From: Samuel Antao <sfantao@us.ibm.com>
Date: Thu, 16 Jun 2016 16:06:22 +0000 (+0000)
Subject: Revert r272900 - [OpenMP] Cast captures by copy when passed to fork call so that... 
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=7658b570923dcdfaaec04d837b7e8862eff895f8;p=clang

Revert r272900 - [OpenMP] Cast captures by copy when passed to fork call so that they are compatible to what the runtime library expects.

Was causing trouble in one of the regression tests for a 32-bit address space.


git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@272908 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/CodeGen/CGOpenMPRuntime.cpp b/lib/CodeGen/CGOpenMPRuntime.cpp
index 059492665d..a76256c93b 100644
--- a/lib/CodeGen/CGOpenMPRuntime.cpp
+++ b/lib/CodeGen/CGOpenMPRuntime.cpp
@@ -4815,7 +4815,8 @@ void CGOpenMPRuntime::emitTargetOutlinedFunctionHelper(
   CGOpenMPTargetRegionInfo CGInfo(CS, CodeGen, EntryFnName);
   CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CGInfo);
 
-  OutlinedFn = CGF.GenerateOpenMPCapturedStmtFunction(CS);
+  OutlinedFn =
+      CGF.GenerateOpenMPCapturedStmtFunction(CS, /*CastValToPtr=*/true);
 
   // If this target outline function is not an offload entry, we don't need to
   // register it.
@@ -5484,6 +5485,7 @@ public:
       MappableExprsHandler::MapValuesArrayTy &CurPointers,
       MappableExprsHandler::MapValuesArrayTy &CurSizes,
       MappableExprsHandler::MapFlagsArrayTy &CurMapTypes) {
+    auto &Ctx = CGF.getContext();
 
     // Do the default mapping.
     if (CI.capturesThis()) {
@@ -5495,17 +5497,36 @@ public:
       CurMapTypes.push_back(MappableExprsHandler::OMP_MAP_TO |
                             MappableExprsHandler::OMP_MAP_FROM);
     } else if (CI.capturesVariableByCopy()) {
-      CurBasePointers.push_back(CV);
-      CurPointers.push_back(CV);
       if (!RI.getType()->isAnyPointerType()) {
-        // We have to signal to the runtime captures passed by value that are
-        // not pointers.
+        // If the field is not a pointer, we need to save the actual value
+        // and load it as a void pointer.
         CurMapTypes.push_back(MappableExprsHandler::OMP_MAP_PRIVATE_VAL);
+        auto DstAddr = CGF.CreateMemTemp(Ctx.getUIntPtrType(),
+                                         Twine(CI.getCapturedVar()->getName()) +
+                                             ".casted");
+        LValue DstLV = CGF.MakeAddrLValue(DstAddr, Ctx.getUIntPtrType());
+
+        auto *SrcAddrVal = CGF.EmitScalarConversion(
+            DstAddr.getPointer(), Ctx.getPointerType(Ctx.getUIntPtrType()),
+            Ctx.getPointerType(RI.getType()), SourceLocation());
+        LValue SrcLV = CGF.MakeNaturalAlignAddrLValue(SrcAddrVal, RI.getType());
+
+        // Store the value using the source type pointer.
+        CGF.EmitStoreThroughLValue(RValue::get(CV), SrcLV);
+
+        // Load the value using the destination type pointer.
+        CurBasePointers.push_back(
+            CGF.EmitLoadOfLValue(DstLV, SourceLocation()).getScalarVal());
+        CurPointers.push_back(CurBasePointers.back());
+
+        // Get the size of the type to be used in the map.
         CurSizes.push_back(CGF.getTypeSize(RI.getType()));
       } else {
         // Pointers are implicitly mapped with a zero size and no flags
         // (other than first map that is added for all implicit maps).
         CurMapTypes.push_back(0u);
+        CurBasePointers.push_back(CV);
+        CurPointers.push_back(CV);
         CurSizes.push_back(llvm::Constant::getNullValue(CGF.SizeTy));
       }
     } else {
diff --git a/lib/CodeGen/CGStmtOpenMP.cpp b/lib/CodeGen/CGStmtOpenMP.cpp
index 061a07791f..89d47e567c 100644
--- a/lib/CodeGen/CGStmtOpenMP.cpp
+++ b/lib/CodeGen/CGStmtOpenMP.cpp
@@ -136,33 +136,10 @@ void CodeGenFunction::GenerateOpenMPCapturedVars(
       CapturedVars.push_back(Val);
     } else if (CurCap->capturesThis())
       CapturedVars.push_back(CXXThisValue);
-    else if (CurCap->capturesVariableByCopy()) {
-      llvm::Value *CV =
-          EmitLoadOfLValue(EmitLValue(*I), SourceLocation()).getScalarVal();
-
-      // If the field is not a pointer, we need to save the actual value
-      // and load it as a void pointer.
-      if (!CurField->getType()->isAnyPointerType()) {
-        auto &Ctx = getContext();
-        auto DstAddr = CreateMemTemp(
-            Ctx.getUIntPtrType(),
-            Twine(CurCap->getCapturedVar()->getName()) + ".casted");
-        LValue DstLV = MakeAddrLValue(DstAddr, Ctx.getUIntPtrType());
-
-        auto *SrcAddrVal = EmitScalarConversion(
-            DstAddr.getPointer(), Ctx.getPointerType(Ctx.getUIntPtrType()),
-            Ctx.getPointerType(CurField->getType()), SourceLocation());
-        LValue SrcLV =
-            MakeNaturalAlignAddrLValue(SrcAddrVal, CurField->getType());
-
-        // Store the value using the source type pointer.
-        EmitStoreThroughLValue(RValue::get(CV), SrcLV);
-
-        // Load the value using the destination type pointer.
-        CV = EmitLoadOfLValue(DstLV, SourceLocation()).getScalarVal();
-      }
-      CapturedVars.push_back(CV);
-    } else {
+    else if (CurCap->capturesVariableByCopy())
+      CapturedVars.push_back(
+          EmitLoadOfLValue(EmitLValue(*I), SourceLocation()).getScalarVal());
+    else {
       assert(CurCap->capturesVariable() && "Expected capture by reference.");
       CapturedVars.push_back(EmitLValue(*I).getAddress().getPointer());
     }
@@ -195,7 +172,8 @@ static Address castValueFromUintptr(CodeGenFunction &CGF, QualType DstType,
 }
 
