ConstantFolding.cpp revision f0443c1eb44d737d9bd78962932fc80f74c6113c
1//===-- ConstantFolding.cpp - Analyze constant folding possibilities ------===//
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 family of functions determines the possibility of performing constant
11// folding.
12//
13//===----------------------------------------------------------------------===//
14
15#include "llvm/Analysis/ConstantFolding.h"
16#include "llvm/Constants.h"
17#include "llvm/DerivedTypes.h"
18#include "llvm/Function.h"
19#include "llvm/GlobalVariable.h"
20#include "llvm/Instructions.h"
21#include "llvm/Intrinsics.h"
22#include "llvm/LLVMContext.h"
23#include "llvm/ADT/SmallVector.h"
24#include "llvm/ADT/StringMap.h"
25#include "llvm/Target/TargetData.h"
26#include "llvm/Support/ErrorHandling.h"
27#include "llvm/Support/GetElementPtrTypeIterator.h"
28#include "llvm/Support/MathExtras.h"
29#include <cerrno>
30#include <cmath>
31using namespace llvm;
32
33//===----------------------------------------------------------------------===//
34// Constant Folding internal helper functions
35//===----------------------------------------------------------------------===//
36
37/// IsConstantOffsetFromGlobal - If this constant is actually a constant offset
38/// from a global, return the global and the constant.  Because of
39/// constantexprs, this function is recursive.
40static bool IsConstantOffsetFromGlobal(Constant *C, GlobalValue *&GV,
41                                       int64_t &Offset, const TargetData &TD) {
42  // Trivial case, constant is the global.
43  if ((GV = dyn_cast<GlobalValue>(C))) {
44    Offset = 0;
45    return true;
46  }
47
48  // Otherwise, if this isn't a constant expr, bail out.
49  ConstantExpr *CE = dyn_cast<ConstantExpr>(C);
50  if (!CE) return false;
51
52  // Look through ptr->int and ptr->ptr casts.
53  if (CE->getOpcode() == Instruction::PtrToInt ||
54      CE->getOpcode() == Instruction::BitCast)
55    return IsConstantOffsetFromGlobal(CE->getOperand(0), GV, Offset, TD);
56
57  // i32* getelementptr ([5 x i32]* @a, i32 0, i32 5)
58  if (CE->getOpcode() == Instruction::GetElementPtr) {
59    // Cannot compute this if the element type of the pointer is missing size
60    // info.
61    if (!cast<PointerType>(CE->getOperand(0)->getType())
62                 ->getElementType()->isSized())
63      return false;
64
65    // If the base isn't a global+constant, we aren't either.
66    if (!IsConstantOffsetFromGlobal(CE->getOperand(0), GV, Offset, TD))
67      return false;
68
69    // Otherwise, add any offset that our operands provide.
70    gep_type_iterator GTI = gep_type_begin(CE);
71    for (User::const_op_iterator i = CE->op_begin() + 1, e = CE->op_end();
72         i != e; ++i, ++GTI) {
73      ConstantInt *CI = dyn_cast<ConstantInt>(*i);
74      if (!CI) return false;  // Index isn't a simple constant?
75      if (CI->getZExtValue() == 0) continue;  // Not adding anything.
76
77      if (const StructType *ST = dyn_cast<StructType>(*GTI)) {
78        // N = N + Offset
79        Offset += TD.getStructLayout(ST)->getElementOffset(CI->getZExtValue());
80      } else {
81        const SequentialType *SQT = cast<SequentialType>(*GTI);
82        Offset += TD.getTypeAllocSize(SQT->getElementType())*CI->getSExtValue();
83      }
84    }
85    return true;
86  }
87
88  return false;
89}
90
91
92/// SymbolicallyEvaluateBinop - One of Op0/Op1 is a constant expression.
93/// Attempt to symbolically evaluate the result of a binary operator merging
94/// these together.  If target data info is available, it is provided as TD,
95/// otherwise TD is null.
96static Constant *SymbolicallyEvaluateBinop(unsigned Opc, Constant *Op0,
97                                           Constant *Op1, const TargetData *TD,
98                                           LLVMContext &Context){
99  // SROA
100
101  // Fold (and 0xffffffff00000000, (shl x, 32)) -> shl.
102  // Fold (lshr (or X, Y), 32) -> (lshr [X/Y], 32) if one doesn't contribute
103  // bits.
104
105
106  // If the constant expr is something like &A[123] - &A[4].f, fold this into a
107  // constant.  This happens frequently when iterating over a global array.
