[AArch64][SVE] Add svboolx2_t and svboolx4_t tuple types

[CarolineConcatto]

Thank you Matt. It LGTM!

[MattDevereau]

a1fae98ba95c18ea6b673fc3c177b917e0f5aa56

clang/include/clang/Basic/AArch64SVEACLETypes.def

 
 SVE_VECTOR_TYPE("__clang_svfloat16x4_t", "svfloat16x4_t", SveFloat16x4, SveFloat16x4Ty, 32, 16, true, true, false)
 SVE_VECTOR_TYPE("__clang_svfloat32x4_t", "svfloat32x4_t", SveFloat32x4, SveFloat32x4Ty, 16, 32, true, true, false)
 SVE_VECTOR_TYPE("__clang_svfloat64x4_t", "svfloat64x4_t", SveFloat64x4, SveFloat64x4Ty, 8, 64, true, true, false)
 
 SVE_VECTOR_TYPE("__clang_svbfloat16x4_t", "svbfloat16x4_t", SveBFloat16x4, SveBFloat16x4Ty, 32, 16, true, false, true)
 
 SVE_PREDICATE_TYPE("__SVBool_t", "__SVBool_t", SveBool, SveBoolTy, 16)
+SVE_PREDICATE_TYPE("__clang_svboolx2_t", "svboolx2_t", SveBoolx2, SveBoolx2Ty, 32)
+SVE_PREDICATE_TYPE("__clang_svboolx4_t", "svboolx4_t", SveBoolx4, SveBoolx4Ty, 64)
 
 SVE_OPAQUE_TYPE("__SVCount_t", "__SVCount_t", SveCount, SveCountTy)
 
 #undef SVE_VECTOR_TYPE
 #undef SVE_PREDICATE_TYPE
 #undef SVE_OPAQUE_TYPE
 #undef SVE_TYPE

clang/lib/AST/ASTContext.cpp

#define SVE_ELTTY(ELTTY, ELTS, NUMVECTORS)                                     \
   case BuiltinType::SveUint64x3:
     return SVE_INT_ELTTY(64, 2, false, 3);
   case BuiltinType::SveInt64x4:
     return SVE_INT_ELTTY(64, 2, true, 4);
   case BuiltinType::SveUint64x4:
     return SVE_INT_ELTTY(64, 2, false, 4);
   case BuiltinType::SveBool:
     return SVE_ELTTY(BoolTy, 16, 1);
+  case BuiltinType::SveBoolx2:
+    return SVE_ELTTY(BoolTy, 16, 2);
+  case BuiltinType::SveBoolx4:
+    return SVE_ELTTY(BoolTy, 16, 4);
   case BuiltinType::SveFloat16:
     return SVE_ELTTY(HalfTy, 8, 1);
   case BuiltinType::SveFloat16x2:
     return SVE_ELTTY(HalfTy, 8, 2);
   case BuiltinType::SveFloat16x3:
     return SVE_ELTTY(HalfTy, 8, 3);
   case BuiltinType::SveFloat16x4:
     return SVE_ELTTY(HalfTy, 8, 4);

clang/lib/AST/Type.cpp

if (const BuiltinType *BT = getAs<BuiltinType>()) {
     case BuiltinType::SveUint16:
     case BuiltinType::SveUint32:
     case BuiltinType::SveUint64:
     case BuiltinType::SveFloat16:
     case BuiltinType::SveFloat32:
     case BuiltinType::SveFloat64:
     case BuiltinType::SveBFloat16:
     case BuiltinType::SveBool:
+    case BuiltinType::SveBoolx2:
+    case BuiltinType::SveBoolx4:
       return true;
     default:
       return false;
     }
   }
   return false;
 }
 

clang/lib/CodeGen/CodeGenTypes.cpp

#include "clang/Basic/OpenCLExtensionTypes.def"
     case BuiltinType::SveUint64:
     case BuiltinType::SveInt64x2:
     case BuiltinType::SveUint64x2:
     case BuiltinType::SveInt64x3:
     case BuiltinType::SveUint64x3:
     case BuiltinType::SveInt64x4:
     case BuiltinType::SveUint64x4:
     case BuiltinType::SveBool:
+    case BuiltinType::SveBoolx2:
+    case BuiltinType::SveBoolx4:
     case BuiltinType::SveFloat16:
     case BuiltinType::SveFloat16x2:
     case BuiltinType::SveFloat16x3:
     case BuiltinType::SveFloat16x4:
     case BuiltinType::SveFloat32:
     case BuiltinType::SveFloat32x2:
     case BuiltinType::SveFloat32x3:
     case BuiltinType::SveFloat32x4:

