AMDGPU: Make hidden argument metadata consistent with amdgpu-implicitarg-num-bytes attribute

[yaxunl]

lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.cpp

258 ↗

(On Diff #154672)

Do we need to check this?

What if a non-OpenCL program wants implicit kernel argument?

Can we remove the dependence on opencl.ocl.version here?

[arsenm]

lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.cpp

258 ↗

(On Diff #154672)

I thought the whole point was to remove this check

[arsenm]

lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.cpp

384 ↗

(On Diff #154672)

Missing space

[kzhuravl]

Drop checks for OpenCL. I was not aware that it was decided to possibly allow this for non-OpenCL languages.

[kzhuravl]

Add missing space.

[yaxunl]

LGTM. Thanks!

[t-tye]

lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.cpp

258 ↗

(On Diff #154683)

Just checking if this is correct for Mesa too?

[arsenm]

lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.cpp

258 ↗

(On Diff #154683)

Clover doesn't use this attribute. I'm not sure what the current state is, but it follow some frankenstein ABI between HSA and what it was doing before. Since it isn't really maintained I don't think we need to worry too much about it

[t-tye]

lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.cpp

258 ↗

(On Diff #154683)

I thought Mesa was using this attribute and setting it to 16. But @kzhuravl points out that this code is only called for the AMDHSA os and so it is not an issue.

llvm/trunk/lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.h

private:
   void emitKernelArg(const Argument &Arg);
 
   void emitKernelArg(const DataLayout &DL, Type *Ty, ValueKind ValueKind,
                      unsigned PointeeAlign = 0,
                      StringRef Name = "", StringRef TypeName = "",
                      StringRef BaseTypeName = "", StringRef AccQual = "",
                      StringRef TypeQual = "");
 
+  void emitHiddenKernelArgs(const Function &Func);
+
 public:
   MetadataStreamer() = default;
   ~MetadataStreamer() = default;
 
   const Metadata &getHSAMetadata() const {
     return HSAMetadata;
   }
 

llvm/trunk/lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.cpp

 //===--- AMDGPUHSAMetadataStreamer.cpp --------------------------*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 //
 /// \file
 /// AMDGPU HSA Metadata Streamer.
 ///
 //
 //===----------------------------------------------------------------------===//
 
 #include "AMDGPUHSAMetadataStreamer.h"
-#include "AMDGPU.h"
+#include "Utils/AMDGPUBaseInfo.h"
 #include "llvm/ADT/StringSwitch.h"
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/Module.h"
 #include "llvm/Support/raw_ostream.h"
 
 namespace llvm {
 
 static cl::opt<bool> DumpHSAMetadata(
Attrs.mRuntimeHandle =
         Func.getFnAttribute("runtime-handle").getValueAsString().str();
   }
 }
 
 void MetadataStreamer::emitKernelArgs(const Function &Func) {
   for (auto &Arg : Func.args())
     emitKernelArg(Arg);
 
-  // TODO: What about other languages?
-  if (!Func.getParent()->getNamedMetadata("opencl.ocl.version"))
-    return;
-
-  auto &DL = Func.getParent()->getDataLayout();
-  auto Int64Ty = Type::getInt64Ty(Func.getContext());
-
-  emitKernelArg(DL, Int64Ty, ValueKind::HiddenGlobalOffsetX);
-  emitKernelArg(DL, Int64Ty, ValueKind::HiddenGlobalOffsetY);
-  emitKernelArg(DL, Int64Ty, ValueKind::HiddenGlobalOffsetZ);
-
-  auto Int8PtrTy = Type::getInt8PtrTy(Func.getContext(),
-                                      AMDGPUASI.GLOBAL_ADDRESS);
-
-  // Emit "printf buffer" argument if printf is used, otherwise emit dummy
-  // "none" argument.
-  if (Func.getParent()->getNamedMetadata("llvm.printf.fmts"))
-    emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenPrintfBuffer);
-  else
-    emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenNone);
-
-  // Emit "default queue" and "completion action" arguments if enqueue kernel is
-  // used, otherwise emit dummy "none" arguments.
-  if (Func.hasFnAttribute("calls-enqueue-kernel")) {
-    emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenDefaultQueue);
-    emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenCompletionAction);
-  } else {
-    emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenNone);
-    emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenNone);
-  }
+  emitHiddenKernelArgs(Func);
 }
 
 void MetadataStreamer::emitKernelArg(const Argument &Arg) {
   auto Func = Arg.getParent();
   auto ArgNo = Arg.getArgNo();
   const MDNode *Node;
 
