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/MCStreamer.h"
41 #include "llvm/MC/MCSymbol.h"
42 #include "llvm/Support/CommandLine.h"
43 #include "llvm/Support/ErrorHandling.h"
44 #include "llvm/Support/FormattedStream.h"
45 #include "llvm/Support/Path.h"
46 #include "llvm/Support/TargetRegistry.h"
47 #include "llvm/Support/TimeValue.h"
48 #include "llvm/Target/TargetLoweringObjectFile.h"
49 #include "llvm/Transforms/Utils/UnrollLoop.h"
50 #include <sstream>
51 using namespace llvm;
52 
53 #define DEPOTNAME "__local_depot"
54 
55 static cl::opt<bool>
56 EmitLineNumbers("nvptx-emit-line-numbers", cl::Hidden,
57                 cl::desc("NVPTX Specific: Emit Line numbers even without -G"),
58                 cl::init(true));
59 
60 static cl::opt<bool>
61 InterleaveSrc("nvptx-emit-src", cl::ZeroOrMore, cl::Hidden,
62               cl::desc("NVPTX Specific: Emit source line in ptx file"),
63               cl::init(false));
64 
65 namespace {
66 /// DiscoverDependentGlobals - Return a set of GlobalVariables on which \p V
67 /// depends.
DiscoverDependentGlobals(const Value * V,DenseSet<const GlobalVariable * > & Globals)68 void DiscoverDependentGlobals(const Value *V,
69                               DenseSet<const GlobalVariable *> &Globals) {
70   if (const GlobalVariable *GV = dyn_cast<GlobalVariable>(V))
71     Globals.insert(GV);
72   else {
73     if (const User *U = dyn_cast<User>(V)) {
74       for (unsigned i = 0, e = U->getNumOperands(); i != e; ++i) {
75         DiscoverDependentGlobals(U->getOperand(i), Globals);
76       }
77     }
78   }
79 }
80 
81 /// VisitGlobalVariableForEmission - Add \p GV to the list of GlobalVariable
82 /// instances to be emitted, but only after any dependents have been added
83 /// first.
VisitGlobalVariableForEmission(const GlobalVariable * GV,SmallVectorImpl<const GlobalVariable * > & Order,DenseSet<const GlobalVariable * > & Visited,DenseSet<const GlobalVariable * > & Visiting)84 void VisitGlobalVariableForEmission(
85     const GlobalVariable *GV, SmallVectorImpl<const GlobalVariable *> &Order,
86     DenseSet<const GlobalVariable *> &Visited,
87     DenseSet<const GlobalVariable *> &Visiting) {
88   // Have we already visited this one?
89   if (Visited.count(GV))
90     return;
91 
92   // Do we have a circular dependency?
93   if (!Visiting.insert(GV).second)
94     report_fatal_error("Circular dependency found in global variable set");
95 
96   // Make sure we visit all dependents first
97   DenseSet<const GlobalVariable *> Others;
98   for (unsigned i = 0, e = GV->getNumOperands(); i != e; ++i)
99     DiscoverDependentGlobals(GV->getOperand(i), Others);
100 
101   for (DenseSet<const GlobalVariable *>::iterator I = Others.begin(),
102                                                   E = Others.end();
103        I != E; ++I)
104     VisitGlobalVariableForEmission(*I, Order, Visited, Visiting);
105 
106   // Now we can visit ourself
107   Order.push_back(GV);
108   Visited.insert(GV);
109   Visiting.erase(GV);
110 }
111 }
112 
emitLineNumberAsDotLoc(const MachineInstr & MI)113 void NVPTXAsmPrinter::emitLineNumberAsDotLoc(const MachineInstr &MI) {
114   if (!EmitLineNumbers)
115     return;
116   if (ignoreLoc(MI))
117     return;
118 
119   DebugLoc curLoc = MI.getDebugLoc();
120 
121   if (!prevDebugLoc && !curLoc)
122     return;
123 
124   if (prevDebugLoc == curLoc)
125     return;
126 
127   prevDebugLoc = curLoc;
128 
129   if (!curLoc)
130     return;
131 
132   auto *Scope = cast_or_null<MDScope>(curLoc.getScope());
133   if (!Scope)
134      return;
135 
136   StringRef fileName(Scope->getFilename());
137   StringRef dirName(Scope->getDirectory());
138   SmallString<128> FullPathName = dirName;
139   if (!dirName.empty() && !sys::path::is_absolute(fileName)) {
140     sys::path::append(FullPathName, fileName);
141     fileName = FullPathName;
142   }
143 
144   if (filenameMap.find(fileName) == filenameMap.end())
145     return;
146 
147   // Emit the line from the source file.
148   if (InterleaveSrc)
149     this->emitSrcInText(fileName, curLoc.getLine());
150 
151   std::stringstream temp;
152   temp << "\t.loc " << filenameMap[fileName] << " " << curLoc.getLine()
153        << " " << curLoc.getCol();
154   OutStreamer.EmitRawText(temp.str());
155 }
156 
EmitInstruction(const MachineInstr * MI)157 void NVPTXAsmPrinter::EmitInstruction(const MachineInstr *MI) {
158   SmallString<128> Str;
159   raw_svector_ostream OS(Str);
160   if (static_cast<NVPTXTargetMachine &>(TM).getDrvInterface() == NVPTX::CUDA)
161     emitLineNumberAsDotLoc(*MI);
162 
163   MCInst Inst;
164   lowerToMCInst(MI, Inst);
165   EmitToStreamer(OutStreamer, Inst);
166 }
167 
168 // Handle symbol backtracking for targets that do not support image handles
lowerImageHandleOperand(const MachineInstr * MI,unsigned OpNo,MCOperand & MCOp)169 bool NVPTXAsmPrinter::lowerImageHandleOperand(const MachineInstr *MI,
170                                            unsigned OpNo, MCOperand &MCOp) {
171   const MachineOperand &MO = MI->getOperand(OpNo);
172   const MCInstrDesc &MCID = MI->getDesc();
173 
174   if (MCID.TSFlags & NVPTXII::IsTexFlag) {
175     // This is a texture fetch, so operand 4 is a texref and operand 5 is
176     // a samplerref
177     if (OpNo == 4 && MO.isImm()) {
178       lowerImageHandleSymbol(MO.getImm(), MCOp);
179       return true;
180     }
181     if (OpNo == 5 && MO.isImm() && !(MCID.TSFlags & NVPTXII::IsTexModeUnifiedFlag)) {
182       lowerImageHandleSymbol(MO.getImm(), MCOp);
183       return true;
184     }
185 
186     return false;
187   } else if (MCID.TSFlags & NVPTXII::IsSuldMask) {
188     unsigned VecSize =
189       1 << (((MCID.TSFlags & NVPTXII::IsSuldMask) >> NVPTXII::IsSuldShift) - 1);
190 
191     // For a surface load of vector size N, the Nth operand will be the surfref
192     if (OpNo == VecSize && MO.isImm()) {
193       lowerImageHandleSymbol(MO.getImm(), MCOp);
194       return true;
195     }
196 
197     return false;
198   } else if (MCID.TSFlags & NVPTXII::IsSustFlag) {
199     // This is a surface store, so operand 0 is a surfref
200     if (OpNo == 0 && MO.isImm()) {
201       lowerImageHandleSymbol(MO.getImm(), MCOp);
202       return true;
203     }
204 
205     return false;
206   } else if (MCID.TSFlags & NVPTXII::IsSurfTexQueryFlag) {
207     // This is a query, so operand 1 is a surfref/texref
208     if (OpNo == 1 && MO.isImm()) {
209       lowerImageHandleSymbol(MO.getImm(), MCOp);
210       return true;
211     }
212 
213     return false;
214   }
215 
216   return false;
217 }
218 
lowerImageHandleSymbol(unsigned Index,MCOperand & MCOp)219 void NVPTXAsmPrinter::lowerImageHandleSymbol(unsigned Index, MCOperand &MCOp) {
220   // Ewwww
221   TargetMachine &TM = const_cast<TargetMachine&>(MF->getTarget());
222   NVPTXTargetMachine &nvTM = static_cast<NVPTXTargetMachine&>(TM);
223   const NVPTXMachineFunctionInfo *MFI = MF->getInfo<NVPTXMachineFunctionInfo>();
224   const char *Sym = MFI->getImageHandleSymbol(Index);
225   std::string *SymNamePtr =
226     nvTM.getManagedStrPool()->getManagedString(Sym);
227   MCOp = GetSymbolRef(OutContext.GetOrCreateSymbol(
228     StringRef(SymNamePtr->c_str())));
229 }
230 
lowerToMCInst(const MachineInstr * MI,MCInst & OutMI)231 void NVPTXAsmPrinter::lowerToMCInst(const MachineInstr *MI, MCInst &OutMI) {
232   OutMI.setOpcode(MI->getOpcode());
233   // Special: Do not mangle symbol operand of CALL_PROTOTYPE
234   if (MI->getOpcode() == NVPTX::CALL_PROTOTYPE) {
235     const MachineOperand &MO = MI->getOperand(0);
236     OutMI.addOperand(GetSymbolRef(
237       OutContext.GetOrCreateSymbol(Twine(MO.getSymbolName()))));
238     return;
239   }
240 
241   for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
242     const MachineOperand &MO = MI->getOperand(i);
243 
244     MCOperand MCOp;
245     if (!nvptxSubtarget->hasImageHandles()) {
246       if (lowerImageHandleOperand(MI, i, MCOp)) {
247         OutMI.addOperand(MCOp);
248         continue;
249       }
250     }
251 
252     if (lowerOperand(MO, MCOp))
253       OutMI.addOperand(MCOp);
254   }
255 }
256 
lowerOperand(const MachineOperand & MO,MCOperand & MCOp)257 bool NVPTXAsmPrinter::lowerOperand(const MachineOperand &MO,
258                                    MCOperand &MCOp) {
259   switch (MO.getType()) {
260   default: llvm_unreachable("unknown operand type");
261   case MachineOperand::MO_Register:
262     MCOp = MCOperand::CreateReg(encodeVirtualRegister(MO.getReg()));
263     break;
264   case MachineOperand::MO_Immediate:
265     MCOp = MCOperand::CreateImm(MO.getImm());
266     break;
267   case MachineOperand::MO_MachineBasicBlock:
268     MCOp = MCOperand::CreateExpr(MCSymbolRefExpr::Create(
269         MO.getMBB()->getSymbol(), OutContext));
270     break;
271   case MachineOperand::MO_ExternalSymbol:
272     MCOp = GetSymbolRef(GetExternalSymbolSymbol(MO.getSymbolName()));
273     break;
274   case MachineOperand::MO_GlobalAddress:
275     MCOp = GetSymbolRef(getSymbol(MO.getGlobal()));
276     break;
277   case MachineOperand::MO_FPImmediate: {
278     const ConstantFP *Cnt = MO.getFPImm();
279     APFloat Val = Cnt->getValueAPF();
280 
281     switch (Cnt->getType()->getTypeID()) {
282     default: report_fatal_error("Unsupported FP type"); break;
283     case Type::FloatTyID:
284       MCOp = MCOperand::CreateExpr(
285         NVPTXFloatMCExpr::CreateConstantFPSingle(Val, OutContext));
286       break;
287     case Type::DoubleTyID:
288       MCOp = MCOperand::CreateExpr(
289         NVPTXFloatMCExpr::CreateConstantFPDouble(Val, OutContext));
290       break;
291     }
292     break;
293   }
294   }
295   return true;
296 }
297 
encodeVirtualRegister(unsigned Reg)298 unsigned NVPTXAsmPrinter::encodeVirtualRegister(unsigned Reg) {
299   if (TargetRegisterInfo::isVirtualRegister(Reg)) {
300     const TargetRegisterClass *RC = MRI->getRegClass(Reg);
301 
302     DenseMap<unsigned, unsigned> &RegMap = VRegMapping[RC];
303     unsigned RegNum = RegMap[Reg];
304 
305     // Encode the register class in the upper 4 bits
306     // Must be kept in sync with NVPTXInstPrinter::printRegName
307     unsigned Ret = 0;
308     if (RC == &NVPTX::Int1RegsRegClass) {
309       Ret = (1 << 28);
310     } else if (RC == &NVPTX::Int16RegsRegClass) {
311       Ret = (2 << 28);
312     } else if (RC == &NVPTX::Int32RegsRegClass) {
313       Ret = (3 << 28);
314     } else if (RC == &NVPTX::Int64RegsRegClass) {
315       Ret = (4 << 28);
316     } else if (RC == &NVPTX::Float32RegsRegClass) {
317       Ret = (5 << 28);
318     } else if (RC == &NVPTX::Float64RegsRegClass) {
319       Ret = (6 << 28);
320     } else {
321       report_fatal_error("Bad register class");
322     }
323 
324     // Insert the vreg number
325     Ret |= (RegNum & 0x0FFFFFFF);
326     return Ret;
327   } else {
328     // Some special-use registers are actually physical registers.
