[OpenCL] Add integer functions to builtin functions

[svenvh]

Summary

This patch adds the integer builtin functions from the OpenCL C
specification.

[Anastasia]

LGTM! Thanks!

clang/lib/Sema/OpenCLBuiltins.td

 def VecAndScalar: IntList<"VecAndScalar", [1, 2, 3, 4, 8, 16]>;
 def VecNoScalar : IntList<"VecNoScalar", [2, 3, 4, 8, 16]>;
 def Vec1        : IntList<"Vec1", [1]>;
 
 // Type lists.
 def TLAll   : TypeList<"TLAll", [Char, UChar, Short, UShort, Int, UInt, Long, ULong, Float, Double, Half]>;
 def TLFloat : TypeList<"TLFloat", [Float, Double, Half]>;
 
+// All unsigned integer types twice, to facilitate unsigned return types for e.g.
+// uchar abs(char) and
+// uchar abs(uchar).
+def TLAllUIntsTwice : TypeList<"TLAllUIntsTwice", [UChar, UChar, UShort, UShort, UInt, UInt, ULong, ULong]>;
+
 def TLAllInts : TypeList<"TLAllInts", [Char, UChar, Short, UShort, Int, UInt, Long, ULong]>;
 
 // GenType definitions for multiple base types (e.g. all floating point types,
 // or all integer types).
 // All types
 def AGenTypeN              : GenericType<"AGenTypeN", TLAll, VecAndScalar>;
 def AGenTypeNNoScalar      : GenericType<"AGenTypeNNoScalar", TLAll, VecNoScalar>;
 // All integer
 def AIGenType1             : GenericType<"AIGenType1", TLAllInts, Vec1>;
 def AIGenTypeN             : GenericType<"AIGenTypeN", TLAllInts, VecAndScalar>;
 def AIGenTypeNNoScalar     : GenericType<"AIGenTypeNNoScalar", TLAllInts, VecNoScalar>;
+// All integer to unsigned
+def AI2UGenTypeN           : GenericType<"AI2UGenTypeN", TLAllUIntsTwice, VecAndScalar>;
 // Float
 def FGenTypeN              : GenericType<"FGenTypeN", TLFloat, VecAndScalar>;
 
 // 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 name = ["half_cos",
   def : Builtin<name, [GenTypeFloatVecAndScalar, GenTypeFloatVecAndScalar], Attr.Const>;
 }
 foreach name = ["half_divide", "half_powr",
                 "native_divide", "native_powr"] in {
   def : Builtin<name, [GenTypeFloatVecAndScalar, GenTypeFloatVecAndScalar, GenTypeFloatVecAndScalar], Attr.Const>;
 }
 
 //--------------------------------------------------------------------
+// OpenCL v1.1 s6.11.3, v1.2 s6.12.3, v2.0 s6.13.3 - Integer Functions
+// --- Table 10 ---
+// --- 1 argument ---
+foreach name = ["abs"] in {
+  def : Builtin<name, [AI2UGenTypeN, AIGenTypeN], Attr.Const>;
+}
+foreach name = ["clz", "popcount"] in {
+  def : Builtin<name, [AIGenTypeN, AIGenTypeN], Attr.Const>;
+}
+let MinVersion = CL20 in {
+  foreach name = ["ctz"] in {
+    def : Builtin<name, [AIGenTypeN, AIGenTypeN]>;
+  }
+}
+
+// --- 2 arguments ---
+foreach name = ["abs_diff"] in {
+  def : Builtin<name, [AI2UGenTypeN, AIGenTypeN, AIGenTypeN], Attr.Const>;
+}
+foreach name = ["add_sat", "hadd", "rhadd", "mul_hi", "rotate", "sub_sat"] in {
+  def : Builtin<name, [AIGenTypeN, AIGenTypeN, AIGenTypeN], Attr.Const>;
+}
+foreach name = ["max", "min"] in {
+  def : Builtin<name, [AIGenTypeN, AIGenTypeN, AIGenTypeN], Attr.Const>;
+  def : Builtin<name, [AIGenTypeNNoScalar, AIGenTypeNNoScalar, AIGenType1], Attr.Const>;
+}
+foreach name = ["upsample"] in {
+  def : Builtin<name, [GenTypeShortVecAndScalar, GenTypeCharVecAndScalar, GenTypeUCharVecAndScalar], Attr.Const>;
+  def : Builtin<name, [GenTypeUShortVecAndScalar, GenTypeUCharVecAndScalar, GenTypeUCharVecAndScalar], Attr.Const>;
+  def : Builtin<name, [GenTypeIntVecAndScalar, GenTypeShortVecAndScalar, GenTypeUShortVecAndScalar], Attr.Const>;
+  def : Builtin<name, [GenTypeUIntVecAndScalar, GenTypeUShortVecAndScalar, GenTypeUShortVecAndScalar], Attr.Const>;
+  def : Builtin<name, [GenTypeLongVecAndScalar, GenTypeIntVecAndScalar, GenTypeUIntVecAndScalar], Attr.Const>;
+  def : Builtin<name, [GenTypeULongVecAndScalar, GenTypeUIntVecAndScalar, GenTypeUIntVecAndScalar], Attr.Const>;
+}
+
+// --- 3 arguments ---
+foreach name = ["clamp"] in {
+  def : Builtin<name, [AIGenTypeN, AIGenTypeN, AIGenTypeN, AIGenTypeN], Attr.Const>;
+  def : Builtin<name, [AIGenTypeNNoScalar, AIGenTypeNNoScalar, AIGenType1, AIGenType1], Attr.Const>;
+}
+foreach name = ["mad_hi", "mad_sat"] in {
+  def : Builtin<name, [AIGenTypeN, AIGenTypeN, AIGenTypeN, AIGenTypeN], Attr.Const>;
+}
+
+// --- Table 11 ---
+foreach name = ["mad24"] in {
+  def : Builtin<name, [GenTypeIntVecAndScalar, GenTypeIntVecAndScalar, GenTypeIntVecAndScalar, GenTypeIntVecAndScalar], Attr.Const>;
+  def : Builtin<name, [GenTypeUIntVecAndScalar, GenTypeUIntVecAndScalar, GenTypeUIntVecAndScalar, GenTypeUIntVecAndScalar], Attr.Const>;
+}
+foreach name = ["mul24"] in {
+  def : Builtin<name, [GenTypeIntVecAndScalar, GenTypeIntVecAndScalar, GenTypeIntVecAndScalar], Attr.Const>;
+  def : Builtin<name, [GenTypeUIntVecAndScalar, GenTypeUIntVecAndScalar, GenTypeUIntVecAndScalar], Attr.Const>;
+}
+
+//--------------------------------------------------------------------
 // OpenCL v1.1 s6.11.4, v1.2 s6.12.4, v2.0 s6.13.4 - Common Functions
 // OpenCL Extension v2.0 s5.1.3 and s6.1.3 - Common Functions
 // --- Table 12 ---
 // --- 1 argument ---
 foreach name = ["degrees", "radians", "sign"] in {
   def : Builtin<name, [FGenTypeN, FGenTypeN], Attr.Const>;
 }
 
