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

Summary:
This patch fixes an issue detected when firstprivate variables are passed to an OpenMP outlined function vararg list. Currently they are not compatible with what the runtime library expects causing malfunction in some targets.

This patch fixes the issue by moving the casting logic already in place for offloading to the common code that creates the outline function and arguments and updates the regression tests accordingly.

Reviewers: hfinkel, arpith-jacob, carlo.bertolli, kkwli0, ABataev

Subscribers: cfe-commits, caomhin

Differential Revision: http://reviews.llvm.org/D21150

llvm-svn: 272900

1 files changed, 53 insertions, 29 deletions

diff --git a/clang/test/OpenMP/parallel_firstprivate_codegen.cpp b/clang/test/OpenMP/parallel_firstprivate_codegen.cpp
index 809edf8fc39..7334b9dc64f 100644
--- a/clang/test/OpenMP/parallel_firstprivate_codegen.cpp
+++ b/clang/test/OpenMP/parallel_firstprivate_codegen.cpp
@@ -130,48 +130,52 @@ 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]]*, i32, i32, i32, [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]]*, i64, i64, i64, [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]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}}, [4 x i{{[0-9]+}}]* {{.+}})
+    // 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-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]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}})
+    // LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i64 {{.+}}, i64 {{.+}}, i64 {{.+}})
     // LAMBDA: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
     // LAMBDA: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
     // LAMBDA: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
-    // LAMBDA: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
-    // LAMBDA: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
+    // 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-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_PRIV]],
+    // LAMBDA-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_CONV]],
     // LAMBDA-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
-    // LAMBDA-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
+    // LAMBDA-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_CONV]],
     // 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) %{{.+}}, i32 {{.*}}%{{.+}})
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, i64 {{.*}}%{{.+}})
     // 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_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]],
     // 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_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
@@ -198,22 +202,23 @@ 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) %{{.+}}, i32 {{.*}}%{{.+}})
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, i64 {{.*}}%{{.+}})
     // 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_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]],
     // 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_ADDR]]
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR_PRIVATE_CONV]]
     // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call {{.*}}void {{%.+}}(i8
     ^{
@@ -240,27 +245,30 @@ 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]]*, i32, i32, i32, [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]]*, i64, i64, i64, [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]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}}, [4 x i{{[0-9]+}}]* {{.+}})
+// 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-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]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}})
+// BLOCKS: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i64 {{.+}}, i64 {{.+}}, i64 {{.+}})
 // BLOCKS: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
 // BLOCKS: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
 // BLOCKS: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
-// BLOCKS: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
-// BLOCKS: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
+// 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-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_PRIV]],
+// BLOCKS-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_CONV]],
 // BLOCKS-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
-// BLOCKS-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
+// BLOCKS-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_CONV]],
 // 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
@@ -286,13 +294,24 @@ 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: 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: [[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 {{.*}}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) %{{.+}}, i32 {{.*}}%{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}, i32 {{.*}}[[SIVAR:%.+]])
+// 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: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[SIVAR7_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
@@ -302,9 +321,11 @@ 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]],
@@ -322,7 +343,7 @@ 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_PRIV]],
+// CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR7_CONV]],
 
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
@@ -343,10 +364,10 @@ 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]]*, i32, i32, i32, [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]]*, i64, i64, i64, [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]]* %{{.+}}, i32 {{.+}}, i32 {{.+}}, i32 {{.+}}, [4 x i{{[0-9]+}}]* {{.+}})
+// 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: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
@@ -354,8 +375,11 @@ 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: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
-// CHECK: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
+// 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: bitcast [4 x i{{[0-9]+}}]* [[E_PRIV]] to i8*
 // CHECK: bitcast [4 x i{{[0-9]+}}]* %{{.+}} to i8*
 // CHECK: call void @llvm.memcpy
@@ -364,9 +388,9 @@ 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_PRIV]],
+// CHECK-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_CONV]],
 // CHECK-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
-// CHECK-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
+// CHECK-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_CONV]],
 // 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