329     // Encode this as the register class ID of 0 and the real register ID.
330     return Reg & 0x0FFFFFFF;
331   }
332 }
333 
GetSymbolRef(const MCSymbol * Symbol)334 MCOperand NVPTXAsmPrinter::GetSymbolRef(const MCSymbol *Symbol) {
335   const MCExpr *Expr;
336   Expr = MCSymbolRefExpr::Create(Symbol, MCSymbolRefExpr::VK_None,
337                                  OutContext);
338   return MCOperand::CreateExpr(Expr);
339 }
340 
printReturnValStr(const Function * F,raw_ostream & O)341 void NVPTXAsmPrinter::printReturnValStr(const Function *F, raw_ostream &O) {
342   const DataLayout *TD = TM.getDataLayout();
343   const TargetLowering *TLI = nvptxSubtarget->getTargetLowering();
344 
345   Type *Ty = F->getReturnType();
346 
347   bool isABI = (nvptxSubtarget->getSmVersion() >= 20);
348 
349   if (Ty->getTypeID() == Type::VoidTyID)
350     return;
351 
352   O << " (";
353 
354   if (isABI) {
355     if (Ty->isFloatingPointTy() || Ty->isIntegerTy()) {
356       unsigned size = 0;
357       if (const IntegerType *ITy = dyn_cast<IntegerType>(Ty)) {
358         size = ITy->getBitWidth();
359         if (size < 32)
360           size = 32;
361       } else {
362         assert(Ty->isFloatingPointTy() && "Floating point type expected here");
363         size = Ty->getPrimitiveSizeInBits();
364       }
365 
366       O << ".param .b" << size << " func_retval0";
367     } else if (isa<PointerType>(Ty)) {
368       O << ".param .b" << TLI->getPointerTy().getSizeInBits()
369         << " func_retval0";
370     } else if ((Ty->getTypeID() == Type::StructTyID) || isa<VectorType>(Ty)) {
371        unsigned totalsz = TD->getTypeAllocSize(Ty);
372        unsigned retAlignment = 0;
373        if (!llvm::getAlign(*F, 0, retAlignment))
374          retAlignment = TD->getABITypeAlignment(Ty);
375        O << ".param .align " << retAlignment << " .b8 func_retval0[" << totalsz
376          << "]";
377     } else
378       llvm_unreachable("Unknown return type");
379   } else {
380     SmallVector<EVT, 16> vtparts;
381     ComputeValueVTs(*TLI, Ty, vtparts);
382     unsigned idx = 0;
383     for (unsigned i = 0, e = vtparts.size(); i != e; ++i) {
384       unsigned elems = 1;
385       EVT elemtype = vtparts[i];
386       if (vtparts[i].isVector()) {
387         elems = vtparts[i].getVectorNumElements();
388         elemtype = vtparts[i].getVectorElementType();
389       }
390 
391       for (unsigned j = 0, je = elems; j != je; ++j) {
392         unsigned sz = elemtype.getSizeInBits();
393         if (elemtype.isInteger() && (sz < 32))
394           sz = 32;
395         O << ".reg .b" << sz << " func_retval" << idx;
396         if (j < je - 1)
397           O << ", ";
398         ++idx;
399       }
400       if (i < e - 1)
401         O << ", ";
402     }
403   }
404   O << ") ";
405   return;
406 }
407 
printReturnValStr(const MachineFunction & MF,raw_ostream & O)408 void NVPTXAsmPrinter::printReturnValStr(const MachineFunction &MF,
409                                         raw_ostream &O) {
410   const Function *F = MF.getFunction();
411   printReturnValStr(F, O);
412 }
413 
414 // Return true if MBB is the header of a loop marked with
415 // llvm.loop.unroll.disable.
416 // TODO: consider "#pragma unroll 1" which is equivalent to "#pragma nounroll".
isLoopHeaderOfNoUnroll(const MachineBasicBlock & MBB) const417 bool NVPTXAsmPrinter::isLoopHeaderOfNoUnroll(
418     const MachineBasicBlock &MBB) const {
419   MachineLoopInfo &LI = getAnalysis<MachineLoopInfo>();
420   // TODO: isLoopHeader() should take "const MachineBasicBlock *".
421   // We insert .pragma "nounroll" only to the loop header.
422   if (!LI.isLoopHeader(const_cast<MachineBasicBlock *>(&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 
EmitBasicBlockStart(const MachineBasicBlock & MBB) const444 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 
EmitFunctionEntryLabel()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   O << *CurrentFnSym;
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 
EmitFunctionBodyStart()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 
EmitFunctionBodyEnd()493 void NVPTXAsmPrinter::EmitFunctionBodyEnd() {
494   OutStreamer.EmitRawText(StringRef("}\n"));
495   VRegMapping.clear();
496 }
497 
emitImplicitDef(const MachineInstr * MI) const498 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 
emitKernelFunctionDirectives(const Function & F,raw_ostream & O) const510 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
getVirtualRegisterName(unsigned Reg) const562 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 
emitVirtualRegister(unsigned int vr,raw_ostream & O)582 void NVPTXAsmPrinter::emitVirtualRegister(unsigned int vr,
583                                           raw_ostream &O) {
584   O << getVirtualRegisterName(vr);
585 }
586 
printVecModifiedImmediate(const MachineOperand & MO,const char * Modifier,raw_ostream & O)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 
emitDeclaration(const Function * F,raw_ostream & O)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   O << *getSymbol(F) << "\n";
628   emitFunctionParamList(F, O);
629   O << ";\n";
630 }
631 
usedInGlobalVarDef(const Constant * C)632 static bool usedInGlobalVarDef(const Constant *C) {
633   if (!C)
634     return false;
635 
636   if (const GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
637     if (GV->getName() == "llvm.used")
638       return false;
639     return true;
640   }
641 
642   for (const User *U : C->users())
643     if (const Constant *C = dyn_cast<Constant>(U))
644       if (usedInGlobalVarDef(C))
645         return true;
646 
647   return false;
648 }
649 
usedInOneFunc(const User * U,Function const * & oneFunc)650 static bool usedInOneFunc(const User *U, Function const *&oneFunc) {
651   if (const GlobalVariable *othergv = dyn_cast<GlobalVariable>(U)) {
652     if (othergv->getName() == "llvm.used")
653       return true;
654   }
655 
656   if (const Instruction *instr = dyn_cast<Instruction>(U)) {
657     if (instr->getParent() && instr->getParent()->getParent()) {
658       const Function *curFunc = instr->getParent()->getParent();
659       if (oneFunc && (curFunc != oneFunc))
660         return false;
661       oneFunc = curFunc;
662       return true;
663     } else
664       return false;
665   }
666 
667   for (const User *UU : U->users())
668     if (!usedInOneFunc(UU, oneFunc))
669       return false;
670 
671   return true;
672 }
673 
674 /* Find out if a global variable can be demoted to local scope.
675  * Currently, this is valid for CUDA shared variables, which have local
676  * scope and global lifetime. So the conditions to check are :
677  * 1. Is the global variable in shared address space?
678  * 2. Does it have internal linkage?
