[OpenCL] Add support of language builtins for OpenCL C 3.0

OpenCL C 3.0 introduces optionality to some builtins, in particularly
to those which are conditionally supported with pipe, device enqueue
and generic address space features.

The idea is to conditionally support such builtins depending on the language options
being set for a certain feature. This allows users to define functions with names
of those optional builtins in OpenCL (as such names are not reserved).

Reviewed By: Anastasia

Differential Revision: https://reviews.llvm.org/D118605

clang/include/clang/Basic/Builtins.def

@@ -1627,50 +1627,50 @@ LANGBUILTIN(__builtin_coro_suspend, "cIb", "n", COR_LANG)
 
 // OpenCL v2.0 s6.13.16, s9.17.3.5 - Pipe functions.
 // We need the generic prototype, since the packet type could be anything.
-LANGBUILTIN(read_pipe, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(write_pipe, "i.", "tn", OCLC20_LANG)
+LANGBUILTIN(read_pipe, "i.", "tn", OCL_PIPE)
+LANGBUILTIN(write_pipe, "i.", "tn", OCL_PIPE)
 
-LANGBUILTIN(reserve_read_pipe, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(reserve_write_pipe, "i.", "tn", OCLC20_LANG)
+LANGBUILTIN(reserve_read_pipe, "i.", "tn", OCL_PIPE)
+LANGBUILTIN(reserve_write_pipe, "i.", "tn", OCL_PIPE)
 
-LANGBUILTIN(commit_write_pipe, "v.", "tn", OCLC20_LANG)
-LANGBUILTIN(commit_read_pipe, "v.", "tn", OCLC20_LANG)
+LANGBUILTIN(commit_write_pipe, "v.", "tn", OCL_PIPE)
+LANGBUILTIN(commit_read_pipe, "v.", "tn", OCL_PIPE)
 
-LANGBUILTIN(sub_group_reserve_read_pipe, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(sub_group_reserve_write_pipe, "i.", "tn", OCLC20_LANG)
+LANGBUILTIN(sub_group_reserve_read_pipe, "i.", "tn", OCL_PIPE)
+LANGBUILTIN(sub_group_reserve_write_pipe, "i.", "tn", OCL_PIPE)
 
-LANGBUILTIN(sub_group_commit_read_pipe, "v.", "tn", OCLC20_LANG)
-LANGBUILTIN(sub_group_commit_write_pipe, "v.", "tn", OCLC20_LANG)
+LANGBUILTIN(sub_group_commit_read_pipe, "v.", "tn", OCL_PIPE)
+LANGBUILTIN(sub_group_commit_write_pipe, "v.", "tn", OCL_PIPE)
 
-LANGBUILTIN(work_group_reserve_read_pipe, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(work_group_reserve_write_pipe, "i.", "tn", OCLC20_LANG)
+LANGBUILTIN(work_group_reserve_read_pipe, "i.", "tn", OCL_PIPE)
+LANGBUILTIN(work_group_reserve_write_pipe, "i.", "tn", OCL_PIPE)
 
-LANGBUILTIN(work_group_commit_read_pipe, "v.", "tn", OCLC20_LANG)
-LANGBUILTIN(work_group_commit_write_pipe, "v.", "tn", OCLC20_LANG)
+LANGBUILTIN(work_group_commit_read_pipe, "v.", "tn", OCL_PIPE)
+LANGBUILTIN(work_group_commit_write_pipe, "v.", "tn", OCL_PIPE)
 
-LANGBUILTIN(get_pipe_num_packets, "Ui.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_pipe_max_packets, "Ui.", "tn", OCLC20_LANG)
+LANGBUILTIN(get_pipe_num_packets, "Ui.", "tn", OCL_PIPE)
+LANGBUILTIN(get_pipe_max_packets, "Ui.", "tn", OCL_PIPE)
 
 // OpenCL v2.0 s6.13.17 - Enqueue kernel functions.
 // Custom builtin check allows to perform special check of passed block arguments.
-LANGBUILTIN(enqueue_kernel, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_kernel_work_group_size, "Ui.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_kernel_preferred_work_group_size_multiple, "Ui.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_kernel_max_sub_group_size_for_ndrange, "Ui.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_kernel_sub_group_count_for_ndrange, "Ui.", "tn", OCLC20_LANG)
+LANGBUILTIN(enqueue_kernel, "i.", "tn", OCL_DSE)
+LANGBUILTIN(get_kernel_work_group_size, "Ui.", "tn", OCL_DSE)
+LANGBUILTIN(get_kernel_preferred_work_group_size_multiple, "Ui.", "tn", OCL_DSE)
+LANGBUILTIN(get_kernel_max_sub_group_size_for_ndrange, "Ui.", "tn", OCL_DSE)
+LANGBUILTIN(get_kernel_sub_group_count_for_ndrange, "Ui.", "tn", OCL_DSE)
 
