Skip to content

[LLVM] Make more use of IRBuilder::CreateIntrinsic. NFC. #112706

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Oct 17, 2024
Merged

[LLVM] Make more use of IRBuilder::CreateIntrinsic. NFC. #112706

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 17 additions & 20 deletions llvm/lib/CodeGen/ExpandVectorPredication.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -235,13 +235,12 @@ Value *CachingVPExpander::convertEVLToMask(IRBuilder<> &Builder,
// TODO add caching
// Scalable vector %evl conversion.
if (ElemCount.isScalable()) {
auto *M = Builder.GetInsertBlock()->getModule();
Type *BoolVecTy = VectorType::get(Builder.getInt1Ty(), ElemCount);
Function *ActiveMaskFunc = Intrinsic::getOrInsertDeclaration(
M, Intrinsic::get_active_lane_mask, {BoolVecTy, EVLParam->getType()});
// `get_active_lane_mask` performs an implicit less-than comparison.
Value *ConstZero = Builder.getInt32(0);
return Builder.CreateCall(ActiveMaskFunc, {ConstZero, EVLParam});
return Builder.CreateIntrinsic(Intrinsic::get_active_lane_mask,
{BoolVecTy, EVLParam->getType()},
{ConstZero, EVLParam});
}

// Fixed vector %evl conversion.
Expand Down Expand Up @@ -299,18 +298,18 @@ Value *CachingVPExpander::expandPredicationToIntCall(
case Intrinsic::umin: {
Value *Op0 = VPI.getOperand(0);
Value *Op1 = VPI.getOperand(1);
Function *Fn = Intrinsic::getOrInsertDeclaration(
VPI.getModule(), UnpredicatedIntrinsicID, {VPI.getType()});
Value *NewOp = Builder.CreateCall(Fn, {Op0, Op1}, VPI.getName());
Value *NewOp = Builder.CreateIntrinsic(
UnpredicatedIntrinsicID, {VPI.getType()}, {Op0, Op1},
/*FMFSource=*/nullptr, VPI.getName());
replaceOperation(*NewOp, VPI);
return NewOp;
}
case Intrinsic::bswap:
case Intrinsic::bitreverse: {
Value *Op = VPI.getOperand(0);
Function *Fn = Intrinsic::getOrInsertDeclaration(
VPI.getModule(), UnpredicatedIntrinsicID, {VPI.getType()});
Value *NewOp = Builder.CreateCall(Fn, {Op}, VPI.getName());
Value *NewOp =
Builder.CreateIntrinsic(UnpredicatedIntrinsicID, {VPI.getType()}, {Op},
/*FMFSource=*/nullptr, VPI.getName());
replaceOperation(*NewOp, VPI);
return NewOp;
}
Expand All @@ -327,19 +326,19 @@ Value *CachingVPExpander::expandPredicationToFPCall(
case Intrinsic::fabs:
case Intrinsic::sqrt: {
Value *Op0 = VPI.getOperand(0);
Function *Fn = Intrinsic::getOrInsertDeclaration(
VPI.getModule(), UnpredicatedIntrinsicID, {VPI.getType()});
Value *NewOp = Builder.CreateCall(Fn, {Op0}, VPI.getName());
Value *NewOp =
Builder.CreateIntrinsic(UnpredicatedIntrinsicID, {VPI.getType()}, {Op0},
/*FMFSource=*/nullptr, VPI.getName());
replaceOperation(*NewOp, VPI);
return NewOp;
}
case Intrinsic::maxnum:
case Intrinsic::minnum: {
Value *Op0 = VPI.getOperand(0);
Value *Op1 = VPI.getOperand(1);
Function *Fn = Intrinsic::getOrInsertDeclaration(
VPI.getModule(), UnpredicatedIntrinsicID, {VPI.getType()});
Value *NewOp = Builder.CreateCall(Fn, {Op0, Op1}, VPI.getName());
Value *NewOp = Builder.CreateIntrinsic(
UnpredicatedIntrinsicID, {VPI.getType()}, {Op0, Op1},
/*FMFSource=*/nullptr, VPI.getName());
replaceOperation(*NewOp, VPI);
return NewOp;
}
Expand Down Expand Up @@ -592,12 +591,10 @@ bool CachingVPExpander::discardEVLParameter(VPIntrinsic &VPI) {
Type *Int32Ty = Type::getInt32Ty(VPI.getContext());
if (StaticElemCount.isScalable()) {
// TODO add caching
auto *M = VPI.getModule();
Function *VScaleFunc =
Intrinsic::getOrInsertDeclaration(M, Intrinsic::vscale, Int32Ty);
IRBuilder<> Builder(VPI.getParent(), VPI.getIterator());
Value *FactorConst = Builder.getInt32(StaticElemCount.getKnownMinValue());
Value *VScale = Builder.CreateCall(VScaleFunc, {}, "vscale");
Value *VScale = Builder.CreateIntrinsic(Intrinsic::vscale, Int32Ty, {},
/*FMFSource=*/nullptr, "vscale");
MaxEVL = Builder.CreateMul(VScale, FactorConst, "scalable_size",
/*NUW*/ true, /*NSW*/ false);
} else {
Expand Down
13 changes: 5 additions & 8 deletions llvm/lib/CodeGen/HardwareLoops.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -512,8 +512,7 @@ Value* HardwareLoop::InsertIterationSetup(Value *LoopCountInit) {
: Intrinsic::test_set_loop_iterations)
: (UsePhi ? Intrinsic::start_loop_iterations
: Intrinsic::set_loop_iterations);
Function *LoopIter = Intrinsic::getOrInsertDeclaration(M, ID, Ty);
Value *LoopSetup = Builder.CreateCall(LoopIter, LoopCountInit);
Value *LoopSetup = Builder.CreateIntrinsic(ID, Ty, LoopCountInit);

