struct._image2d_t") ||
1959                                  !PI->second.compare("struct._image3d_t")))
1960    return true;
1961
1962  return false;
1963}
1964
1965/// PrintAsmOperand - Print out an operand for an inline asm expression.
1966///
1967bool NVPTXAsmPrinter::PrintAsmOperand(const MachineInstr *MI, unsigned OpNo,
1968                                      unsigned AsmVariant,
1969                                      const char *ExtraCode, raw_ostream &O) {
1970  if (ExtraCode && ExtraCode[0]) {
1971    if (ExtraCode[1] != 0)
1972      return true; // Unknown modifier.
1973
1974    switch (ExtraCode[0]) {
1975    default:
1976      // See if this is a generic print operand
1977      return AsmPrinter::PrintAsmOperand(MI, OpNo, AsmVariant, ExtraCode, O);
1978    case 'r':
1979      break;
1980    }
1981  }
1982
1983  printOperand(MI, OpNo, O);
1984
1985  return false;
1986}
1987
1988bool NVPTXAsmPrinter::PrintAsmMemoryOperand(
1989    const MachineInstr *MI, unsigned OpNo, unsigned AsmVariant,
1990    const char *ExtraCode, raw_ostream &O) {
1991  if (ExtraCode && ExtraCode[0])
1992    return true; // Unknown modifier
1993
1994  O << '[';
1995  printMemOperand(MI, OpNo, O);
1996  O << ']';
1997
1998  return false;
1999}
2000
2001bool NVPTXAsmPrinter::ignoreLoc(const MachineInstr &MI) {
2002  switch (MI.getOpcode()) {
2003  default:
2004    return false;
2005  case NVPTX::CallArgBeginInst:
2006  case NVPTX::CallArgEndInst0:
2007  case NVPTX::CallArgEndInst1:
2008  case NVPTX::CallArgF32:
2009  case NVPTX::CallArgF64:
2010  case NVPTX::CallArgI16:
2011  case NVPTX::CallArgI32:
2012  case NVPTX::CallArgI32imm:
2013  case NVPTX::CallArgI64:
2014  case NVPTX::CallArgI8:
2015  case NVPTX::CallArgParam:
2016  case NVPTX::CallVoidInst:
2017  case NVPTX::CallVoidInstReg:
2018  case NVPTX::Callseq_End:
2019  case NVPTX::CallVoidInstReg64:
2020  case NVPTX::DeclareParamInst:
2021  case NVPTX::DeclareRetMemInst:
2022  case NVPTX::DeclareRetRegInst:
2023  case NVPTX::DeclareRetScalarInst:
2024  case NVPTX::DeclareScalarParamInst:
2025  case NVPTX::DeclareScalarRegInst:
2026  case NVPTX::StoreParamF32:
2027  case NVPTX::StoreParamF64:
2028  case NVPTX::StoreParamI16:
2029  case NVPTX::StoreParamI32:
2030  case NVPTX::StoreParamI64:
2031  case NVPTX::StoreParamI8:
2032  case NVPTX::StoreParamS32I8:
2033  case NVPTX::StoreParamU32I8:
2034  case NVPTX::StoreParamS32I16:
2035  case NVPTX::StoreParamU32I16:
2036  case NVPTX::StoreRetvalF32:
2037  case NVPTX::StoreRetvalF64:
2038  case NVPTX::StoreRetvalI16:
2039  case NVPTX::StoreRetvalI32:
2040  case NVPTX::StoreRetvalI64:
2041  case NVPTX::StoreRetvalI8:
2042  case NVPTX::LastCallArgF32:
2043  case NVPTX::LastCallArgF64:
2044  case NVPTX::LastCallArgI16:
2045  case NVPTX::LastCallArgI32:
2046  case NVPTX::LastCallArgI32imm:
2047  case NVPTX::LastCallArgI64:
2048  case NVPTX::LastCallArgI8:
2049  case NVPTX::LastCallArgParam:
2050  case NVPTX::LoadParamMemF32:
2051  case NVPTX::LoadParamMemF64:
2052  case NVPTX::LoadParamMemI16:
2053  case NVPTX::LoadParamMemI32:
2054  case NVPTX::LoadParamMemI64:
2055  case NVPTX::LoadParamMemI8:
2056  case NVPTX::LoadParamRegF32:
2057  case NVPTX::LoadParamRegF64:
2058  case NVPTX::LoadParamRegI16:
2059  case NVPTX::LoadParamRegI32:
2060  case NVPTX::LoadParamRegI64:
2061  case NVPTX::LoadParamRegI8:
2062  case NVPTX::PrototypeInst:
2063  case NVPTX::DBG_VALUE:
2064    return true;
2065  }
2066  return false;
2067}
2068
2069// Force static initialization.
2070extern "C" void LLVMInitializeNVPTXBackendAsmPrinter() {
2071  RegisterAsmPrinter<NVPTXAsmPrinter> X(TheNVPTXTarget32);
2072  RegisterAsmPrinter<NVPTXAsmPrinter> Y(TheNVPTXTarget64);
2073}
2074
2075void NVPTXAsmPrinter::emitSrcInText(StringRef filename, unsigned line) {
2076  std::stringstream temp;
2077  LineReader *reader = this->getReader(filename.str());
2078  temp << "\n//";
2079  temp << filename.str();
2080  temp << ":";
2081  temp << line;
2082  temp << " ";
2083  temp << reader->readLine(line);
2084  temp << "\n";
2085  this->OutStreamer.EmitRawText(Twine(temp.str()));
2086}
2087
2088LineReader *NVPTXAsmPrinter::getReader(std::string filename) {
2089  if (reader == NULL) {
2090    reader = new LineReader(filename);
2091  }
2092
2093  if (reader->fileName() != filename) {
2094    delete reader;
2095    reader = new LineReader(filename);
2096  }
2097
2098  return reader;
2099}
2100
2101std::string LineReader::readLine(unsigned lineNum) {
2102  if (lineNum < theCurLine) {
2103    theCurLine = 0;
2104    fstr.seekg(0, std::ios::beg);
2105  }
2106  while (theCurLine < lineNum) {
2107    fstr.getline(buff, 500);
2108    theCurLine++;
2109  }
2110  return buff;
2111}
2112
2113// Force static initialization.
2114extern "C" void LLVMInitializeNVPTXAsmPrinter() {
2115  RegisterAsmPrinter<NVPTXAsmPrinter> X(TheNVPTXTarget32);
2116  RegisterAsmPrinter<NVPTXAsmPrinter> Y(TheNVPTXTarget64);
2117}
2118