[OpenCL] Add cl_khr_subgroup_extended_types to TableGen BIFs

[svenvh]

Summary

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

[Anastasia]

LGTM! Thanks

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 FuncExtKhrSubgroupExtendedTypes      : FunctionExtension<"cl_khr_subgroup_extended_types">;
 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 FuncExtKhrSubgroupClusteredReduce    : FunctionExtension<"cl_khr_subgroup_clustered_reduce">;
 def FuncExtKhrGlobalInt32BaseAtomics     : FunctionExtension<"cl_khr_global_int32_base_atomics">;
 def FuncExtKhrGlobalInt32ExtendedAtomics : FunctionExtension<"cl_khr_global_int32_extended_atomics">;

 // Signed integer
 def SGenTypeN              : GenericType<"SGenTypeN", TLSignedInts, VecAndScalar>;
 // Unsigned integer
 def UGenTypeN              : GenericType<"UGenTypeN", TLUnsignedInts, VecAndScalar>;
 // Float
 def FGenTypeN              : GenericType<"FGenTypeN", TLFloat, VecAndScalar>;
 // (u)int, (u)long, and all floats
 def IntLongFloatGenType1   : GenericType<"IntLongFloatGenType1", TLIntLongFloats, Vec1>;
+// (u)char and (u)short
+def CharShortGenType1      : GenericType<"CharShortGenType1",
+                                 TypeList<[Char, UChar, Short, UShort]>, Vec1>;
 
 // GenType definitions for every single base type (e.g. fp32 only).
 // Names are like: GenTypeFloatVecAndScalar.
 foreach Type = [Char, UChar, Short, UShort,
                 Int, UInt, Long, ULong,
                 Float, Double, Half] in {
   foreach VecSizes = [VecAndScalar, VecNoScalar] in {
     def "GenType" # Type # VecSizes :
foreach op = ["add", "min", "max"] in {
       def : Builtin<name # op, [IntLongFloatGenType1, IntLongFloatGenType1], Attr.Convergent>;
     }
   }
 }
 
 // OpenCL Extension v3.0 s38 - Extended Subgroup Functions
 
 // Section 38.4.1 - cl_khr_subgroup_extended_types
-// TODO
+let Extension = FuncExtKhrSubgroupExtendedTypes in {
+  // For sub_group_broadcast, add scalar char, uchar, short, and ushort support,
+  def : Builtin<"sub_group_broadcast", [CharShortGenType1, CharShortGenType1, UInt], Attr.Convergent>;
+  // gentype may additionally be one of the supported built-in vector data types.
+  def : Builtin<"sub_group_broadcast", [AGenTypeNNoScalar, AGenTypeNNoScalar, UInt], Attr.Convergent>;
+
+  foreach name = ["sub_group_reduce_", "sub_group_scan_exclusive_",
+                  "sub_group_scan_inclusive_"] in {
+    foreach op = ["add", "min", "max"] in {
+      def : Builtin<name # op, [CharShortGenType1, CharShortGenType1], Attr.Convergent>;
+    }
+  }
+}
 
 // Section 38.5.1 - cl_khr_subgroup_non_uniform_vote
 let Extension = FuncExtKhrSubgroupNonUniformVote in {
   def : Builtin<"sub_group_elect", [Int]>;
   def : Builtin<"sub_group_non_uniform_all", [Int, Int]>;
   def : Builtin<"sub_group_non_uniform_any", [Int, Int]>;
   def : Builtin<"sub_group_non_uniform_all_equal", [Int, AGenType1]>;
 }

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

 typedef uint uint4 __attribute__((ext_vector_type(4)));
 typedef long long2 __attribute__((ext_vector_type(2)));
 
 typedef uint cl_mem_fence_flags;
 #define CLK_GLOBAL_MEM_FENCE 0x02
 
 // Enable extensions that are enabled in opencl-c-base.h.
 #if (defined(__OPENCL_CPP_VERSION__) || __OPENCL_C_VERSION__ >= 200)
+#define cl_khr_subgroup_extended_types 1
 #define cl_khr_subgroup_ballot 1
 #define cl_khr_subgroup_non_uniform_arithmetic 1
 #define cl_khr_subgroup_clustered_reduce 1
 #endif
 #endif
 
 kernel void test_pointers(volatile global void *global_p, global const int4 *a) {
   int i;

 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, global int *scalar) {
+kernel void extended_subgroup(global uint4 *out, global int *scalar, global char2 *c2) {
   out[0] = get_sub_group_eq_mask();
   scalar[0] = sub_group_non_uniform_scan_inclusive_or(3);
   scalar[1] = sub_group_clustered_reduce_logical_xor(2, 4);
+  *c2 = sub_group_broadcast(*c2, 2);
 #if __OPENCL_C_VERSION__ < CL_VERSION_2_0 && !defined(__OPENCL_CPP_VERSION__)
-  // expected-error@-4{{implicit declaration of function 'get_sub_group_eq_mask' is invalid in OpenCL}}
-  // expected-error@-5{{implicit conversion changes signedness}}
-  // expected-error@-5{{implicit declaration of function 'sub_group_non_uniform_scan_inclusive_or' is invalid in OpenCL}}
-  // expected-error@-5{{implicit declaration of function 'sub_group_clustered_reduce_logical_xor' is invalid in OpenCL}}
+  // expected-error@-5{{implicit declaration of function 'get_sub_group_eq_mask' is invalid in OpenCL}}
+  // expected-error@-6{{implicit conversion changes signedness}}
+  // expected-error@-6{{implicit declaration of function 'sub_group_non_uniform_scan_inclusive_or' is invalid in OpenCL}}
+  // expected-error@-6{{implicit declaration of function 'sub_group_clustered_reduce_logical_xor' is invalid in OpenCL}}
+  // expected-error@-6{{implicit declaration of function 'sub_group_broadcast' is invalid in OpenCL}}
+  // expected-error@-7{{implicit conversion loses integer precision}}
 #endif
 }
 
 kernel void basic_vector_data() {
 #if __OPENCL_C_VERSION__ >= CL_VERSION_2_0
   generic void *generic_p;
 #endif
   constant void *constant_p;