[HLSL] [clang] Add vector version of abs for HLSL

[python3kgae]

Summary

Add vector version of abs as

__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
int2 abs (int2 );
__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
int3 abs (int3 );

To make this work.
Allowed custom type checking builtins to be recelareable.

[fhahn]

I'm not familiar with HLSL but I think making use of the existing vector builtins makes sense here!

[beanz]

In D133737#3785987, @fhahn wrote:

I'm not familiar with HLSL but I think making use of the existing vector builtins makes sense here!

Does it make sense to be using the vector builtin for non-vectors?

[RKSimon]

The elementwise builtins should work for scalars as well (and IIRC we do test this as well).

[fhahn]

In D133737#3787066, @RKSimon wrote:

The elementwise builtins should work for scalars as well (and IIRC we do test this as well).

Yep they should, it is specially specified here: https://clang.llvm.org/docs/LanguageExtensions.html#vector-builtins

For scalar types, consider the operation applied to a vector with a single element.

[python3kgae]

Only allow redecl custom type checking builtin for HLSL.

[RKSimon]

LGTM

clang/lib/AST/ASTContext.cpp

 TypedefDecl *ASTContext::getBuiltinMSVaListDecl() const {
   if (!BuiltinMSVaListDecl)
     BuiltinMSVaListDecl = CreateMSVaListDecl(this);
 
   return BuiltinMSVaListDecl;
 }
 
 bool ASTContext::canBuiltinBeRedeclared(const FunctionDecl *FD) const {
+  // Allow redecl custom type checking builtin for HLSL.
+  if (LangOpts.HLSL && FD->getBuiltinID() != Builtin::NotBuiltin &&
+      BuiltinInfo.hasCustomTypechecking(FD->getBuiltinID()))
+    return true;
   return BuiltinInfo.canBeRedeclared(FD->getBuiltinID());
 }
 
 void ASTContext::setObjCConstantStringInterface(ObjCInterfaceDecl *Decl) {
   assert(ObjCConstantStringType.isNull() &&
          "'NSConstantString' type already set!");
 
   ObjCConstantStringType = getObjCInterfaceType(Decl);

clang/lib/Headers/hlsl/hlsl_intrinsics.h

 //===----- hlsl_intrinsics.h - HLSL definitions for intrinsics ----------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef _HLSL_HLSL_INTRINSICS_H_
 #define _HLSL_HLSL_INTRINSICS_H_
 
 __attribute__((availability(shadermodel, introduced = 6.0)))
 __attribute__((clang_builtin_alias(__builtin_hlsl_wave_active_count_bits))) uint
 WaveActiveCountBits(bool bBit);
 
-__attribute__((clang_builtin_alias(__builtin_abs))) int abs(int In);
-__attribute__((clang_builtin_alias(__builtin_labs))) int64_t abs(int64_t In);
-__attribute__((clang_builtin_alias(__builtin_fabsf))) float abs(float In);
-__attribute__((clang_builtin_alias(__builtin_fabs))) double abs(double In);
-
 #ifdef __HLSL_ENABLE_16_BIT
-__attribute__((clang_builtin_alias(__builtin_fabsf16))) half abs(half In);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+int16_t abs(int16_t);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+int16_t2 abs(int16_t2);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+int16_t3 abs(int16_t3);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+int16_t4 abs(int16_t4);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs))) half abs(half);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+half2 abs(half2);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+half3 abs(half3);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+half4 abs(half4);
 #endif
 
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs))) int abs(int);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs))) int2 abs(int2);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs))) int3 abs(int3);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs))) int4 abs(int4);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs))) float
+abs(float);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+float2 abs(float2);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+float3 abs(float3);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+float4 abs(float4);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+int64_t abs(int64_t);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+int64_t2 abs(int64_t2);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+int64_t3 abs(int64_t3);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+int64_t4 abs(int64_t4);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs))) double
+abs(double);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+double2 abs(double2);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+double3 abs(double3);
+__attribute__((clang_builtin_alias(__builtin_elementwise_abs)))
+double4 abs(double4);
+
 #endif //_HLSL_HLSL_INTRINSICS_H_