// Use the return value of the intrinsic to control the entry of the loop.
if (UseLoopGuard) {
Expand Down Expand Up @@ -541,10 +540,9 @@ void HardwareLoop::InsertLoopDec() {
Attribute::StrictFP))
CondBuilder.setIsFPConstrained(true);

Function *DecFunc = Intrinsic::getOrInsertDeclaration(
M, Intrinsic::loop_decrement, LoopDecrement->getType());
Value *Ops[] = { LoopDecrement };
Value *NewCond = CondBuilder.CreateCall(DecFunc, Ops);
Value *NewCond = CondBuilder.CreateIntrinsic(Intrinsic::loop_decrement,
LoopDecrement->getType(), Ops);
Value *OldCond = ExitBranch->getCondition();
ExitBranch->setCondition(NewCond);

Expand All @@ -565,10 +563,9 @@ Instruction* HardwareLoop::InsertLoopRegDec(Value *EltsRem) {
Attribute::StrictFP))
CondBuilder.setIsFPConstrained(true);

Function *DecFunc = Intrinsic::getOrInsertDeclaration(
M, Intrinsic::loop_decrement_reg, {EltsRem->getType()});
Value *Ops[] = { EltsRem, LoopDecrement };
Value *Call = CondBuilder.CreateCall(DecFunc, Ops);
Value *Call = CondBuilder.CreateIntrinsic(Intrinsic::loop_decrement_reg,
{EltsRem->getType()}, Ops);

LLVM_DEBUG(dbgs() << "HWLoops: Inserted loop dec: " << *Call << "\n");
return cast<Instruction>(Call);
Expand Down
68 changes: 25 additions & 43 deletions llvm/lib/IR/AutoUpgrade.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1757,8 +1757,7 @@ static Value *upgradeX86BinaryIntrinsics(IRBuilder<> &Builder, CallBase &CI,
Type *Ty = CI.getType();
Value *Op0 = CI.getOperand(0);
Value *Op1 = CI.getOperand(1);
Function *Intrin = Intrinsic::getOrInsertDeclaration(CI.getModule(), IID, Ty);
Value *Res = Builder.CreateCall(Intrin, {Op0, Op1});
Value *Res = Builder.CreateIntrinsic(IID, Ty, {Op0, Op1});

if (CI.arg_size() == 4) { // For masked intrinsics.
Value *VecSrc = CI.getOperand(2);
Expand All @@ -1784,8 +1783,7 @@ static Value *upgradeX86Rotate(IRBuilder<> &Builder, CallBase &CI,
}