108  if (Opc == Instruction::Sub && TD) {
109    GlobalValue *GV1, *GV2;
110    int64_t Offs1, Offs2;
111
112    if (IsConstantOffsetFromGlobal(Op0, GV1, Offs1, *TD))
113      if (IsConstantOffsetFromGlobal(Op1, GV2, Offs2, *TD) &&
114          GV1 == GV2) {
115        // (&GV+C1) - (&GV+C2) -> C1-C2, pointer arithmetic cannot overflow.
116        return ConstantInt::get(Op0->getType(), Offs1-Offs2);
117      }
118  }
119
120  return 0;
121}
122
123/// SymbolicallyEvaluateGEP - If we can symbolically evaluate the specified GEP
124/// constant expression, do so.
125static Constant *SymbolicallyEvaluateGEP(Constant* const* Ops, unsigned NumOps,
126                                         const Type *ResultTy,
127                                         LLVMContext &Context,
128                                         const TargetData *TD) {
129  Constant *Ptr = Ops[0];
130  if (!TD || !cast<PointerType>(Ptr->getType())->getElementType()->isSized())
131    return 0;
132
133  uint64_t BasePtr = 0;
134  if (!Ptr->isNullValue()) {
135    // If this is a inttoptr from a constant int, we can fold this as the base,
136    // otherwise we can't.
137    if (ConstantExpr *CE = dyn_cast<ConstantExpr>(Ptr))
138      if (CE->getOpcode() == Instruction::IntToPtr)
139        if (ConstantInt *Base = dyn_cast<ConstantInt>(CE->getOperand(0)))
140          BasePtr = Base->getZExtValue();
141
142    if (BasePtr == 0)
143      return 0;
144  }
145
146  // If this is a constant expr gep that is effectively computing an
147  // "offsetof", fold it into 'cast int Size to T*' instead of 'gep 0, 0, 12'
148  for (unsigned i = 1; i != NumOps; ++i)
149    if (!isa<ConstantInt>(Ops[i]))
150      return false;
151
152  uint64_t Offset = TD->getIndexedOffset(Ptr->getType(),
153                                         (Value**)Ops+1, NumOps-1);
154  Constant *C = ConstantInt::get(TD->getIntPtrType(), Offset+BasePtr);
155  return Context.getConstantExprIntToPtr(C, ResultTy);
156}
157
158/// FoldBitCast - Constant fold bitcast, symbolically evaluating it with
159/// targetdata.  Return 0 if unfoldable.
160static Constant *FoldBitCast(Constant *C, const Type *DestTy,
161                             const TargetData &TD, LLVMContext &Context) {
162  // If this is a bitcast from constant vector -> vector, fold it.
163  if (ConstantVector *CV = dyn_cast<ConstantVector>(C)) {
164    if (const VectorType *DestVTy = dyn_cast<VectorType>(DestTy)) {
165      // If the element types match, VMCore can fold it.
166      unsigned NumDstElt = DestVTy->getNumElements();
167      unsigned NumSrcElt = CV->getNumOperands();
168      if (NumDstElt == NumSrcElt)
169        return 0;
170
171      const Type *SrcEltTy = CV->getType()->getElementType();
172      const Type *DstEltTy = DestVTy->getElementType();
173
174      // Otherwise, we're changing the number of elements in a vector, which
175      // requires endianness information to do the right thing.  For example,
176      //    bitcast (<2 x i64> <i64 0, i64 1> to <4 x i32>)
177      // folds to (little endian):
178      //    <4 x i32> <i32 0, i32 0, i32 1, i32 0>
179      // and to (big endian):
180      //    <4 x i32> <i32 0, i32 0, i32 0, i32 1>
181
182      // First thing is first.  We only want to think about integer here, so if
183      // we have something in FP form, recast it as integer.
184      if (DstEltTy->isFloatingPoint()) {
185        // Fold to an vector of integers with same size as our FP type.
186        unsigned FPWidth = DstEltTy->getPrimitiveSizeInBits();
187        const Type *DestIVTy = Context.getVectorType(
188                                   Context.getIntegerType(FPWidth), NumDstElt);
189        // Recursively handle this integer conversion, if possible.
190        C = FoldBitCast(C, DestIVTy, TD, Context);
191        if (!C) return 0;
192
193        // Finally, VMCore can handle this now that #elts line up.
194        return Context.getConstantExprBitCast(C, DestTy);
195      }
196
197      // Okay, we know the destination is integer, if the input is FP, convert
198      // it to integer first.
199      if (SrcEltTy->isFloatingPoint()) {
200        unsigned FPWidth = SrcEltTy->getPrimitiveSizeInBits();
201        const Type *SrcIVTy = Context.getVectorType(
202                                   Context.getIntegerType(FPWidth), NumSrcElt);
203        // Ask VMCore to do the conversion now that #elts line up.