679  * 3. Is the global variable referenced only in one function?
680  */
canDemoteGlobalVar(const GlobalVariable * gv,Function const * & f)681 static bool canDemoteGlobalVar(const GlobalVariable *gv, Function const *&f) {
682   if (!gv->hasInternalLinkage())
683     return false;
684   const PointerType *Pty = gv->getType();
685   if (Pty->getAddressSpace() != llvm::ADDRESS_SPACE_SHARED)
686     return false;
687 
688   const Function *oneFunc = nullptr;
689 
690   bool flag = usedInOneFunc(gv, oneFunc);
691   if (!flag)
692     return false;
693   if (!oneFunc)
694     return false;
695   f = oneFunc;
696   return true;
697 }
698 
useFuncSeen(const Constant * C,llvm::DenseMap<const Function *,bool> & seenMap)699 static bool useFuncSeen(const Constant *C,
700                         llvm::DenseMap<const Function *, bool> &seenMap) {
701   for (const User *U : C->users()) {
702     if (const Constant *cu = dyn_cast<Constant>(U)) {
703       if (useFuncSeen(cu, seenMap))
704         return true;
705     } else if (const Instruction *I = dyn_cast<Instruction>(U)) {
706       const BasicBlock *bb = I->getParent();
707       if (!bb)
708         continue;
709       const Function *caller = bb->getParent();
710       if (!caller)
711         continue;
712       if (seenMap.find(caller) != seenMap.end())
713         return true;
714     }
715   }
716   return false;
717 }
718 
emitDeclarations(const Module & M,raw_ostream & O)719 void NVPTXAsmPrinter::emitDeclarations(const Module &M, raw_ostream &O) {
720   llvm::DenseMap<const Function *, bool> seenMap;
721   for (Module::const_iterator FI = M.begin(), FE = M.end(); FI != FE; ++FI) {
722     const Function *F = FI;
723 
724     if (F->isDeclaration()) {
725       if (F->use_empty())
726         continue;
727       if (F->getIntrinsicID())
728         continue;
729       emitDeclaration(F, O);
730       continue;
731     }
732     for (const User *U : F->users()) {
733       if (const Constant *C = dyn_cast<Constant>(U)) {
734         if (usedInGlobalVarDef(C)) {
735           // The use is in the initialization of a global variable
736           // that is a function pointer, so print a declaration
737           // for the original function
738           emitDeclaration(F, O);
739           break;
740         }
741         // Emit a declaration of this function if the function that
742         // uses this constant expr has already been seen.
743         if (useFuncSeen(C, seenMap)) {
744           emitDeclaration(F, O);
745           break;
746         }
747       }
748 
749       if (!isa<Instruction>(U))
750         continue;
751       const Instruction *instr = cast<Instruction>(U);
752       const BasicBlock *bb = instr->getParent();
753       if (!bb)
754         continue;
755       const Function *caller = bb->getParent();
756       if (!caller)
757         continue;
758 
759       // If a caller has already been seen, then the caller is
760       // appearing in the module before the callee. so print out
761       // a declaration for the callee.
762       if (seenMap.find(caller) != seenMap.end()) {
763         emitDeclaration(F, O);
764         break;
765       }
766     }
767     seenMap[F] = true;
768   }
769 }
770 
recordAndEmitFilenames(Module & M)771 void NVPTXAsmPrinter::recordAndEmitFilenames(Module &M) {
772   DebugInfoFinder DbgFinder;
773   DbgFinder.processModule(M);
774 
775   unsigned i = 1;
776   for (const MDCompileUnit *DIUnit : DbgFinder.compile_units()) {
777     StringRef Filename = DIUnit->getFilename();
778     StringRef Dirname = DIUnit->getDirectory();
779     SmallString<128> FullPathName = Dirname;
780     if (!Dirname.empty() && !sys::path::is_absolute(Filename)) {
781       sys::path::append(FullPathName, Filename);
782       Filename = FullPathName;
783     }
784     if (filenameMap.find(Filename) != filenameMap.end())
785       continue;
786     filenameMap[Filename] = i;
787     OutStreamer.EmitDwarfFileDirective(i, "", Filename);
788     ++i;
789   }
790 
791   for (MDSubprogram *SP : DbgFinder.subprograms()) {
792     StringRef Filename = SP->getFilename();
793     StringRef Dirname = SP->getDirectory();
794     SmallString<128> FullPathName = Dirname;
795     if (!Dirname.empty() && !sys::path::is_absolute(Filename)) {
796       sys::path::append(FullPathName, Filename);
797       Filename = FullPathName;
798     }
799     if (filenameMap.find(Filename) != filenameMap.end())
800       continue;
801     filenameMap[Filename] = i;
802     ++i;
803   }
804 }
805 
doInitialization(Module & M)806 bool NVPTXAsmPrinter::doInitialization(Module &M) {
807   // Construct a default subtarget off of the TargetMachine defaults. The
808   // rest of NVPTX isn't friendly to change subtargets per function and
809   // so the default TargetMachine will have all of the options.
810   StringRef TT = TM.getTargetTriple();
811   StringRef CPU = TM.getTargetCPU();
812   StringRef FS = TM.getTargetFeatureString();
813   const NVPTXTargetMachine &NTM = static_cast<const NVPTXTargetMachine &>(TM);
814   const NVPTXSubtarget STI(TT, CPU, FS, NTM);
815 
816   SmallString<128> Str1;
817   raw_svector_ostream OS1(Str1);
818 
819   MMI = getAnalysisIfAvailable<MachineModuleInfo>();
820   MMI->AnalyzeModule(M);
821 
822   // We need to call the parent's one explicitly.
823   //bool Result = AsmPrinter::doInitialization(M);
824 
825   // Initialize TargetLoweringObjectFile.
826   const_cast<TargetLoweringObjectFile &>(getObjFileLowering())
827       .Initialize(OutContext, TM);
828 
829   Mang = new Mangler(TM.getDataLayout());
830 
831   // Emit header before any dwarf directives are emitted below.
832   emitHeader(M, OS1, STI);
833   OutStreamer.EmitRawText(OS1.str());
834 
835   // Already commented out
836   //bool Result = AsmPrinter::doInitialization(M);
837 
838   // Emit module-level inline asm if it exists.
839   if (!M.getModuleInlineAsm().empty()) {
840     OutStreamer.AddComment("Start of file scope inline assembly");
841     OutStreamer.AddBlankLine();
842     OutStreamer.EmitRawText(StringRef(M.getModuleInlineAsm()));
843     OutStreamer.AddBlankLine();
844     OutStreamer.AddComment("End of file scope inline assembly");
845     OutStreamer.AddBlankLine();
846   }
847 
848   // If we're not NVCL we're CUDA, go ahead and emit filenames.
849   if (Triple(TM.getTargetTriple()).getOS() != Triple::NVCL)
850     recordAndEmitFilenames(M);
851 
852   GlobalsEmitted = false;
853 
854   return false; // success
855 }
856 
emitGlobals(const Module & M)857 void NVPTXAsmPrinter::emitGlobals(const Module &M) {
858   SmallString<128> Str2;
859   raw_svector_ostream OS2(Str2);
860 
861   emitDeclarations(M, OS2);
862 
863   // As ptxas does not support forward references of globals, we need to first
864   // sort the list of module-level globals in def-use order. We visit each
865   // global variable in order, and ensure that we emit it *after* its dependent
866   // globals. We use a little extra memory maintaining both a set and a list to
867   // have fast searches while maintaining a strict ordering.
868   SmallVector<const GlobalVariable *, 8> Globals;
869   DenseSet<const GlobalVariable *> GVVisited;
870   DenseSet<const GlobalVariable *> GVVisiting;
871 
872   // Visit each global variable, in order
873   for (Module::const_global_iterator I = M.global_begin(), E = M.global_end();
874        I != E; ++I)
875     VisitGlobalVariableForEmission(I, Globals, GVVisited, GVVisiting);
876 
877   assert(GVVisited.size() == M.getGlobalList().size() &&
878          "Missed a global variable");
879   assert(GVVisiting.size() == 0 && "Did not fully process a global variable");
880 
881   // Print out module-level global variables in proper order
882   for (unsigned i = 0, e = Globals.size(); i != e; ++i)
883     printModuleLevelGV(Globals[i], OS2);
884 
885   OS2 << '\n';
886 
887   OutStreamer.EmitRawText(OS2.str());
888 }
889 
emitHeader(Module & M,raw_ostream & O,const NVPTXSubtarget & STI)890 void NVPTXAsmPrinter::emitHeader(Module &M, raw_ostream &O,
891                                  const NVPTXSubtarget &STI) {
892   O << "//\n";
893   O << "// Generated by LLVM NVPTX Back-End\n";
894   O << "//\n";
895   O << "\n";
896 
897   unsigned PTXVersion = STI.getPTXVersion();
898   O << ".version " << (PTXVersion / 10) << "." << (PTXVersion % 10) << "\n";
899 
900   O << ".target ";
901   O << STI.getTargetName();
902 
903   const NVPTXTargetMachine &NTM = static_cast<const NVPTXTargetMachine &>(TM);
904   if (NTM.getDrvInterface() == NVPTX::NVCL)
905     O << ", texmode_independent";
906   else {
907     if (!STI.hasDouble())
908       O << ", map_f64_to_f32";
909   }
910 
911   if (MAI->doesSupportDebugInformation())
912     O << ", debug";
913 
914   O << "\n";
915 
916   O << ".address_size ";
917   if (NTM.is64Bit())
918     O << "64";
919   else
920     O << "32";
921   O << "\n";
922 
923   O << "\n";
924 }
925 
doFinalization(Module & M)926 bool NVPTXAsmPrinter::doFinalization(Module &M) {
927   // If we did not emit any functions, then the global declarations have not
928   // yet been emitted.
929   if (!GlobalsEmitted) {
930     emitGlobals(M);
931     GlobalsEmitted = true;
932   }
933 
934   // XXX Temproarily remove global variables so that doFinalization() will not
935   // emit them again (global variables are emitted at beginning).