   StringRef Name;
auto P = StringSwitch<bool*>(Key)
                  .Case("volatile", &Arg.mIsVolatile)
                  .Case("pipe",     &Arg.mIsPipe)
                  .Default(nullptr);
     if (P)
       *P = true;
   }
 }
 
+void MetadataStreamer::emitHiddenKernelArgs(const Function &Func) {
+  int HiddenArgNumBytes =
+      getIntegerAttribute(Func, "amdgpu-implicitarg-num-bytes", 0);
+
+  if (!HiddenArgNumBytes)
+    return;
+
+  auto &DL = Func.getParent()->getDataLayout();
+  auto Int64Ty = Type::getInt64Ty(Func.getContext());
+
+  if (HiddenArgNumBytes >= 8)
+    emitKernelArg(DL, Int64Ty, ValueKind::HiddenGlobalOffsetX);
+  if (HiddenArgNumBytes >= 16)
+    emitKernelArg(DL, Int64Ty, ValueKind::HiddenGlobalOffsetY);
+  if (HiddenArgNumBytes >= 24)
+    emitKernelArg(DL, Int64Ty, ValueKind::HiddenGlobalOffsetZ);
+
+  auto Int8PtrTy = Type::getInt8PtrTy(Func.getContext(),
+                                      AMDGPUASI.GLOBAL_ADDRESS);
+
+  // Emit "printf buffer" argument if printf is used, otherwise emit dummy
+  // "none" argument.
+  if (HiddenArgNumBytes >= 32) {
+    if (Func.getParent()->getNamedMetadata("llvm.printf.fmts"))
+      emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenPrintfBuffer);
+    else
+      emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenNone);
+  }
+
+  // Emit "default queue" and "completion action" arguments if enqueue kernel is
+  // used, otherwise emit dummy "none" arguments.
+  if (HiddenArgNumBytes >= 48) {
+    if (Func.hasFnAttribute("calls-enqueue-kernel")) {
+      emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenDefaultQueue);
+      emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenCompletionAction);
+    } else {
+      emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenNone);
+      emitKernelArg(DL, Int8PtrTy, ValueKind::HiddenNone);
+    }
+  }
+}
+
 void MetadataStreamer::begin(const Module &Mod) {
   AMDGPUASI = getAMDGPUAS(Mod);
   emitVersion();
   emitPrintf(Mod);
 }
 
 void MetadataStreamer::end() {
   std::string HSAMetadataString;

llvm/trunk/test/CodeGen/AMDGPU/hsa-metadata-enqueue-kernel.ll

 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetY
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NOT:        ValueKind:     HiddenDefaultQueue
 ; CHECK-NOT:        ValueKind:     HiddenCompletionAction
-define amdgpu_kernel void @test_non_enqueue_kernel_caller(i8 %a)
+define amdgpu_kernel void @test_non_enqueue_kernel_caller(i8 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_enqueue_kernel_caller
 ; CHECK-NEXT:   SymbolName:      'test_enqueue_kernel_caller@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       ValueKind:     HiddenDefaultQueue
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenCompletionAction
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_enqueue_kernel_caller(i8 %a) #0
+define amdgpu_kernel void @test_enqueue_kernel_caller(i8 %a) #1
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 {
   ret void
 }
 
-attributes #0 = { "calls-enqueue-kernel" }
+attributes #0 = { "amdgpu-implicitarg-num-bytes"="48" }
+attributes #1 = { "calls-enqueue-kernel" "amdgpu-implicitarg-num-bytes"="48" }
 
