[llvm-commits] [llvm] r158358 - in /llvm/trunk: include/llvm/Constants.h include/llvm/Instruction.h lib/Transforms/Scalar/Reassociate.cpp lib/VMCore/Constants.cpp lib/VMCore/Instruction.cpp test/Transforms/Reassociate/repeats.ll

Matt Beaumont-Gay matthewbg at google.com
Tue Jun 12 11:17:28 PDT 2012


This might also help:

==31759== ERROR: AddressSanitizer attempting double-free on 0x7f4fc6415580:
  #0 0x40b3c02 operator delete[]()
  #1 0x4006c54 llvm::APInt::AssignSlowCase()
  #2 0x39048eb std::pair<>::operator=()
  #3 0x39045c9 llvm::FlatArrayMap<>::insertInternal()
  #4 0x3902ee3 llvm::FlatArrayMap<>::insert()
  #5 0x3902a0c llvm::MultiImplMap<>::insert()
  #6 0x39013d1 llvm::MultiImplMap<>::operator[]()
  #7 0x38f12bf LinearizeExprTree()
  #8 0x38eff0c (anonymous namespace)::Reassociate::ReassociateExpression()
  #9 0x38ef101 (anonymous namespace)::Reassociate::OptimizeInst()
  #10 0x38ee154 (anonymous namespace)::Reassociate::runOnFunction()
  #11 0x3f02ac5 llvm::FPPassManager::runOnFunction()
  #12 0x3b9f498 (anonymous namespace)::CGPassManager::RunPassOnSCC()
  #13 0x3b9ed0e (anonymous namespace)::CGPassManager::RunAllPassesOnSCC()
  #14 0x3b9e4e2 (anonymous namespace)::CGPassManager::runOnModule()
  #15 0x3f0328c llvm::MPPassManager::runOnModule()
  #16 0x3f03c96 llvm::PassManagerImpl::run()
  #17 0x3f03e99 llvm::PassManager::run()
  #18 0x14b92a1 (anonymous namespace)::EmitAssemblyHelper::EmitAssembly()
  #19 0x14b8df7 clang::EmitBackendOutput()
  #20 0x14b25b3 clang::BackendConsumer::HandleTranslationUnit()
  #21 0x2193ba2 clang::ParseAST()
  #22 0x14b06b4 clang::CodeGenAction::ExecuteAction()
  #23 0x2005fb3 clang::FrontendAction::Execute()
  #24 0x1e960fa clang::CompilerInstance::ExecuteAction()
  #25 0x14abf1a clang::ExecuteCompilerInvocation()
  #26 0x148f499 cc1_main()
  #27 0x14a2b9b main
  #28 0x7f4fc9656d5d __libc_start_main
0x7f4fc6415580 is located 0 bytes inside of 16-byte region
[0x7f4fc6415580,0x7f4fc6415590)
freed by thread T0 here:
  #0 0x40b3c02 operator delete[]()
  #1 0x39022bf llvm::FlatArrayMap<>::erase()
  #2 0x3902200 llvm::MultiImplMap<>::erase()
  #3 0x38f0ffc LinearizeExprTree()
  #4 0x38eff0c (anonymous namespace)::Reassociate::ReassociateExpression()
  #5 0x38ef101 (anonymous namespace)::Reassociate::OptimizeInst()
  #6 0x38ee154 (anonymous namespace)::Reassociate::runOnFunction()
  #7 0x3f02ac5 llvm::FPPassManager::runOnFunction()
  #8 0x3b9f498 (anonymous namespace)::CGPassManager::RunPassOnSCC()
  #9 0x3b9ed0e (anonymous namespace)::CGPassManager::RunAllPassesOnSCC()
  #10 0x3b9e4e2 (anonymous namespace)::CGPassManager::runOnModule()
  #11 0x3f0328c llvm::MPPassManager::runOnModule()
  #12 0x3f03c96 llvm::PassManagerImpl::run()
  #13 0x3f03e99 llvm::PassManager::run()
  #14 0x14b92a1 (anonymous namespace)::EmitAssemblyHelper::EmitAssembly()
  #15 0x14b8df7 clang::EmitBackendOutput()
  #16 0x14b25b3 clang::BackendConsumer::HandleTranslationUnit()
  #17 0x2193ba2 clang::ParseAST()
  #18 0x14b06b4 clang::CodeGenAction::ExecuteAction()
  #19 0x2005fb3 clang::FrontendAction::Execute()
  #20 0x1e960fa clang::CompilerInstance::ExecuteAction()
  #21 0x14abf1a clang::ExecuteCompilerInvocation()
  #22 0x148f499 cc1_main()
  #23 0x14a2b9b main
  #24 0x7f4fc9656d5d __libc_start_main
previously allocated by thread T0 here:
  #0 0x40b3a82 operator new[]()
  #1 0x4005c7f getMemory()
  #2 0x4006a2f llvm::APInt::AssignSlowCase()
  #3 0x38f12d2 LinearizeExprTree()
  #4 0x38eff0c (anonymous namespace)::Reassociate::ReassociateExpression()
  #5 0x38ef101 (anonymous namespace)::Reassociate::OptimizeInst()
  #6 0x38ee154 (anonymous namespace)::Reassociate::runOnFunction()
  #7 0x3f02ac5 llvm::FPPassManager::runOnFunction()
  #8 0x3b9f498 (anonymous namespace)::CGPassManager::RunPassOnSCC()
  #9 0x3b9ed0e (anonymous namespace)::CGPassManager::RunAllPassesOnSCC()
  #10 0x3b9e4e2 (anonymous namespace)::CGPassManager::runOnModule()
  #11 0x3f0328c llvm::MPPassManager::runOnModule()
  #12 0x3f03c96 llvm::PassManagerImpl::run()
  #13 0x3f03e99 llvm::PassManager::run()
  #14 0x14b92a1 (anonymous namespace)::EmitAssemblyHelper::EmitAssembly()
  #15 0x14b8df7 clang::EmitBackendOutput()
  #16 0x14b25b3 clang::BackendConsumer::HandleTranslationUnit()
  #17 0x2193ba2 clang::ParseAST()
  #18 0x14b06b4 clang::CodeGenAction::ExecuteAction()
  #19 0x2005fb3 clang::FrontendAction::Execute()
  #20 0x1e960fa clang::CompilerInstance::ExecuteAction()
  #21 0x14abf1a clang::ExecuteCompilerInvocation()
  #22 0x148f499 cc1_main()
  #23 0x14a2b9b main

On Tue, Jun 12, 2012 at 11:04 AM, Matt Beaumont-Gay
<matthewbg at google.com> wrote:
> This seems to have caused some heap corruption when building OpenSSL at -O1:
>
> #14 0x00007f45009c7806 in malloc_printerr (action=3,
>    str=0x7f4500a9b2f0 "double free or corruption (fasttop)",
>    ptr=<optimized out>) at malloc.c:6266
> #15 0x00007f45009ce0d3 in *__GI___libc_free (mem=<optimized out>)
>    at malloc.c:3738
> #16 0x0000000002821bb9 in llvm::APInt::AssignSlowCase (this=0x7fffa46d55b8,
>    RHS=...) at llvm/lib/Support/APInt.cpp:143
> #17 0x00000000008f1f59 in llvm::APInt::operator= (this=0x7fffa46d55b8, RHS=...)
