#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Module.h"
#include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/Regex.h"
#include <cstring>
using namespace llvm;
F->arg_begin()->getType());
return true;
}
+ Regex vldRegex("^arm\\.neon\\.vld([1234]|[234]lane)\\.v[a-z0-9]*$");
+ if (vldRegex.match(Name)) {
+ auto fArgs = F->getFunctionType()->params();
+ SmallVector<Type *, 4> Tys(fArgs.begin(), fArgs.end());
+ // Can't use Intrinsic::getDeclaration here as the return types might
+ // then only be structurally equal.
+ FunctionType* fType = FunctionType::get(F->getReturnType(), Tys, false);
+ NewFn = Function::Create(fType, F->getLinkage(),
+ "llvm." + Name + ".p0i8", F->getParent());
+ return true;
+ }
+ Regex vstRegex("^arm\\.neon\\.vst([1234]|[234]lane)\\.v[a-z0-9]*$");
+ if (vstRegex.match(Name)) {
+ static const Intrinsic::ID StoreInts[] = {Intrinsic::arm_neon_vst1,
+ Intrinsic::arm_neon_vst2,
+ Intrinsic::arm_neon_vst3,
+ Intrinsic::arm_neon_vst4};
+
+ static const Intrinsic::ID StoreLaneInts[] = {
+ Intrinsic::arm_neon_vst2lane, Intrinsic::arm_neon_vst3lane,
+ Intrinsic::arm_neon_vst4lane
+ };
+
+ auto fArgs = F->getFunctionType()->params();
+ Type *Tys[] = {fArgs[0], fArgs[1]};
+ if (Name.find("lane") == StringRef::npos)
+ NewFn = Intrinsic::getDeclaration(F->getParent(),
+ StoreInts[fArgs.size() - 3], Tys);
+ else
+ NewFn = Intrinsic::getDeclaration(F->getParent(),
+ StoreLaneInts[fArgs.size() - 5], Tys);
+ return true;
+ }
break;
}
+
case 'c': {
if (Name.startswith("ctlz.") && F->arg_size() == 1) {
F->setName(Name + ".old");
Name.startswith("x86.avx2.vbroadcast") ||
Name.startswith("x86.avx2.pbroadcast") ||
Name.startswith("x86.avx.vpermil.") ||
+ Name.startswith("x86.sse41.pmovsx") ||
Name == "x86.avx.vinsertf128.pd.256" ||
Name == "x86.avx.vinsertf128.ps.256" ||
Name == "x86.avx.vinsertf128.si.256" ||
Name == "x86.avx2.pblendd.128" ||
Name == "x86.avx2.pblendd.256" ||
Name == "x86.avx2.vbroadcasti128" ||
+ Name == "x86.xop.vpcmov" ||
(Name.startswith("x86.xop.vpcom") && F->arg_size() == 2)) {
NewFn = nullptr;
return true;
Function *F = CI->getCalledFunction();
LLVMContext &C = CI->getContext();
IRBuilder<> Builder(C);
- Builder.SetInsertPoint(CI->getParent(), CI);
+ Builder.SetInsertPoint(CI->getParent(), CI->getIterator());
assert(F && "Intrinsic call is not direct?");
Name == "llvm.x86.avx.movnt.ps.256" ||
Name == "llvm.x86.avx.movnt.pd.256") {
IRBuilder<> Builder(C);
- Builder.SetInsertPoint(CI->getParent(), CI);
+ Builder.SetInsertPoint(CI->getParent(), CI->getIterator());
Module *M = F->getParent();
SmallVector<Metadata *, 1> Elts;
"cast");
StoreInst *SI = Builder.CreateStore(Arg1, BC);
SI->setMetadata(M->getMDKindID("nontemporal"), Node);
- SI->setAlignment(16);
+ SI->setAlignment(32);
// Remove intrinsic.