 !1 = !{i32 0}
 !2 = !{!"none"}
 !3 = !{!"char"}
 !4 = !{!""}
 
 !opencl.ocl.version = !{!90}
 !90 = !{i32 2, i32 0}
 
-
 ; PARSER: AMDGPU HSA Metadata Parser Test: PASS

llvm/trunk/test/CodeGen/AMDGPU/hsa-metadata-from-llvm-ir-full.ll

 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
 ; CHECK-NOT:        ValueKind:     HiddenDefaultQueue
 ; CHECK-NOT:        ValueKind:     HiddenCompletionAction
-define amdgpu_kernel void @test_char(i8 %a)
+define amdgpu_kernel void @test_char(i8 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !9
     !kernel_arg_base_type !9 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_ushort2
 ; CHECK-NEXT:   SymbolName:      'test_ushort2@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_ushort2(<2 x i16> %a)
+define amdgpu_kernel void @test_ushort2(<2 x i16> %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !10
     !kernel_arg_base_type !10 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_int3
 ; CHECK-NEXT:   SymbolName:      'test_int3@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_int3(<3 x i32> %a)
+define amdgpu_kernel void @test_int3(<3 x i32> %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !11
     !kernel_arg_base_type !11 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_ulong4
 ; CHECK-NEXT:   SymbolName:      'test_ulong4@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_ulong4(<4 x i64> %a)
+define amdgpu_kernel void @test_ulong4(<4 x i64> %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !12
     !kernel_arg_base_type !12 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_half8
 ; CHECK-NEXT:   SymbolName:      'test_half8@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_half8(<8 x half> %a)
+define amdgpu_kernel void @test_half8(<8 x half> %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !13
     !kernel_arg_base_type !13 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_float16
 ; CHECK-NEXT:   SymbolName:      'test_float16@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_float16(<16 x float> %a)
+define amdgpu_kernel void @test_float16(<16 x float> %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !14
     !kernel_arg_base_type !14 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_double16
 ; CHECK-NEXT:   SymbolName:      'test_double16@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_double16(<16 x double> %a)
+define amdgpu_kernel void @test_double16(<16 x double> %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !15
     !kernel_arg_base_type !15 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_pointer
 ; CHECK-NEXT:   SymbolName:      'test_pointer@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_pointer(i32 addrspace(1)* %a)
+define amdgpu_kernel void @test_pointer(i32 addrspace(1)* %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !16
     !kernel_arg_base_type !16 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_image
 ; CHECK-NEXT:   SymbolName:      'test_image@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_image(%opencl.image2d_t addrspace(1)* %a)
+define amdgpu_kernel void @test_image(%opencl.image2d_t addrspace(1)* %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !17
     !kernel_arg_base_type !17 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_sampler
 ; CHECK-NEXT:   SymbolName:      'test_sampler@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_sampler(i32 %a)
+define amdgpu_kernel void @test_sampler(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !18
     !kernel_arg_base_type !18 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_queue
 ; CHECK-NEXT:   SymbolName:      'test_queue@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_queue(%opencl.queue_t addrspace(1)* %a)
+define amdgpu_kernel void @test_queue(%opencl.queue_t addrspace(1)* %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !19
     !kernel_arg_base_type !19 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_struct
 ; CHECK-NEXT:   SymbolName:      'test_struct@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_struct(%struct.A addrspace(5)* byval %a)
+define amdgpu_kernel void @test_struct(%struct.A addrspace(5)* byval %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !20
     !kernel_arg_base_type !20 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_i128
 ; CHECK-NEXT:   SymbolName:      'test_i128@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_i128(i128 %a)
+define amdgpu_kernel void @test_i128(i128 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !21
     !kernel_arg_base_type !21 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_multi_arg
 ; CHECK-NEXT:   SymbolName:      'test_multi_arg@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_multi_arg(i32 %a, <2 x i16> %b, <3 x i8> %c)
+define amdgpu_kernel void @test_multi_arg(i32 %a, <2 x i16> %b, <3 x i8> %c) #0
     !kernel_arg_addr_space !22 !kernel_arg_access_qual !23 !kernel_arg_type !24
     !kernel_arg_base_type !24 !kernel_arg_type_qual !25 {
   ret void
 }
 
 ; CHECK:      - Name:            test_addr_space
 ; CHECK-NEXT:   SymbolName:      'test_addr_space@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
 define amdgpu_kernel void @test_addr_space(i32 addrspace(1)* %g,
                                            i32 addrspace(4)* %c,
-                                           i32 addrspace(3)* %l)
+                                           i32 addrspace(3)* %l) #0
     !kernel_arg_addr_space !50 !kernel_arg_access_qual !23 !kernel_arg_type !51
     !kernel_arg_base_type !51 !kernel_arg_type_qual !25 {
   ret void
 }
 