Intrinsic::ID IID = IsRotateRight ? Intrinsic::fshr : Intrinsic::fshl;
Function *Intrin = Intrinsic::getOrInsertDeclaration(CI.getModule(), IID, Ty);
Value *Res = Builder.CreateCall(Intrin, {Src, Src, Amt});
Value *Res = Builder.CreateIntrinsic(IID, Ty, {Src, Src, Amt});

if (CI.arg_size() == 4) { // For masked intrinsics.
Value *VecSrc = CI.getOperand(2);
Expand Down Expand Up @@ -1854,8 +1852,7 @@ static Value *upgradeX86ConcatShift(IRBuilder<> &Builder, CallBase &CI,
}

Intrinsic::ID IID = IsShiftRight ? Intrinsic::fshr : Intrinsic::fshl;
Function *Intrin = Intrinsic::getOrInsertDeclaration(CI.getModule(), IID, Ty);
Value *Res = Builder.CreateCall(Intrin, {Op0, Op1, Amt});
Value *Res = Builder.CreateIntrinsic(IID, Ty, {Op0, Op1, Amt});

unsigned NumArgs = CI.arg_size();
if (NumArgs >= 4) { // For masked intrinsics.
Expand Down Expand Up @@ -1915,9 +1912,8 @@ static Value *upgradeMaskedLoad(IRBuilder<> &Builder, Value *Ptr,
static Value *upgradeAbs(IRBuilder<> &Builder, CallBase &CI) {
Type *Ty = CI.getType();
Value *Op0 = CI.getArgOperand(0);
Function *F =
Intrinsic::getOrInsertDeclaration(CI.getModule(), Intrinsic::abs, Ty);
Value *Res = Builder.CreateCall(F, {Op0, Builder.getInt1(false)});
Value *Res = Builder.CreateIntrinsic(Intrinsic::abs, Ty,
{Op0, Builder.getInt1(false)});
if (CI.arg_size() == 3)
Res = emitX86Select(Builder, CI.getArgOperand(2), Res, CI.getArgOperand(1));
return Res;
Expand Down Expand Up @@ -2009,9 +2005,8 @@ static Value *upgradeMaskedCompare(IRBuilder<> &Builder, CallBase &CI,
// Replace a masked intrinsic with an older unmasked intrinsic.
static Value *upgradeX86MaskedShift(IRBuilder<> &Builder, CallBase &CI,
Intrinsic::ID IID) {
Function *Intrin = Intrinsic::getOrInsertDeclaration(CI.getModule(), IID);
Value *Rep = Builder.CreateCall(Intrin,
{ CI.getArgOperand(0), CI.getArgOperand(1) });
Value *Rep = Builder.CreateIntrinsic(
IID, {}, {CI.getArgOperand(0), CI.getArgOperand(1)});
return emitX86Select(Builder, CI.getArgOperand(3), Rep, CI.getArgOperand(2));
}