foreach Type = [Int, UInt] in {
   foreach name = ["atom_inc", "atom_dec"] in {
     def : Builtin<name, [Type, PointerType<VolatileType<Type>, GlobalAS>]>;
   }
   foreach name = ["atom_cmpxchg"] in {
     def : Builtin<name, [Type, PointerType<VolatileType<Type>, GlobalAS>, Type, Type]>;
   }
 }
 
-// OpenCL v1.1 s6.11.3, v1.2 s6.12.3, v2.0 s6.13.3 - Integer Functions
-foreach name = ["max", "min"] in {
-  def : Builtin<name, [AIGenTypeN, AIGenTypeN, AIGenTypeN], Attr.Const>;
-  def : Builtin<name, [AIGenTypeNNoScalar, AIGenTypeNNoScalar, AIGenType1], Attr.Const>;
-}
-
 //--------------------------------------------------------------------
 // OpenCL v1.1 s6.11.3, v1.2 s6.12.14, v2.0 s6.13.14: Image Read and Write Functions
 // OpenCL Extension v2.0 s5.1.8 and s6.1.8: Image Read and Write Functions
 // --- Table 22: Image Read Functions with Samplers ---
 foreach imgTy = [Image1d] in {
   foreach coordTy = [Int, Float] in {
     def : Builtin<"read_imagef", [VectorType<Float, 4>, ImageType<imgTy, "RO">, Sampler, coordTy], Attr.Pure>;
     def : Builtin<"read_imagei", [VectorType<Int, 4>, ImageType<imgTy, "RO">, Sampler, coordTy], Attr.Pure>;

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

 
 #pragma OPENCL EXTENSION cl_khr_fp16 : enable
 #if __OPENCL_C_VERSION__ < CL_VERSION_1_2
 #pragma OPENCL EXTENSION cl_khr_fp64 : enable
 #endif
 
 // Provide typedefs when invoking clang without -finclude-default-header.
 #ifdef NO_HEADER
+typedef unsigned char uchar;
+typedef unsigned int uint;
+typedef unsigned long ulong;
+typedef unsigned short ushort;
+typedef __SIZE_TYPE__ size_t;
 typedef char char2 __attribute__((ext_vector_type(2)));
 typedef char char4 __attribute__((ext_vector_type(4)));
+typedef uchar uchar4 __attribute__((ext_vector_type(4)));
 typedef float float4 __attribute__((ext_vector_type(4)));
 typedef half half4 __attribute__((ext_vector_type(4)));
 typedef int int2 __attribute__((ext_vector_type(2)));
 typedef int int4 __attribute__((ext_vector_type(4)));
 typedef long long2 __attribute__((ext_vector_type(2)));
-typedef unsigned char uchar;
-typedef unsigned int uint;
-typedef unsigned long ulong;
-typedef unsigned short ushort;
-typedef __SIZE_TYPE__ size_t;
 #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_conversion() {
   d = convert_double_sat_rtp(f);
   l2 = convert_long2_rtz(c2);
   i4 = convert_int4_sat(f4);
 }
 
 char4 test_int(char c, char4 c4) {
   char m = max(c, c);
   char4 m4 = max(c4, c4);
+  uchar4 abs1 = abs(c4);
+  uchar4 abs2 = abs(abs1);
   return max(c4, c);
 }
 
 kernel void basic_image_readonly(read_only image2d_t image_read_only_image2d) {
   int2 i2;
   sampler_t sampler;
   half4 res;
   float4 resf;