204        C = Context.getConstantExprBitCast(C, SrcIVTy);
205        CV = dyn_cast<ConstantVector>(C);
206        if (!CV) return 0;  // If VMCore wasn't able to fold it, bail out.
207      }
208
209      // Now we know that the input and output vectors are both integer vectors
210      // of the same size, and that their #elements is not the same.  Do the
211      // conversion here, which depends on whether the input or output has
212      // more elements.
213      bool isLittleEndian = TD.isLittleEndian();
214
215      SmallVector<Constant*, 32> Result;
216      if (NumDstElt < NumSrcElt) {
217        // Handle: bitcast (<4 x i32> <i32 0, i32 1, i32 2, i32 3> to <2 x i64>)
218        Constant *Zero = Context.getNullValue(DstEltTy);
219        unsigned Ratio = NumSrcElt/NumDstElt;
220        unsigned SrcBitSize = SrcEltTy->getPrimitiveSizeInBits();
221        unsigned SrcElt = 0;
222        for (unsigned i = 0; i != NumDstElt; ++i) {
223          // Build each element of the result.
224          Constant *Elt = Zero;
225          unsigned ShiftAmt = isLittleEndian ? 0 : SrcBitSize*(Ratio-1);
226          for (unsigned j = 0; j != Ratio; ++j) {
227            Constant *Src = dyn_cast<ConstantInt>(CV->getOperand(SrcElt++));
228            if (!Src) return 0;  // Reject constantexpr elements.
229
230            // Zero extend the element to the right size.
231            Src = Context.getConstantExprZExt(Src, Elt->getType());
232
233            // Shift it to the right place, depending on endianness.
234            Src = Context.getConstantExprShl(Src,
235                             ConstantInt::get(Src->getType(), ShiftAmt));
236            ShiftAmt += isLittleEndian ? SrcBitSize : -SrcBitSize;
237
238            // Mix it in.
239            Elt = Context.getConstantExprOr(Elt, Src);
240          }
241          Result.push_back(Elt);
242        }
243      } else {
244        // Handle: bitcast (<2 x i64> <i64 0, i64 1> to <4 x i32>)
245        unsigned Ratio = NumDstElt/NumSrcElt;
246        unsigned DstBitSize = DstEltTy->getPrimitiveSizeInBits();
247
248        // Loop over each source value, expanding into multiple results.
249        for (unsigned i = 0; i != NumSrcElt; ++i) {
250          Constant *Src = dyn_cast<ConstantInt>(CV->getOperand(i));
251          if (!Src) return 0;  // Reject constantexpr elements.
252
253          unsigned ShiftAmt = isLittleEndian ? 0 : DstBitSize*(Ratio-1);
254          for (unsigned j = 0; j != Ratio; ++j) {
255            // Shift the piece of the value into the right place, depending on
256            // endianness.
257            Constant *Elt = Context.getConstantExprLShr(Src,
258                            ConstantInt::get(Src->getType(), ShiftAmt));
259            ShiftAmt += isLittleEndian ? DstBitSize : -DstBitSize;
260
261            // Truncate and remember this piece.
262            Result.push_back(Context.getConstantExprTrunc(Elt, DstEltTy));
263          }
264        }
265      }
266
267      return Context.getConstantVector(Result.data(), Result.size());
268    }
269  }
270
271  return 0;
272}
273
274
275//===----------------------------------------------------------------------===//
276// Constant Folding public APIs
277//===----------------------------------------------------------------------===//
278
279
280/// ConstantFoldInstruction - Attempt to constant fold the specified
281/// instruction.  If successful, the constant result is returned, if not, null
282/// is returned.  Note that this function can only fail when attempting to fold
283/// instructions like loads and stores, which have no constant expression form.
284///
285Constant *llvm::ConstantFoldInstruction(Instruction *I, LLVMContext &Context,
286                                        const TargetData *TD) {
287  if (PHINode *PN = dyn_cast<PHINode>(I)) {
288    if (PN->getNumIncomingValues() == 0)
289      return Context.getUndef(PN->getType());
290
291    Constant *Result = dyn_cast<Constant>(PN->getIncomingValue(0));
292    if (Result == 0) return 0;
293
294    // Handle PHI nodes specially here...
295    for (unsigned i = 1, e = PN->getNumIncomingValues(); i != e; ++i)
296      if (PN->getIncomingValue(i) != Result && PN->getIncomingValue(i) != PN)
297        return 0;   // Not all the same incoming constants...
298
299    // If we reach here, all incoming values are the same constant.
300    return Result;
301  }
302
303  // Scan the operand list, checking to see if they are all constants, if so,
304  // hand off to ConstantFoldInstOperands.
