[llvm-commits] [llvm] r151365 - in /llvm/trunk: lib/Transforms/Scalar/GVN.cpp test/Transforms/GVN/commute.ll

[Dan Gohman]

On Feb 24, 2012, at 7:16 AM, Duncan Sands wrote:
> 
> +  if (I->isCommutative()) {
> +    // Ensure that commutative instructions that only differ by a permutation
> +    // of their operands get the same value number by sorting the operand value
> +    // numbers.  Since all commutative instructions have two operands it is more
> +    // efficient to sort by hand rather than using, say, std::sort.
> +    assert(I->getNumOperands() == 2 && "Unsupported commutative instruction!");
> +    if (e.varargs[0] > e.varargs[1])
> +      std::swap(e.varargs[0], e.varargs[1]);
> +  }

The reassociate pass already canonicalizes the order of operands in
add instructions. Is it really desirable for GVN to redo it?

> 
>   if (CmpInst *C = dyn_cast<CmpInst>(I)) {
> -    e.opcode = (C->getOpcode() << 8) | C->getPredicate();
> +    // Sort the operand value numbers so x<y and y>x get the same value number.
> +    CmpInst::Predicate Predicate = C->getPredicate();
> +    if (e.varargs[0] > e.varargs[1]) {
> +      std::swap(e.varargs[0], e.varargs[1]);
> +      Predicate = CmpInst::getSwappedPredicate(Predicate);
> +    }
> +    e.opcode = (C->getOpcode() << 8) | Predicate;

Reassociate doesn't canonicalize comparisons like it does adds,
but it easily could.

Dan