936 
937   Module::GlobalListType &global_list = M.getGlobalList();
938   int i, n = global_list.size();
939   GlobalVariable **gv_array = new GlobalVariable *[n];
940 
941   // first, back-up GlobalVariable in gv_array
942   i = 0;
943   for (Module::global_iterator I = global_list.begin(), E = global_list.end();
944        I != E; ++I)
945     gv_array[i++] = &*I;
946 
947   // second, empty global_list
948   while (!global_list.empty())
949     global_list.remove(global_list.begin());
950 
951   // call doFinalization
952   bool ret = AsmPrinter::doFinalization(M);
953 
954   // now we restore global variables
955   for (i = 0; i < n; i++)
956     global_list.insert(global_list.end(), gv_array[i]);
957 
958   clearAnnotationCache(&M);
959 
960   delete[] gv_array;
961   return ret;
962 
963   //bool Result = AsmPrinter::doFinalization(M);
964   // Instead of calling the parents doFinalization, we may
965   // clone parents doFinalization and customize here.
966   // Currently, we if NVISA out the EmitGlobals() in
967   // parent's doFinalization, which is too intrusive.
968   //
969   // Same for the doInitialization.
970   //return Result;
971 }
972 
973 // This function emits appropriate linkage directives for
974 // functions and global variables.
975 //
976 // extern function declaration            -> .extern
977 // extern function definition             -> .visible
978 // external global variable with init     -> .visible
979 // external without init                  -> .extern
980 // appending                              -> not allowed, assert.
981 // for any linkage other than
982 // internal, private, linker_private,
983 // linker_private_weak, linker_private_weak_def_auto,
984 // we emit                                -> .weak.
985 
emitLinkageDirective(const GlobalValue * V,raw_ostream & O)986 void NVPTXAsmPrinter::emitLinkageDirective(const GlobalValue *V,
987                                            raw_ostream &O) {
988   if (static_cast<NVPTXTargetMachine &>(TM).getDrvInterface() == NVPTX::CUDA) {
989     if (V->hasExternalLinkage()) {
990       if (isa<GlobalVariable>(V)) {
991         const GlobalVariable *GVar = cast<GlobalVariable>(V);
992         if (GVar) {
993           if (GVar->hasInitializer())
994             O << ".visible ";
995           else
996             O << ".extern ";
997         }
998       } else if (V->isDeclaration())
999         O << ".extern ";
1000       else
1001         O << ".visible ";
1002     } else if (V->hasAppendingLinkage()) {
1003       std::string msg;
1004       msg.append("Error: ");
1005       msg.append("Symbol ");
1006       if (V->hasName())
1007         msg.append(V->getName());
1008       msg.append("has unsupported appending linkage type");
1009       llvm_unreachable(msg.c_str());
1010     } else if (!V->hasInternalLinkage() &&
1011                !V->hasPrivateLinkage()) {
1012       O << ".weak ";
1013     }
1014   }
1015 }
1016 
printModuleLevelGV(const GlobalVariable * GVar,raw_ostream & O,bool processDemoted)1017 void NVPTXAsmPrinter::printModuleLevelGV(const GlobalVariable *GVar,
1018                                          raw_ostream &O,
1019                                          bool processDemoted) {
1020 
1021   // Skip meta data
1022   if (GVar->hasSection()) {
1023     if (GVar->getSection() == StringRef("llvm.metadata"))
1024       return;
1025   }
1026 
1027   // Skip LLVM intrinsic global variables
1028   if (GVar->getName().startswith("llvm.") ||
1029       GVar->getName().startswith("nvvm."))
1030     return;
1031 
1032   const DataLayout *TD = TM.getDataLayout();
1033 
1034   // GlobalVariables are always constant pointers themselves.
1035   const PointerType *PTy = GVar->getType();
1036   Type *ETy = PTy->getElementType();
1037 
1038   if (GVar->hasExternalLinkage()) {
1039     if (GVar->hasInitializer())
1040       O << ".visible ";
1041     else
1042       O << ".extern ";
1043   } else if (GVar->hasLinkOnceLinkage() || GVar->hasWeakLinkage() ||
1044              GVar->hasAvailableExternallyLinkage() ||
1045              GVar->hasCommonLinkage()) {
1046     O << ".weak ";
1047   }
1048 
1049   if (llvm::isTexture(*GVar)) {
1050     O << ".global .texref " << llvm::getTextureName(*GVar) << ";\n";
1051     return;
1052   }
1053 
1054   if (llvm::isSurface(*GVar)) {
1055     O << ".global .surfref " << llvm::getSurfaceName(*GVar) << ";\n";
1056     return;
1057   }
1058 
1059   if (GVar->isDeclaration()) {
1060     // (extern) declarations, no definition or initializer
1061     // Currently the only known declaration is for an automatic __local
1062     // (.shared) promoted to global.
1063     emitPTXGlobalVariable(GVar, O);
1064     O << ";\n";
1065     return;
1066   }
1067 
1068   if (llvm::isSampler(*GVar)) {
1069     O << ".global .samplerref " << llvm::getSamplerName(*GVar);
1070 
1071     const Constant *Initializer = nullptr;
1072     if (GVar->hasInitializer())
1073       Initializer = GVar->getInitializer();
1074     const ConstantInt *CI = nullptr;
1075     if (Initializer)
1076       CI = dyn_cast<ConstantInt>(Initializer);
1077     if (CI) {
1078       unsigned sample = CI->getZExtValue();
1079 
1080       O << " = { ";
1081 
1082       for (int i = 0,
1083                addr = ((sample & __CLK_ADDRESS_MASK) >> __CLK_ADDRESS_BASE);
1084            i < 3; i++) {
1085         O << "addr_mode_" << i << " = ";
1086         switch (addr) {
1087         case 0:
1088           O << "wrap";
1089           break;
1090         case 1:
1091           O << "clamp_to_border";
1092           break;
1093         case 2:
1094           O << "clamp_to_edge";
1095           break;
1096         case 3:
1097           O << "wrap";
1098           break;
1099         case 4:
1100           O << "mirror";
1101           break;
1102         }
1103         O << ", ";
1104       }
1105       O << "filter_mode = ";
1106       switch ((sample & __CLK_FILTER_MASK) >> __CLK_FILTER_BASE) {
1107       case 0:
1108         O << "nearest";
1109         break;
1110       case 1:
1111         O << "linear";
1112         break;
1113       case 2:
1114         llvm_unreachable("Anisotropic filtering is not supported");
1115       default:
1116         O << "nearest";
1117         break;
1118       }
1119       if (!((sample & __CLK_NORMALIZED_MASK) >> __CLK_NORMALIZED_BASE)) {
1120         O << ", force_unnormalized_coords = 1";
1121       }
1122       O << " }";
1123     }
1124 
1125     O << ";\n";
1126     return;
1127   }
1128 
1129   if (GVar->hasPrivateLinkage()) {
1130 
1131     if (!strncmp(GVar->getName().data(), "unrollpragma", 12))
1132       return;
1133 
1134     // FIXME - need better way (e.g. Metadata) to avoid generating this global
1135     if (!strncmp(GVar->getName().data(), "filename", 8))
1136       return;
1137     if (GVar->use_empty())
1138       return;
1139   }
1140 
1141   const Function *demotedFunc = nullptr;
1142   if (!processDemoted && canDemoteGlobalVar(GVar, demotedFunc)) {
1143     O << "// " << GVar->getName() << " has been demoted\n";
1144     if (localDecls.find(demotedFunc) != localDecls.end())
1145       localDecls[demotedFunc].push_back(GVar);
1146     else {
1147       std::vector<const GlobalVariable *> temp;
1148       temp.push_back(GVar);
1149       localDecls[demotedFunc] = temp;
1150     }
1151     return;
1152   }
1153 
1154   O << ".";
1155   emitPTXAddressSpace(PTy->getAddressSpace(), O);
1156 
1157   if (isManaged(*GVar)) {
1158     O << " .attribute(.managed)";
1159   }
1160 
1161   if (GVar->getAlignment() == 0)
1162     O << " .align " << (int) TD->getPrefTypeAlignment(ETy);
1163   else
1164     O << " .align " << GVar->getAlignment();
1165 
1166   if (ETy->isFloatingPointTy() || ETy->isIntegerTy() || ETy->isPointerTy()) {
1167     O << " .";
1168     // Special case: ABI requires that we use .u8 for predicates
1169     if (ETy->isIntegerTy(1))
1170       O << "u8";
1171     else
1172       O << getPTXFundamentalTypeStr(ETy, false);
1173     O << " ";
1174     O << *getSymbol(GVar);
1175 
1176     // Ptx allows variable initilization only for constant and global state
1177     // spaces.
1178     if (GVar->hasInitializer()) {
1179       if ((PTy->getAddressSpace() == llvm::ADDRESS_SPACE_GLOBAL) ||
1180           (PTy->getAddressSpace() == llvm::ADDRESS_SPACE_CONST)) {
1181         const Constant *Initializer = GVar->getInitializer();
1182         // 'undef' is treated as there is no value spefied.
1183         if (!Initializer->isNullValue() && !isa<UndefValue>(Initializer)) {
1184           O << " = ";
1185           printScalarConstant(Initializer, O);
1186         }
1187       } else {
1188         // The frontend adds zero-initializer to variables that don't have an
1189         // initial value, so skip warning for this case.
1190         if (!GVar->getInitializer()->isNullValue()) {
1191           std::string warnMsg =
1192               ("initial value of '" + GVar->getName() +
1193                "' is not allowed in addrspace(" +
1194                Twine(llvm::utostr_32(PTy->getAddressSpace())) + ")").str();
1195           report_fatal_error(warnMsg.c_str());
1196         }
1197       }
1198     }
1199   } else {
1200     unsigned int ElementSize = 0;
1201 
1202     // Although PTX has direct support for struct type and array type and
1203     // LLVM IR is very similar to PTX, the LLVM CodeGen does not support for
1204     // targets that support these high level field accesses. Structs, arrays
1205     // and vectors are lowered into arrays of bytes.