>    at llvm/include/llvm/ADT/APInt.h:595
> #18 0x0000000002401d5e in std::pair<llvm::Value*, llvm::APInt>::operator= (
>    this=0x7fffa46d55b0)
>    at /usr/lib/gcc/x86_64-linux-gnu/4.4/../../../../include/c++/4.4/bits/stl_pair.h:67
> #19 0x0000000002402ae6 in llvm::FlatArrayMap<llvm::Value*,
> llvm::APInt, 8u>::insertInternal (this=0x7fffa46d5598, Ptr=0x4767698,
>    Val=<error reading variable: Unhandled dwarf expression opcode 0x0>,
>    Item=@0x7fffa46d4fd8: 0x4767698)
>    at llvm/include/llvm/ADT/FlatArrayMap.h:117
> #20 0x0000000002401fed in llvm::FlatArrayMap<llvm::Value*,
> llvm::APInt, 8u>::insert (this=0x7fffa46d5598, KV=...)
>    at llvm/include/llvm/ADT/FlatArrayMap.h:188
> #21 0x0000000002401e57 in
> llvm::MultiImplMap<llvm::FlatArrayMap<llvm::Value*, llvm::APInt, 8u>,
> llvm::DenseMap<llvm::Value*, llvm::APInt,
> llvm::DenseMapInfo<llvm::Value*> >, 8u, false,
> llvm::MultiImplMapIteratorsFactory<llvm::FlatArrayMap<llvm::Value*,
> llvm::APInt, 8u>, llvm::DenseMap<llvm::Value*, llvm::APInt,
> llvm::DenseMapInfo<llvm::Value*> > > >::insert (this=0x7fffa46d5598,
> KV=...)
>    at llvm/include/llvm/ADT/MultiImplMap.h:218
> #22 0x0000000002400e81 in
> llvm::MultiImplMap<llvm::FlatArrayMap<llvm::Value*, llvm::APInt, 8u>,
> llvm::DenseMap<llvm::Value*, llvm::APInt,
> llvm::DenseMapInfo<llvm::Value*> >, 8u, false,
> llvm::MultiImplMapIteratorsFactory<llvm::FlatArrayMap<llvm::Value*,
> llvm::APInt, 8u>, llvm::DenseMap<llvm::Value*, llvm::APInt,
> llvm::DenseMapInfo<llvm::Value*> > > >::operator[]
> (this=0x7fffa46d5598,
>    Key=@0x7fffa46d54a0: 0x4767698)
>    at llvm/include/llvm/ADT/MultiImplMap.h:281
> #23 0x00000000023f6626 in LinearizeExprTree (I=0x47679b0, Ops=...)
>    at llvm/lib/Transforms/Scalar/Reassociate.cpp:589
>
> I'll throw delta at it and give you a real bug report soon.
>
> On Tue, Jun 12, 2012 at 7:33 AM, Duncan Sands <baldrick at free.fr> wrote:
>> Author: baldrick
>> Date: Tue Jun 12 09:33:56 2012
>> New Revision: 158358
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=158358&view=rev
>> Log:
>> Now that Reassociate's LinearizeExprTree can look through arbitrary expression
>> topologies, it is quite possible for a leaf node to have huge multiplicity, for
>> example: x0 = x*x, x1 = x0*x0, x2 = x1*x1, ... rapidly gives a value which is x
>> raised to a vast power (the multiplicity, or weight, of x).  This patch fixes
>> the computation of weights by correctly computing them no matter how big they
>> are, rather than just overflowing and getting a wrong value.  It turns out that
>> the weight for a value never needs more bits to represent than the value itself,
>> so it is enough to represent weights as APInts of the same bitwidth and do the
>> right overflow-avoiding dance steps when computing weights.  As a side-effect it
>> reduces the number of multiplies needed in some cases of large powers.  While
>> there, in view of external uses (eg by the vectorizer) I made LinearizeExprTree
>> static, pushing the rank computation out into users.  This is progress towards
>> fixing PR13021.