 ; CHECK:      - Name:            test_type_qual
 ; CHECK-NEXT:   SymbolName:      'test_type_qual@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
 define amdgpu_kernel void @test_type_qual(i32 addrspace(1)* %a,
                                           i32 addrspace(1)* %b,
-                                          %opencl.pipe_t addrspace(1)* %c)
+                                          %opencl.pipe_t addrspace(1)* %c) #0
     !kernel_arg_addr_space !22 !kernel_arg_access_qual !23 !kernel_arg_type !51
     !kernel_arg_base_type !51 !kernel_arg_type_qual !70 {
   ret void
 }
 
 ; CHECK:      - Name:            test_access_qual
 ; CHECK-NEXT:   SymbolName:      'test_access_qual@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
 define amdgpu_kernel void @test_access_qual(%opencl.image1d_t addrspace(1)* %ro,
                                             %opencl.image2d_t addrspace(1)* %wo,
-                                            %opencl.image3d_t addrspace(1)* %rw)
+                                            %opencl.image3d_t addrspace(1)* %rw) #0
     !kernel_arg_addr_space !60 !kernel_arg_access_qual !61 !kernel_arg_type !62
     !kernel_arg_base_type !62 !kernel_arg_type_qual !25 {
   ret void
 }
 
 ; CHECK:      - Name:            test_vec_type_hint_half
 ; CHECK-NEXT:   SymbolName:      'test_vec_type_hint_half@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_vec_type_hint_half(i32 %a)
+define amdgpu_kernel void @test_vec_type_hint_half(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 !vec_type_hint !26 {
   ret void
 }
 
 ; CHECK:      - Name:            test_vec_type_hint_float
 ; CHECK-NEXT:   SymbolName:      'test_vec_type_hint_float@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_vec_type_hint_float(i32 %a)
+define amdgpu_kernel void @test_vec_type_hint_float(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 !vec_type_hint !27 {
   ret void
 }
 
 ; CHECK:      - Name:            test_vec_type_hint_double
 ; CHECK-NEXT:   SymbolName:      'test_vec_type_hint_double@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_vec_type_hint_double(i32 %a)
+define amdgpu_kernel void @test_vec_type_hint_double(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 !vec_type_hint !28 {
   ret void
 }
 
 ; CHECK:      - Name:            test_vec_type_hint_char
 ; CHECK-NEXT:   SymbolName:      'test_vec_type_hint_char@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_vec_type_hint_char(i32 %a)
+define amdgpu_kernel void @test_vec_type_hint_char(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 !vec_type_hint !29 {
   ret void
 }
 
 ; CHECK:      - Name:            test_vec_type_hint_short
 ; CHECK-NEXT:   SymbolName:      'test_vec_type_hint_short@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_vec_type_hint_short(i32 %a)
+define amdgpu_kernel void @test_vec_type_hint_short(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 !vec_type_hint !30 {
   ret void
 }
 
 ; CHECK:      - Name:            test_vec_type_hint_long
 ; CHECK-NEXT:   SymbolName:      'test_vec_type_hint_long@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_vec_type_hint_long(i32 %a)
+define amdgpu_kernel void @test_vec_type_hint_long(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 !vec_type_hint !31 {
   ret void
 }
 
 ; CHECK:      - Name:            test_vec_type_hint_unknown
 ; CHECK-NEXT:   SymbolName:      'test_vec_type_hint_unknown@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_vec_type_hint_unknown(i32 %a)
+define amdgpu_kernel void @test_vec_type_hint_unknown(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 !vec_type_hint !32 {
   ret void
 }
 
 ; CHECK:      - Name:            test_reqd_wgs_vec_type_hint
 ; CHECK-NEXT:   SymbolName:      'test_reqd_wgs_vec_type_hint@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:             8
 ; CHECK-NEXT:       ValueKind:         HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:         I64
 ; CHECK-NEXT:     - Size:              8
 ; CHECK-NEXT:       Align:             8
 ; CHECK-NEXT:       ValueKind:         HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:         I8
 ; CHECK-NEXT:       AddrSpaceQual:     Global
-define amdgpu_kernel void @test_reqd_wgs_vec_type_hint(i32 %a)
+define amdgpu_kernel void @test_reqd_wgs_vec_type_hint(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 !vec_type_hint !5
     !reqd_work_group_size !6 {
   ret void
 }
 
 ; CHECK:      - Name:            test_wgs_hint_vec_type_hint
 ; CHECK-NEXT:   SymbolName:      'test_wgs_hint_vec_type_hint@kd'