305  SmallVector<Constant*, 8> Ops;
306  for (User::op_iterator i = I->op_begin(), e = I->op_end(); i != e; ++i)
307    if (Constant *Op = dyn_cast<Constant>(*i))
308      Ops.push_back(Op);
309    else
310      return 0;  // All operands not constant!
311
312  if (const CmpInst *CI = dyn_cast<CmpInst>(I))
313    return ConstantFoldCompareInstOperands(CI->getPredicate(),
314                                           Ops.data(), Ops.size(),
315                                           Context, TD);
316  else
317    return ConstantFoldInstOperands(I->getOpcode(), I->getType(),
318                                    Ops.data(), Ops.size(), Context, TD);
319}
320
321/// ConstantFoldConstantExpression - Attempt to fold the constant expression
322/// using the specified TargetData.  If successful, the constant result is
323/// result is returned, if not, null is returned.
324Constant *llvm::ConstantFoldConstantExpression(ConstantExpr *CE,
325                                               LLVMContext &Context,
326                                               const TargetData *TD) {
327  SmallVector<Constant*, 8> Ops;
328  for (User::op_iterator i = CE->op_begin(), e = CE->op_end(); i != e; ++i)
329    Ops.push_back(cast<Constant>(*i));
330
331  if (CE->isCompare())
332    return ConstantFoldCompareInstOperands(CE->getPredicate(),
333                                           Ops.data(), Ops.size(),
334                                           Context, TD);
335  else
336    return ConstantFoldInstOperands(CE->getOpcode(), CE->getType(),
337                                    Ops.data(), Ops.size(), Context, TD);
338}
339
340/// ConstantFoldInstOperands - Attempt to constant fold an instruction with the
341/// specified opcode and operands.  If successful, the constant result is
342/// returned, if not, null is returned.  Note that this function can fail when
343/// attempting to fold instructions like loads and stores, which have no
344/// constant expression form.
345///
346Constant *llvm::ConstantFoldInstOperands(unsigned Opcode, const Type *DestTy,
347                                         Constant* const* Ops, unsigned NumOps,
348                                         LLVMContext &Context,
349                                         const TargetData *TD) {
350  // Handle easy binops first.
351  if (Instruction::isBinaryOp(Opcode)) {
352    if (isa<ConstantExpr>(Ops[0]) || isa<ConstantExpr>(Ops[1]))
353      if (Constant *C = SymbolicallyEvaluateBinop(Opcode, Ops[0], Ops[1], TD,
354                                                  Context))
355        return C;
356
357    return Context.getConstantExpr(Opcode, Ops[0], Ops[1]);
358  }
359
360  switch (Opcode) {
361  default: return 0;
362  case Instruction::Call:
363    if (Function *F = dyn_cast<Function>(Ops[0]))
364      if (canConstantFoldCallTo(F))
365        return ConstantFoldCall(F, Ops+1, NumOps-1);
366    return 0;
367  case Instruction::ICmp:
368  case Instruction::FCmp:
369    llvm_unreachable("This function is invalid for compares: no predicate specified");
370  case Instruction::PtrToInt:
371    // If the input is a inttoptr, eliminate the pair.  This requires knowing
372    // the width of a pointer, so it can't be done in ConstantExpr::getCast.
373    if (ConstantExpr *CE = dyn_cast<ConstantExpr>(Ops[0])) {
374      if (TD && CE->getOpcode() == Instruction::IntToPtr) {
375        Constant *Input = CE->getOperand(0);
376        unsigned InWidth = Input->getType()->getScalarSizeInBits();
377        if (TD->getPointerSizeInBits() < InWidth) {
378          Constant *Mask =
379            ConstantInt::get(Context, APInt::getLowBitsSet(InWidth,
380                                                  TD->getPointerSizeInBits()));
381          Input = Context.getConstantExprAnd(Input, Mask);
382        }
383        // Do a zext or trunc to get to the dest size.
384        return Context.getConstantExprIntegerCast(Input, DestTy, false);
385      }
386    }
387    return Context.getConstantExprCast(Opcode, Ops[0], DestTy);
388  case Instruction::IntToPtr:
389    // If the input is a ptrtoint, turn the pair into a ptr to ptr bitcast if
390    // the int size is >= the ptr size.  This requires knowing the width of a
391    // pointer, so it can't be done in ConstantExpr::getCast.
392    if (ConstantExpr *CE = dyn_cast<ConstantExpr>(Ops[0])) {
393      if (TD &&
394          TD->getPointerSizeInBits() <=
395          CE->getType()->getScalarSizeInBits()) {
396        if (CE->getOpcode() == Instruction::PtrToInt) {
397          Constant *Input = CE->getOperand(0);
398          Constant *C = FoldBitCast(Input, DestTy, *TD, Context);
399          return C ? C : Context.getConstantExprBitCast(Input, DestTy);
400        }
401        // If there's a constant offset added to the integer value before
402        // it is casted back to a pointer, see if the expression can be
403        // converted into a GEP.