>>
>> Added:
>>    llvm/trunk/test/Transforms/Reassociate/repeats.ll
>> Modified:
>>    llvm/trunk/include/llvm/Constants.h
>>    llvm/trunk/include/llvm/Instruction.h
>>    llvm/trunk/lib/Transforms/Scalar/Reassociate.cpp
>>    llvm/trunk/lib/VMCore/Constants.cpp
>>    llvm/trunk/lib/VMCore/Instruction.cpp
>>
>> Modified: llvm/trunk/include/llvm/Constants.h
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Constants.h?rev=158358&r1=158357&r2=158358&view=diff
>> ==============================================================================
>> --- llvm/trunk/include/llvm/Constants.h (original)
>> +++ llvm/trunk/include/llvm/Constants.h Tue Jun 12 09:33:56 2012
>> @@ -917,6 +917,11 @@
>>     return getLShr(C1, C2, true);
>>   }
>>
>> +  /// getBinOpIdentity - Return the identity for the given binary operation,
>> +  /// i.e. a constant C such that X op C = X and C op X = X for every X.  It
>> +  /// is an error to call this for an operation that doesn't have an identity.
>> +  static Constant *getBinOpIdentity(unsigned Opcode, Type *Ty);
>> +
>>   /// Transparently provide more efficient getOperand methods.
>>   DECLARE_TRANSPARENT_OPERAND_ACCESSORS(Constant);
>>
>>
>> Modified: llvm/trunk/include/llvm/Instruction.h
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Instruction.h?rev=158358&r1=158357&r2=158358&view=diff
>> ==============================================================================
>> --- llvm/trunk/include/llvm/Instruction.h (original)
>> +++ llvm/trunk/include/llvm/Instruction.h Tue Jun 12 09:33:56 2012
>> @@ -215,6 +215,27 @@
>>   bool isCommutative() const { return isCommutative(getOpcode()); }
>>   static bool isCommutative(unsigned op);
>>
>> +  /// isIdempotent - Return true if the instruction is idempotent:
>> +  ///
>> +  ///   Idempotent operators satisfy:  x op x === x
>> +  ///
>> +  /// In LLVM, the And and Or operators are idempotent.
>> +  ///
>> +  bool isIdempotent() const { return isIdempotent(getOpcode()); }
>> +  static bool isIdempotent(unsigned op);
>> +
>> +  /// isNilpotent - Return true if the instruction is nilpotent:
>> +  ///
>> +  ///   Nilpotent operators satisfy:  x op x === Id,
>> +  ///
>> +  ///   where Id is the identity for the operator, i.e. a constant such that
>> +  ///     x op Id === x and Id op x === x for all x.
>> +  ///
>> +  /// In LLVM, the Xor operator is nilpotent.
>> +  ///
>> +  bool isNilpotent() const { return isNilpotent(getOpcode()); }
>> +  static bool isNilpotent(unsigned op);
>> +
>>   /// mayWriteToMemory - Return true if this instruction may modify memory.
>>   ///
>>   bool mayWriteToMemory() const;
>>
>> Modified: llvm/trunk/lib/Transforms/Scalar/Reassociate.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/Reassociate.cpp?rev=158358&r1=158357&r2=158358&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/Transforms/Scalar/Reassociate.cpp (original)
>> +++ llvm/trunk/lib/Transforms/Scalar/Reassociate.cpp Tue Jun 12 09:33:56 2012
>> @@ -143,7 +143,6 @@
>>     Value *buildMinimalMultiplyDAG(IRBuilder<> &Builder,
>>                                    SmallVectorImpl<Factor> &Factors);
>>     Value *OptimizeMul(BinaryOperator *I, SmallVectorImpl<ValueEntry> &Ops);
>> -    void LinearizeExprTree(BinaryOperator *I, SmallVectorImpl<ValueEntry> &Ops);
>>     Value *RemoveFactorFromExpression(Value *V, Value *Factor);
>>     void EraseInst(Instruction *I);
>>     void OptimizeInst(Instruction *I);
>> @@ -251,10 +250,148 @@
>>   return Res;
>>  }
>>
>> +/// CarmichaelShift - Returns k such that lambda(2^Bitwidth) = 2^k, where lambda
>> +/// is the Carmichael function. This means that x^(2^k) === 1 mod 2^Bitwidth for
>> +/// every odd x, i.e. x^(2^k) = 1 for every odd x in Bitwidth-bit arithmetic.
>> +/// Note that 0 <= k < Bitwidth, and if Bitwidth > 3 then x^(2^k) = 0 for every
>> +/// even x in Bitwidth-bit arithmetic.
>> +static unsigned CarmichaelShift(unsigned Bitwidth) {
>> +  if (Bitwidth < 3)
>> +    return Bitwidth - 1;
>> +  return Bitwidth - 2;
>> +}
>> +
>> +/// IncorporateWeight - Add the extra weight 'RHS' to the existing weight 'LHS',
>> +/// reducing the combined weight using any special properties of the operation.
>> +/// The existing weight LHS represents the computation X op X op ... op X where
>> +/// X occurs LHS times.  The combined weight represents  X op X op ... op X with
>> +/// X occurring LHS + RHS times.  If op is "Xor" for example then the combined
>> +/// operation is equivalent to X if LHS + RHS is odd, or 0 if LHS + RHS is even;
>> +/// the routine returns 1 in LHS in the first case, and 0 in LHS in the second.
>> +static void IncorporateWeight(APInt &LHS, const APInt &RHS, unsigned Opcode) {
>> +  // If we were working with infinite precision arithmetic then the combined
>> +  // weight would be LHS + RHS.  But we are using finite precision arithmetic,
>> +  // and the APInt sum LHS + RHS may not be correct if it wraps (it is correct
>> +  // for nilpotent operations and addition, but not for idempotent operations
>> +  // and multiplication), so it is important to correctly reduce the combined
>> +  // weight back into range if wrapping would be wrong.
