Analysis.cpp revision 7f9bb1d223d6ab9f65f6b620e0e5329eef75ba6c
1//===-- Analysis.cpp - CodeGen LLVM IR Analysis Utilities --*- C++ ------*-===//
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 defines several CodeGen-specific LLVM IR analysis utilties.
11//
12//===----------------------------------------------------------------------===//
13
14#include "llvm/CodeGen/Analysis.h"
15#include "llvm/DerivedTypes.h"
16#include "llvm/Function.h"
17#include "llvm/Instructions.h"
18#include "llvm/IntrinsicInst.h"
19#include "llvm/LLVMContext.h"
20#include "llvm/Module.h"
21#include "llvm/CodeGen/MachineFunction.h"
22#include "llvm/Target/TargetData.h"
23#include "llvm/Target/TargetLowering.h"
24#include "llvm/Target/TargetOptions.h"
25#include "llvm/Support/ErrorHandling.h"
26#include "llvm/Support/MathExtras.h"
27using namespace llvm;
28
29/// ComputeLinearIndex - Given an LLVM IR aggregate type and a sequence
30/// of insertvalue or extractvalue indices that identify a member, return
31/// the linearized index of the start of the member.
32///
33unsigned llvm::ComputeLinearIndex(const TargetLowering &TLI, const Type *Ty,
34                                  const unsigned *Indices,
35                                  const unsigned *IndicesEnd,
36                                  unsigned CurIndex) {
37  // Base case: We're done.
38  if (Indices && Indices == IndicesEnd)
39    return CurIndex;
40
41  // Given a struct type, recursively traverse the elements.
42  if (const StructType *STy = dyn_cast<StructType>(Ty)) {
43    for (StructType::element_iterator EB = STy->element_begin(),
44                                      EI = EB,
45                                      EE = STy->element_end();
46        EI != EE; ++EI) {
47      if (Indices && *Indices == unsigned(EI - EB))
48        return ComputeLinearIndex(TLI, *EI, Indices+1, IndicesEnd, CurIndex);
49      CurIndex = ComputeLinearIndex(TLI, *EI, 0, 0, CurIndex);
50    }
51    return CurIndex;
52  }
53  // Given an array type, recursively traverse the elements.
54  else if (const ArrayType *ATy = dyn_cast<ArrayType>(Ty)) {
55    const Type *EltTy = ATy->getElementType();
56    for (unsigned i = 0, e = ATy->getNumElements(); i != e; ++i) {
57      if (Indices && *Indices == i)
58        return ComputeLinearIndex(TLI, EltTy, Indices+1, IndicesEnd, CurIndex);
59      CurIndex = ComputeLinearIndex(TLI, EltTy, 0, 0, CurIndex);
60    }
61    return CurIndex;
62  }
63  // We haven't found the type we're looking for, so keep searching.
64  return CurIndex + 1;
65}
66
67/// ComputeValueVTs - Given an LLVM IR type, compute a sequence of
68/// EVTs that represent all the individual underlying
69/// non-aggregate types that comprise it.
70///
71/// If Offsets is non-null, it points to a vector to be filled in
72/// with the in-memory offsets of each of the individual values.
73///
74void llvm::ComputeValueVTs(const TargetLowering &TLI, const Type *Ty,
75                           SmallVectorImpl<EVT> &ValueVTs,
76                           SmallVectorImpl<uint64_t> *Offsets,
77                           uint64_t StartingOffset) {
78  // Given a struct type, recursively traverse the elements.
79  if (const StructType *STy = dyn_cast<StructType>(Ty)) {
80    const StructLayout *SL = TLI.getTargetData()->getStructLayout(STy);
81    for (StructType::element_iterator EB = STy->element_begin(),
82                                      EI = EB,
83                                      EE = STy->element_end();
84         EI != EE; ++EI)
85      ComputeValueVTs(TLI, *EI, ValueVTs, Offsets,
86                      StartingOffset + SL->getElementOffset(EI - EB));
87    return;
88  }
89  // Given an array type, recursively traverse the elements.
90  if (const ArrayType *ATy = dyn_cast<ArrayType>(Ty)) {
91    const Type *EltTy = ATy->getElementType();
92    uint64_t EltSize = TLI.getTargetData()->getTypeAllocSize(EltTy);
93    for (unsigned i = 0, e = ATy->getNumElements(); i != e; ++i)
94      ComputeValueVTs(TLI, EltTy, ValueVTs, Offsets,
95                      StartingOffset + i * EltSize);
96    return;
97  }
98  // Interpret void as zero return values.
99  if (Ty->isVoidTy())
100    return;
101  // Base case: we can get an EVT for this LLVM IR type.
