[clang] [HLSL] add extra scalar vector overloads for clamp (PR #129939)
Sarah Spall via cfe-commits
cfe-commits at lists.llvm.org
Wed Mar 5 13:21:06 PST 2025
https://github.com/spall created https://github.com/llvm/llvm-project/pull/129939
Add additional vector scalar overloads for clamp
Revamp Macros which generate overloads
Add Tests
Fix one test which now errors instead of warns.
Closes #128230
>From b464aec7a7686a696006d23f16541ff2ca0bc1b2 Mon Sep 17 00:00:00 2001
From: Sarah Spall <sarahspall at microsoft.com>
Date: Tue, 4 Mar 2025 09:53:56 -0800
Subject: [PATCH] extra scalar vector overloads for clamp
---
.../lib/Headers/hlsl/hlsl_alias_intrinsics.h | 99 ++++++++++++-------
clang/test/CodeGenHLSL/builtins/clamp.hlsl | 32 ++++++
.../test/SemaHLSL/BuiltIns/clamp-errors.hlsl | 2 +-
3 files changed, 96 insertions(+), 37 deletions(-)
diff --git a/clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h b/clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h
index 75b0c95440461..62262535130cc 100644
--- a/clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h
+++ b/clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h
@@ -35,26 +35,44 @@ namespace hlsl {
#define _HLSL_16BIT_AVAILABILITY_STAGE(environment, version, stage)
#endif
-#define GEN_VEC_SCALAR_OVERLOADS(FUNC_NAME, BASE_TYPE, AVAIL) \
- GEN_BOTH_OVERLOADS(FUNC_NAME, BASE_TYPE, BASE_TYPE##2, AVAIL) \
- GEN_BOTH_OVERLOADS(FUNC_NAME, BASE_TYPE, BASE_TYPE##3, AVAIL) \
- GEN_BOTH_OVERLOADS(FUNC_NAME, BASE_TYPE, BASE_TYPE##4, AVAIL)
-
-#define GEN_BOTH_OVERLOADS(FUNC_NAME, BASE_TYPE, VECTOR_TYPE, AVAIL) \
- IF_TRUE_##AVAIL( \
- _HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)) constexpr VECTOR_TYPE \
- FUNC_NAME(VECTOR_TYPE p0, BASE_TYPE p1) { \
- return __builtin_elementwise_##FUNC_NAME(p0, (VECTOR_TYPE)p1); \
+#define _HLSL_CAT(a,b) a##b
+#define _HLSL_VEC_SCALAR_OVERLOADS(NAME, BASE_T, AVAIL) \
+ _HLSL_ALL_OVERLOADS(NAME, BASE_T, AVAIL, _HLSL_CAT(_HLSL_NUM_ARGS_,NAME))
+
+#define _HLSL_ALL_OVERLOADS(NAME, BASE_T, AVAIL, NUM_ARGS) \
+ _HLSL_CAT(_HLSL_BOTH_OVERLOADS_,NUM_ARGS)(NAME, BASE_T, _HLSL_CAT(BASE_T,2), AVAIL) \
+ _HLSL_CAT(_HLSL_BOTH_OVERLOADS_,NUM_ARGS)(NAME, BASE_T, _HLSL_CAT(BASE_T,3), AVAIL) \
+ _HLSL_CAT(_HLSL_BOTH_OVERLOADS_,NUM_ARGS)(NAME, BASE_T, _HLSL_CAT(BASE_T,4), AVAIL)
+
+#define _HLSL_BOTH_OVERLOADS_2(NAME, BASE_T, VECTOR_T, AVAIL) \
+ _HLSL_CAT(_HLSL_IF_TRUE_,AVAIL)( \
+ _HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)) constexpr VECTOR_T \
+ NAME(VECTOR_T p0, BASE_T p1) { \
+ return _HLSL_CAT(__builtin_elementwise_,NAME)(p0, (VECTOR_T)p1); \
} \
- IF_TRUE_##AVAIL( \
- _HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)) constexpr VECTOR_TYPE \
- FUNC_NAME(BASE_TYPE p0, VECTOR_TYPE p1) { \
- return __builtin_elementwise_##FUNC_NAME((VECTOR_TYPE)p0, p1); \
+ _HLSL_CAT(_HLSL_IF_TRUE_,AVAIL)( \
+ _HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)) constexpr VECTOR_T \
+ NAME(BASE_T p0, VECTOR_T p1) { \
+ return _HLSL_CAT(__builtin_elementwise_,NAME)((VECTOR_T)p0, p1); \
}
-#define IF_TRUE_0(EXPR)
-#define IF_TRUE_1(EXPR) EXPR
+#define _HLSL_BOTH_OVERLOADS_3(NAME, BASE_T, VECTOR_T, AVAIL) \
+ _HLSL_CAT(_HLSL_IF_TRUE_,AVAIL)(_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)) \
+ constexpr VECTOR_T NAME(VECTOR_T p0, VECTOR_T p1, BASE_T p2) { \
+ return _HLSL_CAT(__builtin_hlsl_elementwise_,NAME)(p0, p1, (VECTOR_T)p2); \