 llvm::Function *
-CodeGenFunction::GenerateOpenMPCapturedStmtFunction(const CapturedStmt &S) {
+CodeGenFunction::GenerateOpenMPCapturedStmtFunction(const CapturedStmt &S,
+                                                    bool CastValToPtr) {
   assert(
       CapturedStmtInfo &&
       "CapturedStmtInfo should be set when generating the captured function");
@@ -219,9 +197,11 @@ CodeGenFunction::GenerateOpenMPCapturedStmtFunction(const CapturedStmt &S) {
     // uintptr. This is necessary given that the runtime library is only able to
     // deal with pointers. We can pass in the same way the VLA type sizes to the
     // outlined function.
-    if ((I->capturesVariableByCopy() && !ArgType->isAnyPointerType()) ||
-        I->capturesVariableArrayType())
-      ArgType = Ctx.getUIntPtrType();
+    if (CastValToPtr) {
+      if ((I->capturesVariableByCopy() && !ArgType->isAnyPointerType()) ||
+          I->capturesVariableArrayType())
+        ArgType = Ctx.getUIntPtrType();
+    }
 
     if (I->capturesVariable() || I->capturesVariableByCopy()) {
       CapVar = I->getCapturedVar();
@@ -275,9 +255,12 @@ CodeGenFunction::GenerateOpenMPCapturedStmtFunction(const CapturedStmt &S) {
                        AlignmentSource::Decl);
     if (FD->hasCapturedVLAType()) {
       LValue CastedArgLVal =
-          MakeAddrLValue(castValueFromUintptr(*this, FD->getType(),
-                                              Args[Cnt]->getName(), ArgLVal),
-                         FD->getType(), AlignmentSource::Decl);
+          CastValToPtr
+              ? MakeAddrLValue(castValueFromUintptr(*this, FD->getType(),
+                                                    Args[Cnt]->getName(),
+                                                    ArgLVal),
+                               FD->getType(), AlignmentSource::Decl)
+              : ArgLVal;
       auto *ExprArg =
           EmitLoadOfLValue(CastedArgLVal, SourceLocation()).getScalarVal();
       auto VAT = FD->getCapturedVLAType();
@@ -297,9 +280,16 @@ CodeGenFunction::GenerateOpenMPCapturedStmtFunction(const CapturedStmt &S) {
              "Not expecting a captured pointer.");
       auto *Var = I->getCapturedVar();
       QualType VarTy = Var->getType();
-      setAddrOfLocalVar(Var, castValueFromUintptr(*this, FD->getType(),
-                                                  Args[Cnt]->getName(), ArgLVal,
-                                                  VarTy->isReferenceType()));
+      if (!CastValToPtr && VarTy->isReferenceType()) {
+        Address Temp = CreateMemTemp(VarTy);
+        Builder.CreateStore(ArgLVal.getPointer(), Temp);
+        ArgLVal = MakeAddrLValue(Temp, VarTy);
+      }
+      setAddrOfLocalVar(Var, CastValToPtr ? castValueFromUintptr(
+                                                *this, FD->getType(),
+                                                Args[Cnt]->getName(), ArgLVal,
+                                                VarTy->isReferenceType())
+                                          : ArgLVal.getAddress());
     } else {
       // If 'this' is captured, load it into CXXThisValue.
       assert(I->capturesThis());
diff --git a/lib/CodeGen/CodeGenFunction.h b/lib/CodeGen/CodeGenFunction.h
index b96e6bad8a..478093aa0a 100644