>> +
>> +  // If RHS is zero then the weight didn't change.
>> +  if (RHS.isMinValue())
>> +    return;
>> +  // If LHS is zero then the combined weight is RHS.
>> +  if (LHS.isMinValue()) {
>> +    LHS = RHS;
>> +    return;
>> +  }
>> +  // From this point on we know that neither LHS nor RHS is zero.
>> +
>> +  if (Instruction::isIdempotent(Opcode)) {
>> +    // Idempotent means X op X === X, so any non-zero weight is equivalent to a
>> +    // weight of 1.  Keeping weights at zero or one also means that wrapping is
>> +    // not a problem.
>> +    assert(LHS == 1 && RHS == 1 && "Weights not reduced!");
>> +    return; // Return a weight of 1.
>> +  }
>> +  if (Instruction::isNilpotent(Opcode)) {
>> +    // Nilpotent means X op X === 0, so reduce weights modulo 2.
>> +    assert(LHS == 1 && RHS == 1 && "Weights not reduced!");
>> +    LHS = 0; // 1 + 1 === 0 modulo 2.
>> +    return;
>> +  }
>> +  if (Opcode == Instruction::Add) {
>> +    // TODO: Reduce the weight by exploiting nsw/nuw?
>> +    LHS += RHS;
>> +    return;
>> +  }
>> +
>> +  assert(Opcode == Instruction::Mul && "Unknown associative operation!");
>> +  unsigned Bitwidth = LHS.getBitWidth();
>> +  // If CM is the Carmichael number then a weight W satisfying W >= CM+Bitwidth
>> +  // can be replaced with W-CM.  That's because x^W=x^(W-CM) for every Bitwidth
>> +  // bit number x, since either x is odd in which case x^CM = 1, or x is even in
>> +  // which case both x^W and x^(W - CM) are zero.  By subtracting off multiples
>> +  // of CM like this weights can always be reduced to the range [0, CM+Bitwidth)
>> +  // which by a happy accident means that they can always be represented using
>> +  // Bitwidth bits.
>> +  // TODO: Reduce the weight by exploiting nsw/nuw?  (Could do much better than
>> +  // the Carmichael number).
>> +  if (Bitwidth > 3) {
>> +    /// CM - The value of Carmichael's lambda function.
>> +    APInt CM = APInt::getOneBitSet(Bitwidth, CarmichaelShift(Bitwidth));
>> +    // Any weight W >= Threshold can be replaced with W - CM.
>> +    APInt Threshold = CM + Bitwidth;
>> +    assert(LHS.ult(Threshold) && RHS.ult(Threshold) && "Weights not reduced!");
>> +    // For Bitwidth 4 or more the following sum does not overflow.
>> +    LHS += RHS;
>> +    while (LHS.uge(Threshold))
>> +      LHS -= CM;
>> +  } else {
>> +    // To avoid problems with overflow do everything the same as above but using
>> +    // a larger type.
>> +    unsigned CM = 1U << CarmichaelShift(Bitwidth);
>> +    unsigned Threshold = CM + Bitwidth;
>> +    assert(LHS.getZExtValue() < Threshold && RHS.getZExtValue() < Threshold &&
>> +           "Weights not reduced!");
>> +    unsigned Total = LHS.getZExtValue() + RHS.getZExtValue();
>> +    while (Total >= Threshold)
>> +      Total -= CM;
>> +    LHS = Total;
>> +  }
>> +}
>> +
>> +/// EvaluateRepeatedConstant - Compute C op C op ... op C where the constant C
>> +/// is repeated Weight times.
>> +static Constant *EvaluateRepeatedConstant(unsigned Opcode, Constant *C,
>> +                                          APInt Weight) {
>> +  // For addition the result can be efficiently computed as the product of the
>> +  // constant and the weight.
>> +  if (Opcode == Instruction::Add)
>> +    return ConstantExpr::getMul(C, ConstantInt::get(C->getContext(), Weight));
>> +
>> +  // The weight might be huge, so compute by repeated squaring to ensure that
>> +  // compile time is proportional to the logarithm of the weight.
>> +  Constant *Result = 0;
>> +  Constant *Power = C; // Successively C, C op C, (C op C) op (C op C) etc.
>> +  // Visit the bits in Weight.
>> +  while (Weight != 0) {
>> +    // If the current bit in Weight is non-zero do Result = Result op Power.
>> +    if (Weight[0])
>> +      Result = Result ? ConstantExpr::get(Opcode, Result, Power) : Power;
>> +    // Move on to the next bit if any more are non-zero.
>> +    Weight = Weight.lshr(1);
>> +    if (Weight.isMinValue())
>> +      break;
>> +    // Square the power.
>> +    Power = ConstantExpr::get(Opcode, Power, Power);
>> +  }
>> +
>> +  assert(Result && "Only positive weights supported!");
>> +  return Result;
>> +}
>> +
>> +typedef std::pair<Value*, APInt> RepeatedValue;
>> +
>>  /// LinearizeExprTree - Given an associative binary expression, return the leaf
>> -/// nodes in Ops.  The original expression is the same as Ops[0] op ... Ops[N].
>> -/// Note that a node may occur multiple times in Ops, but if so all occurrences
>> -/// are consecutive in the vector.
>> +/// nodes in Ops along with their weights (how many times the leaf occurs).  The
>> +/// original expression is the same as
>> +///   (Ops[0].first op Ops[0].first op ... Ops[0].first)  <- Ops[0].second times
>> +/// op
>> +///   (Ops[1].first op Ops[1].first op ... Ops[1].first)  <- Ops[1].second times
>> +/// op
>> +///   ...