+ } \
+ _HLSL_CAT(_HLSL_IF_TRUE_,AVAIL)(_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)) \
+ constexpr VECTOR_T NAME(VECTOR_T p0, BASE_T p1, VECTOR_T p2) { \
+ return _HLSL_CAT(__builtin_hlsl_elementwise_,NAME)(p0, (VECTOR_T)p1, p2); \
+ }
+
+#define _HLSL_IF_TRUE_0(EXPR)
+#define _HLSL_IF_TRUE_1(EXPR) EXPR
+#define _HLSL_NUM_ARGS_min 2
+#define _HLSL_NUM_ARGS_max 2
+#define _HLSL_NUM_ARGS_clamp 3
+
//===----------------------------------------------------------------------===//
// abs builtins
//===----------------------------------------------------------------------===//
@@ -582,7 +600,8 @@ half3 clamp(half3, half3, half3);
_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
half4 clamp(half4, half4, half4);
-
+_HLSL_VEC_SCALAR_OVERLOADS(clamp, half, 1)
+
#ifdef __HLSL_ENABLE_16_BIT
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
@@ -596,7 +615,8 @@ int16_t3 clamp(int16_t3, int16_t3, int16_t3);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int16_t4 clamp(int16_t4, int16_t4, int16_t4);
-
+_HLSL_VEC_SCALAR_OVERLOADS(clamp, int16_t, 1)
+
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint16_t clamp(uint16_t, uint16_t, uint16_t);
@@ -609,6 +629,7 @@ uint16_t3 clamp(uint16_t3, uint16_t3, uint16_t3);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint16_t4 clamp(uint16_t4, uint16_t4, uint16_t4);
+_HLSL_VEC_SCALAR_OVERLOADS(clamp, uint16_t, 1)
#endif
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
@@ -619,6 +640,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int3 clamp(int3, int3, int3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int4 clamp(int4, int4, int4);
+_HLSL_VEC_SCALAR_OVERLOADS(clamp, int, 0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint clamp(uint, uint, uint);
@@ -628,6 +650,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint3 clamp(uint3, uint3, uint3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint4 clamp(uint4, uint4, uint4);
+_HLSL_VEC_SCALAR_OVERLOADS(clamp, uint, 0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int64_t clamp(int64_t, int64_t, int64_t);
@@ -637,6 +660,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int64_t3 clamp(int64_t3, int64_t3, int64_t3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int64_t4 clamp(int64_t4, int64_t4, int64_t4);
+_HLSL_VEC_SCALAR_OVERLOADS(clamp, int64_t, 0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint64_t clamp(uint64_t, uint64_t, uint64_t);
@@ -646,6 +670,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint64_t3 clamp(uint64_t3, uint64_t3, uint64_t3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint64_t4 clamp(uint64_t4, uint64_t4, uint64_t4);
+_HLSL_VEC_SCALAR_OVERLOADS(clamp, uint64_t, 0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
float clamp(float, float, float);
@@ -655,6 +680,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
float3 clamp(float3, float3, float3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
float4 clamp(float4, float4, float4);
+_HLSL_VEC_SCALAR_OVERLOADS(clamp, float, 0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
double clamp(double, double, double);
@@ -664,6 +690,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
double3 clamp(double3, double3, double3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
double4 clamp(double4, double4, double4);
+_HLSL_VEC_SCALAR_OVERLOADS(clamp, double, 0)