--- a/lib/CodeGen/CodeGenFunction.h
+++ b/lib/CodeGen/CodeGenFunction.h
@@ -2230,7 +2230,8 @@ public:
   llvm::Function *EmitCapturedStmt(const CapturedStmt &S, CapturedRegionKind K);
   llvm::Function *GenerateCapturedStmtFunction(const CapturedStmt &S);
   Address GenerateCapturedStmtArgument(const CapturedStmt &S);
-  llvm::Function *GenerateOpenMPCapturedStmtFunction(const CapturedStmt &S);
+  llvm::Function *GenerateOpenMPCapturedStmtFunction(const CapturedStmt &S,
+                                                     bool CastValToPtr = false);
   void GenerateOpenMPCapturedVars(const CapturedStmt &S,
                                   SmallVectorImpl<llvm::Value *> &CapturedVars);
   void emitOMPSimpleStore(LValue LVal, RValue RVal, QualType RValTy,
diff --git a/test/OpenMP/for_firstprivate_codegen.cpp b/test/OpenMP/for_firstprivate_codegen.cpp
index 3c6f37229f..e0380e3b02 100644
--- a/test/OpenMP/for_firstprivate_codegen.cpp
+++ b/test/OpenMP/for_firstprivate_codegen.cpp
@@ -262,18 +262,12 @@ int main() {
 
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
-// CHECK: [[TVAR:%.+]] = alloca i32,
-// CHECK: [[TVAR_CAST:%.+]] = alloca i64,
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: [[TVAR_VAL:%.+]] = load i32, i32* [[TVAR]],
-// CHECK: [[TVAR_CONV:%.+]] = bitcast i64* [[TVAR_CAST]] to i32*
-// CHECK: store i32 [[TVAR_VAL]], i32* [[TVAR_CONV]],
-// CHECK: [[PVT_CASTVAL:%[^,]+]] = load i64, i64* [[TVAR_CAST]],
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i64, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void  (i32*, i32*, ...)*), i64 [[PVT_CASTVAL]],
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i64 {{.*}}%{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32 {{.*}}%{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}})
 // Skip temp vars for loop
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
@@ -285,7 +279,6 @@ int main() {
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
-// CHECK: %{{.+}} = bitcast i64* [[T_VAR_PRIV]] to i32*
 
 // CHECK-NOT: load i{{[0-9]+}}*, i{{[0-9]+}}** %
 // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
diff --git a/test/OpenMP/parallel_firstprivate_codegen.cpp b/test/OpenMP/parallel_firstprivate_codegen.cpp
index 7334b9dc64..809edf8fc3 100644
--- a/test/OpenMP/parallel_firstprivate_codegen.cpp
+++ b/test/OpenMP/parallel_firstprivate_codegen.cpp
@@ -130,52 +130,48 @@ int main() {
     // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0
     // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1
     // LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2
-    // LAMBDA: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*, i64, i64, i64, [4 x i{{[0-9]+}}]*)* [[SS_MICROTASK:@.+]] to void
+    // LAMBDA: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*, i32, i32, i32, [4 x i{{[0-9]+}}]*)* [[SS_MICROTASK:@.+]] to void
     // LAMBDA: ret
 
-    // LAMBDA: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i64 {{.+}}, i64 {{.+}}, i64 {{.+}}, [4 x i{{[0-9]+}}]* {{.+}})
+    // LAMBDA: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}}, [4 x i{{[0-9]+}}]* {{.+}})
     // LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %
     // LAMBDA: call{{.*}} void
     // LAMBDA: ret void
 