Expand Down Expand Up @@ -2480,9 +2475,7 @@ static Value *upgradeX86IntrinsicCall(StringRef Name, CallBase *CI, Function *F,
} else if (Name == "sse.sqrt.ss" || Name == "sse2.sqrt.sd") {
Value *Vec = CI->getArgOperand(0);
Value *Elt0 = Builder.CreateExtractElement(Vec, (uint64_t)0);
Function *Intr = Intrinsic::getOrInsertDeclaration(
F->getParent(), Intrinsic::sqrt, Elt0->getType());
Elt0 = Builder.CreateCall(Intr, Elt0);
Elt0 = Builder.CreateIntrinsic(Intrinsic::sqrt, Elt0->getType(), Elt0);
Rep = Builder.CreateInsertElement(Vec, Elt0, (uint64_t)0);
} else if (Name.starts_with("avx.sqrt.p") ||
Name.starts_with("sse2.sqrt.p") ||
Expand Down Expand Up @@ -2770,9 +2763,8 @@ static Value *upgradeX86IntrinsicCall(StringRef Name, CallBase *CI, Function *F,
cast<ConstantInt>(CI->getArgOperand(3))->getZExtValue() != 4)) {
Intrinsic::ID IID = IsUnsigned ? Intrinsic::x86_avx512_uitofp_round
: Intrinsic::x86_avx512_sitofp_round;
Function *F = Intrinsic::getOrInsertDeclaration(CI->getModule(), IID,
{DstTy, SrcTy});
Rep = Builder.CreateCall(F, {Rep, CI->getArgOperand(3)});
Rep = Builder.CreateIntrinsic(IID, {DstTy, SrcTy},
{Rep, CI->getArgOperand(3)});
} else {
Rep = IsUnsigned ? Builder.CreateUIToFP(Rep, DstTy, "cvt")
: Builder.CreateSIToFP(Rep, DstTy, "cvt");
Expand Down Expand Up @@ -2813,9 +2805,8 @@ static Value *upgradeX86IntrinsicCall(StringRef Name, CallBase *CI, Function *F,
Value *MaskVec = getX86MaskVec(Builder, CI->getArgOperand(2),
ResultTy->getNumElements());

Function *ELd = Intrinsic::getOrInsertDeclaration(
F->getParent(), Intrinsic::masked_expandload, ResultTy);
Rep = Builder.CreateCall(ELd, {Ptr, MaskVec, CI->getOperand(1)});
Rep = Builder.CreateIntrinsic(Intrinsic::masked_expandload, ResultTy,
{Ptr, MaskVec, CI->getOperand(1)});
} else if (Name.starts_with("avx512.mask.compress.store.")) {
auto *ResultTy = cast<VectorType>(CI->getArgOperand(1)->getType());
Type *PtrTy = ResultTy->getElementType();
Expand All @@ -2828,9 +2819,8 @@ static Value *upgradeX86IntrinsicCall(StringRef Name, CallBase *CI, Function *F,
getX86MaskVec(Builder, CI->getArgOperand(2),
cast<FixedVectorType>(ResultTy)->getNumElements());