//===----------------------------------------------------------------------===//
// clip builtins
@@ -1576,7 +1603,7 @@ half3 max(half3, half3);
_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
half4 max(half4, half4);
-GEN_VEC_SCALAR_OVERLOADS(max, half, 1)
+_HLSL_VEC_SCALAR_OVERLOADS(max, half, 1)
#ifdef __HLSL_ENABLE_16_BIT
_HLSL_AVAILABILITY(shadermodel, 6.2)
@@ -1591,7 +1618,7 @@ int16_t3 max(int16_t3, int16_t3);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
int16_t4 max(int16_t4, int16_t4);
-GEN_VEC_SCALAR_OVERLOADS(max, int16_t, 1)
+_HLSL_VEC_SCALAR_OVERLOADS(max, int16_t, 1)
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
@@ -1605,7 +1632,7 @@ uint16_t3 max(uint16_t3, uint16_t3);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
uint16_t4 max(uint16_t4, uint16_t4);
-GEN_VEC_SCALAR_OVERLOADS(max, uint16_t, 1)
+_HLSL_VEC_SCALAR_OVERLOADS(max, uint16_t, 1)
#endif
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
@@ -1616,7 +1643,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
int3 max(int3, int3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
int4 max(int4, int4);
-GEN_VEC_SCALAR_OVERLOADS(max, int, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(max, int, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
uint max(uint, uint);
@@ -1626,7 +1653,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
uint3 max(uint3, uint3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
uint4 max(uint4, uint4);
-GEN_VEC_SCALAR_OVERLOADS(max, uint, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(max, uint, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
int64_t max(int64_t, int64_t);
@@ -1636,7 +1663,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
int64_t3 max(int64_t3, int64_t3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
int64_t4 max(int64_t4, int64_t4);
-GEN_VEC_SCALAR_OVERLOADS(max, int64_t, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(max, int64_t, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
uint64_t max(uint64_t, uint64_t);
@@ -1646,7 +1673,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
uint64_t3 max(uint64_t3, uint64_t3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
uint64_t4 max(uint64_t4, uint64_t4);
-GEN_VEC_SCALAR_OVERLOADS(max, uint64_t, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(max, uint64_t, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
float max(float, float);
@@ -1656,7 +1683,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
float3 max(float3, float3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
float4 max(float4, float4);
-GEN_VEC_SCALAR_OVERLOADS(max, float, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(max, float, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
double max(double, double);
@@ -1666,7 +1693,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
double3 max(double3, double3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_max)
double4 max(double4, double4);
-GEN_VEC_SCALAR_OVERLOADS(max, double, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(max, double, 0)
//===----------------------------------------------------------------------===//
// min builtins
@@ -1689,7 +1716,7 @@ half3 min(half3, half3);
_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
half4 min(half4, half4);
-GEN_VEC_SCALAR_OVERLOADS(min, half, 1)
+_HLSL_VEC_SCALAR_OVERLOADS(min, half, 1)