 // OpenCL v2.0 s6.13.9 - Address space qualifier functions.
 // FIXME: Pointer parameters of OpenCL builtins should have their address space
 // requirement defined.
-LANGBUILTIN(to_global, "v*v*", "tn", OCLC20_LANG)
-LANGBUILTIN(to_local, "v*v*", "tn", OCLC20_LANG)
-LANGBUILTIN(to_private, "v*v*", "tn", OCLC20_LANG)
+LANGBUILTIN(to_global, "v*v*", "tn", OCL_GAS)
+LANGBUILTIN(to_local, "v*v*", "tn", OCL_GAS)
+LANGBUILTIN(to_private, "v*v*", "tn", OCL_GAS)
 
 // OpenCL half load/store builtin
-LANGBUILTIN(__builtin_store_half, "vdh*", "n", ALL_OCLC_LANGUAGES)
-LANGBUILTIN(__builtin_store_halff, "vfh*", "n", ALL_OCLC_LANGUAGES)
-LANGBUILTIN(__builtin_load_half, "dhC*", "nc", ALL_OCLC_LANGUAGES)
-LANGBUILTIN(__builtin_load_halff, "fhC*", "nc", ALL_OCLC_LANGUAGES)
+LANGBUILTIN(__builtin_store_half, "vdh*", "n", ALL_OCL_LANGUAGES)
+LANGBUILTIN(__builtin_store_halff, "vfh*", "n", ALL_OCL_LANGUAGES)
+LANGBUILTIN(__builtin_load_half, "dhC*", "nc", ALL_OCL_LANGUAGES)
+LANGBUILTIN(__builtin_load_halff, "fhC*", "nc", ALL_OCL_LANGUAGES)
 
 // Builtins for os_log/os_trace
 BUILTIN(__builtin_os_log_format_buffer_size, "zcC*.", "p:0:nut")

clang/include/clang/Basic/Builtins.h

@@ -28,20 +28,21 @@ class IdentifierTable;
 class LangOptions;
 
 enum LanguageID {
-  GNU_LANG = 0x1,     // builtin requires GNU mode.
-  C_LANG = 0x2,       // builtin for c only.
-  CXX_LANG = 0x4,     // builtin for cplusplus only.
-  OBJC_LANG = 0x8,    // builtin for objective-c and objective-c++
-  MS_LANG = 0x10,     // builtin requires MS mode.
-  OCLC20_LANG = 0x20, // builtin for OpenCL C 2.0 only.
-  OCLC1X_LANG = 0x40, // builtin for OpenCL C 1.x only.
-  OMP_LANG = 0x80,    // builtin requires OpenMP.
-  CUDA_LANG = 0x100,  // builtin requires CUDA.
-  COR_LANG = 0x200,   // builtin requires use of 'fcoroutine-ts' option.
+  GNU_LANG = 0x1,            // builtin requires GNU mode.
+  C_LANG = 0x2,              // builtin for c only.
+  CXX_LANG = 0x4,            // builtin for cplusplus only.
+  OBJC_LANG = 0x8,           // builtin for objective-c and objective-c++
+  MS_LANG = 0x10,            // builtin requires MS mode.
+  OMP_LANG = 0x20,           // builtin requires OpenMP.
+  CUDA_LANG = 0x40,          // builtin requires CUDA.
+  COR_LANG = 0x80,           // builtin requires use of 'fcoroutine-ts' option.
+  OCL_GAS = 0x100,           // builtin requires OpenCL generic address space.
+  OCL_PIPE = 0x200,          // builtin requires OpenCL pipe.
+  OCL_DSE = 0x400,           // builtin requires OpenCL device side enqueue.
+  ALL_OCL_LANGUAGES = 0x800, // builtin for OCL languages.
   ALL_LANGUAGES = C_LANG | CXX_LANG | OBJC_LANG, // builtin for all languages.
   ALL_GNU_LANGUAGES = ALL_LANGUAGES | GNU_LANG,  // builtin requires GNU mode.
-  ALL_MS_LANGUAGES = ALL_LANGUAGES | MS_LANG,    // builtin requires MS mode.
-  ALL_OCLC_LANGUAGES = OCLC1X_LANG | OCLC20_LANG // builtin for OCLC languages.
+  ALL_MS_LANGUAGES = ALL_LANGUAGES | MS_LANG     // builtin requires MS mode.
 };
 