1206     switch (ETy->getTypeID()) {
1207     case Type::StructTyID:
1208     case Type::ArrayTyID:
1209     case Type::VectorTyID:
1210       ElementSize = TD->getTypeStoreSize(ETy);
1211       // Ptx allows variable initilization only for constant and
1212       // global state spaces.
1213       if (((PTy->getAddressSpace() == llvm::ADDRESS_SPACE_GLOBAL) ||
1214            (PTy->getAddressSpace() == llvm::ADDRESS_SPACE_CONST)) &&
1215           GVar->hasInitializer()) {
1216         const Constant *Initializer = GVar->getInitializer();
1217         if (!isa<UndefValue>(Initializer) && !Initializer->isNullValue()) {
1218           AggBuffer aggBuffer(ElementSize, O, *this);
1219           bufferAggregateConstant(Initializer, &aggBuffer);
1220           if (aggBuffer.numSymbols) {
1221             if (static_cast<const NVPTXTargetMachine &>(TM).is64Bit()) {
1222               O << " .u64 " << *getSymbol(GVar) << "[";
1223               O << ElementSize / 8;
1224             } else {
1225               O << " .u32 " << *getSymbol(GVar) << "[";
1226               O << ElementSize / 4;
1227             }
1228             O << "]";
1229           } else {
1230             O << " .b8 " << *getSymbol(GVar) << "[";
1231             O << ElementSize;
1232             O << "]";
1233           }
1234           O << " = {";
1235           aggBuffer.print();
1236           O << "}";
1237         } else {
1238           O << " .b8 " << *getSymbol(GVar);
1239           if (ElementSize) {
1240             O << "[";
1241             O << ElementSize;
1242             O << "]";
1243           }
1244         }
1245       } else {
1246         O << " .b8 " << *getSymbol(GVar);
1247         if (ElementSize) {
1248           O << "[";
1249           O << ElementSize;
1250           O << "]";
1251         }
1252       }
1253       break;
1254     default:
1255       llvm_unreachable("type not supported yet");
1256     }
1257 
1258   }
1259   O << ";\n";
1260 }
1261 
emitDemotedVars(const Function * f,raw_ostream & O)1262 void NVPTXAsmPrinter::emitDemotedVars(const Function *f, raw_ostream &O) {
1263   if (localDecls.find(f) == localDecls.end())
1264     return;
1265 
1266   std::vector<const GlobalVariable *> &gvars = localDecls[f];
1267 
1268   for (unsigned i = 0, e = gvars.size(); i != e; ++i) {
1269     O << "\t// demoted variable\n\t";
1270     printModuleLevelGV(gvars[i], O, true);
1271   }
1272 }
1273 
emitPTXAddressSpace(unsigned int AddressSpace,raw_ostream & O) const1274 void NVPTXAsmPrinter::emitPTXAddressSpace(unsigned int AddressSpace,
1275                                           raw_ostream &O) const {
1276   switch (AddressSpace) {
1277   case llvm::ADDRESS_SPACE_LOCAL:
1278     O << "local";
1279     break;
1280   case llvm::ADDRESS_SPACE_GLOBAL:
1281     O << "global";
1282     break;
1283   case llvm::ADDRESS_SPACE_CONST:
1284     O << "const";
1285     break;
1286   case llvm::ADDRESS_SPACE_SHARED:
1287     O << "shared";
1288     break;
1289   default:
1290     report_fatal_error("Bad address space found while emitting PTX");
1291     break;
1292   }
1293 }
1294 
1295 std::string
getPTXFundamentalTypeStr(const Type * Ty,bool useB4PTR) const1296 NVPTXAsmPrinter::getPTXFundamentalTypeStr(const Type *Ty, bool useB4PTR) const {
1297   switch (Ty->getTypeID()) {
1298   default:
1299     llvm_unreachable("unexpected type");
1300     break;
1301   case Type::IntegerTyID: {
1302     unsigned NumBits = cast<IntegerType>(Ty)->getBitWidth();
1303     if (NumBits == 1)
1304       return "pred";
1305     else if (NumBits <= 64) {
1306       std::string name = "u";
1307       return name + utostr(NumBits);
1308     } else {
1309       llvm_unreachable("Integer too large");
1310       break;
1311     }
1312     break;
1313   }
1314   case Type::FloatTyID:
1315     return "f32";
1316   case Type::DoubleTyID:
1317     return "f64";
1318   case Type::PointerTyID:
1319     if (static_cast<const NVPTXTargetMachine &>(TM).is64Bit())
1320       if (useB4PTR)
1321         return "b64";
1322       else
1323         return "u64";
1324     else if (useB4PTR)
1325       return "b32";
1326     else
1327       return "u32";
1328   }
1329   llvm_unreachable("unexpected type");
1330   return nullptr;
1331 }
1332 
emitPTXGlobalVariable(const GlobalVariable * GVar,raw_ostream & O)1333 void NVPTXAsmPrinter::emitPTXGlobalVariable(const GlobalVariable *GVar,
1334                                             raw_ostream &O) {
1335 
1336   const DataLayout *TD = TM.getDataLayout();
1337 
1338   // GlobalVariables are always constant pointers themselves.
1339   const PointerType *PTy = GVar->getType();
1340   Type *ETy = PTy->getElementType();
1341 
1342   O << ".";
1343   emitPTXAddressSpace(PTy->getAddressSpace(), O);
1344   if (GVar->getAlignment() == 0)
1345     O << " .align " << (int) TD->getPrefTypeAlignment(ETy);
1346   else
1347     O << " .align " << GVar->getAlignment();
1348 
1349   if (ETy->isFloatingPointTy() || ETy->isIntegerTy() || ETy->isPointerTy()) {
1350     O << " .";
1351     O << getPTXFundamentalTypeStr(ETy);
1352     O << " ";
1353     O << *getSymbol(GVar);
1354     return;
1355   }
1356 
1357   int64_t ElementSize = 0;
1358 
1359   // Although PTX has direct support for struct type and array type and LLVM IR
1360   // is very similar to PTX, the LLVM CodeGen does not support for targets that
1361   // support these high level field accesses. Structs and arrays are lowered
1362   // into arrays of bytes.
1363   switch (ETy->getTypeID()) {
1364   case Type::StructTyID:
1365   case Type::ArrayTyID:
1366   case Type::VectorTyID:
1367     ElementSize = TD->getTypeStoreSize(ETy);
1368     O << " .b8 " << *getSymbol(GVar) << "[";
1369     if (ElementSize) {
1370       O << itostr(ElementSize);
1371     }
1372     O << "]";
1373     break;
1374   default:
1375     llvm_unreachable("type not supported yet");
1376   }
1377   return;
1378 }
1379 
getOpenCLAlignment(const DataLayout * TD,Type * Ty)1380 static unsigned int getOpenCLAlignment(const DataLayout *TD, Type *Ty) {
1381   if (Ty->isSingleValueType())
1382     return TD->getPrefTypeAlignment(Ty);
1383 
1384   const ArrayType *ATy = dyn_cast<ArrayType>(Ty);
1385   if (ATy)
1386     return getOpenCLAlignment(TD, ATy->getElementType());
1387 
1388   const StructType *STy = dyn_cast<StructType>(Ty);
1389   if (STy) {
1390     unsigned int alignStruct = 1;
1391     // Go through each element of the struct and find the
1392     // largest alignment.