>> +/// op
>> +///   (Ops[N].first op Ops[N].first op ... Ops[N].first)  <- Ops[N].second times
>> +///
>> +/// Note that the values Ops[0].first, ..., Ops[N].first are all distinct, and
>> +/// they are all non-constant except possibly for the last one, which if it is
>> +/// constant will have weight one (Ops[N].second === 1).
>> +///
>> +/// This routine may modify the function, in which case it returns 'true'.  The
>> +/// changes it makes may well be destructive, changing the value computed by 'I'
>> +/// to something completely different.  Thus if the routine returns 'true' then
>> +/// you MUST either replace I with a new expression computed from the Ops array,
>> +/// or use RewriteExprTree to put the values back in.
>>  ///
>>  /// A leaf node is either not a binary operation of the same kind as the root
>>  /// node 'I' (i.e. is not a binary operator at all, or is, but with a different
>> @@ -276,7 +413,7 @@
>>  ///                   +   *      |      F,  G
>>  ///
>>  /// The leaf nodes are C, E, F and G.  The Ops array will contain (maybe not in
>> -/// that order) C, E, F, F, G, G.
>> +/// that order) (C, 1), (E, 1), (F, 2), (G, 2).
>>  ///
>>  /// The expression is maximal: if some instruction is a binary operator of the
>>  /// same kind as 'I', and all of its uses are non-leaf nodes of the expression,
>> @@ -287,7 +424,8 @@
>>  /// order to ensure that every non-root node in the expression has *exactly one*
>>  /// use by a non-leaf node of the expression.  This destruction means that the
>>  /// caller MUST either replace 'I' with a new expression or use something like
>> -/// RewriteExprTree to put the values back in.
>> +/// RewriteExprTree to put the values back in if the routine indicates that it
>> +/// made a change by returning 'true'.
>>  ///
>>  /// In the above example either the right operand of A or the left operand of B
>>  /// will be replaced by undef.  If it is B's operand then this gives:
>> @@ -310,9 +448,14 @@
>>  /// of the expression) if it can turn them into binary operators of the right
>>  /// type and thus make the expression bigger.
>>
>> -void Reassociate::LinearizeExprTree(BinaryOperator *I,
>> -                                    SmallVectorImpl<ValueEntry> &Ops) {
>> +static bool LinearizeExprTree(BinaryOperator *I,
>> +                              SmallVectorImpl<RepeatedValue> &Ops) {
>>   DEBUG(dbgs() << "LINEARIZE: " << *I << '\n');
>> +  unsigned Bitwidth = I->getType()->getScalarType()->getPrimitiveSizeInBits();
>> +  unsigned Opcode = I->getOpcode();
>> +  assert(Instruction::isAssociative(Opcode) &&
>> +         Instruction::isCommutative(Opcode) &&
>> +         "Expected an associative and commutative operation!");
>>
>>   // Visit all operands of the expression, keeping track of their weight (the
>>   // number of paths from the expression root to the operand, or if you like
>> @@ -324,9 +467,9 @@
>>   // with their weights, representing a certain number of paths to the operator.
>>   // If an operator occurs in the worklist multiple times then we found multiple
>>   // ways to get to it.
>> -  SmallVector<std::pair<BinaryOperator*, unsigned>, 8> Worklist; // (Op, Weight)
>> -  Worklist.push_back(std::make_pair(I, 1));
>> -  unsigned Opcode = I->getOpcode();
>> +  SmallVector<std::pair<BinaryOperator*, APInt>, 8> Worklist; // (Op, Weight)
>> +  Worklist.push_back(std::make_pair(I, APInt(Bitwidth, 1)));
>> +  bool MadeChange = false;
>>
>>   // Leaves of the expression are values that either aren't the right kind of
>>   // operation (eg: a constant, or a multiply in an add tree), or are, but have
>> @@ -343,7 +486,7 @@
>>
>>   // Leaves - Keeps track of the set of putative leaves as well as the number of
>>   // paths to each leaf seen so far.
>> -  typedef SmallMap<Value*, unsigned, 8> LeafMap;
>> +  typedef SmallMap<Value*, APInt, 8> LeafMap;
>>   LeafMap Leaves; // Leaf -> Total weight so far.
>>   SmallVector<Value*, 8> LeafOrder; // Ensure deterministic leaf output order.
>>
>> @@ -351,13 +494,12 @@
>>   SmallPtrSet<Value*, 8> Visited; // For sanity checking the iteration scheme.
>>  #endif
>>   while (!Worklist.empty()) {
>> -    std::pair<BinaryOperator*, unsigned> P = Worklist.pop_back_val();
>> +    std::pair<BinaryOperator*, APInt> P = Worklist.pop_back_val();
>>     I = P.first; // We examine the operands of this binary operator.
>> -    assert(P.second >= 1 && "No paths to here, so how did we get here?!");
>>
>>     for (unsigned OpIdx = 0; OpIdx < 2; ++OpIdx) { // Visit operands.
>>       Value *Op = I->getOperand(OpIdx);
>> -      unsigned Weight = P.second; // Number of paths to this operand.
>> +      APInt Weight = P.second; // Number of paths to this operand.
>>       DEBUG(dbgs() << "OPERAND: " << *Op << " (" << Weight << ")\n");
>>       assert(!Op->use_empty() && "No uses, so how did we get to it?!");
>>
>> @@ -389,7 +531,7 @@
>>         assert(Visited.count(Op) && "In leaf map but not visited!");
>>
>>         // Update the number of paths to the leaf.
>> -        It->second += Weight;
>> +        IncorporateWeight(It->second, Weight, Opcode);
>>
>>         // The leaf already has one use from inside the expression.  As we want
>>         // exactly one such use, drop this new use of the leaf.
>> @@ -450,21 +592,44 @@
>>
>>   // The leaves, repeated according to their weights, represent the linearized
>>   // form of the expression.
>> +  Constant *Cst = 0; // Accumulate constants here.