Function *CSt = Intrinsic::getOrInsertDeclaration(
F->getParent(), Intrinsic::masked_compressstore, ResultTy);
Rep = Builder.CreateCall(CSt, {CI->getArgOperand(1), Ptr, MaskVec});
Rep = Builder.CreateIntrinsic(Intrinsic::masked_compressstore, ResultTy,
{CI->getArgOperand(1), Ptr, MaskVec});
} else if (Name.starts_with("avx512.mask.compress.") ||
Name.starts_with("avx512.mask.expand.")) {
auto *ResultTy = cast<FixedVectorType>(CI->getType());
Expand All @@ -2841,10 +2831,8 @@ static Value *upgradeX86IntrinsicCall(StringRef Name, CallBase *CI, Function *F,
bool IsCompress = Name[12] == 'c';
Intrinsic::ID IID = IsCompress ? Intrinsic::x86_avx512_mask_compress
: Intrinsic::x86_avx512_mask_expand;
Function *Intr =
Intrinsic::getOrInsertDeclaration(F->getParent(), IID, ResultTy);
Rep = Builder.CreateCall(Intr,
{CI->getOperand(0), CI->getOperand(1), MaskVec});
Rep = Builder.CreateIntrinsic(
IID, ResultTy, {CI->getOperand(0), CI->getOperand(1), MaskVec});
} else if (Name.starts_with("xop.vpcom")) {
bool IsSigned;
if (Name.ends_with("ub") || Name.ends_with("uw") || Name.ends_with("ud") ||
Expand Down Expand Up @@ -2905,11 +2893,10 @@ static Value *upgradeX86IntrinsicCall(StringRef Name, CallBase *CI, Function *F,
bool ZeroMask = Name[11] == 'z';
Rep = upgradeX86ConcatShift(Builder, *CI, true, ZeroMask);
} else if (Name == "sse42.crc32.64.8") {
Function *CRC32 = Intrinsic::getOrInsertDeclaration(
F->getParent(), Intrinsic::x86_sse42_crc32_32_8);
Value *Trunc0 =
Builder.CreateTrunc(CI->getArgOperand(0), Type::getInt32Ty(C));
Rep = Builder.CreateCall(CRC32, {Trunc0, CI->getArgOperand(1)});
Rep = Builder.CreateIntrinsic(Intrinsic::x86_sse42_crc32_32_8, {},
{Trunc0, CI->getArgOperand(1)});
Rep = Builder.CreateZExt(Rep, CI->getType(), "");
} else if (Name.starts_with("avx.vbroadcast.s") ||
Name.starts_with("avx512.vbroadcast.s")) {
Expand Down Expand Up @@ -3769,12 +3756,9 @@ static Value *upgradeX86IntrinsicCall(StringRef Name, CallBase *CI, Function *F,
IID = Intrinsic::x86_avx512_vfmadd_f64;
else
IID = Intrinsic::x86_avx512_vfmadd_f32;
Function *FMA = Intrinsic::getOrInsertDeclaration(CI->getModule(), IID);
Rep = Builder.CreateCall(FMA, Ops);
Rep = Builder.CreateIntrinsic(IID, {}, Ops);
} else {
Function *FMA = Intrinsic::getOrInsertDeclaration(
CI->getModule(), Intrinsic::fma, A->getType());
Rep = Builder.CreateCall(FMA, {A, B, C});
Rep = Builder.CreateIntrinsic(Intrinsic::fma, A->getType(), {A, B, C});
}

Value *PassThru = IsMaskZ ? Constant::getNullValue(Rep->getType())
Expand Down Expand Up @@ -3827,9 +3811,7 @@ static Value *upgradeX86IntrinsicCall(StringRef Name, CallBase *CI, Function *F,

Rep = Builder.CreateIntrinsic(IID, {}, {A, B, C, CI->getArgOperand(4)});
} else {
Function *FMA = Intrinsic::getOrInsertDeclaration(
CI->getModule(), Intrinsic::fma, A->getType());
Rep = Builder.CreateCall(FMA, {A, B, C});
Rep = Builder.CreateIntrinsic(Intrinsic::fma, A->getType(), {A, B, C});
}

Value *PassThru = IsMaskZ ? llvm::Constant::getNullValue(CI->getType())
Expand Down Expand Up @@ -4088,8 +4070,8 @@ static Value *upgradeAArch64IntrinsicCall(StringRef Name, CallBase *CI,
Args[1] = Builder.CreateIntrinsic(Intrinsic::aarch64_sve_convert_from_svbool,
GoodPredTy, Args[1]);

Function *NewF = Intrinsic::getOrInsertDeclaration(CI->getModule(), NewID);
return Builder.CreateCall(NewF, Args, CI->getName());
return Builder.CreateIntrinsic(NewID, {}, Args, /*FMFSource=*/nullptr,
CI->getName());
}

static Value *upgradeARMIntrinsicCall(StringRef Name, CallBase *CI, Function *F,
Expand Down Expand Up @@ -4171,8 +4153,8 @@ static Value *upgradeARMIntrinsicCall(StringRef Name, CallBase *CI, Function *F,
Ops.push_back(Op);
}

Function *Fn = Intrinsic::getOrInsertDeclaration(F->getParent(), ID, Tys);
return Builder.CreateCall(Fn, Ops, CI->getName());
return Builder.CreateIntrinsic(ID, Tys, Ops, /*FMFSource=*/nullptr,
CI->getName());
}
llvm_unreachable("Unknown function for ARM CallBase upgrade.");
}
Expand Down
Loading
Loading