-    // LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i64 {{.+}}, i64 {{.+}}, i64 {{.+}})
+    // LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}})
     // LAMBDA: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
     // LAMBDA: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
     // LAMBDA: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
-    // LAMBDA: [[A_CONV:%.+]] = bitcast i64* [[A_PRIV]] to i32*
-    // LAMBDA: store i32* [[A_CONV]], i32** [[REFA:%.+]],
-    // LAMBDA: [[B_CONV:%.+]] = bitcast i64* [[B_PRIV]] to i32*
-    // LAMBDA: [[C_CONV:%.+]] = bitcast i64* [[C_PRIV]] to i32*
-    // LAMBDA: store i32* [[C_CONV]], i32** [[REFC:%.+]],
+    // LAMBDA: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
+    // LAMBDA: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
     // LAMBDA-NEXT: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]],
     // LAMBDA-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
     // LAMBDA-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
     // LAMBDA-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
-    // LAMBDA-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_CONV]],
+    // LAMBDA-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]],
     // LAMBDA-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
-    // LAMBDA-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_CONV]],
+    // LAMBDA-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
     // LAMBDA-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]],
     // LAMBDA-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]],
     // LAMBDA-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1
     // LAMBDA-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]],
     // LAMBDA-NEXT: ret void
 
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, i64 {{.*}}%{{.+}})
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, i32 {{.*}}%{{.+}})
     // LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128
     // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR:%.+]]
-    // LAMBDA: [[SIVAR_PRIVATE_CONV:%.+]] = bitcast i64* [[SIVAR_PRIVATE_ADDR]] to i32*
     // LAMBDA: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G_REF]], align 128
     // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128
     // LAMBDA-NOT: call {{.*}}void @__kmpc_barrier(
     g = 1;
     sivar = 2;
     // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
-    // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]],
+    // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
     // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
-    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
@@ -202,23 +198,22 @@ int main() {
   // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 2, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]], {{.+}})
 #pragma omp parallel firstprivate(g, sivar)
   {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, i64 {{.*}}%{{.+}})
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, i32 {{.*}}%{{.+}})
     // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128
     // BLOCKS: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR:%.+]]
-    // BLOCKS: [[SIVAR_PRIVATE_CONV:%.+]] = bitcast i64* [[SIVAR_PRIVATE_ADDR]] to i32*
     // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G_REF]], align 128
     // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128
     // BLOCKS-NOT: call {{.*}}void @__kmpc_barrier(
     g = 1;
     sivar = 2;
     // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
-    // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]],
+    // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
-    // BLOCKS: i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]]
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call {{.*}}void {{%.+}}(i8
     ^{
@@ -245,30 +240,27 @@ int main() {
 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0
 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1
 // BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2
-// BLOCKS: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*, i64, i64, i64, [4 x i{{[0-9]+}}]*)* [[SS_MICROTASK:@.+]] to void
+// BLOCKS: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*, i32, i32, i32, [4 x i{{[0-9]+}}]*)* [[SS_MICROTASK:@.+]] to void
 // BLOCKS: ret
 
-// BLOCKS: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i64 {{.+}}, i64 {{.+}}, i64 {{.+}}, [4 x i{{[0-9]+}}]* {{.+}})
+// BLOCKS: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}}, [4 x i{{[0-9]+}}]* {{.+}})
 // BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %
 // BLOCKS: call{{.*}} void
 // BLOCKS: ret void
 
-// BLOCKS: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i64 {{.+}}, i64 {{.+}}, i64 {{.+}})
+// BLOCKS: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}})
 // BLOCKS: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
 // BLOCKS: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
 // BLOCKS: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
-// BLOCKS: [[A_CONV:%.+]] = bitcast i64* [[A_PRIV]] to i32*
-// BLOCKS: store i32* [[A_CONV]], i32** [[REFA:%.+]],
-// BLOCKS: [[B_CONV:%.+]] = bitcast i64* [[B_PRIV]] to i32*
-// BLOCKS: [[C_CONV:%.+]] = bitcast i64* [[C_PRIV]] to i32*
-// BLOCKS: store i32* [[C_CONV]], i32** [[REFC:%.+]],
+// BLOCKS: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
+// BLOCKS: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
 // BLOCKS-NEXT: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]],
 // BLOCKS-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
 // BLOCKS-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
 // BLOCKS-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
-// BLOCKS-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_CONV]],
+// BLOCKS-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]],
 // BLOCKS-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
-// BLOCKS-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_CONV]],
+// BLOCKS-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
 // BLOCKS-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]],
 // BLOCKS-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]],
 // BLOCKS-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1
@@ -294,24 +286,13 @@ int main() {
 