 namespace Builtin {

clang/lib/Basic/Builtins.cpp

@@ -69,22 +69,26 @@ bool Builtin::Context::builtinIsSupported(const Builtin::Info &BuiltinInfo,
   bool MSModeUnsupported =
       !LangOpts.MicrosoftExt && (BuiltinInfo.Langs & MS_LANG);
   bool ObjCUnsupported = !LangOpts.ObjC && BuiltinInfo.Langs == OBJC_LANG;
-  bool OclC1Unsupported = (LangOpts.OpenCLVersion / 100) != 1 &&
-                          (BuiltinInfo.Langs & ALL_OCLC_LANGUAGES ) ==  OCLC1X_LANG;
-  bool OclC2Unsupported =
-      (LangOpts.getOpenCLCompatibleVersion() != 200) &&
-      (BuiltinInfo.Langs & ALL_OCLC_LANGUAGES) == OCLC20_LANG;
-  bool OclCUnsupported = !LangOpts.OpenCL &&
-                         (BuiltinInfo.Langs & ALL_OCLC_LANGUAGES);
+  bool OclCUnsupported =
+      !LangOpts.OpenCL && (BuiltinInfo.Langs & ALL_OCL_LANGUAGES);
+  bool OclGASUnsupported =
+      !LangOpts.OpenCLGenericAddressSpace && (BuiltinInfo.Langs & OCL_GAS);
+  bool OclPipeUnsupported =
+      !LangOpts.OpenCLPipes && (BuiltinInfo.Langs & OCL_PIPE);
+  // Device side enqueue is not supported until OpenCL 2.0. In 2.0 and higher
+  // support is indicated with language option for blocks.
+  bool OclDSEUnsupported =
+      (LangOpts.getOpenCLCompatibleVersion() < 200 || !LangOpts.Blocks) &&
+      (BuiltinInfo.Langs & OCL_DSE);
   bool OpenMPUnsupported = !LangOpts.OpenMP && BuiltinInfo.Langs == OMP_LANG;
   bool CUDAUnsupported = !LangOpts.CUDA && BuiltinInfo.Langs == CUDA_LANG;
   bool CPlusPlusUnsupported =
       !LangOpts.CPlusPlus && BuiltinInfo.Langs == CXX_LANG;
   return !BuiltinsUnsupported && !CorBuiltinsUnsupported &&
-         !MathBuiltinsUnsupported && !OclCUnsupported && !OclC1Unsupported &&
-         !OclC2Unsupported && !OpenMPUnsupported && !GnuModeUnsupported &&
-         !MSModeUnsupported && !ObjCUnsupported && !CPlusPlusUnsupported &&
-         !CUDAUnsupported;
+         !MathBuiltinsUnsupported && !OclCUnsupported && !OclGASUnsupported &&
+         !OclPipeUnsupported && !OclDSEUnsupported && !OpenMPUnsupported &&
+         !GnuModeUnsupported && !MSModeUnsupported && !ObjCUnsupported &&
+         !CPlusPlusUnsupported && !CUDAUnsupported;
 }
 
 /// initializeBuiltins - Mark the identifiers for all the builtins with their

clang/test/CodeGenOpenCL/cl20-device-side-enqueue.cl

@@ -1,6 +1,9 @@
 // RUN: %clang_cc1 -disable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir-unknown-unknown" | FileCheck %s --check-prefix=COMMON --check-prefix=B32
 // RUN: %clang_cc1 -disable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefix=COMMON --check-prefix=B64
 // RUN: %clang_cc1 -disable-noundef-analysis %s -cl-std=CL2.0 -ffake-address-space-map -O1 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefix=CHECK-LIFETIMES
+// RUN: %clang_cc1 -disable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir-unknown-unknown" | FileCheck %s --check-prefix=COMMON --check-prefix=B32
+// RUN: %clang_cc1 -disable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefix=COMMON --check-prefix=B64
+// RUN: %clang_cc1 -disable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O1 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefix=CHECK-LIFETIMES
 
 #pragma OPENCL EXTENSION cl_khr_subgroups : enable
 

clang/test/CodeGenOpenCL/pipe_types.cl

@@ -1,4 +1,4 @@
-// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -O0 -cl-std=CL2.0 -DTEST_STRUCT -o - %s | FileCheck --check-prefixes=CHECK,CHECK-STRUCT %s
+// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -O0 -cl-std=CL2.0 -o - %s | FileCheck --check-prefixes=CHECK,CHECK-STRUCT %s
 // RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -O0 -cl-std=CL3.0 -cl-ext=-all,+__opencl_c_pipes,+__opencl_c_generic_address_space,+__opencl_c_program_scope_global_variables -o - %s | FileCheck --check-prefixes=CHECK %s
 // RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -O0 -cl-std=CL3.0 -cl-ext=-all,+__opencl_c_pipes,+__opencl_c_generic_address_space -o - %s | FileCheck --check-prefixes=CHECK %s
 // RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -O0 -cl-std=clc++2021 -cl-ext=-all,+__opencl_c_pipes,+__opencl_c_generic_address_space,+__opencl_c_program_scope_global_variables -o - %s | FileCheck --check-prefixes=CHECK %s
@@ -36,8 +36,6 @@ kernel void test6(MyPipe p) {
 // CHECK: define{{.*}} spir_kernel void @test6(%opencl.pipe_ro_t* %p)
 }
 