404        if (CE->getOpcode() == Instruction::Add)
405          if (ConstantInt *L = dyn_cast<ConstantInt>(CE->getOperand(0)))
406            if (ConstantExpr *R = dyn_cast<ConstantExpr>(CE->getOperand(1)))
407              if (R->getOpcode() == Instruction::PtrToInt)
408                if (GlobalVariable *GV =
409                      dyn_cast<GlobalVariable>(R->getOperand(0))) {
410                  const PointerType *GVTy = cast<PointerType>(GV->getType());
411                  if (const ArrayType *AT =
412                        dyn_cast<ArrayType>(GVTy->getElementType())) {
413                    const Type *ElTy = AT->getElementType();
414                    uint64_t AllocSize = TD->getTypeAllocSize(ElTy);
415                    APInt PSA(L->getValue().getBitWidth(), AllocSize);
416                    if (ElTy == cast<PointerType>(DestTy)->getElementType() &&
417                        L->getValue().urem(PSA) == 0) {
418                      APInt ElemIdx = L->getValue().udiv(PSA);
419                      if (ElemIdx.ult(APInt(ElemIdx.getBitWidth(),
420                                            AT->getNumElements()))) {
421                        Constant *Index[] = {
422                          Context.getNullValue(CE->getType()),
423                          ConstantInt::get(Context, ElemIdx)
424                        };
425                        return
426                        Context.getConstantExprGetElementPtr(GV, &Index[0], 2);
427                      }
428                    }
429                  }
430                }
431      }
432    }
433    return Context.getConstantExprCast(Opcode, Ops[0], DestTy);
434  case Instruction::Trunc:
435  case Instruction::ZExt:
436  case Instruction::SExt:
437  case Instruction::FPTrunc:
438  case Instruction::FPExt:
439  case Instruction::UIToFP:
440  case Instruction::SIToFP:
441  case Instruction::FPToUI:
442  case Instruction::FPToSI:
443      return Context.getConstantExprCast(Opcode, Ops[0], DestTy);
444  case Instruction::BitCast:
445    if (TD)
446      if (Constant *C = FoldBitCast(Ops[0], DestTy, *TD, Context))
447        return C;
448    return Context.getConstantExprBitCast(Ops[0], DestTy);
449  case Instruction::Select:
450    return Context.getConstantExprSelect(Ops[0], Ops[1], Ops[2]);
451  case Instruction::ExtractElement:
452    return Context.getConstantExprExtractElement(Ops[0], Ops[1]);
453  case Instruction::InsertElement:
454    return Context.getConstantExprInsertElement(Ops[0], Ops[1], Ops[2]);
455  case Instruction::ShuffleVector:
456    return Context.getConstantExprShuffleVector(Ops[0], Ops[1], Ops[2]);
457  case Instruction::GetElementPtr:
458    if (Constant *C = SymbolicallyEvaluateGEP(Ops, NumOps, DestTy, Context, TD))
459      return C;
460
461    return Context.getConstantExprGetElementPtr(Ops[0], Ops+1, NumOps-1);
462  }
463}
464
465/// ConstantFoldCompareInstOperands - Attempt to constant fold a compare
466/// instruction (icmp/fcmp) with the specified operands.  If it fails, it
467/// returns a constant expression of the specified operands.
468///
469Constant *llvm::ConstantFoldCompareInstOperands(unsigned Predicate,
470                                                Constant*const * Ops,
471                                                unsigned NumOps,
472                                                LLVMContext &Context,
473                                                const TargetData *TD) {
474  // fold: icmp (inttoptr x), null         -> icmp x, 0
475  // fold: icmp (ptrtoint x), 0            -> icmp x, null
476  // fold: icmp (inttoptr x), (inttoptr y) -> icmp trunc/zext x, trunc/zext y
477  // fold: icmp (ptrtoint x), (ptrtoint y) -> icmp x, y
478  //
479  // ConstantExpr::getCompare cannot do this, because it doesn't have TD
480  // around to know if bit truncation is happening.
481  if (ConstantExpr *CE0 = dyn_cast<ConstantExpr>(Ops[0])) {
482    if (TD && Ops[1]->isNullValue()) {
483      const Type *IntPtrTy = TD->getIntPtrType();
484      if (CE0->getOpcode() == Instruction::IntToPtr) {
485        // Convert the integer value to the right size to ensure we get the
486        // proper extension or truncation.