 // CHECK: define {{.*}}i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
-// CHECK: [[T_VAR:%.+]] = alloca i32,
-// CHECK: [[T_VARCAST:%.+]] = alloca i64,
-// CHECK: [[SIVARCAST:%.+]] = alloca i64,
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: [[T_VARVAL:%.+]] = load i32, i32* [[T_VAR]],
-// CHECK: [[T_VARCONV:%.+]] = bitcast i64* [[T_VARCAST]] to i32*
-// CHECK: store i32 [[T_VARVAL]], i32* [[T_VARCONV]],
-// CHECK: [[T_VARPVT:%.+]] = load i64, i64* [[T_VARCAST]],
-// CHECK: [[SIVARVAL:%.+]] = load i32, i32* @{{.+}},
-// CHECK: [[SIVARCONV:%.+]] = bitcast i64* [[SIVARCAST]] to i32*
-// CHECK: store i32 [[SIVARVAL]], i32* [[SIVARCONV]],
-// CHECK: [[SIVARPVT:%.+]] = load i64, i64* [[SIVARCAST]],
-// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, i64, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i{{[0-9]+}})* [[MAIN_MICROTASK:@.+]] to void {{.*}}i64 [[T_VARPVT]],{{.*}}i64 [[SIVARPVT]]
+// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, i32, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i{{[0-9]+}})* [[MAIN_MICROTASK:@.+]] to void
 // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, i64 {{.*}}%{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}, i64 {{.*}}[[SIVAR:%.+]])
+// CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, i32 {{.*}}%{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}, i32 {{.*}}[[SIVAR:%.+]])
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[SIVAR7_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
@@ -321,11 +302,9 @@ int main() {
 
 // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
 // CHECK-NOT: load i{{[0-9]+}}*, i{{[0-9]+}}** %
-// CHECK: [[T_VAR_CONV:%.+]] = bitcast i64* [[T_VAR_PRIV]] to i32*
 // CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** %
 // CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
 // CHECK-NOT: load i{{[0-9]+}}*, i{{[0-9]+}}** %
-// CHECK: [[SIVAR7_CONV:%.+]] = bitcast i64* [[SIVAR7_PRIV]] to i32*
 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8*
 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]],
@@ -343,7 +322,7 @@ int main() {
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
 
-// CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR7_CONV]],
+// CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR7_PRIV]],
 
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
@@ -364,10 +343,10 @@ int main() {
 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0
 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1
 // CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*, i64, i64, i64, [4 x i32]*)* [[SS_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*, i32, i32, i32, [4 x i32]*)* [[SS_MICROTASK:@.+]] to void
 // CHECK: ret
 
-// CHECK: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i64 {{.+}}, i64 {{.+}}, i64 {{.+}}, [4 x i{{[0-9]+}}]* {{.+}})
+// CHECK: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}}, [4 x i{{[0-9]+}}]* {{.+}})
 // CHECK: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
@@ -375,11 +354,8 @@ int main() {
 // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[A_PRIV]]
 // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[B_PRIV]]
 // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[C_PRIV]]
-// CHECK: [[A_CONV:%.+]] = bitcast i64* [[A_PRIV:%.+]] to i32*
-// CHECK: store i32* [[A_CONV]], i32** [[REFA:%.+]],
-// CHECK: [[B_CONV:%.+]] = bitcast i64* [[B_PRIV:%.+]] to i32*
-// CHECK: [[C_CONV:%.+]] = bitcast i64* [[C_PRIV:%.+]] to i32*
-// CHECK: store i32* [[C_CONV]], i32** [[REFC:%.+]],
+// CHECK: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
+// CHECK: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
 // CHECK: bitcast [4 x i{{[0-9]+}}]* [[E_PRIV]] to i8*
 // CHECK: bitcast [4 x i{{[0-9]+}}]* %{{.+}} to i8*
 // CHECK: call void @llvm.memcpy
@@ -388,9 +364,9 @@ int main() {
 // CHECK-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
 // CHECK-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
 // CHECK-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
-// CHECK-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_CONV]],
+// CHECK-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]],
 // CHECK-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
-// CHECK-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_CONV]],
+// CHECK-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
 // CHECK-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]],
 // CHECK-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]],
 // CHECK-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1
diff --git a/test/OpenMP/sections_firstprivate_codegen.cpp b/test/OpenMP/sections_firstprivate_codegen.cpp
index 51d0c7b618..23358a86e3 100644
--- a/test/OpenMP/sections_firstprivate_codegen.cpp
+++ b/test/OpenMP/sections_firstprivate_codegen.cpp
@@ -257,11 +257,7 @@ int main() {
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: [[T_VARVAL:%.+]] = load i32, i32* %{{.+}},
-// CHECK: [[T_VARCONV:%.+]] = bitcast i64* [[T_VARCAST:%.+]] to i32*
-// CHECK: store i32  [[T_VARVAL]], i32* [[T_VARCONV]],
-// CHECK: [[T_VARPVT:%.+]] = load i64, i64* [[T_VARCAST]],
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i64, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void {{.*}}i64 [[T_VARPVT]],
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
diff --git a/test/OpenMP/single_codegen.cpp b/test/OpenMP/single_codegen.cpp
index a2140c2eda..ab57cf727e 100644
--- a/test/OpenMP/single_codegen.cpp
+++ b/test/OpenMP/single_codegen.cpp
@@ -225,24 +225,21 @@ void array_func(int n, int a[n], St s[2]) {
 #endif
 