#ifdef __HLSL_ENABLE_16_BIT
_HLSL_AVAILABILITY(shadermodel, 6.2)
@@ -1704,7 +1731,7 @@ int16_t3 min(int16_t3, int16_t3);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
int16_t4 min(int16_t4, int16_t4);
-GEN_VEC_SCALAR_OVERLOADS(min, int16_t, 1)
+_HLSL_VEC_SCALAR_OVERLOADS(min, int16_t, 1)
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
@@ -1718,7 +1745,7 @@ uint16_t3 min(uint16_t3, uint16_t3);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
uint16_t4 min(uint16_t4, uint16_t4);
-GEN_VEC_SCALAR_OVERLOADS(min, uint16_t, 1)
+_HLSL_VEC_SCALAR_OVERLOADS(min, uint16_t, 1)
#endif
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
@@ -1729,7 +1756,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
int3 min(int3, int3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
int4 min(int4, int4);
-GEN_VEC_SCALAR_OVERLOADS(min, int, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(min, int, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
uint min(uint, uint);
@@ -1739,7 +1766,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
uint3 min(uint3, uint3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
uint4 min(uint4, uint4);
-GEN_VEC_SCALAR_OVERLOADS(min, uint, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(min, uint, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
float min(float, float);
@@ -1749,7 +1776,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
float3 min(float3, float3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
float4 min(float4, float4);
-GEN_VEC_SCALAR_OVERLOADS(min, float, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(min, float, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
int64_t min(int64_t, int64_t);
@@ -1759,7 +1786,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
int64_t3 min(int64_t3, int64_t3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
int64_t4 min(int64_t4, int64_t4);
-GEN_VEC_SCALAR_OVERLOADS(min, int64_t, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(min, int64_t, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
uint64_t min(uint64_t, uint64_t);
@@ -1769,7 +1796,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
uint64_t3 min(uint64_t3, uint64_t3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
uint64_t4 min(uint64_t4, uint64_t4);
-GEN_VEC_SCALAR_OVERLOADS(min, uint64_t, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(min, uint64_t, 0)
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
double min(double, double);
@@ -1779,7 +1806,7 @@ _HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
double3 min(double3, double3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_min)
double4 min(double4, double4);
-GEN_VEC_SCALAR_OVERLOADS(min, double, 0)
+_HLSL_VEC_SCALAR_OVERLOADS(min, double, 0)
//===----------------------------------------------------------------------===//
// normalize builtins
diff --git a/clang/test/CodeGenHLSL/builtins/clamp.hlsl b/clang/test/CodeGenHLSL/builtins/clamp.hlsl
index d01c2a45c43c8..51454d6479708 100644
--- a/clang/test/CodeGenHLSL/builtins/clamp.hlsl
+++ b/clang/test/CodeGenHLSL/builtins/clamp.hlsl
@@ -28,6 +28,9 @@ int16_t3 test_clamp_short3(int16_t3 p0, int16_t3 p1) { return clamp(p0, p1,p1);
// NATIVE_HALF: define [[FNATTRS]] <4 x i16> @_Z17test_clamp_short4
// NATIVE_HALF: call <4 x i16> @llvm.[[TARGET]].sclamp.v4i16
int16_t4 test_clamp_short4(int16_t4 p0, int16_t4 p1) { return clamp(p0, p1,p1); }
