[HLSL] add extra scalar vector overloads for clamp #129939

spall · 2025-03-05T21:21:00Z

Add additional vector scalar overloads for clamp using templates
Add Tests
fixup tests which have changed.
Closes #128230

llvmbot · 2025-03-05T21:21:32Z

@llvm/pr-subscribers-backend-x86

Author: Sarah Spall (spall)

Changes

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

Full diff: https://github.com/llvm/llvm-project/pull/129939.diff

3 Files Affected:

(modified) clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h (+63-36)
(modified) clang/test/CodeGenHLSL/builtins/clamp.hlsl (+32)
(modified) clang/test/SemaHLSL/BuiltIns/clamp-errors.hlsl (+1-1)

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@-1 {{implicit conversion truncates vector: 'float3' (aka 'vector<float, 3>') to 'vector<float, 2>' (vector of 2 'float' values)}}
+  // expected-error@-1 {{call to 'clamp' is ambiguous}}
 }
 
 float2 test_clamp_builtin_vector_size_mismatch(float3 p0, float2 p1) {

llvmbot · 2025-03-05T21:21:32Z

@llvm/pr-subscribers-clang

Author: Sarah Spall (spall)

Changes

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

Full diff: https://github.com/llvm/llvm-project/pull/129939.diff

3 Files Affected:

(modified) clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h (+63-36)
(modified) clang/test/CodeGenHLSL/builtins/clamp.hlsl (+32)
(modified) clang/test/SemaHLSL/BuiltIns/clamp-errors.hlsl (+1-1)

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@-1 {{implicit conversion truncates vector: 'float3' (aka 'vector<float, 3>') to 'vector<float, 2>' (vector of 2 'float' values)}}
+  // expected-error@-1 {{call to 'clamp' is ambiguous}}
 }
 
 float2 test_clamp_builtin_vector_size_mismatch(float3 p0, float2 p1) {

llvmbot · 2025-03-05T21:21:33Z

@llvm/pr-subscribers-hlsl

Author: Sarah Spall (spall)

Changes

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

Full diff: https://github.com/llvm/llvm-project/pull/129939.diff

3 Files Affected:

(modified) clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h (+63-36)
(modified) clang/test/CodeGenHLSL/builtins/clamp.hlsl (+32)
(modified) clang/test/SemaHLSL/BuiltIns/clamp-errors.hlsl (+1-1)

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@-1 {{implicit conversion truncates vector: 'float3' (aka 'vector<float, 3>') to 'vector<float, 2>' (vector of 2 'float' values)}}
+  // expected-error@-1 {{call to 'clamp' is ambiguous}}
 }
 
 float2 test_clamp_builtin_vector_size_mismatch(float3 p0, float2 p1) {

clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h

bob80905

Ok, given the specific conditions that:

there are 3 parameters
the first parameter should always be the vector type
and there should be at most one base type among all the parameters

This macro should work for that specific case.
I think it may need to be adjusted / more parameters may need to be added to the macro if there are later intrinsics that don't satisfy these specific conditions (that is, if there are 2 base types in the parameters that need to be cast to vector types).
Might need to double check with someone whether the vector base base case is allowed or not. But otherwise LGTM.

… test with new error message

github-actions · 2025-03-10T18:58:41Z

✅ With the latest revision this PR passed the C/C++ code formatter.

clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h

clang/test/SemaHLSL/BuiltIns/clamp-errors.hlsl

spall · 2025-03-10T19:46:04Z

I need to add more tests for certain odd cases.

farzonl · 2025-03-12T00:38:07Z

LGTM thanks for adding the vec5 test case!

clang/test/CodeGenHLSL/builtins/clamp.hlsl

clang/lib/Headers/hlsl/hlsl_compat_overloads.h

…lementwise_clamp, to ensure clamp can't be called on 16 bit types in a shader model before 6.2. Add tests to show this

llvm-beanz

One small nit on formatting, otherwise looks good.