>>   for (unsigned i = 0, e = LeafOrder.size(); i != e; ++i) {
>>     Value *V = LeafOrder[i];
>>     LeafMap::iterator It = Leaves.find(V);
>>     if (It == Leaves.end())
>> -      // Leaf already output, or node initially thought to be a leaf wasn't.
>> +      // Node initially thought to be a leaf wasn't.
>>       continue;
>>     assert(!isReassociableOp(V, Opcode) && "Shouldn't be a leaf!");
>> -    unsigned Weight = It->second;
>> -    assert(Weight > 0 && "No paths to this value!");
>> -    // FIXME: Rather than repeating values Weight times, use a vector of
>> -    // (ValueEntry, multiplicity) pairs.
>> -    Ops.append(Weight, ValueEntry(getRank(V), V));
>> +    APInt Weight = It->second;
>> +    if (Weight.isMinValue())
>> +      // Leaf already output or weight reduction eliminated it.
>> +      continue;
>>     // Ensure the leaf is only output once.
>> -    Leaves.erase(It);
>> +    It->second = 0;
>> +    // Glob all constants together into Cst.
>> +    if (Constant *C = dyn_cast<Constant>(V)) {
>> +      C = EvaluateRepeatedConstant(Opcode, C, Weight);
>> +      Cst = Cst ? ConstantExpr::get(Opcode, Cst, C) : C;
>> +      continue;
>> +    }
>> +    // Add non-constant
>> +    Ops.push_back(std::make_pair(V, Weight));
>> +  }
>> +
>> +  // Add any constants back into Ops, all globbed together and reduced to having
>> +  // weight 1 for the convenience of users.
>> +  if (Cst && Cst != ConstantExpr::getBinOpIdentity(Opcode, I->getType()))
>> +    Ops.push_back(std::make_pair(Cst, APInt(Bitwidth, 1)));
>> +
>> +  // For nilpotent operations or addition there may be no operands, for example
>> +  // because the expression was "X xor X" or consisted of 2^Bitwidth additions:
>> +  // in both cases the weight reduces to 0 causing the value to be skipped.
>> +  if (Ops.empty()) {
>> +    Constant *Identity = ConstantExpr::getBinOpIdentity(Opcode, I->getType());
>> +    Ops.push_back(std::make_pair(Identity, APInt(Bitwidth, 1)));
>>   }
>> +
>> +  return MadeChange;
>>  }
>>
>>  // RewriteExprTree - Now that the operands for this expression tree are
>> @@ -775,8 +940,15 @@
>>   BinaryOperator *BO = isReassociableOp(V, Instruction::Mul);
>>   if (!BO) return 0;
>>
>> +  SmallVector<RepeatedValue, 8> Tree;
>> +  MadeChange |= LinearizeExprTree(BO, Tree);
>>   SmallVector<ValueEntry, 8> Factors;
>> -  LinearizeExprTree(BO, Factors);
>> +  Factors.reserve(Tree.size());
>> +  for (unsigned i = 0, e = Tree.size(); i != e; ++i) {
>> +    RepeatedValue E = Tree[i];
>> +    Factors.append(E.second.getZExtValue(),
>> +                   ValueEntry(getRank(E.first), E.first));
>> +  }
>>
>>   bool FoundFactor = false;
>>   bool NeedsNegate = false;
>> @@ -1439,8 +1611,15 @@
>>
>>   // First, walk the expression tree, linearizing the tree, collecting the
>>   // operand information.
>> +  SmallVector<RepeatedValue, 8> Tree;
>> +  MadeChange |= LinearizeExprTree(I, Tree);
>>   SmallVector<ValueEntry, 8> Ops;
>> -  LinearizeExprTree(I, Ops);
>> +  Ops.reserve(Tree.size());
>> +  for (unsigned i = 0, e = Tree.size(); i != e; ++i) {
>> +    RepeatedValue E = Tree[i];
>> +    Ops.append(E.second.getZExtValue(),
>> +               ValueEntry(getRank(E.first), E.first));
>> +  }
>>
>>   DEBUG(dbgs() << "RAIn:\t"; PrintOps(I, Ops); dbgs() << '\n');
>>
>>
>> Modified: llvm/trunk/lib/VMCore/Constants.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Constants.cpp?rev=158358&r1=158357&r2=158358&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/VMCore/Constants.cpp (original)
>> +++ llvm/trunk/lib/VMCore/Constants.cpp Tue Jun 12 09:33:56 2012
>> @@ -2007,6 +2007,26 @@
>>              isExact ? PossiblyExactOperator::IsExact : 0);
>>  }
>>
>> +/// getBinOpIdentity - Return the identity for the given binary operation,
>> +/// i.e. a constant C such that X op C = X and C op X = X for every X.  It
>> +/// is an error to call this for an operation that doesn't have an identity.
>> +Constant *ConstantExpr::getBinOpIdentity(unsigned Opcode, Type *Ty) {
>> +  switch (Opcode) {
>> +  default:
>> +    llvm_unreachable("Not a binary operation with identity");
>> +  case Instruction::Add:
>> +  case Instruction::Or:
>> +  case Instruction::Xor:
>> +    return Constant::getNullValue(Ty);
>> +
>> +  case Instruction::Mul:
>> +    return ConstantInt::get(Ty, 1);
>> +
>> +  case Instruction::And:
>> +    return Constant::getAllOnesValue(Ty);
>> +  }
>> +}
>> +
>>  // destroyConstant - Remove the constant from the constant table...
