[X86] Replace avx2 broadcast intrinsics with native IR.

[oota-llvm.git] / lib / IR / AutoUpgrade.cpp

diff --git a/lib/IR/AutoUpgrade.cpp b/lib/IR/AutoUpgrade.cpp
index 8de69203811d451e94d8c7cbd4f85ac21f9a4920..aeefa38f74d6cb6ebe4f1092020915339ce33700 100644 (file)
--- a/lib/IR/AutoUpgrade.cpp
+++ b/lib/IR/AutoUpgrade.cpp
@@ -129,6 +129,8 @@ static bool UpgradeIntrinsicFunction1(Function *F, Function *&NewFn) {
         Name.startswith("x86.sse2.pcmpgt.") ||
         Name.startswith("x86.avx2.pcmpeq.") ||
         Name.startswith("x86.avx2.pcmpgt.") ||
+        Name.startswith("x86.avx2.vbroadcast") ||
+        Name.startswith("x86.avx2.pbroadcast") ||
         Name.startswith("x86.avx.vpermil.") ||
         Name == "x86.avx.vinsertf128.pd.256" ||
         Name == "x86.avx.vinsertf128.ps.256" ||
@@ -229,13 +231,13 @@ static bool UpgradeIntrinsicFunction1(Function *F, Function *&NewFn) {
 bool llvm::UpgradeIntrinsicFunction(Function *F, Function *&NewFn) {
   NewFn = nullptr;
   bool Upgraded = UpgradeIntrinsicFunction1(F, NewFn);
+  assert(F != NewFn && "Intrinsic function upgraded to the same function");
 
   // Upgrade intrinsic attributes.  This does not change the function.
   if (NewFn)
     F = NewFn;
-  if (unsigned id = F->getIntrinsicID())
-    F->setAttributes(Intrinsic::getAttributes(F->getContext(),
-                                              (Intrinsic::ID)id));
+  if (Intrinsic::ID id = F->getIntrinsicID())
+    F->setAttributes(Intrinsic::getAttributes(F->getContext(), id));
   return Upgraded;
 }
 
@@ -416,13 +418,14 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
         llvm_unreachable("Unknown condition");
 
       Function *VPCOM = Intrinsic::getDeclaration(F->getParent(), intID);
-      Rep = Builder.CreateCall3(VPCOM, CI->getArgOperand(0),
-                                CI->getArgOperand(1), Builder.getInt8(Imm));
+      Rep =
+          Builder.CreateCall(VPCOM, {CI->getArgOperand(0), CI->getArgOperand(1),
+                                     Builder.getInt8(Imm)});
     } else if (Name == "llvm.x86.sse42.crc32.64.8") {
       Function *CRC32 = Intrinsic::getDeclaration(F->getParent(),
                                                Intrinsic::x86_sse42_crc32_32_8);
       Value *Trunc0 = Builder.CreateTrunc(CI->getArgOperand(0), Type::getInt32Ty(C));
-      Rep = Builder.CreateCall2(CRC32, Trunc0, CI->getArgOperand(1));
+      Rep = Builder.CreateCall(CRC32, {Trunc0, CI->getArgOperand(1)});
       Rep = Builder.CreateZExt(Rep, CI->getType(), "");
     } else if (Name.startswith("llvm.x86.avx.vbroadcast")) {
       // Replace broadcasts with a series of insertelements.
@@ -431,7 +434,7 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
       unsigned EltNum = VecTy->getVectorNumElements();
       Value *Cast = Builder.CreateBitCast(CI->getArgOperand(0),
                                           EltTy->getPointerTo());
-      Value *Load = Builder.CreateLoad(Cast);
+      Value *Load = Builder.CreateLoad(EltTy, Cast);
       Type *I32Ty = Type::getInt32Ty(C);
       Rep = UndefValue::get(VecTy);
       for (unsigned I = 0; I < EltNum; ++I)
@@ -439,13 +442,21 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
                                           ConstantInt::get(I32Ty, I));
     } else if (Name == "llvm.x86.avx2.vbroadcasti128") {
       // Replace vbroadcasts with a vector shuffle.
-      Value *Op = Builder.CreatePointerCast(
-          CI->getArgOperand(0),
-          PointerType::getUnqual(VectorType::get(Type::getInt64Ty(C), 2)));
-      Value *Load = Builder.CreateLoad(Op);
+      Type *VT = VectorType::get(Type::getInt64Ty(C), 2);
+      Value *Op = Builder.CreatePointerCast(CI->getArgOperand(0),
+                                            PointerType::getUnqual(VT));
+      Value *Load = Builder.CreateLoad(VT, Op);
       const int Idxs[4] = { 0, 1, 0, 1 };
       Rep = Builder.CreateShuffleVector(Load, UndefValue::get(Load->getType()),
                                         Idxs);
+    } else if (Name.startswith("llvm.x86.avx2.pbroadcast") ||
+               Name.startswith("llvm.x86.avx2.vbroadcast")) {
+      // Replace vp?broadcasts with a vector shuffle.
+      Value *Op = CI->getArgOperand(0);
+      unsigned NumElts = CI->getType()->getVectorNumElements();
+      Type *MaskTy = VectorType::get(Type::getInt32Ty(C), NumElts);
+      Rep = Builder.CreateShuffleVector(Op, UndefValue::get(Op->getType()),
+                                        Constant::getNullValue(MaskTy));
     } else if (Name == "llvm.x86.sse2.psll.dq") {
       // 128-bit shift left specified in bits.
       unsigned Shift = cast<ConstantInt>(CI->getArgOperand(1))->getZExtValue();
@@ -630,29 +641,27 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
   case Intrinsic::cttz:
     assert(CI->getNumArgOperands() == 1 &&
            "Mismatch between function args and call args");
-    CI->replaceAllUsesWith(Builder.CreateCall2(NewFn, CI->getArgOperand(0),
-                                               Builder.getFalse(), Name));
+    CI->replaceAllUsesWith(Builder.CreateCall(
+        NewFn, {CI->getArgOperand(0), Builder.getFalse()}, Name));
     CI->eraseFromParent();
     return;
 