 // CHECK-LABEL:@_ZN2SSC2ERi(
-// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* @{{.+}}, i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [[SS_TY]]*, i64, i64, i64)* [[SS_MICROTASK:@.+]] to void
+// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* @{{.+}}, i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [[SS_TY]]*, i32, i32, i32)* [[SS_MICROTASK:@.+]] to void
 // CHECK-NEXT: ret void
 
-// CHECK: define internal void [[SS_MICROTASK]](i32* {{[^,]+}}, i32* {{[^,]+}}, [[SS_TY]]* {{.+}}, i64 {{.+}}, i64 {{.+}}, i64 {{.+}})
+// CHECK: define internal void [[SS_MICROTASK]](i32* {{[^,]+}}, i32* {{[^,]+}}, [[SS_TY]]* {{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}})
 // Private a
-// CHECK: alloca i64,
+// CHECK: alloca i32,
 // Private b
-// CHECK: alloca i64,
+// CHECK: alloca i32,
 // Private c
-// CHECK: alloca i64,
+// CHECK: alloca i32,
 // CHECK: alloca i32*,
 // CHECK: alloca i32*,
 // CHECK: alloca i32*,
 // CHECK: alloca i32*,
 // CHECK: [[DID_IT:%.+]] = alloca i32,
-// CHECK: bitcast i64* %{{.+}} to i32*
-// CHECK: bitcast i64* %{{.+}} to i32*
-// CHECK: bitcast i64* %{{.+}} to i32*
 // CHECK: store i32 0, i32* [[DID_IT]],
 // CHECK: [[RES:%.+]] = call i32 @__kmpc_single([[IDENT_T_TY]]* @{{.+}}, i32 %{{.+}})
 // CHECK-NEXT: icmp ne i32 [[RES]], 0
@@ -302,42 +299,30 @@ void array_func(int n, int a[n], St s[2]) {
 // CHECK-NEXT: getelementptr inbounds [[CAP_TY]], [[CAP_TY]]* [[CAP]], i32 0, i32 1
 // CHECK-NEXT: load i32*, i32** %
 // CHECK-NEXT: load i32, i32* %
-// CHECK-NEXT: bitcast i64* %
-// CHECK-NEXT: store i32 %{{.+}}, i32* %
-// CHECK-NEXT: load i64, i64* %
 // CHECK-NEXT: getelementptr inbounds [[CAP_TY]], [[CAP_TY]]* [[CAP]], i32 0, i32 2
 // CHECK-NEXT: load i32*, i32** %
 // CHECK-NEXT: load i32, i32* %
-// CHECK-NEXT: bitcast i64* %
-// CHECK-NEXT: store i32 %{{.+}}, i32* %
-// CHECK-NEXT: load i64, i64* %
 // CHECK-NEXT: getelementptr inbounds [[CAP_TY]], [[CAP_TY]]* [[CAP]], i32 0, i32 3
 // CHECK-NEXT: load i32*, i32** %
 // CHECK-NEXT: load i32, i32* %
-// CHECK-NEXT: bitcast i64* %
-// CHECK-NEXT: store i32 %{{.+}}, i32* %
-// CHECK-NEXT: load i64, i64* %
-// CHECK-NEXT: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* @{{.+}}, i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [[SS_TY]]*, i64, i64, i64)* [[SS_MICROTASK1:@.+]] to void
+// CHECK-NEXT: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* @{{.+}}, i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [[SS_TY]]*, i32, i32, i32)* [[SS_MICROTASK1:@.+]] to void
 // CHECK-NEXT: ret void
 