+// NATIVE_HALF: define [[FNATTRS]] <4 x i16> {{.*}}test_clamp_short4_mismatch
+// NATIVE_HALF: call <4 x i16> @llvm.[[TARGET]].sclamp.v4i16
+int16_t4 test_clamp_short4_mismatch(int16_t4 p0, int16_t p1) { return clamp(p0, p0,p1); }
// NATIVE_HALF: define [[FNATTRS]] i16 @_Z17test_clamp_ushort
// NATIVE_HALF: call i16 @llvm.[[TARGET]].uclamp.i16(
@@ -41,6 +44,9 @@ uint16_t3 test_clamp_ushort3(uint16_t3 p0, uint16_t3 p1) { return clamp(p0, p1,p
// NATIVE_HALF: define [[FNATTRS]] <4 x i16> @_Z18test_clamp_ushort4
// NATIVE_HALF: call <4 x i16> @llvm.[[TARGET]].uclamp.v4i16
uint16_t4 test_clamp_ushort4(uint16_t4 p0, uint16_t4 p1) { return clamp(p0, p1,p1); }
+// NATIVE_HALF: define [[FNATTRS]] <4 x i16> {{.*}}test_clamp_ushort4_mismatch
+// NATIVE_HALF: call <4 x i16> @llvm.[[TARGET]].uclamp.v4i16
+uint16_t4 test_clamp_ushort4_mismatch(uint16_t4 p0, uint16_t p1) { return clamp(p0, p0,p1); }
#endif
// CHECK: define [[FNATTRS]] i32 @_Z14test_clamp_int
@@ -55,6 +61,9 @@ int3 test_clamp_int3(int3 p0, int3 p1) { return clamp(p0, p1,p1); }
// CHECK: define [[FNATTRS]] <4 x i32> @_Z15test_clamp_int4
// CHECK: call <4 x i32> @llvm.[[TARGET]].sclamp.v4i32
int4 test_clamp_int4(int4 p0, int4 p1) { return clamp(p0, p1,p1); }
+// CHECK: define [[FNATTRS]] <4 x i32> {{.*}}test_clamp_int4_mismatch
+// CHECK: call <4 x i32> @llvm.[[TARGET]].sclamp.v4i32
+int4 test_clamp_int4_mismatch(int4 p0, int p1) { return clamp(p0, p0,p1); }
// CHECK: define [[FNATTRS]] i32 @_Z15test_clamp_uint
// CHECK: call i32 @llvm.[[TARGET]].uclamp.i32(
@@ -68,6 +77,9 @@ uint3 test_clamp_uint3(uint3 p0, uint3 p1) { return clamp(p0, p1,p1); }
// CHECK: define [[FNATTRS]] <4 x i32> @_Z16test_clamp_uint4
// CHECK: call <4 x i32> @llvm.[[TARGET]].uclamp.v4i32
uint4 test_clamp_uint4(uint4 p0, uint4 p1) { return clamp(p0, p1,p1); }
+// CHECK: define [[FNATTRS]] <4 x i32> {{.*}}test_clamp_uint4_mismatch
+// CHECK: call <4 x i32> @llvm.[[TARGET]].uclamp.v4i32
+uint4 test_clamp_uint4_mismatch(uint4 p0, uint p1) { return clamp(p0, p0,p1); }
// CHECK: define [[FNATTRS]] i64 @_Z15test_clamp_long
// CHECK: call i64 @llvm.[[TARGET]].sclamp.i64(
@@ -81,6 +93,9 @@ int64_t3 test_clamp_long3(int64_t3 p0, int64_t3 p1) { return clamp(p0, p1,p1); }
// CHECK: define [[FNATTRS]] <4 x i64> @_Z16test_clamp_long4
// CHECK: call <4 x i64> @llvm.[[TARGET]].sclamp.v4i64
int64_t4 test_clamp_long4(int64_t4 p0, int64_t4 p1) { return clamp(p0, p1,p1); }
+// CHECK: define [[FNATTRS]] <4 x i64> {{.*}}test_clamp_long4_mismatch
+// CHECK: call <4 x i64> @llvm.[[TARGET]].sclamp.v4i64
+int64_t4 test_clamp_long4_mismatch(int64_t4 p0, int64_t4 p1) { return clamp(p0, p0,p1); }
// CHECK: define [[FNATTRS]] i64 @_Z16test_clamp_ulong
// CHECK: call i64 @llvm.[[TARGET]].uclamp.i64(
@@ -94,6 +109,9 @@ uint64_t3 test_clamp_ulong3(uint64_t3 p0, uint64_t3 p1) { return clamp(p0, p1,p1
// CHECK: define [[FNATTRS]] <4 x i64> @_Z17test_clamp_ulong4
// CHECK: call <4 x i64> @llvm.[[TARGET]].uclamp.v4i64
uint64_t4 test_clamp_ulong4(uint64_t4 p0, uint64_t4 p1) { return clamp(p0, p1,p1); }
+// CHECK: define [[FNATTRS]] <4 x i64> {{.*}}test_clamp_ulong4_mismatch
+// CHECK: call <4 x i64> @llvm.[[TARGET]].uclamp.v4i64
+uint64_t4 test_clamp_ulong4_mismatch(uint64_t4 p0, uint64_t4 p1) { return clamp(p0, p0,p1); }
// NATIVE_HALF: define [[FNATTRS]] [[FFNATTRS]] half @_Z15test_clamp_half