clang/test/CodeGen/svboolx2_t.cpp

+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -S -emit-llvm -o - %s | FileCheck %s
+
+// CHECK-LABEL: @_Z3foo10svboolx2_t(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ARG_ADDR:%.*]] = alloca <vscale x 32 x i1>, align 2
+// CHECK-NEXT:    store <vscale x 32 x i1> [[ARG:%.*]], ptr [[ARG_ADDR]], align 2
+// CHECK-NEXT:    [[TMP0:%.*]] = load <vscale x 32 x i1>, ptr [[ARG_ADDR]], align 2
+// CHECK-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+__clang_svboolx2_t foo(__clang_svboolx2_t arg) { return arg; }
+
+__clang_svboolx2_t bar();
+// CHECK-LABEL: @_Z4foo2v(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[CALL:%.*]] = call <vscale x 32 x i1> @_Z3barv()
+// CHECK-NEXT:    ret <vscale x 32 x i1> [[CALL]]
+//
+__clang_svboolx2_t foo2() { return bar(); }
+
+__clang_svboolx2_t bar2(__clang_svboolx2_t);
+// CHECK-LABEL: @_Z4foo310svboolx2_t(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ARG_ADDR:%.*]] = alloca <vscale x 32 x i1>, align 2
+// CHECK-NEXT:    store <vscale x 32 x i1> [[ARG:%.*]], ptr [[ARG_ADDR]], align 2
+// CHECK-NEXT:    [[TMP0:%.*]] = load <vscale x 32 x i1>, ptr [[ARG_ADDR]], align 2
+// CHECK-NEXT:    [[CALL:%.*]] = call <vscale x 32 x i1> @_Z4bar210svboolx2_t(<vscale x 32 x i1> [[TMP0]])
+// CHECK-NEXT:    ret <vscale x 32 x i1> [[CALL]]
+//
+__clang_svboolx2_t foo3(__clang_svboolx2_t arg) { return bar2(arg); }
+

clang/test/CodeGen/svboolx4_t.cpp

+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -S -emit-llvm -o - %s | FileCheck %s
+
+// CHECK-LABEL: @_Z3foo10svboolx4_t(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ARG_ADDR:%.*]] = alloca <vscale x 64 x i1>, align 2
+// CHECK-NEXT:    store <vscale x 64 x i1> [[ARG:%.*]], ptr [[ARG_ADDR]], align 2
+// CHECK-NEXT:    [[TMP0:%.*]] = load <vscale x 64 x i1>, ptr [[ARG_ADDR]], align 2
+// CHECK-NEXT:    ret <vscale x 64 x i1> [[TMP0]]
+//
+__clang_svboolx4_t foo(__clang_svboolx4_t arg) { return arg; }
+
+__clang_svboolx4_t bar();
+// CHECK-LABEL: @_Z4foo2v(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[CALL:%.*]] = call <vscale x 64 x i1> @_Z3barv()
+// CHECK-NEXT:    ret <vscale x 64 x i1> [[CALL]]
+//
+__clang_svboolx4_t foo2() { return bar(); }
+
+__clang_svboolx4_t bar2(__clang_svboolx4_t);
+// CHECK-LABEL: @_Z4foo310svboolx4_t(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ARG_ADDR:%.*]] = alloca <vscale x 64 x i1>, align 2
+// CHECK-NEXT:    store <vscale x 64 x i1> [[ARG:%.*]], ptr [[ARG_ADDR]], align 2
+// CHECK-NEXT:    [[TMP0:%.*]] = load <vscale x 64 x i1>, ptr [[ARG_ADDR]], align 2
+// CHECK-NEXT:    [[CALL:%.*]] = call <vscale x 64 x i1> @_Z4bar210svboolx4_t(<vscale x 64 x i1> [[TMP0]])
+// CHECK-NEXT:    ret <vscale x 64 x i1> [[CALL]]
+//
+__clang_svboolx4_t foo3(__clang_svboolx4_t arg) { return bar2(arg); }
+

clang/test/CodeGenCXX/aarch64-mangle-sve-vectors.cpp

 // CHECK: _Z3f451SI13svfloat64x4_tE
 void f45(S<__clang_svfloat64x4_t>) {}
 // CHECK: _Z3f461SI14svbfloat16x2_tE
 void f46(S<__clang_svbfloat16x2_t>) {}
 // CHECK: _Z3f471SI14svbfloat16x3_tE
 void f47(S<__clang_svbfloat16x3_t>) {}
 // CHECK: _Z3f481SI14svbfloat16x4_tE
 void f48(S<__clang_svbfloat16x4_t>) {}
+// CHECK: _Z3f491SI10svboolx2_tE
+void f49(S<__clang_svboolx2_t>) {}
+// CHECK: _Z3f501SI10svboolx4_tE
+void f50(S<__clang_svboolx4_t>) {}

clang/utils/TableGen/SveEmitter.cpp

void SVEEmitter::createHeader(raw_ostream &OS) {
   OS << "typedef __clang_svint64x4_t svint64x4_t;\n";
   OS << "typedef __clang_svuint8x4_t svuint8x4_t;\n";
   OS << "typedef __clang_svuint16x4_t svuint16x4_t;\n";
   OS << "typedef __clang_svuint32x4_t svuint32x4_t;\n";
   OS << "typedef __clang_svuint64x4_t svuint64x4_t;\n";
   OS << "typedef __clang_svfloat16x4_t svfloat16x4_t;\n";
   OS << "typedef __clang_svfloat32x4_t svfloat32x4_t;\n";
   OS << "typedef __clang_svfloat64x4_t svfloat64x4_t;\n";
-  OS << "typedef __SVBool_t  svbool_t;\n\n";
+  OS << "typedef __SVBool_t  svbool_t;\n";
+  OS << "typedef __clang_svboolx2_t  svboolx2_t;\n";
+  OS << "typedef __clang_svboolx4_t  svboolx4_t;\n\n";
 
   OS << "typedef __clang_svbfloat16x2_t svbfloat16x2_t;\n";
   OS << "typedef __clang_svbfloat16x3_t svbfloat16x3_t;\n";
   OS << "typedef __clang_svbfloat16x4_t svbfloat16x4_t;\n";
 
   OS << "enum svpattern\n";
   OS << "{\n";
   OS << "  SV_POW2 = 0,\n";