 // CHECK: define internal void [[COPY_FUNC]](i8*, i8*)
 // CHECK: ret void
 
-// CHECK: define internal void [[SS_MICROTASK1]](i32* {{[^,]+}}, i32* {{[^,]+}}, [[SS_TY]]* {{.+}}, i64 {{.+}}, i64 {{.+}}, i64 {{.+}})
+// CHECK: define internal void [[SS_MICROTASK1]](i32* {{[^,]+}}, i32* {{[^,]+}}, [[SS_TY]]* {{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}})
 // Private a
-// CHECK: alloca i64,
+// CHECK: alloca i32,
 // Private b
-// CHECK: alloca i64,
+// CHECK: alloca i32,
 // Private c
-// CHECK: alloca i64,
+// CHECK: alloca i32,
 // CHECK: alloca i32*,
 // CHECK: alloca i32*,
 // CHECK: alloca i32*,
 // CHECK: alloca i32*,
 // CHECK: [[DID_IT:%.+]] = alloca i32,
-// CHECK: bitcast i64* %{{.+}} to i32*
-// CHECK: bitcast i64* %{{.+}} to i32*
-// CHECK: bitcast i64* %{{.+}} to i32*
 // CHECK: [[RES:%.+]] = call i32 @__kmpc_single([[IDENT_T_TY]]* @{{.+}}, i32 %{{.+}})
 // CHECK-NEXT: icmp ne i32 [[RES]], 0
 // CHECK-NEXT: br i1
@@ -383,16 +368,13 @@ void array_func(int n, int a[n], St s[2]) {
 // CHECK: getelementptr inbounds [[SST_TY]], [[SST_TY]]* %{{.+}}, i32 0, i32 0
 // CHECK-NEXT: store double 0.000000e+00, double* %
 // CHECK-NEXT: getelementptr inbounds [[SST_TY]], [[SST_TY]]* %{{.+}}, i32 0, i32 0
-// CHECK-NEXT: store double* %{{.+}}, double** %
+// CHECK-NEXT: store double* %
 // CHECK-NEXT: load double*, double** %
 // CHECK-NEXT: load double, double* %
-// CHECK-NEXT: bitcast i64* %{{.+}} to double*
-// CHECK-NEXT: store double %{{.+}}, double* %
-// CHECK-NEXT: load i64, i64* %
-// CHECK-NEXT: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* @{{.+}}, i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [[SST_TY]]*, i64)* [[SST_MICROTASK:@.+]] to void
+// CHECK-NEXT: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* @{{.+}}, i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [[SST_TY]]*, double)* [[SST_MICROTASK:@.+]] to void
 // CHECK-NEXT: ret void
 