// NATIVE_HALF: call reassoc nnan ninf nsz arcp afn half @llvm.[[TARGET]].nclamp.f16(
@@ -115,6 +133,11 @@ half3 test_clamp_half3(half3 p0, half3 p1) { return clamp(p0, p1,p1); }
// NO_HALF: define [[FNATTRS]] [[FFNATTRS]] <4 x float> @_Z16test_clamp_half4
// NO_HALF: call reassoc nnan ninf nsz arcp afn <4 x float> @llvm.[[TARGET]].nclamp.v4f32(
half4 test_clamp_half4(half4 p0, half4 p1) { return clamp(p0, p1,p1); }
+// NATIVE_HALF: define [[FNATTRS]] [[FFNATTRS]] <4 x half> {{.*}}test_clamp_half4_mismatch
+// NATIVE_HALF: call reassoc nnan ninf nsz arcp afn <4 x half> @llvm.[[TARGET]].nclamp.v4f16
+// NO_HALF: define [[FNATTRS]] [[FFNATTRS]] <4 x float> {{.*}}test_clamp_half4_mismatch
+// NO_HALF: call reassoc nnan ninf nsz arcp afn <4 x float> @llvm.[[TARGET]].nclamp.v4f32(
+half4 test_clamp_half4_mismatch(half4 p0, half p1) { return clamp(p0, p0,p1); }
// CHECK: define [[FNATTRS]] [[FFNATTRS]] float @_Z16test_clamp_float
// CHECK: call reassoc nnan ninf nsz arcp afn float @llvm.[[TARGET]].nclamp.f32(
@@ -128,6 +151,9 @@ float3 test_clamp_float3(float3 p0, float3 p1) { return clamp(p0, p1,p1); }
// CHECK: define [[FNATTRS]] [[FFNATTRS]] <4 x float> @_Z17test_clamp_float4
// CHECK: call reassoc nnan ninf nsz arcp afn <4 x float> @llvm.[[TARGET]].nclamp.v4f32
float4 test_clamp_float4(float4 p0, float4 p1) { return clamp(p0, p1,p1); }
+// CHECK: define [[FNATTRS]] [[FFNATTRS]] <4 x float> {{.*}}test_clamp_float4_mismatch
+// CHECK: call reassoc nnan ninf nsz arcp afn <4 x float> @llvm.[[TARGET]].nclamp.v4f32
+float4 test_clamp_float4_mismatch(float4 p0, float p1) { return clamp(p0, p0,p1); }
// CHECK: define [[FNATTRS]] [[FFNATTRS]] double @_Z17test_clamp_double
// CHECK: call reassoc nnan ninf nsz arcp afn double @llvm.[[TARGET]].nclamp.f64(
@@ -141,3 +167,9 @@ double3 test_clamp_double3(double3 p0, double3 p1) { return clamp(p0, p1,p1); }
// CHECK: define [[FNATTRS]] [[FFNATTRS]] <4 x double> @_Z18test_clamp_double4
// CHECK: call reassoc nnan ninf nsz arcp afn <4 x double> @llvm.[[TARGET]].nclamp.v4f64
double4 test_clamp_double4(double4 p0, double4 p1) { return clamp(p0, p1,p1); }
+// CHECK: define [[FNATTRS]] [[FFNATTRS]] <4 x double> {{.*}}test_clamp_double4_mismatch
+// CHECK: call reassoc nnan ninf nsz arcp afn <4 x double> @llvm.[[TARGET]].nclamp.v4f64
+double4 test_clamp_double4_mismatch(double4 p0, double p1) { return clamp(p0, p0,p1); }
+// CHECK: define [[FNATTRS]] [[FFNATTRS]] <4 x double> {{.*}}test_clamp_double4_mismatch2
+// CHECK: call reassoc nnan ninf nsz arcp afn <4 x double> @llvm.[[TARGET]].nclamp.v4f64
+double4 test_clamp_double4_mismatch2(double4 p0, double p1) { return clamp(p0, p1,p0); }
diff --git a/clang/test/SemaHLSL/BuiltIns/clamp-errors.hlsl b/clang/test/SemaHLSL/BuiltIns/clamp-errors.hlsl
index 036f04cdac0b5..a1850d47b105d 100644
--- a/clang/test/SemaHLSL/BuiltIns/clamp-errors.hlsl
+++ b/clang/test/SemaHLSL/BuiltIns/clamp-errors.hlsl
@@ -22,7 +22,7 @@ float2 test_clamp_no_second_arg(float2 p0) {
float2 test_clamp_vector_size_mismatch(float3 p0, float2 p1) {
return clamp(p0, p0, p1);
- // expected-warning at -1 {{implicit conversion truncates vector: 'float3' (aka 'vector<float, 3>') to 'vector<float, 2>' (vector of 2 'float' values)}}
+ // expected-error at -1 {{call to 'clamp' is ambiguous}}
}
float2 test_clamp_builtin_vector_size_mismatch(float3 p0, float2 p1) {
More information about the cfe-commits
mailing list