487        Constant *C = Context.getConstantExprIntegerCast(CE0->getOperand(0),
488                                                   IntPtrTy, false);
489        Constant *NewOps[] = { C, Context.getNullValue(C->getType()) };
490        return ConstantFoldCompareInstOperands(Predicate, NewOps, 2,
491                                               Context, TD);
492      }
493
494      // Only do this transformation if the int is intptrty in size, otherwise
495      // there is a truncation or extension that we aren't modeling.
496      if (CE0->getOpcode() == Instruction::PtrToInt &&
497          CE0->getType() == IntPtrTy) {
498        Constant *C = CE0->getOperand(0);
499        Constant *NewOps[] = { C, Context.getNullValue(C->getType()) };
500        // FIXME!
501        return ConstantFoldCompareInstOperands(Predicate, NewOps, 2,
502                                               Context, TD);
503      }
504    }
505
506    if (ConstantExpr *CE1 = dyn_cast<ConstantExpr>(Ops[1])) {
507      if (TD && CE0->getOpcode() == CE1->getOpcode()) {
508        const Type *IntPtrTy = TD->getIntPtrType();
509
510        if (CE0->getOpcode() == Instruction::IntToPtr) {
511          // Convert the integer value to the right size to ensure we get the
512          // proper extension or truncation.
513          Constant *C0 = Context.getConstantExprIntegerCast(CE0->getOperand(0),
514                                                      IntPtrTy, false);
515          Constant *C1 = Context.getConstantExprIntegerCast(CE1->getOperand(0),
516                                                      IntPtrTy, false);
517          Constant *NewOps[] = { C0, C1 };
518          return ConstantFoldCompareInstOperands(Predicate, NewOps, 2,
519                                                 Context, TD);
520        }
521
522        // Only do this transformation if the int is intptrty in size, otherwise
523        // there is a truncation or extension that we aren't modeling.
524        if ((CE0->getOpcode() == Instruction::PtrToInt &&
525             CE0->getType() == IntPtrTy &&
526             CE0->getOperand(0)->getType() == CE1->getOperand(0)->getType())) {
527          Constant *NewOps[] = {
528            CE0->getOperand(0), CE1->getOperand(0)
529          };
530          return ConstantFoldCompareInstOperands(Predicate, NewOps, 2,
531                                                 Context, TD);
532        }
533      }
534    }
535  }
536  return Context.getConstantExprCompare(Predicate, Ops[0], Ops[1]);
537}
538
539
540/// ConstantFoldLoadThroughGEPConstantExpr - Given a constant and a
541/// getelementptr constantexpr, return the constant value being addressed by the
542/// constant expression, or null if something is funny and we can't decide.
543Constant *llvm::ConstantFoldLoadThroughGEPConstantExpr(Constant *C,
544                                                       ConstantExpr *CE,
545                                                       LLVMContext &Context) {
546  if (CE->getOperand(1) != Context.getNullValue(CE->getOperand(1)->getType()))
547    return 0;  // Do not allow stepping over the value!
548
549  // Loop over all of the operands, tracking down which value we are
550  // addressing...
551  gep_type_iterator I = gep_type_begin(CE), E = gep_type_end(CE);
552  for (++I; I != E; ++I)
553    if (const StructType *STy = dyn_cast<StructType>(*I)) {
554      ConstantInt *CU = cast<ConstantInt>(I.getOperand());
555      assert(CU->getZExtValue() < STy->getNumElements() &&
556             "Struct index out of range!");
557      unsigned El = (unsigned)CU->getZExtValue();
558      if (ConstantStruct *CS = dyn_cast<ConstantStruct>(C)) {
559        C = CS->getOperand(El);
560      } else if (isa<ConstantAggregateZero>(C)) {
561        C = Context.getNullValue(STy->getElementType(El));
562      } else if (isa<UndefValue>(C)) {
563        C = Context.getUndef(STy->getElementType(El));
564      } else {
565        return 0;
566      }
567    } else if (ConstantInt *CI = dyn_cast<ConstantInt>(I.getOperand())) {
568      if (const ArrayType *ATy = dyn_cast<ArrayType>(*I)) {
569        if (CI->getZExtValue() >= ATy->getNumElements())
570         return 0;
571        if (ConstantArray *CA = dyn_cast<ConstantArray>(C))
572          C = CA->getOperand(CI->getZExtValue());
573        else if (isa<ConstantAggregateZero>(C))
574          C = Context.getNullValue(ATy->getElementType());
575        else if (isa<UndefValue>(C))
576          C = Context.getUndef(ATy->getElementType());
577        else
578          return 0;
579      } else if (const VectorType *PTy = dyn_cast<VectorType>(*I)) {
580        if (CI->getZExtValue() >= PTy->getNumElements())
581          return 0;
582        if (ConstantVector *CP = dyn_cast<ConstantVector>(C))
583          C = CP->getOperand(CI->getZExtValue());
584        else if (isa<ConstantAggregateZero>(C))
585          C = Context.getNullValue(PTy->getElementType());
586        else if (isa<UndefValue>(C))
587          C = Context.getUndef(PTy->getElementType());
588        else
589          return 0;
590      } else {
591        return 0;
592      }
593    } else {
594      return 0;
595    }
596  return C;
597}
598
599
600//===----------------------------------------------------------------------===//
601//  Constant Folding for Calls
602//
603
604/// canConstantFoldCallTo - Return true if its even possible to fold a call to
605/// the specified function.