1393     for (unsigned i = 0, e = STy->getNumElements(); i != e; i++) {
1394       Type *ETy = STy->getElementType(i);
1395       unsigned int align = getOpenCLAlignment(TD, ETy);
1396       if (align > alignStruct)
1397         alignStruct = align;
1398     }
1399     return alignStruct;
1400   }
1401 
1402   const FunctionType *FTy = dyn_cast<FunctionType>(Ty);
1403   if (FTy)
1404     return TD->getPointerPrefAlignment();
1405   return TD->getPrefTypeAlignment(Ty);
1406 }
1407 
printParamName(Function::const_arg_iterator I,int paramIndex,raw_ostream & O)1408 void NVPTXAsmPrinter::printParamName(Function::const_arg_iterator I,
1409                                      int paramIndex, raw_ostream &O) {
1410   O << *getSymbol(I->getParent()) << "_param_" << paramIndex;
1411 }
1412 
printParamName(int paramIndex,raw_ostream & O)1413 void NVPTXAsmPrinter::printParamName(int paramIndex, raw_ostream &O) {
1414   O << *CurrentFnSym << "_param_" << paramIndex;
1415 }
1416 
emitFunctionParamList(const Function * F,raw_ostream & O)1417 void NVPTXAsmPrinter::emitFunctionParamList(const Function *F, raw_ostream &O) {
1418   const DataLayout *TD = TM.getDataLayout();
1419   const AttributeSet &PAL = F->getAttributes();
1420   const TargetLowering *TLI = nvptxSubtarget->getTargetLowering();
1421   Function::const_arg_iterator I, E;
1422   unsigned paramIndex = 0;
1423   bool first = true;
1424   bool isKernelFunc = llvm::isKernelFunction(*F);
1425   bool isABI = (nvptxSubtarget->getSmVersion() >= 20);
1426   MVT thePointerTy = TLI->getPointerTy();
1427 
1428   O << "(\n";
1429 
1430   for (I = F->arg_begin(), E = F->arg_end(); I != E; ++I, paramIndex++) {
1431     Type *Ty = I->getType();
1432 
1433     if (!first)
1434       O << ",\n";
1435 
1436     first = false;
1437 
1438     // Handle image/sampler parameters
1439     if (isKernelFunction(*F)) {
1440       if (isSampler(*I) || isImage(*I)) {
1441         if (isImage(*I)) {
1442           std::string sname = I->getName();
1443           if (isImageWriteOnly(*I) || isImageReadWrite(*I)) {
1444             if (nvptxSubtarget->hasImageHandles())
1445               O << "\t.param .u64 .ptr .surfref ";
1446             else
1447               O << "\t.param .surfref ";
1448             O << *CurrentFnSym << "_param_" << paramIndex;
1449           }
1450           else { // Default image is read_only
1451             if (nvptxSubtarget->hasImageHandles())
1452               O << "\t.param .u64 .ptr .texref ";
1453             else
1454               O << "\t.param .texref ";
1455             O << *CurrentFnSym << "_param_" << paramIndex;
1456           }
1457         } else {
1458           if (nvptxSubtarget->hasImageHandles())
1459             O << "\t.param .u64 .ptr .samplerref ";
1460           else
1461             O << "\t.param .samplerref ";
1462           O << *CurrentFnSym << "_param_" << paramIndex;
1463         }
1464         continue;
1465       }
1466     }
1467 
1468     if (!PAL.hasAttribute(paramIndex + 1, Attribute::ByVal)) {
1469       if (Ty->isAggregateType() || Ty->isVectorTy()) {
1470         // Just print .param .align <a> .b8 .param[size];
1471         // <a> = PAL.getparamalignment
1472         // size = typeallocsize of element type
1473         unsigned align = PAL.getParamAlignment(paramIndex + 1);
1474         if (align == 0)
1475           align = TD->getABITypeAlignment(Ty);
1476 
1477         unsigned sz = TD->getTypeAllocSize(Ty);
1478         O << "\t.param .align " << align << " .b8 ";
1479         printParamName(I, paramIndex, O);
1480         O << "[" << sz << "]";
1481 
1482         continue;
1483       }
1484       // Just a scalar
1485       const PointerType *PTy = dyn_cast<PointerType>(Ty);
1486       if (isKernelFunc) {
1487         if (PTy) {
1488           // Special handling for pointer arguments to kernel
1489           O << "\t.param .u" << thePointerTy.getSizeInBits() << " ";
1490 
1491           if (static_cast<NVPTXTargetMachine &>(TM).getDrvInterface() !=
1492               NVPTX::CUDA) {
1493             Type *ETy = PTy->getElementType();
1494             int addrSpace = PTy->getAddressSpace();
1495             switch (addrSpace) {
1496             default:
1497               O << ".ptr ";
1498               break;
1499             case llvm::ADDRESS_SPACE_CONST:
1500               O << ".ptr .const ";
1501               break;
1502             case llvm::ADDRESS_SPACE_SHARED:
1503               O << ".ptr .shared ";
1504               break;
1505             case llvm::ADDRESS_SPACE_GLOBAL:
1506               O << ".ptr .global ";
1507               break;
1508             }
1509             O << ".align " << (int) getOpenCLAlignment(TD, ETy) << " ";
1510           }
1511           printParamName(I, paramIndex, O);
1512           continue;
1513         }
1514 
1515         // non-pointer scalar to kernel func
1516         O << "\t.param .";
1517         // Special case: predicate operands become .u8 types
1518         if (Ty->isIntegerTy(1))
1519           O << "u8";
1520         else
1521           O << getPTXFundamentalTypeStr(Ty);
1522         O << " ";
1523         printParamName(I, paramIndex, O);
1524         continue;
1525       }
1526       // Non-kernel function, just print .param .b<size> for ABI
1527       // and .reg .b<size> for non-ABI
1528       unsigned sz = 0;
1529       if (isa<IntegerType>(Ty)) {
1530         sz = cast<IntegerType>(Ty)->getBitWidth();
1531         if (sz < 32)
1532           sz = 32;
1533       } else if (isa<PointerType>(Ty))
1534         sz = thePointerTy.getSizeInBits();
1535       else
1536         sz = Ty->getPrimitiveSizeInBits();
1537       if (isABI)
1538         O << "\t.param .b" << sz << " ";
1539       else
1540         O << "\t.reg .b" << sz << " ";
1541       printParamName(I, paramIndex, O);
1542       continue;
1543     }
1544 
1545     // param has byVal attribute. So should be a pointer
1546     const PointerType *PTy = dyn_cast<PointerType>(Ty);
1547     assert(PTy && "Param with byval attribute should be a pointer type");
1548     Type *ETy = PTy->getElementType();
1549 
1550     if (isABI || isKernelFunc) {
1551       // Just print .param .align <a> .b8 .param[size];
1552       // <a> = PAL.getparamalignment
1553       // size = typeallocsize of element type
1554       unsigned align = PAL.getParamAlignment(paramIndex + 1);
1555       if (align == 0)
1556         align = TD->getABITypeAlignment(ETy);
1557 
1558       unsigned sz = TD->getTypeAllocSize(ETy);
1559       O << "\t.param .align " << align << " .b8 ";
1560       printParamName(I, paramIndex, O);
1561       O << "[" << sz << "]";
1562       continue;
1563     } else {
1564       // Split the ETy into constituent parts and
1565       // print .param .b<size> <name> for each part.
1566       // Further, if a part is vector, print the above for
1567       // each vector element.
1568       SmallVector<EVT, 16> vtparts;
1569       ComputeValueVTs(*TLI, ETy, vtparts);
1570       for (unsigned i = 0, e = vtparts.size(); i != e; ++i) {
1571         unsigned elems = 1;
1572         EVT elemtype = vtparts[i];
1573         if (vtparts[i].isVector()) {
1574           elems = vtparts[i].getVectorNumElements();
1575           elemtype = vtparts[i].getVectorElementType();
1576         }
1577 
1578         for (unsigned j = 0, je = elems; j != je; ++j) {
1579           unsigned sz = elemtype.getSizeInBits();
1580           if (elemtype.isInteger() && (sz < 32))
1581             sz = 32;
1582           O << "\t.reg .b" << sz << " ";
1583           printParamName(I, paramIndex, O);
1584           if (j < je - 1)
1585             O << ",\n";
1586           ++paramIndex;
1587         }
1588         if (i < e - 1)
1589           O << ",\n";
1590       }
1591       --paramIndex;
1592       continue;
1593     }
1594   }
1595 
1596   O << "\n)\n";
1597 }
1598 
emitFunctionParamList(const MachineFunction & MF,raw_ostream & O)1599 void NVPTXAsmPrinter::emitFunctionParamList(const MachineFunction &MF,
1600                                             raw_ostream &O) {
1601   const Function *F = MF.getFunction();
1602   emitFunctionParamList(F, O);
1603 }
1604 
setAndEmitFunctionVirtualRegisters(const MachineFunction & MF)1605 void NVPTXAsmPrinter::setAndEmitFunctionVirtualRegisters(
1606     const MachineFunction &MF) {
1607   SmallString<128> Str;
1608   raw_svector_ostream O(Str);
1609 
1610   // Map the global virtual register number to a register class specific
1611   // virtual register number starting from 1 with that class.
1612   const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo();
1613   //unsigned numRegClasses = TRI->getNumRegClasses();
1614 
1615   // Emit the Fake Stack Object
1616   const MachineFrameInfo *MFI = MF.getFrameInfo();
1617   int NumBytes = (int) MFI->getStackSize();
1618   if (NumBytes) {
1619     O << "\t.local .align " << MFI->getMaxAlignment() << " .b8 \t" << DEPOTNAME
1620       << getFunctionNumber() << "[" << NumBytes << "];\n";
1621     if (static_cast<const NVPTXTargetMachine &>(MF.getTarget()).is64Bit()) {
1622       O << "\t.reg .b64 \t%SP;\n";
1623       O << "\t.reg .b64 \t%SPL;\n";
1624     } else {
1625       O << "\t.reg .b32 \t%SP;\n";
1626       O << "\t.reg .b32 \t%SPL;\n";
1627     }
1628   }
1629 
1630   // Go through all virtual registers to establish the mapping between the
1631   // global virtual
1632   // register number and the per class virtual register number.
1633   // We use the per class virtual register number in the ptx output.
1634   unsigned int numVRs = MRI->getNumVirtRegs();
1635   for (unsigned i = 0; i < numVRs; i++) {
1636     unsigned int vr = TRI->index2VirtReg(i);
1637     const TargetRegisterClass *RC = MRI->getRegClass(vr);
1638     DenseMap<unsigned, unsigned> &regmap = VRegMapping[RC];
1639     int n = regmap.size();
1640     regmap.insert(std::make_pair(vr, n + 1));
1641   }
1642 
1643   // Emit register declarations
1644   // @TODO: Extract out the real register usage
1645   // O << "\t.reg .pred %p<" << NVPTXNumRegisters << ">;\n";
1646   // O << "\t.reg .s16 %rc<" << NVPTXNumRegisters << ">;\n";
1647   // O << "\t.reg .s16 %rs<" << NVPTXNumRegisters << ">;\n";
1648   // O << "\t.reg .s32 %r<" << NVPTXNumRegisters << ">;\n";
1649   // O << "\t.reg .s64 %rd<" << NVPTXNumRegisters << ">;\n";
1650   // O << "\t.reg .f32 %f<" << NVPTXNumRegisters << ">;\n";
1651   // O << "\t.reg .f64 %fd<" << NVPTXNumRegisters << ">;\n";
1652 
1653   // Emit declaration of the virtual registers or 'physical' registers for
1654   // each register class
1655   for (unsigned i=0; i< TRI->getNumRegClasses(); i++) {
1656     const TargetRegisterClass *RC = TRI->getRegClass(i);
1657     DenseMap<unsigned, unsigned> &regmap = VRegMapping[RC];
1658     std::string rcname = getNVPTXRegClassName(RC);
1659     std::string rcStr = getNVPTXRegClassStr(RC);
1660     int n = regmap.size();
1661 
1662     // Only declare those registers that may be used.