-#ifdef TEST_STRUCT
-// FIXME: not supported for OpenCL C 3.0 as language built-ins not supported yet
 struct Person {
   const char *Name;
   bool isFemale;
@@ -52,4 +50,3 @@ void test_reserved_read_pipe(global struct Person *SDst,
   read_pipe (SPipe, SDst);
   // CHECK-STRUCT: call i32 @__read_pipe_2(%opencl.pipe_ro_t* %{{.*}}, i8* %{{.*}}, i32 16, i32 8)
 }
-#endif // TEST_STRUCT

clang/test/CodeGenOpenCL/to_addr_builtin.cl

@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -triple spir-unknown-unknown -emit-llvm -O0 -cl-std=clc++ -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple spir-unknown-unknown -emit-llvm -O0 -cl-std=cl2.0 -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple spir-unknown-unknown -emit-llvm -O0 -cl-std=cl3.0 -o - %s | FileCheck %s
 
 // CHECK: %[[A:.*]] = type { float, float, float }
 typedef struct {

clang/test/SemaOpenCL/cl20-device-side-enqueue.cl

@@ -3,6 +3,11 @@
 // RUN: %clang_cc1 %s -cl-std=CL2.0 -triple "spir-unknown-unknown" -verify -pedantic -fsyntax-only -DB32 -DQUALS="const volatile"
 // RUN: %clang_cc1 %s -cl-std=CL2.0 -triple "spir64-unknown-unknown" -verify -pedantic -fsyntax-only -Wconversion -DWCONV -DQUALS=
 // RUN: %clang_cc1 %s -cl-std=CL2.0 -triple "spir64-unknown-unknown" -verify -pedantic -fsyntax-only -Wconversion -DWCONV -DQUALS="const volatile"
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -triple "spir-unknown-unknown" -verify -pedantic -fsyntax-only -DB32 -DQUALS=
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -triple "spir-unknown-unknown" -verify -pedantic -fsyntax-only -DB32 -DQUALS= -cl-ext=-cl_khr_subgroups
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -triple "spir-unknown-unknown" -verify -pedantic -fsyntax-only -DB32 -DQUALS="const volatile"
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -triple "spir64-unknown-unknown" -verify -pedantic -fsyntax-only -Wconversion -DWCONV -DQUALS=
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -triple "spir64-unknown-unknown" -verify -pedantic -fsyntax-only -Wconversion -DWCONV -DQUALS="const volatile"
 
 typedef struct {int a;} ndrange_t;
 // Diagnostic tests for different overloads of enqueue_kernel from Table 6.13.17.1 of OpenCL 2.0 Spec.

clang/test/SemaOpenCL/clang-builtin-version.cl

@@ -1,6 +1,8 @@
 // RUN: %clang_cc1 %s -fblocks -verify -pedantic -fsyntax-only -ferror-limit 100
+// RUN: %clang_cc1 %s -fblocks -verify -pedantic -fsyntax-only -ferror-limit 100 -cl-std=CL1.2
+// RUN: %clang_cc1 %s -fblocks -verify -pedantic -fsyntax-only -ferror-limit 100 -cl-std=CL3.0 -cl-ext=-__opencl_c_device_enqueue,-__opencl_c_generic_address_space,-__opencl_c_pipes
 
-// Confirm CL2.0 Clang builtins are not available in earlier versions
+// Confirm CL2.0 Clang builtins are not available in earlier versions and in OpenCL C 3.0 without required features.
 
 kernel void dse_builtins() {
   int tmp;
@@ -13,6 +15,11 @@ kernel void dse_builtins() {
   size = get_kernel_preferred_work_group_size_multiple(^(void) { // expected-error{{implicit declaration of function 'get_kernel_preferred_work_group_size_multiple' is invalid in OpenCL}}
     return;
   });
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0) && !defined(__opencl_c_device_enqueue)
+// expected-error@-10{{support disabled - compile with -fblocks or for OpenCL C 2.0 or OpenCL C 3.0 with __opencl_c_device_enqueue feature}}
+// expected-error@-8{{support disabled - compile with -fblocks or for OpenCL C 2.0 or OpenCL C 3.0 with __opencl_c_device_enqueue feature}}
+// expected-error@-6{{support disabled - compile with -fblocks or for OpenCL C 2.0 or OpenCL C 3.0 with __opencl_c_device_enqueue feature}}
+#endif
 }
 
 void pipe_builtins() {