>>  //
>>  void ConstantExpr::destroyConstant() {
>>
>> Modified: llvm/trunk/lib/VMCore/Instruction.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Instruction.cpp?rev=158358&r1=158357&r2=158358&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/VMCore/Instruction.cpp (original)
>> +++ llvm/trunk/lib/VMCore/Instruction.cpp Tue Jun 12 09:33:56 2012
>> @@ -395,6 +395,29 @@
>>   }
>>  }
>>
>> +/// isIdempotent - Return true if the instruction is idempotent:
>> +///
>> +///   Idempotent operators satisfy:  x op x === x
>> +///
>> +/// In LLVM, the And and Or operators are idempotent.
>> +///
>> +bool Instruction::isIdempotent(unsigned Opcode) {
>> +  return Opcode == And || Opcode == Or;
>> +}
>> +
>> +/// isNilpotent - Return true if the instruction is nilpotent:
>> +///
>> +///   Nilpotent operators satisfy:  x op x === Id,
>> +///
>> +///   where Id is the identity for the operator, i.e. a constant such that
>> +///     x op Id === x and Id op x === x for all x.
>> +///
>> +/// In LLVM, the Xor operator is nilpotent.
>> +///
>> +bool Instruction::isNilpotent(unsigned Opcode) {
>> +  return Opcode == Xor;
>> +}
>> +
>>  Instruction *Instruction::clone() const {
>>   Instruction *New = clone_impl();
>>   New->SubclassOptionalData = SubclassOptionalData;
>>
>> Added: llvm/trunk/test/Transforms/Reassociate/repeats.ll
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/Reassociate/repeats.ll?rev=158358&view=auto
>> ==============================================================================
>> --- llvm/trunk/test/Transforms/Reassociate/repeats.ll (added)
>> +++ llvm/trunk/test/Transforms/Reassociate/repeats.ll Tue Jun 12 09:33:56 2012
>> @@ -0,0 +1,252 @@
>> +; RUN: opt < %s -reassociate -S | FileCheck %s
>> +
>> +; Tests involving repeated operations on the same value.
>> +
>> +define i8 @nilpotent(i8 %x) {
>> +; CHECK: @nilpotent
>> +  %tmp = xor i8 %x, %x
>> +  ret i8 %tmp
>> +; CHECK: ret i8 0
>> +}
>> +
>> +define i2 @idempotent(i2 %x) {
>> +; CHECK: @idempotent
>> +  %tmp1 = and i2 %x, %x
>> +  %tmp2 = and i2 %tmp1, %x
>> +  %tmp3 = and i2 %tmp2, %x
>> +  ret i2 %tmp3
>> +; CHECK: ret i2 %x
>> +}
>> +
>> +define i2 @add(i2 %x) {
>> +; CHECK: @add
>> +  %tmp1 = add i2 %x, %x
>> +  %tmp2 = add i2 %tmp1, %x
>> +  %tmp3 = add i2 %tmp2, %x
>> +  ret i2 %tmp3
>> +; CHECK: ret i2 0
>> +}
>> +
>> +define i2 @cst_add() {
>> +; CHECK: @cst_add
>> +  %tmp1 = add i2 1, 1
>> +  %tmp2 = add i2 %tmp1, 1
>> +  ret i2 %tmp2
>> +; CHECK: ret i2 -1
>> +}
>> +
>> +define i8 @cst_mul() {
>> +; CHECK: @cst_mul
>> +  %tmp1 = mul i8 3, 3
>> +  %tmp2 = mul i8 %tmp1, 3
>> +  %tmp3 = mul i8 %tmp2, 3
>> +  %tmp4 = mul i8 %tmp3, 3
>> +  ret i8 %tmp4
>> +; CHECK: ret i8 -13
>> +}
>> +
>> +define i3 @foo3x5(i3 %x) {
>> +; Can be done with two multiplies.
>> +; CHECK: @foo3x5
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i3 %x, %x
>> +  %tmp2 = mul i3 %tmp1, %x
>> +  %tmp3 = mul i3 %tmp2, %x
>> +  %tmp4 = mul i3 %tmp3, %x
>> +  ret i3 %tmp4
>> +}
>> +
>> +define i3 @foo3x6(i3 %x) {
>> +; Can be done with two multiplies.
>> +; CHECK: @foo3x6
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i3 %x, %x
>> +  %tmp2 = mul i3 %tmp1, %x
>> +  %tmp3 = mul i3 %tmp2, %x
>> +  %tmp4 = mul i3 %tmp3, %x
>> +  %tmp5 = mul i3 %tmp4, %x
>> +  ret i3 %tmp5
>> +}
>> +
>> +define i3 @foo3x7(i3 %x) {
>> +; Can be done with two multiplies.
>> +; CHECK: @foo3x7
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i3 %x, %x
>> +  %tmp2 = mul i3 %tmp1, %x
>> +  %tmp3 = mul i3 %tmp2, %x
>> +  %tmp4 = mul i3 %tmp3, %x
>> +  %tmp5 = mul i3 %tmp4, %x
>> +  %tmp6 = mul i3 %tmp5, %x
>> +  ret i3 %tmp6
>> +}
>> +
>> +define i4 @foo4x8(i4 %x) {
>> +; Can be done with two multiplies.
>> +; CHECK: @foo4x8
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i4 %x, %x
>> +  %tmp2 = mul i4 %tmp1, %x
>> +  %tmp3 = mul i4 %tmp2, %x
>> +  %tmp4 = mul i4 %tmp3, %x
>> +  %tmp5 = mul i4 %tmp4, %x
>> +  %tmp6 = mul i4 %tmp5, %x
>> +  %tmp7 = mul i4 %tmp6, %x
>> +  ret i4 %tmp7
>> +}
>> +
>> +define i4 @foo4x9(i4 %x) {
>> +; Can be done with three multiplies.