1663     if (n) {
1664        O << "\t.reg " << rcname << " \t" << rcStr << "<" << (n+1)
1665          << ">;\n";
1666     }
1667   }
1668 
1669   OutStreamer.EmitRawText(O.str());
1670 }
1671 
printFPConstant(const ConstantFP * Fp,raw_ostream & O)1672 void NVPTXAsmPrinter::printFPConstant(const ConstantFP *Fp, raw_ostream &O) {
1673   APFloat APF = APFloat(Fp->getValueAPF()); // make a copy
1674   bool ignored;
1675   unsigned int numHex;
1676   const char *lead;
1677 
1678   if (Fp->getType()->getTypeID() == Type::FloatTyID) {
1679     numHex = 8;
1680     lead = "0f";
1681     APF.convert(APFloat::IEEEsingle, APFloat::rmNearestTiesToEven, &ignored);
1682   } else if (Fp->getType()->getTypeID() == Type::DoubleTyID) {
1683     numHex = 16;
1684     lead = "0d";
1685     APF.convert(APFloat::IEEEdouble, APFloat::rmNearestTiesToEven, &ignored);
1686   } else
1687     llvm_unreachable("unsupported fp type");
1688 
1689   APInt API = APF.bitcastToAPInt();
1690   std::string hexstr(utohexstr(API.getZExtValue()));
1691   O << lead;
1692   if (hexstr.length() < numHex)
1693     O << std::string(numHex - hexstr.length(), '0');
1694   O << utohexstr(API.getZExtValue());
1695 }
1696 
printScalarConstant(const Constant * CPV,raw_ostream & O)1697 void NVPTXAsmPrinter::printScalarConstant(const Constant *CPV, raw_ostream &O) {
1698   if (const ConstantInt *CI = dyn_cast<ConstantInt>(CPV)) {
1699     O << CI->getValue();
1700     return;
1701   }
1702   if (const ConstantFP *CFP = dyn_cast<ConstantFP>(CPV)) {
1703     printFPConstant(CFP, O);
1704     return;
1705   }
1706   if (isa<ConstantPointerNull>(CPV)) {
1707     O << "0";
1708     return;
1709   }
1710   if (const GlobalValue *GVar = dyn_cast<GlobalValue>(CPV)) {
1711     PointerType *PTy = dyn_cast<PointerType>(GVar->getType());
1712     bool IsNonGenericPointer = false;
1713     if (PTy && PTy->getAddressSpace() != 0) {
1714       IsNonGenericPointer = true;
1715     }
1716     if (EmitGeneric && !isa<Function>(CPV) && !IsNonGenericPointer) {
1717       O << "generic(";
1718       O << *getSymbol(GVar);
1719       O << ")";
1720     } else {
1721       O << *getSymbol(GVar);
1722     }
1723     return;
1724   }
1725   if (const ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
1726     const Value *v = Cexpr->stripPointerCasts();
1727     PointerType *PTy = dyn_cast<PointerType>(Cexpr->getType());
1728     bool IsNonGenericPointer = false;
1729     if (PTy && PTy->getAddressSpace() != 0) {
1730       IsNonGenericPointer = true;
1731     }
1732     if (const GlobalValue *GVar = dyn_cast<GlobalValue>(v)) {
1733       if (EmitGeneric && !isa<Function>(v) && !IsNonGenericPointer) {
1734         O << "generic(";
1735         O << *getSymbol(GVar);
1736         O << ")";
1737       } else {
1738         O << *getSymbol(GVar);
1739       }
1740       return;
1741     } else {
1742       O << *lowerConstant(CPV);
1743       return;
1744     }
1745   }
1746   llvm_unreachable("Not scalar type found in printScalarConstant()");
1747 }
1748 
bufferLEByte(const Constant * CPV,int Bytes,AggBuffer * aggBuffer)1749 void NVPTXAsmPrinter::bufferLEByte(const Constant *CPV, int Bytes,
1750                                    AggBuffer *aggBuffer) {
1751 
1752   const DataLayout *TD = TM.getDataLayout();
1753 
1754   if (isa<UndefValue>(CPV) || CPV->isNullValue()) {
1755     int s = TD->getTypeAllocSize(CPV->getType());
1756     if (s < Bytes)
1757       s = Bytes;
1758     aggBuffer->addZeros(s);
1759     return;
1760   }
1761 
1762   unsigned char *ptr;
1763   switch (CPV->getType()->getTypeID()) {
1764 
1765   case Type::IntegerTyID: {
1766     const Type *ETy = CPV->getType();
1767     if (ETy == Type::getInt8Ty(CPV->getContext())) {
1768       unsigned char c = (unsigned char)cast<ConstantInt>(CPV)->getZExtValue();
1769       ptr = &c;
1770       aggBuffer->addBytes(ptr, 1, Bytes);
1771     } else if (ETy == Type::getInt16Ty(CPV->getContext())) {
1772       short int16 = (short)cast<ConstantInt>(CPV)->getZExtValue();
1773       ptr = (unsigned char *)&int16;
1774       aggBuffer->addBytes(ptr, 2, Bytes);
1775     } else if (ETy == Type::getInt32Ty(CPV->getContext())) {
1776       if (const ConstantInt *constInt = dyn_cast<ConstantInt>(CPV)) {
1777         int int32 = (int)(constInt->getZExtValue());
1778         ptr = (unsigned char *)&int32;
1779         aggBuffer->addBytes(ptr, 4, Bytes);
1780         break;
1781       } else if (const ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
1782         if (const ConstantInt *constInt = dyn_cast<ConstantInt>(
1783                 ConstantFoldConstantExpression(Cexpr, *TD))) {
1784           int int32 = (int)(constInt->getZExtValue());
1785           ptr = (unsigned char *)&int32;
1786           aggBuffer->addBytes(ptr, 4, Bytes);
1787           break;
1788         }
1789         if (Cexpr->getOpcode() == Instruction::PtrToInt) {
1790           Value *v = Cexpr->getOperand(0)->stripPointerCasts();
1791           aggBuffer->addSymbol(v);
1792           aggBuffer->addZeros(4);
1793           break;
1794         }
1795       }
1796       llvm_unreachable("unsupported integer const type");
1797     } else if (ETy == Type::getInt64Ty(CPV->getContext())) {
1798       if (const ConstantInt *constInt = dyn_cast<ConstantInt>(CPV)) {
1799         long long int64 = (long long)(constInt->getZExtValue());
1800         ptr = (unsigned char *)&int64;
1801         aggBuffer->addBytes(ptr, 8, Bytes);
1802         break;
1803       } else if (const ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
1804         if (const ConstantInt *constInt = dyn_cast<ConstantInt>(
1805                 ConstantFoldConstantExpression(Cexpr, *TD))) {
1806           long long int64 = (long long)(constInt->getZExtValue());
1807           ptr = (unsigned char *)&int64;
1808           aggBuffer->addBytes(ptr, 8, Bytes);
1809           break;
1810         }
1811         if (Cexpr->getOpcode() == Instruction::PtrToInt) {
1812           Value *v = Cexpr->getOperand(0)->stripPointerCasts();
1813           aggBuffer->addSymbol(v);
1814           aggBuffer->addZeros(8);
1815           break;
1816         }
1817       }
1818       llvm_unreachable("unsupported integer const type");
1819     } else
1820       llvm_unreachable("unsupported integer const type");
1821     break;
1822   }
1823   case Type::FloatTyID:
1824   case Type::DoubleTyID: {
1825     const ConstantFP *CFP = dyn_cast<ConstantFP>(CPV);
1826     const Type *Ty = CFP->getType();
1827     if (Ty == Type::getFloatTy(CPV->getContext())) {
1828       float float32 = (float) CFP->getValueAPF().convertToFloat();
1829       ptr = (unsigned char *)&float32;
1830       aggBuffer->addBytes(ptr, 4, Bytes);
1831     } else if (Ty == Type::getDoubleTy(CPV->getContext())) {
1832       double float64 = CFP->getValueAPF().convertToDouble();
1833       ptr = (unsigned char *)&float64;
1834       aggBuffer->addBytes(ptr, 8, Bytes);
1835     } else {
1836       llvm_unreachable("unsupported fp const type");
1837     }
1838     break;
1839   }
1840   case Type::PointerTyID: {
1841     if (const GlobalValue *GVar = dyn_cast<GlobalValue>(CPV)) {
1842       aggBuffer->addSymbol(GVar);
1843     } else if (const ConstantExpr *Cexpr = dyn_cast<ConstantExpr>(CPV)) {
1844       const Value *v = Cexpr->stripPointerCasts();
1845       aggBuffer->addSymbol(v);
1846     }
1847     unsigned int s = TD->getTypeAllocSize(CPV->getType());
1848     aggBuffer->addZeros(s);
1849     break;
1850   }
1851 
1852   case Type::ArrayTyID:
1853   case Type::VectorTyID:
1854   case Type::StructTyID: {
1855     if (isa<ConstantArray>(CPV) || isa<ConstantVector>(CPV) ||
1856         isa<ConstantStruct>(CPV) || isa<ConstantDataSequential>(CPV)) {
1857       int ElementSize = TD->getTypeAllocSize(CPV->getType());
1858       bufferAggregateConstant(CPV, aggBuffer);
1859       if (Bytes > ElementSize)
1860         aggBuffer->addZeros(Bytes - ElementSize);
1861     } else if (isa<ConstantAggregateZero>(CPV))
1862       aggBuffer->addZeros(Bytes);
1863     else
1864       llvm_unreachable("Unexpected Constant type");
1865     break;
1866   }
1867 
1868   default:
1869     llvm_unreachable("unsupported type");
1870   }
1871 }
1872 
bufferAggregateConstant(const Constant * CPV,AggBuffer * aggBuffer)1873 void NVPTXAsmPrinter::bufferAggregateConstant(const Constant *CPV,
1874                                               AggBuffer *aggBuffer) {
1875   const DataLayout *TD = TM.getDataLayout();
1876   int Bytes;
1877 
1878   // Old constants
1879   if (isa<ConstantArray>(CPV) || isa<ConstantVector>(CPV)) {
1880     if (CPV->getNumOperands())
1881       for (unsigned i = 0, e = CPV->getNumOperands(); i != e; ++i)
1882         bufferLEByte(cast<Constant>(CPV->getOperand(i)), 0, aggBuffer);
1883     return;
1884   }
1885 
1886   if (const ConstantDataSequential *CDS =
1887           dyn_cast<ConstantDataSequential>(CPV)) {
1888     if (CDS->getNumElements())
1889       for (unsigned i = 0; i < CDS->getNumElements(); ++i)
1890         bufferLEByte(cast<Constant>(CDS->getElementAsConstant(i)), 0,
1891                      aggBuffer);
1892     return;
1893   }
1894 
1895   if (isa<ConstantStruct>(CPV)) {
1896     if (CPV->getNumOperands()) {
1897       StructType *ST = cast<StructType>(CPV->getType());
1898       for (unsigned i = 0, e = CPV->getNumOperands(); i != e; ++i) {
1899         if (i == (e - 1))
1900           Bytes = TD->getStructLayout(ST)->getElementOffset(0) +
1901                   TD->getTypeAllocSize(ST) -
1902                   TD->getStructLayout(ST)->getElementOffset(i);
1903         else
1904           Bytes = TD->getStructLayout(ST)->getElementOffset(i + 1) -
1905                   TD->getStructLayout(ST)->getElementOffset(i);
1906         bufferLEByte(cast<Constant>(CPV->getOperand(i)), Bytes, aggBuffer);
1907       }
1908     }
1909     return;
1910   }
1911   llvm_unreachable("unsupported constant type in printAggregateConstant()");
1912 }
1913 
1914 // buildTypeNameMap - Run through symbol table looking for type names.