clang/test/CodeGenHLSL/builtins/abs.hlsl

 // RUN: %clang_cc1 -std=hlsl2021 -finclude-default-header -x hlsl -triple \
 // RUN:   dxil-pc-shadermodel6.3-library %s -fnative-half-type \
-// RUN:   -emit-llvm -disable-llvm-passes -o - | FileCheck %s
+// RUN:   -emit-llvm -disable-llvm-passes -O3 -o - | FileCheck %s
 // RUN: %clang_cc1 -std=hlsl2021 -finclude-default-header -x hlsl -triple \
 // RUN:   dxil-pc-shadermodel6.3-library %s -emit-llvm -disable-llvm-passes \
-// RUN:   -o - | FileCheck %s --check-prefix=NO_HALF
+// RUN:   -D__HLSL_ENABLE_16_BIT -o - | FileCheck %s --check-prefix=NO_HALF
 
 
-float abs_float(float X) {
-  return abs(X);
+// CHECK: define noundef signext i16 @
+// FIXME: int16_t is promoted to i32 now. Change to abs.i16 once it is fixed.
+// CHECK: call i32 @llvm.abs.i32(
+int16_t test_abs_int16_t ( int16_t p0 ) {
+  return abs ( p0 );
 }
-
-// CHECK: define noundef float @"?abs_float@@YAMM@Z"(
-// CHECK: call float @llvm.fabs.f32(float %0)
-
-double abs_double(double X) {
-  return abs(X);
+// CHECK: define noundef <2 x i16> @
+// CHECK: call <2 x i16> @llvm.abs.v2i16(
+int16_t2 test_abs_int16_t2 ( int16_t2 p0 ) {
+  return abs ( p0 );
 }
-
-// CHECK: define noundef double @"?abs_double@@YANN@Z"(
-// CHECK: call double @llvm.fabs.f64(double %0)
-
-half abs_half(half X) {
-  return abs(X);
+// CHECK: define noundef <3 x i16> @
+// CHECK: call <3 x i16> @llvm.abs.v3i16(
+int16_t3 test_abs_int16_t3 ( int16_t3 p0 ) {
+  return abs ( p0 );
 }
-
-// CHECK: define noundef half @"?abs_half@@YA$f16@$f16@@Z"(
-// CHECK: call half @llvm.fabs.f16(half %0)
-// NO_HALF: define noundef float @"?abs_half@@YA$halff@$halff@@Z"(
+// CHECK: define noundef <4 x i16> @
+// CHECK: call <4 x i16> @llvm.abs.v4i16(
+int16_t4 test_abs_int16_t4 ( int16_t4 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef half @
+// CHECK: call half @llvm.fabs.f16(
+// NO_HALF: define noundef float @"?test_abs_half@@YA$halff@$halff@@Z"(
 // NO_HALF: call float @llvm.fabs.f32(float %0)
-
-int abs_int(int X) {
-  return abs(X);
+half test_abs_half ( half p0 ) {
+  return abs ( p0 );
 }
-
-// NO_HALF: define noundef i32 @"?abs_int@@YAHH@Z"(i32
-// CHECK: define noundef i32 @"?abs_int@@YAHH@Z"(i32
-// CHECK:         [[A_ADDR:%.*]] = alloca i32, align 4
-// CHECK-NEXT:    store i32 [[A:%.*]], ptr [[A_ADDR]], align 4
-// CHECK-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
-// CHECK-NEXT:    [[NEG:%.*]] = sub nsw i32 0, [[TMP0]]
-// CHECK-NEXT:    [[ABSCOND:%.*]] = icmp slt i32 [[TMP0]], 0
-// CHECK-NEXT:    [[ABS:%.*]] = select i1 [[ABSCOND]], i32 [[NEG]], i32 [[TMP0]]
-// CHECK-NEXT:    ret i32 [[ABS]]
-
-int64_t abs_int64(int64_t X) {
-  return abs(X);
+// CHECK: define noundef <2 x half> @
+// CHECK: call <2 x half> @llvm.fabs.v2f16(
+// NO_HALF: define noundef <2 x float> @"?test_abs_half2@@YAT?$__vector@$halff@$01@__clang@@T12@@Z"(
+// NO_HALF: call <2 x float> @llvm.fabs.v2f32(
+half2 test_abs_half2 ( half2 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <3 x half> @
+// CHECK: call <3 x half> @llvm.fabs.v3f16(
+// NO_HALF: define noundef <3 x float> @"?test_abs_half3@@YAT?$__vector@$halff@$02@__clang@@T12@@Z"(
+// NO_HALF: call <3 x float> @llvm.fabs.v3f32(