606bool
607llvm::canConstantFoldCallTo(const Function *F) {
608  switch (F->getIntrinsicID()) {
609  case Intrinsic::sqrt:
610  case Intrinsic::powi:
611  case Intrinsic::bswap:
612  case Intrinsic::ctpop:
613  case Intrinsic::ctlz:
614  case Intrinsic::cttz:
615    return true;
616  default: break;
617  }
618
619  if (!F->hasName()) return false;
620  StringRef Name = F->getName();
621
622  // In these cases, the check of the length is required.  We don't want to
623  // return true for a name like "cos\0blah" which strcmp would return equal to
624  // "cos", but has length 8.
625  switch (Name[0]) {
626  default: return false;
627  case 'a':
628    return Name == "acos" || Name == "asin" ||
629      Name == "atan" || Name == "atan2";
630  case 'c':
631    return Name == "cos" || Name == "ceil" || Name == "cosf" || Name == "cosh";
632  case 'e':
633    return Name == "exp";
634  case 'f':
635    return Name == "fabs" || Name == "fmod" || Name == "floor";
636  case 'l':
637    return Name == "log" || Name == "log10";
638  case 'p':
639    return Name == "pow";
640  case 's':
641    return Name == "sin" || Name == "sinh" || Name == "sqrt" ||
642      Name == "sinf" || Name == "sqrtf";
643  case 't':
644    return Name == "tan" || Name == "tanh";
645  }
646}
647
648static Constant *ConstantFoldFP(double (*NativeFP)(double), double V,
649                                const Type *Ty, LLVMContext &Context) {
650  errno = 0;
651  V = NativeFP(V);
652  if (errno != 0) {
653    errno = 0;
654    return 0;
655  }
656
657  if (Ty == Type::FloatTy)
658    return Context.getConstantFP(APFloat((float)V));
659  if (Ty == Type::DoubleTy)
660    return Context.getConstantFP(APFloat(V));
661  llvm_unreachable("Can only constant fold float/double");
662  return 0; // dummy return to suppress warning
663}
664
665static Constant *ConstantFoldBinaryFP(double (*NativeFP)(double, double),
666                                      double V, double W,
667                                      const Type *Ty,
668                                      LLVMContext &Context) {
669  errno = 0;
670  V = NativeFP(V, W);
671  if (errno != 0) {
672    errno = 0;
673    return 0;
674  }
675
676  if (Ty == Type::FloatTy)
677    return Context.getConstantFP(APFloat((float)V));
678  if (Ty == Type::DoubleTy)
679    return Context.getConstantFP(APFloat(V));
680  llvm_unreachable("Can only constant fold float/double");
681  return 0; // dummy return to suppress warning
682}
683
684/// ConstantFoldCall - Attempt to constant fold a call to the specified function
685/// with the specified arguments, returning null if unsuccessful.
686
687Constant *
688llvm::ConstantFoldCall(Function *F,
689                       Constant* const* Operands, unsigned NumOperands) {
690  if (!F->hasName()) return 0;
691  LLVMContext &Context = F->getContext();
692  StringRef Name = F->getName();
693
694  const Type *Ty = F->getReturnType();
695  if (NumOperands == 1) {
696    if (ConstantFP *Op = dyn_cast<ConstantFP>(Operands[0])) {
697      if (Ty!=Type::FloatTy && Ty!=Type::DoubleTy)
698        return 0;
699      /// Currently APFloat versions of these functions do not exist, so we use
700      /// the host native double versions.  Float versions are not called
701      /// directly but for all these it is true (float)(f((double)arg)) ==
702      /// f(arg).  Long double not supported yet.