clang/lib/Headers/hlsl/hlsl_compat_overloads.h

clang/test/SemaHLSL/BuiltIns/clamp-errors-half.hlsl

clang/test/SemaHLSL/BuiltIns/clamp-errors-16bit.hlsl

llvmbot added clang Clang issues not falling into any other category backend:X86 clang:headers Headers provided by Clang, e.g. for intrinsics HLSL HLSL Language Support labels Mar 5, 2025

github-project-automation bot added this to HLSL Support Mar 5, 2025

bob80905 reviewed Mar 5, 2025

View reviewed changes

clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h Outdated Show resolved Hide resolved

bob80905 approved these changes Mar 6, 2025

View reviewed changes

spall marked this pull request as draft March 7, 2025 00:15

spall added 4 commits March 10, 2025 08:18

extra scalar vector overloads for clamp

23debaf

make clang format happy

44a41cc

implement clamp overloads with templates instead of macros

ac260e6

update semantic checking for __builtin_hlsl_elementwise_clamp; update…

3311cbd

… test with new error message

spall force-pushed the issue-128230 branch from 3b00312 to 3311cbd Compare March 10, 2025 18:54

make clang format happy

fec28cf

spall marked this pull request as ready for review March 10, 2025 18:55

llvmbot added the clang:frontend Language frontend issues, e.g. anything involving "Sema" label Mar 10, 2025

make clang format happy again

101beb2

farzonl reviewed Mar 10, 2025

View reviewed changes

clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h Outdated Show resolved Hide resolved

farzonl reviewed Mar 10, 2025

View reviewed changes

clang/test/SemaHLSL/BuiltIns/clamp-errors.hlsl Show resolved Hide resolved

farzonl approved these changes Mar 10, 2025

View reviewed changes

spall added 4 commits March 11, 2025 12:42

self review + respond to pr comments

69dab19

add vector size checks to clamp templates

47c90c4

make clang format happy

dd0765a

test for float5 error

819cae5

bob80905 reviewed Mar 13, 2025

View reviewed changes

clang/test/CodeGenHLSL/builtins/clamp.hlsl Outdated Show resolved Hide resolved

clang/test/CodeGenHLSL/builtins/clamp.hlsl Outdated Show resolved Hide resolved

fix two tests

8e01d99

bob80905 approved these changes Mar 13, 2025

View reviewed changes

farzonl reviewed Mar 13, 2025

View reviewed changes

clang/lib/Headers/hlsl/hlsl_compat_overloads.h Show resolved Hide resolved

spall added 2 commits March 14, 2025 17:03

change clamp templates so they call clamp instead of __builtin_hlsl_e…

4ed0507

…lementwise_clamp, to ensure clamp can't be called on 16 bit types in a shader model before 6.2. Add tests to show this

put test for each type in own file

2807d0e

llvm-beanz reviewed Mar 17, 2025

View reviewed changes

clang/lib/Headers/hlsl/hlsl_compat_overloads.h Outdated Show resolved Hide resolved

farzonl reviewed Mar 17, 2025

View reviewed changes

clang/test/SemaHLSL/BuiltIns/clamp-errors-half.hlsl Outdated Show resolved Hide resolved

farzonl reviewed Mar 17, 2025

View reviewed changes

clang/test/SemaHLSL/BuiltIns/clamp-errors-half.hlsl Outdated Show resolved Hide resolved

spall added 2 commits March 17, 2025 08:10

address pr comments

9636c1f

fix typo

65c4967

farzonl reviewed Mar 17, 2025

View reviewed changes

clang/test/SemaHLSL/BuiltIns/clamp-errors-16bit.hlsl Outdated Show resolved Hide resolved

fix return type in test

eaa7cb9

spall merged commit af5abd9 into llvm:main Mar 17, 2025
12 checks passed

damyanp moved this to Closed in HLSL Support Apr 25, 2025

damyanp removed this from HLSL Support Jun 25, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[HLSL] add extra scalar vector overloads for clamp #129939

[HLSL] add extra scalar vector overloads for clamp #129939

Uh oh!

spall commented Mar 5, 2025 •

edited

Loading

Uh oh!

llvmbot commented Mar 5, 2025

Uh oh!

llvmbot commented Mar 5, 2025

Uh oh!

llvmbot commented Mar 5, 2025

Uh oh!

Uh oh!

bob80905 left a comment •

edited

Loading

Uh oh!

github-actions bot commented Mar 10, 2025 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

spall commented Mar 10, 2025

Uh oh!

farzonl commented Mar 12, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

llvm-beanz left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

[HLSL] add extra scalar vector overloads for clamp #129939

[HLSL] add extra scalar vector overloads for clamp #129939

Uh oh!

Conversation

spall commented Mar 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

llvmbot commented Mar 5, 2025

Uh oh!

llvmbot commented Mar 5, 2025

Uh oh!

llvmbot commented Mar 5, 2025

Uh oh!

Uh oh!

bob80905 left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented Mar 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

spall commented Mar 10, 2025

Uh oh!

farzonl commented Mar 12, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

llvm-beanz left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

spall commented Mar 5, 2025 •

edited

Loading

bob80905 left a comment •

edited

Loading

github-actions bot commented Mar 10, 2025 •

edited

Loading