+half3 test_abs_half3 ( half3 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <4 x half> @
+// CHECK: call <4 x half> @llvm.fabs.v4f16(
+// NO_HALF: define noundef <4 x float> @"?test_abs_half4@@YAT?$__vector@$halff@$03@__clang@@T12@@Z"(
+// NO_HALF: call <4 x float> @llvm.fabs.v4f32(
+half4 test_abs_half4 ( half4 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef i32 @
+// CHECK: call i32 @llvm.abs.i32(
+// NO_HALF: define noundef i32 @"?test_abs_int@@YAHH@Z"
+int test_abs_int ( int p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <2 x i32> @
+// CHECK: call <2 x i32> @llvm.abs.v2i32(
+int2 test_abs_int2 ( int2 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <3 x i32> @
+// CHECK: call <3 x i32> @llvm.abs.v3i32(
+int3 test_abs_int3 ( int3 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <4 x i32> @
+// CHECK: call <4 x i32> @llvm.abs.v4i32(
+int4 test_abs_int4 ( int4 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef float @
+// CHECK: call float @llvm.fabs.f32(
+float test_abs_float ( float p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <2 x float> @
+// CHECK: call <2 x float> @llvm.fabs.v2f32(
+float2 test_abs_float2 ( float2 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <3 x float> @
+// CHECK: call <3 x float> @llvm.fabs.v3f32(
+float3 test_abs_float3 ( float3 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <4 x float> @
+// CHECK: call <4 x float> @llvm.fabs.v4f32(
+float4 test_abs_float4 ( float4 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef i64 @
+// CHECK: call i64 @llvm.abs.i64(
+int64_t test_abs_int64_t ( int64_t p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <2 x i64> @
+// CHECK: call <2 x i64> @llvm.abs.v2i64(
+int64_t2 test_abs_int64_t2 ( int64_t2 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <3 x i64> @
+// CHECK: call <3 x i64> @llvm.abs.v3i64(
+int64_t3 test_abs_int64_t3 ( int64_t3 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <4 x i64> @
+// CHECK: call <4 x i64> @llvm.abs.v4i64(
+int64_t4 test_abs_int64_t4 ( int64_t4 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef double @
+// CHECK: call double @llvm.fabs.f64(
+double test_abs_double ( double p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <2 x double> @
+// CHECK: call <2 x double> @llvm.fabs.v2f64(
+double2 test_abs_double2 ( double2 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <3 x double> @
+// CHECK: call <3 x double> @llvm.fabs.v3f64(
+double3 test_abs_double3 ( double3 p0 ) {
+  return abs ( p0 );
+}
+// CHECK: define noundef <4 x double> @
+// CHECK: call <4 x double> @llvm.fabs.v4f64(
+double4 test_abs_double4 ( double4 p0 ) {
+  return abs ( p0 );
 }
-
-// CHECK: define noundef i64 @"?abs_int64@@YAJJ@Z"(i64
-// CHECK:         [[A_ADDR:%.*]] = alloca i64, align 8
-// CHECK-NEXT:    store i64 [[A:%.*]], ptr [[A_ADDR]], align 8
-// CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr [[A_ADDR]], align 8
-// CHECK-NEXT:    [[NEG:%.*]] = sub nsw i64 0, [[TMP0]]
-// CHECK-NEXT:    [[ABSCOND:%.*]] = icmp slt i64 [[TMP0]], 0
-// CHECK-NEXT:    [[ABS:%.*]] = select i1 [[ABSCOND]], i64 [[NEG]], i64 [[TMP0]]
-// CHECK-NEXT:    ret i64 [[ABS]]