102  ValueVTs.push_back(TLI.getValueType(Ty));
103  if (Offsets)
104    Offsets->push_back(StartingOffset);
105}
106
107/// ExtractTypeInfo - Returns the type info, possibly bitcast, encoded in V.
108GlobalVariable *llvm::ExtractTypeInfo(Value *V) {
109  V = V->stripPointerCasts();
110  GlobalVariable *GV = dyn_cast<GlobalVariable>(V);
111
112  if (GV && GV->getName() == "llvm.eh.catch.all.value") {
113    assert(GV->hasInitializer() &&
114           "The EH catch-all value must have an initializer");
115    Value *Init = GV->getInitializer();
116    GV = dyn_cast<GlobalVariable>(Init);
117    if (!GV) V = cast<ConstantPointerNull>(Init);
118  }
119
120  assert((GV || isa<ConstantPointerNull>(V)) &&
121         "TypeInfo must be a global variable or NULL");
122  return GV;
123}
124
125/// hasInlineAsmMemConstraint - Return true if the inline asm instruction being
126/// processed uses a memory 'm' constraint.
127bool
128llvm::hasInlineAsmMemConstraint(std::vector<InlineAsm::ConstraintInfo> &CInfos,
129                                const TargetLowering &TLI) {
130  for (unsigned i = 0, e = CInfos.size(); i != e; ++i) {
131    InlineAsm::ConstraintInfo &CI = CInfos[i];
132    for (unsigned j = 0, ee = CI.Codes.size(); j != ee; ++j) {
133      TargetLowering::ConstraintType CType = TLI.getConstraintType(CI.Codes[j]);
134      if (CType == TargetLowering::C_Memory)
135        return true;
136    }
137
138    // Indirect operand accesses access memory.
139    if (CI.isIndirect)
140      return true;
141  }
142
143  return false;
144}
145
146/// getFCmpCondCode - Return the ISD condition code corresponding to
147/// the given LLVM IR floating-point condition code.  This includes
148/// consideration of global floating-point math flags.
149///
150ISD::CondCode llvm::getFCmpCondCode(FCmpInst::Predicate Pred) {
151  ISD::CondCode FPC, FOC;
152  switch (Pred) {
153  case FCmpInst::FCMP_FALSE: FOC = FPC = ISD::SETFALSE; break;
154  case FCmpInst::FCMP_OEQ:   FOC = ISD::SETEQ; FPC = ISD::SETOEQ; break;
155  case FCmpInst::FCMP_OGT:   FOC = ISD::SETGT; FPC = ISD::SETOGT; break;
156  case FCmpInst::FCMP_OGE:   FOC = ISD::SETGE; FPC = ISD::SETOGE; break;
157  case FCmpInst::FCMP_OLT:   FOC = ISD::SETLT; FPC = ISD::SETOLT; break;
158  case FCmpInst::FCMP_OLE:   FOC = ISD::SETLE; FPC = ISD::SETOLE; break;
159  case FCmpInst::FCMP_ONE:   FOC = ISD::SETNE; FPC = ISD::SETONE; break;
160  case FCmpInst::FCMP_ORD:   FOC = FPC = ISD::SETO;   break;
161  case FCmpInst::FCMP_UNO:   FOC = FPC = ISD::SETUO;  break;
162  case FCmpInst::FCMP_UEQ:   FOC = ISD::SETEQ; FPC = ISD::SETUEQ; break;
163  case FCmpInst::FCMP_UGT:   FOC = ISD::SETGT; FPC = ISD::SETUGT; break;
164  case FCmpInst::FCMP_UGE:   FOC = ISD::SETGE; FPC = ISD::SETUGE; break;
165  case FCmpInst::FCMP_ULT:   FOC = ISD::SETLT; FPC = ISD::SETULT; break;
166  case FCmpInst::FCMP_ULE:   FOC = ISD::SETLE; FPC = ISD::SETULE; break;
167  case FCmpInst::FCMP_UNE:   FOC = ISD::SETNE; FPC = ISD::SETUNE; break;
168  case FCmpInst::FCMP_TRUE:  FOC = FPC = ISD::SETTRUE; break;
169  default:
170    llvm_unreachable("Invalid FCmp predicate opcode!");
171    FOC = FPC = ISD::SETFALSE;
172    break;
173  }
174  if (NoNaNsFPMath)
175    return FOC;
176  else
177    return FPC;
178}
179
180/// getICmpCondCode - Return the ISD condition code corresponding to
181/// the given LLVM IR integer condition code.