-// CHECK: define internal void [[SST_MICROTASK]](i32* {{[^,]+}}, i32* {{[^,]+}}, [[SST_TY]]* {{.+}}, i64 {{.+}})
+// CHECK: define internal void [[SST_MICROTASK]](i32* {{[^,]+}}, i32* {{[^,]+}}, [[SST_TY]]* {{.+}}, double {{.+}})
 // CHECK: [[RES:%.+]] = call i32 @__kmpc_single([[IDENT_T_TY]]* @{{.+}}, i32 %{{.+}})
 // CHECK-NEXT: icmp ne i32 [[RES]], 0
 // CHECK-NEXT: br i1
diff --git a/test/OpenMP/single_firstprivate_codegen.cpp b/test/OpenMP/single_firstprivate_codegen.cpp
index 537ae76f12..dc1e1f9e7a 100644
--- a/test/OpenMP/single_firstprivate_codegen.cpp
+++ b/test/OpenMP/single_firstprivate_codegen.cpp
@@ -223,16 +223,12 @@ int main() {
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: [[T_VARVAL:%.+]] = load i32, i32* [[T_VAR:%.+]],
-// CHECK: [[T_VARCONV:%.+]] = bitcast i64* [[T_VARCAST:%.+]] to i32*
-// CHECK: store i32 [[T_VARVAL]], i32* [[T_VARCONV]],
-// CHECK: [[T_VARPVT:%.+]] = load i64, i64* [[T_VARCAST]],
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i64, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void {{.*}}i64 [[T_VARPVT:%.+]],
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i64 {{.*}}%{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}})
-// CHECK: [[T_VAR_ARG:%.+]] = alloca i{{[0-9]+}},
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32 {{.*}}%{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}})
+// CHECK: alloca i{{[0-9]+}},
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
@@ -240,7 +236,6 @@ int main() {
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
 // CHECK-NOT: load i{{[0-9]+}}*, i{{[0-9]+}}** %
-// CHECK: [[T_VAR_CONV:%.+]] = bitcast i64* [[T_VAR_ARG]] to i32*
 // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
 // CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** %
 // CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
diff --git a/test/OpenMP/teams_firstprivate_codegen.cpp b/test/OpenMP/teams_firstprivate_codegen.cpp
index 1d87bbbae6..505d76313a 100644
--- a/test/OpenMP/teams_firstprivate_codegen.cpp
+++ b/test/OpenMP/teams_firstprivate_codegen.cpp
@@ -76,23 +76,20 @@ int main() {
   #pragma omp target
   #pragma omp teams firstprivate(g, sivar)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[G_IN:%.+]], i{{64|32}} {{.*}}[[SIVAR_IN:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[G_IN:%.+]], i32 {{.*}}[[SIVAR_IN:%.+]])
     // LAMBDA: store i{{[0-9]+}}* [[G_IN]], i{{[0-9]+}}** [[G_ADDR:%.+]],
     // LAMBDA: store i{{[0-9]+}} [[SIVAR_IN]], i{{[0-9]+}}* [[SIVAR_ADDR:%.+]],
     // LAMBDA: [[G_ADDR_VAL:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_ADDR]],
-    // LAMBDA-64: [[SIVAR_CONV:%.+]] = bitcast i64*  [[SIVAR_ADDR]] to i32*
     // LAMBDA: [[G_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[G_ADDR_VAL]],
     // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_LOCAL:%.+]],
     g = 1;
     sivar = 2;
     // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_LOCAL]],
-    // LAMBDA-64: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_CONV]],
-    // LAMBDA-32: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_LOCAL]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
     // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
-    // LAMBDA-64: store i{{[0-9]+}}* [[SIVAR_CONV]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
-    // LAMBDA-32: store i{{[0-9]+}}* [[SIVAR_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
@@ -130,10 +127,10 @@ int main() {
 }
 
 // CHECK: define internal {{.*}}void [[OMP_OFFLOADING:@.+]](
-// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, i{{32|64}}, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i{{[0-9]+}})* [[OMP_OUTLINED:@.+]] to void
+// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, i32, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i{{[0-9]+}})* [[OMP_OUTLINED:@.+]] to void
 // CHECK: ret
 //
-// CHECK: define internal {{.*}}void [[OMP_OUTLINED]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, i{{32|64}} {{.*}}%{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}, i{{32|64}} {{.*}}[[SIVAR:%.+]])
+// CHECK: define internal {{.*}}void [[OMP_OUTLINED]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, i32 {{.*}}%{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}, i32 {{.*}}[[SIVAR:%.+]])
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[SIVAR7_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
@@ -142,10 +139,8 @@ int main() {
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
 // CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
-// CHECK-64: [[T_VAR_CONV:%.+]] = bitcast i64* [[T_VAR_PRIV]] to i32*
 // CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** %
 // CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
-// CHECK-64: [[SIVAR7_CONV:%.+]] = bitcast i64* [[SIVAR7_PRIV]] to i32*
 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8*
 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]],
@@ -163,8 +158,7 @@ int main() {
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
 
-// CHECK-64: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR7_CONV]],
-// CHECK-32: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR7_PRIV]],
+// CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR7_PRIV]],
 
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
@@ -174,7 +168,7 @@ int main() {
 // CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i{{[0-9]+}})* [[OMP_OUTLINED_1:@.+]] to void
 // CHECK: ret
 
-// CHECK: define internal {{.*}}void [[OMP_OUTLINED_1]](i{{[0-9]+}}* noalias {{%.+}}, i{{[0-9]+}}* noalias {{%.+}}, i{{32|64}} {{.*}}[[T_VAR:%.+]])
+// CHECK: define internal {{.*}}void [[OMP_OUTLINED_1]](i{{[0-9]+}}* noalias {{%.+}}, i{{[0-9]+}}* noalias {{%.+}}, i32 {{.*}}[[T_VAR:%.+]])
 // CHECK: [[T_VAR_LOC:%.+]] = alloca i{{[0-9]+}},
 // CHECK: store i{{[0-9]+}} [[T_VAR]], i{{[0-9]+}}* [[T_VAR_LOC]],
 // CHECK: ret