CI->eraseFromParent();
Rep =
Builder.CreateCall(VPCOM, {CI->getArgOperand(0), CI->getArgOperand(1),
Builder.getInt8(Imm)});
+ } else if (Name == "llvm.x86.xop.vpcmov") {
+ Value *Arg0 = CI->getArgOperand(0);
+ Value *Arg1 = CI->getArgOperand(1);
+ Value *Sel = CI->getArgOperand(2);
+ unsigned NumElts = CI->getType()->getVectorNumElements();
+ Constant *MinusOne = ConstantVector::getSplat(NumElts, Builder.getInt64(-1));
+ Value *NotSel = Builder.CreateXor(Sel, MinusOne);
+ Value *Sel0 = Builder.CreateAnd(Arg0, Sel);
+ Value *Sel1 = Builder.CreateAnd(Arg1, NotSel);
+ Rep = Builder.CreateOr(Sel0, Sel1);
} else if (Name == "llvm.x86.sse42.crc32.64.8") {
Function *CRC32 = Intrinsic::getDeclaration(F->getParent(),
Intrinsic::x86_sse42_crc32_32_8);
for (unsigned I = 0; I < EltNum; ++I)
Rep = Builder.CreateInsertElement(Rep, Load,
ConstantInt::get(I32Ty, I));
+ } else if (Name.startswith("llvm.x86.sse41.pmovsx")) {
+ VectorType *SrcTy = cast<VectorType>(CI->getArgOperand(0)->getType());
+ VectorType *DstTy = cast<VectorType>(CI->getType());
+ unsigned NumDstElts = DstTy->getNumElements();
+
+ // Extract a subvector of the first NumDstElts lanes and sign extend.
+ SmallVector<int, 8> ShuffleMask;
+ for (int i = 0; i != (int)NumDstElts; ++i)
+ ShuffleMask.push_back(i);
+
+ Value *SV = Builder.CreateShuffleVector(
+ CI->getArgOperand(0), UndefValue::get(SrcTy), ShuffleMask);
+ Rep = Builder.CreateSExt(SV, DstTy);
} else if (Name == "llvm.x86.avx2.vbroadcasti128") {
// Replace vbroadcasts with a vector shuffle.
Type *VT = VectorType::get(Type::getInt64Ty(C), 2);
unsigned Imm = cast<ConstantInt>(CI->getArgOperand(2))->getZExtValue();
VectorType *VecTy = cast<VectorType>(CI->getType());
unsigned NumElts = VecTy->getNumElements();
-
+
// Mask off the high bits of the immediate value; hardware ignores those.
Imm = Imm & 1;
-
+
// Extend the second operand into a vector that is twice as big.
Value *UndefV = UndefValue::get(Op1->getType());
SmallVector<Constant*, 8> Idxs;
unsigned Imm = cast<ConstantInt>(CI->getArgOperand(1))->getZExtValue();
VectorType *VecTy = cast<VectorType>(CI->getType());
unsigned NumElts = VecTy->getNumElements();
-
+
// Mask off the high bits of the immediate value; hardware ignores those.
Imm = Imm & 1;
default:
llvm_unreachable("Unknown function for CallInst upgrade.");
+ case Intrinsic::arm_neon_vld1:
+ case Intrinsic::arm_neon_vld2:
+ case Intrinsic::arm_neon_vld3:
+ case Intrinsic::arm_neon_vld4:
+ case Intrinsic::arm_neon_vld2lane:
+ case Intrinsic::arm_neon_vld3lane:
+ case Intrinsic::arm_neon_vld4lane:
+ case Intrinsic::arm_neon_vst1:
+ case Intrinsic::arm_neon_vst2:
+ case Intrinsic::arm_neon_vst3:
+ case Intrinsic::arm_neon_vst4:
+ case Intrinsic::arm_neon_vst2lane:
+ case Intrinsic::arm_neon_vst3lane:
+ case Intrinsic::arm_neon_vst4lane: {
+ SmallVector<Value *, 4> Args(CI->arg_operands().begin(),
+ CI->arg_operands().end());
+ CI->replaceAllUsesWith(Builder.CreateCall(NewFn, Args));
+ CI->eraseFromParent();
+ return;
+ }
+
case Intrinsic::ctlz:
case Intrinsic::cttz:
assert(CI->getNumArgOperands() == 1 &&