182///
183ISD::CondCode llvm::getICmpCondCode(ICmpInst::Predicate Pred) {
184  switch (Pred) {
185  case ICmpInst::ICMP_EQ:  return ISD::SETEQ;
186  case ICmpInst::ICMP_NE:  return ISD::SETNE;
187  case ICmpInst::ICMP_SLE: return ISD::SETLE;
188  case ICmpInst::ICMP_ULE: return ISD::SETULE;
189  case ICmpInst::ICMP_SGE: return ISD::SETGE;
190  case ICmpInst::ICMP_UGE: return ISD::SETUGE;
191  case ICmpInst::ICMP_SLT: return ISD::SETLT;
192  case ICmpInst::ICMP_ULT: return ISD::SETULT;
193  case ICmpInst::ICMP_SGT: return ISD::SETGT;
194  case ICmpInst::ICMP_UGT: return ISD::SETUGT;
195  default:
196    llvm_unreachable("Invalid ICmp predicate opcode!");
197    return ISD::SETNE;
198  }
199}
200
201/// Test if the given instruction is in a position to be optimized
202/// with a tail-call. This roughly means that it's in a block with
203/// a return and there's nothing that needs to be scheduled
204/// between it and the return.
205///
206/// This function only tests target-independent requirements.
207bool llvm::isInTailCallPosition(ImmutableCallSite CS, Attributes CalleeRetAttr,
208                                const TargetLowering &TLI) {
209  const Instruction *I = CS.getInstruction();
210  const BasicBlock *ExitBB = I->getParent();
211  const TerminatorInst *Term = ExitBB->getTerminator();
212  const ReturnInst *Ret = dyn_cast<ReturnInst>(Term);
213  const Function *F = ExitBB->getParent();
214
215  // The block must end in a return statement or unreachable.
216  //
217  // FIXME: Decline tailcall if it's not guaranteed and if the block ends in
218  // an unreachable, for now. The way tailcall optimization is currently
219  // implemented means it will add an epilogue followed by a jump. That is
220  // not profitable. Also, if the callee is a special function (e.g.
221  // longjmp on x86), it can end up causing miscompilation that has not
222  // been fully understood.
223  if (!Ret &&
224      (!GuaranteedTailCallOpt || !isa<UnreachableInst>(Term))) return false;
225
226  // If I will have a chain, make sure no other instruction that will have a
227  // chain interposes between I and the return.
228  if (I->mayHaveSideEffects() || I->mayReadFromMemory() ||
229      !I->isSafeToSpeculativelyExecute())
230    for (BasicBlock::const_iterator BBI = prior(prior(ExitBB->end())); ;
231         --BBI) {
232      if (&*BBI == I)
233        break;
234      // Debug info intrinsics do not get in the way of tail call optimization.
235      if (isa<DbgInfoIntrinsic>(BBI))
236        continue;
237      if (BBI->mayHaveSideEffects() || BBI->mayReadFromMemory() ||
238          !BBI->isSafeToSpeculativelyExecute())
239        return false;
240    }
241
242  // If the block ends with a void return or unreachable, it doesn't matter
243  // what the call's return type is.
244  if (!Ret || Ret->getNumOperands() == 0) return true;
245
246  // If the return value is undef, it doesn't matter what the call's
247  // return type is.
248  if (isa<UndefValue>(Ret->getOperand(0))) return true;
249
250  // Conservatively require the attributes of the call to match those of
251  // the return. Ignore noalias because it doesn't affect the call sequence.
252  unsigned CallerRetAttr = F->getAttributes().getRetAttributes();
253  if ((CalleeRetAttr ^ CallerRetAttr) & ~Attribute::NoAlias)
254    return false;
255
256  // It's not safe to eliminate the sign / zero extension of the return value.
257  if ((CallerRetAttr & Attribute::ZExt) || (CallerRetAttr & Attribute::SExt))
258    return false;
259
260  // Otherwise, make sure the unmodified return value of I is the return value.
261  for (const Instruction *U = dyn_cast<Instruction>(Ret->getOperand(0)); ;
262       U = dyn_cast<Instruction>(U->getOperand(0))) {
263    if (!U)
264      return false;
265    if (!U->hasOneUse())
266      return false;
267    if (U == I)
268      break;
269    // Check for a truly no-op truncate.
270    if (isa<TruncInst>(U) &&
271        TLI.isTruncateFree(U->getOperand(0)->getType(), U->getType()))
272      continue;
273    // Check for a truly no-op bitcast.
274    if (isa<BitCastInst>(U) &&
275        (U->getOperand(0)->getType() == U->getType() ||
276         (U->getOperand(0)->getType()->isPointerTy() &&
277          U->getType()->isPointerTy())))
278      continue;
279    // Otherwise it's not a true no-op.
280    return false;
281  }
282
283  return true;
284}
285
286