[OpenCL] Add cl_khr_subgroup_non_uniform_arithmetic to TableGen BIFs

[svenvh]

Summary

Add the builtin functions brought by the cl_khr_subgroup_non_uniform_arithmetic extension to -fdeclare-opencl-builtins.

[Anastasia]

LGTM! Thanks

clang/lib/Sema/OpenCLBuiltins.td

1529

There is a bit of repetition with FuncExtKhrSubgroups but I guess it's not worth creating common patterns?

[svenvh]

clang/lib/Sema/OpenCLBuiltins.td

1529

Indeed there is some overlap, but factoring out commonality would increase complexity.

clang/lib/Sema/OpenCLBuiltins.td

 // Extension associated to a builtin function.
 class FunctionExtension<string _Ext> : AbstractExtension<_Ext>;
 
 // FunctionExtension definitions.
 def FuncExtNone                          : FunctionExtension<"">;
 def FuncExtKhrSubgroups                  : FunctionExtension<"cl_khr_subgroups">;
 def FuncExtKhrSubgroupNonUniformVote     : FunctionExtension<"cl_khr_subgroup_non_uniform_vote">;
 def FuncExtKhrSubgroupBallot             : FunctionExtension<"cl_khr_subgroup_ballot">;
+def FuncExtKhrSubgroupNonUniformArithmetic: FunctionExtension<"cl_khr_subgroup_non_uniform_arithmetic">;
 def FuncExtKhrSubgroupShuffle            : FunctionExtension<"cl_khr_subgroup_shuffle">;
 def FuncExtKhrSubgroupShuffleRelative    : FunctionExtension<"cl_khr_subgroup_shuffle_relative">;
 def FuncExtKhrGlobalInt32BaseAtomics     : FunctionExtension<"cl_khr_global_int32_base_atomics">;
 def FuncExtKhrGlobalInt32ExtendedAtomics : FunctionExtension<"cl_khr_global_int32_extended_atomics">;
 def FuncExtKhrLocalInt32BaseAtomics      : FunctionExtension<"cl_khr_local_int32_base_atomics">;
 def FuncExtKhrLocalInt32ExtendedAtomics  : FunctionExtension<"cl_khr_local_int32_extended_atomics">;
 def FuncExtKhrInt64BaseAtomics           : FunctionExtension<"cl_khr_int64_base_atomics">;
 def FuncExtKhrInt64ExtendedAtomics       : FunctionExtension<"cl_khr_int64_extended_atomics">;
let Extension = FuncExtKhrSubgroupBallot in {
   def : Builtin<"sub_group_ballot_find_msb", [UInt, VectorType<UInt, 4>]>;
 
   foreach op = ["eq", "ge", "gt", "le", "lt"] in {
     def : Builtin<"get_sub_group_" # op # "_mask", [VectorType<UInt, 4>], Attr.Const>;
   }
 }
 
 // Section 38.7.1 - cl_khr_subgroup_non_uniform_arithmetic
-// TODO
+let Extension = FuncExtKhrSubgroupNonUniformArithmetic in {
+  foreach name = ["reduce_", "scan_exclusive_", "scan_inclusive_"] in {
+    foreach op = ["add", "min", "max", "mul"] in {
+      def : Builtin<"sub_group_non_uniform_" # name # op, [AGenType1, AGenType1]>;
+    }
+    foreach op = ["and", "or", "xor"] in {
+      def : Builtin<"sub_group_non_uniform_" # name # op, [AIGenType1, AIGenType1]>;
+    }
+    foreach op = ["and", "or", "xor"] in {
+      def : Builtin<"sub_group_non_uniform_" # name # "logical_" # op, [Int, Int]>;
+    }
+  }
+}
 
 // Section 38.8.1 - cl_khr_subgroup_shuffle
 let Extension = FuncExtKhrSubgroupShuffle in {
   def : Builtin<"sub_group_shuffle", [AGenType1, AGenType1, UInt]>;
   def : Builtin<"sub_group_shuffle_xor", [AGenType1, AGenType1, UInt]>;
 }
 
 // Section 38.9.1 - cl_khr_subgroup_shuffle_relative

clang/test/SemaOpenCL/fdeclare-opencl-builtins.cl

 typedef int int2 __attribute__((ext_vector_type(2)));
 typedef int int4 __attribute__((ext_vector_type(4)));
 typedef uint uint4 __attribute__((ext_vector_type(4)));
 typedef long long2 __attribute__((ext_vector_type(2)));
 
 // Enable extensions that are enabled in opencl-c-base.h.
 #if (defined(__OPENCL_CPP_VERSION__) || __OPENCL_C_VERSION__ >= 200)
 #define cl_khr_subgroup_ballot 1
+#define cl_khr_subgroup_non_uniform_arithmetic 1
 #endif
 #endif
 
 kernel void test_pointers(volatile global void *global_p, global const int4 *a) {
   int i;
   unsigned int ui;
 
   prefetch(a, 2);

 kernel void basic_subgroup(global uint *out) {
   out[0] = get_sub_group_size();
 #if __OPENCL_C_VERSION__ <= CL_VERSION_1_2 && !defined(__OPENCL_CPP_VERSION__)
   // expected-error@-2{{implicit declaration of function 'get_sub_group_size' is invalid in OpenCL}}
   // expected-error@-3{{implicit conversion changes signedness}}
 #endif
 }
 
-kernel void extended_subgroup(global uint4 *out) {
+kernel void extended_subgroup(global uint4 *out, global int *scalar) {
   out[0] = get_sub_group_eq_mask();
+  scalar[0] = sub_group_non_uniform_scan_inclusive_or(3);
 #if __OPENCL_C_VERSION__ < CL_VERSION_2_0 && !defined(__OPENCL_CPP_VERSION__)
-  // expected-error@-2{{implicit declaration of function 'get_sub_group_eq_mask' is invalid in OpenCL}}
-  // expected-error@-3{{implicit conversion changes signedness}}
+  // expected-error@-3{{implicit declaration of function 'get_sub_group_eq_mask' is invalid in OpenCL}}
+  // expected-error@-4{{implicit conversion changes signedness}}
+  // expected-error@-4{{implicit declaration of function 'sub_group_non_uniform_scan_inclusive_or' is invalid in OpenCL}}
 #endif
 }
 
 kernel void basic_vector_data() {
 #if __OPENCL_C_VERSION__ >= CL_VERSION_2_0
   generic void *generic_p;
 #endif
   constant void *constant_p;