1915 //
1916 
isImageType(const Type * Ty)1917 bool NVPTXAsmPrinter::isImageType(const Type *Ty) {
1918 
1919   std::map<const Type *, std::string>::iterator PI = TypeNameMap.find(Ty);
1920 
1921   if (PI != TypeNameMap.end() && (!PI->second.compare("struct._image1d_t") ||
1922                                   !PI->second.compare("struct._image2d_t") ||
1923                                   !PI->second.compare("struct._image3d_t")))
1924     return true;
1925 
1926   return false;
1927 }
1928 
1929 
ignoreLoc(const MachineInstr & MI)1930 bool NVPTXAsmPrinter::ignoreLoc(const MachineInstr &MI) {
1931   switch (MI.getOpcode()) {
1932   default:
1933     return false;
1934   case NVPTX::CallArgBeginInst:
1935   case NVPTX::CallArgEndInst0:
1936   case NVPTX::CallArgEndInst1:
1937   case NVPTX::CallArgF32:
1938   case NVPTX::CallArgF64:
1939   case NVPTX::CallArgI16:
1940   case NVPTX::CallArgI32:
1941   case NVPTX::CallArgI32imm:
1942   case NVPTX::CallArgI64:
1943   case NVPTX::CallArgParam:
1944   case NVPTX::CallVoidInst:
1945   case NVPTX::CallVoidInstReg:
1946   case NVPTX::Callseq_End:
1947   case NVPTX::CallVoidInstReg64:
1948   case NVPTX::DeclareParamInst:
1949   case NVPTX::DeclareRetMemInst:
1950   case NVPTX::DeclareRetRegInst:
1951   case NVPTX::DeclareRetScalarInst:
1952   case NVPTX::DeclareScalarParamInst:
1953   case NVPTX::DeclareScalarRegInst:
1954   case NVPTX::StoreParamF32:
1955   case NVPTX::StoreParamF64:
1956   case NVPTX::StoreParamI16:
1957   case NVPTX::StoreParamI32:
1958   case NVPTX::StoreParamI64:
1959   case NVPTX::StoreParamI8:
1960   case NVPTX::StoreRetvalF32:
1961   case NVPTX::StoreRetvalF64:
1962   case NVPTX::StoreRetvalI16:
1963   case NVPTX::StoreRetvalI32:
1964   case NVPTX::StoreRetvalI64:
1965   case NVPTX::StoreRetvalI8:
1966   case NVPTX::LastCallArgF32:
1967   case NVPTX::LastCallArgF64:
1968   case NVPTX::LastCallArgI16:
1969   case NVPTX::LastCallArgI32:
1970   case NVPTX::LastCallArgI32imm:
1971   case NVPTX::LastCallArgI64:
1972   case NVPTX::LastCallArgParam:
1973   case NVPTX::LoadParamMemF32:
1974   case NVPTX::LoadParamMemF64:
1975   case NVPTX::LoadParamMemI16:
1976   case NVPTX::LoadParamMemI32:
1977   case NVPTX::LoadParamMemI64:
1978   case NVPTX::LoadParamMemI8:
1979   case NVPTX::PrototypeInst:
1980   case NVPTX::DBG_VALUE:
1981     return true;
1982   }
1983   return false;
1984 }
1985 
1986 /// PrintAsmOperand - Print out an operand for an inline asm expression.
1987 ///
PrintAsmOperand(const MachineInstr * MI,unsigned OpNo,unsigned AsmVariant,const char * ExtraCode,raw_ostream & O)1988 bool NVPTXAsmPrinter::PrintAsmOperand(const MachineInstr *MI, unsigned OpNo,
1989                                       unsigned AsmVariant,
1990                                       const char *ExtraCode, raw_ostream &O) {
1991   if (ExtraCode && ExtraCode[0]) {
1992     if (ExtraCode[1] != 0)
1993       return true; // Unknown modifier.
1994 
1995     switch (ExtraCode[0]) {
1996     default:
1997       // See if this is a generic print operand
1998       return AsmPrinter::PrintAsmOperand(MI, OpNo, AsmVariant, ExtraCode, O);
1999     case 'r':
2000       break;
2001     }
2002   }
2003 
2004   printOperand(MI, OpNo, O);
2005 
2006   return false;
2007 }
2008 
PrintAsmMemoryOperand(const MachineInstr * MI,unsigned OpNo,unsigned AsmVariant,const char * ExtraCode,raw_ostream & O)2009 bool NVPTXAsmPrinter::PrintAsmMemoryOperand(
2010     const MachineInstr *MI, unsigned OpNo, unsigned AsmVariant,
2011     const char *ExtraCode, raw_ostream &O) {
2012   if (ExtraCode && ExtraCode[0])
2013     return true; // Unknown modifier
2014 
2015   O << '[';
2016   printMemOperand(MI, OpNo, O);
2017   O << ']';
2018 
2019   return false;
2020 }
2021 
printOperand(const MachineInstr * MI,int opNum,raw_ostream & O,const char * Modifier)2022 void NVPTXAsmPrinter::printOperand(const MachineInstr *MI, int opNum,
2023                                    raw_ostream &O, const char *Modifier) {
2024   const MachineOperand &MO = MI->getOperand(opNum);
2025   switch (MO.getType()) {
2026   case MachineOperand::MO_Register:
2027     if (TargetRegisterInfo::isPhysicalRegister(MO.getReg())) {
2028       if (MO.getReg() == NVPTX::VRDepot)
2029         O << DEPOTNAME << getFunctionNumber();
2030       else
2031         O << NVPTXInstPrinter::getRegisterName(MO.getReg());
2032     } else {
2033       emitVirtualRegister(MO.getReg(), O);
2034     }
2035     return;
2036 
2037   case MachineOperand::MO_Immediate:
2038     if (!Modifier)
2039       O << MO.getImm();
2040     else if (strstr(Modifier, "vec") == Modifier)
2041       printVecModifiedImmediate(MO, Modifier, O);
2042     else
2043       llvm_unreachable(
2044           "Don't know how to handle modifier on immediate operand");
2045     return;
2046 
2047   case MachineOperand::MO_FPImmediate:
2048     printFPConstant(MO.getFPImm(), O);
2049     break;
2050 
2051   case MachineOperand::MO_GlobalAddress:
2052     O << *getSymbol(MO.getGlobal());
2053     break;
2054 
2055   case MachineOperand::MO_MachineBasicBlock:
2056     O << *MO.getMBB()->getSymbol();
2057     return;
2058 
2059   default:
2060     llvm_unreachable("Operand type not supported.");
2061   }
2062 }
2063 
printMemOperand(const MachineInstr * MI,int opNum,raw_ostream & O,const char * Modifier)2064 void NVPTXAsmPrinter::printMemOperand(const MachineInstr *MI, int opNum,
2065                                       raw_ostream &O, const char *Modifier) {
2066   printOperand(MI, opNum, O);
2067 
2068   if (Modifier && !strcmp(Modifier, "add")) {
2069     O << ", ";
2070     printOperand(MI, opNum + 1, O);
2071   } else {
2072     if (MI->getOperand(opNum + 1).isImm() &&
2073         MI->getOperand(opNum + 1).getImm() == 0)
2074       return; // don't print ',0' or '+0'
2075     O << "+";
2076     printOperand(MI, opNum + 1, O);
2077   }
2078 }
2079 
emitSrcInText(StringRef filename,unsigned line)2080 void NVPTXAsmPrinter::emitSrcInText(StringRef filename, unsigned line) {
2081   std::stringstream temp;
2082   LineReader *reader = this->getReader(filename);
2083   temp << "\n//";
2084   temp << filename.str();
2085   temp << ":";
2086   temp << line;
2087   temp << " ";
2088   temp << reader->readLine(line);
2089   temp << "\n";
2090   this->OutStreamer.EmitRawText(temp.str());
2091 }
2092 
getReader(std::string filename)2093 LineReader *NVPTXAsmPrinter::getReader(std::string filename) {
2094   if (!reader) {
2095     reader = new LineReader(filename);
2096   }
2097 
2098   if (reader->fileName() != filename) {
2099     delete reader;
2100     reader = new LineReader(filename);
2101   }
2102 
2103   return reader;
2104 }
2105 
readLine(unsigned lineNum)2106 std::string LineReader::readLine(unsigned lineNum) {
2107   if (lineNum < theCurLine) {
2108     theCurLine = 0;
2109     fstr.seekg(0, std::ios::beg);
2110   }
2111   while (theCurLine < lineNum) {
2112     fstr.getline(buff, 500);
2113     theCurLine++;
2114   }
2115   return buff;
2116 }
2117 
2118 // Force static initialization.
LLVMInitializeNVPTXAsmPrinter()2119 extern "C" void LLVMInitializeNVPTXAsmPrinter() {
2120   RegisterAsmPrinter<NVPTXAsmPrinter> X(TheNVPTXTarget32);
2121   RegisterAsmPrinter<NVPTXAsmPrinter> Y(TheNVPTXTarget64);
2122 }
2123