>> +; CHECK: @foo4x9
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i4 %x, %x
>> +  %tmp2 = mul i4 %tmp1, %x
>> +  %tmp3 = mul i4 %tmp2, %x
>> +  %tmp4 = mul i4 %tmp3, %x
>> +  %tmp5 = mul i4 %tmp4, %x
>> +  %tmp6 = mul i4 %tmp5, %x
>> +  %tmp7 = mul i4 %tmp6, %x
>> +  %tmp8 = mul i4 %tmp7, %x
>> +  ret i4 %tmp8
>> +}
>> +
>> +define i4 @foo4x10(i4 %x) {
>> +; Can be done with three multiplies.
>> +; CHECK: @foo4x10
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i4 %x, %x
>> +  %tmp2 = mul i4 %tmp1, %x
>> +  %tmp3 = mul i4 %tmp2, %x
>> +  %tmp4 = mul i4 %tmp3, %x
>> +  %tmp5 = mul i4 %tmp4, %x
>> +  %tmp6 = mul i4 %tmp5, %x
>> +  %tmp7 = mul i4 %tmp6, %x
>> +  %tmp8 = mul i4 %tmp7, %x
>> +  %tmp9 = mul i4 %tmp8, %x
>> +  ret i4 %tmp9
>> +}
>> +
>> +define i4 @foo4x11(i4 %x) {
>> +; Can be done with four multiplies.
>> +; CHECK: @foo4x11
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i4 %x, %x
>> +  %tmp2 = mul i4 %tmp1, %x
>> +  %tmp3 = mul i4 %tmp2, %x
>> +  %tmp4 = mul i4 %tmp3, %x
>> +  %tmp5 = mul i4 %tmp4, %x
>> +  %tmp6 = mul i4 %tmp5, %x
>> +  %tmp7 = mul i4 %tmp6, %x
>> +  %tmp8 = mul i4 %tmp7, %x
>> +  %tmp9 = mul i4 %tmp8, %x
>> +  %tmp10 = mul i4 %tmp9, %x
>> +  ret i4 %tmp10
>> +}
>> +
>> +define i4 @foo4x12(i4 %x) {
>> +; Can be done with two multiplies.
>> +; CHECK: @foo4x12
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i4 %x, %x
>> +  %tmp2 = mul i4 %tmp1, %x
>> +  %tmp3 = mul i4 %tmp2, %x
>> +  %tmp4 = mul i4 %tmp3, %x
>> +  %tmp5 = mul i4 %tmp4, %x
>> +  %tmp6 = mul i4 %tmp5, %x
>> +  %tmp7 = mul i4 %tmp6, %x
>> +  %tmp8 = mul i4 %tmp7, %x
>> +  %tmp9 = mul i4 %tmp8, %x
>> +  %tmp10 = mul i4 %tmp9, %x
>> +  %tmp11 = mul i4 %tmp10, %x
>> +  ret i4 %tmp11
>> +}
>> +
>> +define i4 @foo4x13(i4 %x) {
>> +; Can be done with three multiplies.
>> +; CHECK: @foo4x13
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i4 %x, %x
>> +  %tmp2 = mul i4 %tmp1, %x
>> +  %tmp3 = mul i4 %tmp2, %x
>> +  %tmp4 = mul i4 %tmp3, %x
>> +  %tmp5 = mul i4 %tmp4, %x
>> +  %tmp6 = mul i4 %tmp5, %x
>> +  %tmp7 = mul i4 %tmp6, %x
>> +  %tmp8 = mul i4 %tmp7, %x
>> +  %tmp9 = mul i4 %tmp8, %x
>> +  %tmp10 = mul i4 %tmp9, %x
>> +  %tmp11 = mul i4 %tmp10, %x
>> +  %tmp12 = mul i4 %tmp11, %x
>> +  ret i4 %tmp12
>> +}
>> +
>> +define i4 @foo4x14(i4 %x) {
>> +; Can be done with three multiplies.
>> +; CHECK: @foo4x14
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i4 %x, %x
>> +  %tmp2 = mul i4 %tmp1, %x
>> +  %tmp3 = mul i4 %tmp2, %x
>> +  %tmp4 = mul i4 %tmp3, %x
>> +  %tmp5 = mul i4 %tmp4, %x
>> +  %tmp6 = mul i4 %tmp5, %x
>> +  %tmp7 = mul i4 %tmp6, %x
>> +  %tmp8 = mul i4 %tmp7, %x
>> +  %tmp9 = mul i4 %tmp8, %x
>> +  %tmp10 = mul i4 %tmp9, %x
>> +  %tmp11 = mul i4 %tmp10, %x
>> +  %tmp12 = mul i4 %tmp11, %x
>> +  %tmp13 = mul i4 %tmp12, %x
>> +  ret i4 %tmp13
>> +}
>> +
>> +define i4 @foo4x15(i4 %x) {
>> +; Can be done with four multiplies.
>> +; CHECK: @foo4x15
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: mul
>> +; CHECK-NEXT: ret
>> +  %tmp1 = mul i4 %x, %x
>> +  %tmp2 = mul i4 %tmp1, %x
>> +  %tmp3 = mul i4 %tmp2, %x
>> +  %tmp4 = mul i4 %tmp3, %x
>> +  %tmp5 = mul i4 %tmp4, %x
>> +  %tmp6 = mul i4 %tmp5, %x
>> +  %tmp7 = mul i4 %tmp6, %x
>> +  %tmp8 = mul i4 %tmp7, %x
>> +  %tmp9 = mul i4 %tmp8, %x
>> +  %tmp10 = mul i4 %tmp9, %x
>> +  %tmp11 = mul i4 %tmp10, %x
>> +  %tmp12 = mul i4 %tmp11, %x
>> +  %tmp13 = mul i4 %tmp12, %x
>> +  %tmp14 = mul i4 %tmp13, %x
>> +  ret i4 %tmp14
>> +}
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits




More information about the llvm-commits mailing list