   case Intrinsic::objectsize:
-    CI->replaceAllUsesWith(Builder.CreateCall2(NewFn,
-                                               CI->getArgOperand(0),
-                                               CI->getArgOperand(1),
-                                               Name));
+    CI->replaceAllUsesWith(Builder.CreateCall(
+        NewFn, {CI->getArgOperand(0), CI->getArgOperand(1)}, Name));
     CI->eraseFromParent();
     return;
 
   case Intrinsic::ctpop: {
-    CI->replaceAllUsesWith(Builder.CreateCall(NewFn, CI->getArgOperand(0)));
+    CI->replaceAllUsesWith(Builder.CreateCall(NewFn, {CI->getArgOperand(0)}));
     CI->eraseFromParent();
     return;
   }
 
   case Intrinsic::x86_xop_vfrcz_ss:
   case Intrinsic::x86_xop_vfrcz_sd:
-    CI->replaceAllUsesWith(Builder.CreateCall(NewFn, CI->getArgOperand(1),
-                                              Name));
+    CI->replaceAllUsesWith(
+        Builder.CreateCall(NewFn, {CI->getArgOperand(1)}, Name));
     CI->eraseFromParent();
     return;
 
@@ -675,9 +684,7 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
     Value *BC0 = Builder.CreateBitCast(Arg0, NewVecTy, "cast");
     Value *BC1 = Builder.CreateBitCast(Arg1, NewVecTy, "cast");
 
-    Type *Ty[] = {NewVecTy, NewVecTy};
-    CallInst *NewCall = Builder.CreateCall2(
-        FunctionType::get(CI->getType(), Ty, false), NewFn, BC0, BC1, Name);
+    CallInst *NewCall = Builder.CreateCall(NewFn, {BC0, BC1}, Name);
     CI->replaceAllUsesWith(NewCall);
     CI->eraseFromParent();
     return;
@@ -714,16 +721,14 @@ void llvm::UpgradeCallsToIntrinsic(Function* F) {
   // Upgrade the function and check if it is a totaly new function.
   Function *NewFn;
   if (UpgradeIntrinsicFunction(F, NewFn)) {
-    if (NewFn != F) {
-      // Replace all uses to the old function with the new one if necessary.
-      for (Value::user_iterator UI = F->user_begin(), UE = F->user_end();
-           UI != UE; ) {
-        if (CallInst *CI = dyn_cast<CallInst>(*UI++))
-          UpgradeIntrinsicCall(CI, NewFn);
-      }
-      // Remove old function, no longer used, from the module.
-      F->eraseFromParent();
+    // Replace all uses to the old function with the new one if necessary.
+    for (Value::user_iterator UI = F->user_begin(), UE = F->user_end();
+         UI != UE;) {
+      if (CallInst *CI = dyn_cast<CallInst>(*UI++))
+        UpgradeIntrinsicCall(CI, NewFn);
     }
+    // Remove old function, no longer used, from the module.
+    F->eraseFromParent();
   }
 }