703      double V = Ty==Type::FloatTy ? (double)Op->getValueAPF().convertToFloat():
704                                     Op->getValueAPF().convertToDouble();
705      switch (Name[0]) {
706      case 'a':
707        if (Name == "acos")
708          return ConstantFoldFP(acos, V, Ty, Context);
709        else if (Name == "asin")
710          return ConstantFoldFP(asin, V, Ty, Context);
711        else if (Name == "atan")
712          return ConstantFoldFP(atan, V, Ty, Context);
713        break;
714      case 'c':
715        if (Name == "ceil")
716          return ConstantFoldFP(ceil, V, Ty, Context);
717        else if (Name == "cos")
718          return ConstantFoldFP(cos, V, Ty, Context);
719        else if (Name == "cosh")
720          return ConstantFoldFP(cosh, V, Ty, Context);
721        else if (Name == "cosf")
722          return ConstantFoldFP(cos, V, Ty, Context);
723        break;
724      case 'e':
725        if (Name == "exp")
726          return ConstantFoldFP(exp, V, Ty, Context);
727        break;
728      case 'f':
729        if (Name == "fabs")
730          return ConstantFoldFP(fabs, V, Ty, Context);
731        else if (Name == "floor")
732          return ConstantFoldFP(floor, V, Ty, Context);
733        break;
734      case 'l':
735        if (Name == "log" && V > 0)
736          return ConstantFoldFP(log, V, Ty, Context);
737        else if (Name == "log10" && V > 0)
738          return ConstantFoldFP(log10, V, Ty, Context);
739        else if (Name == "llvm.sqrt.f32" ||
740                 Name == "llvm.sqrt.f64") {
741          if (V >= -0.0)
742            return ConstantFoldFP(sqrt, V, Ty, Context);
743          else // Undefined
744            return Context.getNullValue(Ty);
745        }
746        break;
747      case 's':
748        if (Name == "sin")
749          return ConstantFoldFP(sin, V, Ty, Context);
750        else if (Name == "sinh")
751          return ConstantFoldFP(sinh, V, Ty, Context);
752        else if (Name == "sqrt" && V >= 0)
753          return ConstantFoldFP(sqrt, V, Ty, Context);
754        else if (Name == "sqrtf" && V >= 0)
755          return ConstantFoldFP(sqrt, V, Ty, Context);
756        else if (Name == "sinf")
757          return ConstantFoldFP(sin, V, Ty, Context);
758        break;
759      case 't':
760        if (Name == "tan")
761          return ConstantFoldFP(tan, V, Ty, Context);
762        else if (Name == "tanh")
763          return ConstantFoldFP(tanh, V, Ty, Context);
764        break;
765      default:
766        break;
767      }
768    } else if (ConstantInt *Op = dyn_cast<ConstantInt>(Operands[0])) {
769      if (Name.startswith("llvm.bswap"))
770        return ConstantInt::get(Context, Op->getValue().byteSwap());
771      else if (Name.startswith("llvm.ctpop"))
772        return ConstantInt::get(Ty, Op->getValue().countPopulation());
773      else if (Name.startswith("llvm.cttz"))
774        return ConstantInt::get(Ty, Op->getValue().countTrailingZeros());
775      else if (Name.startswith("llvm.ctlz"))
776        return ConstantInt::get(Ty, Op->getValue().countLeadingZeros());
777    }
778  } else if (NumOperands == 2) {
779    if (ConstantFP *Op1 = dyn_cast<ConstantFP>(Operands[0])) {
780      if (Ty!=Type::FloatTy && Ty!=Type::DoubleTy)
781        return 0;
782      double Op1V = Ty==Type::FloatTy ?
783                      (double)Op1->getValueAPF().convertToFloat():
784                      Op1->getValueAPF().convertToDouble();
785      if (ConstantFP *Op2 = dyn_cast<ConstantFP>(Operands[1])) {
786        double Op2V = Ty==Type::FloatTy ?
787                      (double)Op2->getValueAPF().convertToFloat():
788                      Op2->getValueAPF().convertToDouble();
789
790        if (Name == "pow") {
791          return ConstantFoldBinaryFP(pow, Op1V, Op2V, Ty, Context);
792        } else if (Name == "fmod") {
793          return ConstantFoldBinaryFP(fmod, Op1V, Op2V, Ty, Context);
794        } else if (Name == "atan2") {
795          return ConstantFoldBinaryFP(atan2, Op1V, Op2V, Ty, Context);
796        }
797      } else if (ConstantInt *Op2C = dyn_cast<ConstantInt>(Operands[1])) {
798        if (Name == "llvm.powi.f32") {
799          return Context.getConstantFP(APFloat((float)std::pow((float)Op1V,
800                                                 (int)Op2C->getZExtValue())));
801        } else if (Name == "llvm.powi.f64") {
802          return Context.getConstantFP(APFloat((double)std::pow((double)Op1V,
803                                                 (int)Op2C->getZExtValue())));
804        }
805      }
806    }
807  }
808  return 0;
809}
810
811