 ; CHECK-NEXT:       Align:             8
 ; CHECK-NEXT:       ValueKind:         HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:         I64
 ; CHECK-NEXT:     - Size:              8
 ; CHECK-NEXT:       Align:             8
 ; CHECK-NEXT:       ValueKind:         HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:         I8
 ; CHECK-NEXT:       AddrSpaceQual:     Global
-define amdgpu_kernel void @test_wgs_hint_vec_type_hint(i32 %a)
+define amdgpu_kernel void @test_wgs_hint_vec_type_hint(i32 %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !3
     !kernel_arg_base_type !3 !kernel_arg_type_qual !4 !vec_type_hint !7
     !work_group_size_hint !8 {
   ret void
 }
 
 ; CHECK:      - Name:            test_arg_ptr_to_ptr
 ; CHECK-NEXT:   SymbolName:      'test_arg_ptr_to_ptr@kd'

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_arg_ptr_to_ptr(i32 addrspace(5)* addrspace(1)* %a)
+define amdgpu_kernel void @test_arg_ptr_to_ptr(i32 addrspace(5)* addrspace(1)* %a) #0
     !kernel_arg_addr_space !81 !kernel_arg_access_qual !2 !kernel_arg_type !80
     !kernel_arg_base_type !80 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_arg_struct_contains_ptr
 ; CHECK-NEXT:   SymbolName:      'test_arg_struct_contains_ptr@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_arg_struct_contains_ptr(%struct.B addrspace(5)* byval %a)
+define amdgpu_kernel void @test_arg_struct_contains_ptr(%struct.B addrspace(5)* byval %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !82
     !kernel_arg_base_type !82 !kernel_arg_type_qual !4 {
  ret void
 }
 
 ; CHECK:      - Name:            test_arg_vector_of_ptr
 ; CHECK-NEXT:   SymbolName:      'test_arg_vector_of_ptr@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_arg_vector_of_ptr(<2 x i32 addrspace(1)*> %a)
+define amdgpu_kernel void @test_arg_vector_of_ptr(<2 x i32 addrspace(1)*> %a) #0
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !83
     !kernel_arg_base_type !83 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_arg_unknown_builtin_type
 ; CHECK-NEXT:   SymbolName:      'test_arg_unknown_builtin_type@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
 define amdgpu_kernel void @test_arg_unknown_builtin_type(
-    %opencl.clk_event_t addrspace(1)* %a)
+    %opencl.clk_event_t addrspace(1)* %a) #0
     !kernel_arg_addr_space !81 !kernel_arg_access_qual !2 !kernel_arg_type !84
     !kernel_arg_base_type !84 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_pointee_align
 ; CHECK-NEXT:   SymbolName:      'test_pointee_align@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       AddrSpaceQual: Global
 define amdgpu_kernel void @test_pointee_align(i64 addrspace(1)* %a,
                                               i8 addrspace(3)* %b,
                                               <2 x i8> addrspace(3)* %c,
                                               <3 x i8> addrspace(3)* %d,
                                               <4 x i8> addrspace(3)* %e,
                                               <8 x i8> addrspace(3)* %f,
                                               <16 x i8> addrspace(3)* %g,
-                                              {} addrspace(3)* %h)
+                                              {} addrspace(3)* %h) #0
     !kernel_arg_addr_space !91 !kernel_arg_access_qual !92 !kernel_arg_type !93
     !kernel_arg_base_type !93 !kernel_arg_type_qual !94 {
   ret void
 }
 
 ; CHECK:      - Name:            test_pointee_align_attribute
 ; CHECK-NEXT:   SymbolName:      'test_pointee_align_attribute@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       AddrSpaceQual: Global
 define amdgpu_kernel void @test_pointee_align_attribute(i64 addrspace(1)* align 16 %a,
                                                         i8 addrspace(3)* align 8 %b,
                                                         <2 x i8> addrspace(3)* align 32 %c,
                                                         <3 x i8> addrspace(3)* align 64 %d,
                                                         <4 x i8> addrspace(3)* align 256 %e,
                                                         <8 x i8> addrspace(3)* align 128 %f,
                                                         <16 x i8> addrspace(3)* align 1024 %g,
-                                                        {} addrspace(3)* align 16 %h)
+                                                        {} addrspace(3)* align 16 %h) #0
     !kernel_arg_addr_space !91 !kernel_arg_access_qual !92 !kernel_arg_type !93
     !kernel_arg_base_type !93 !kernel_arg_type_qual !94 {
   ret void
 }
 
 
 ; CHECK:      - Name:            __test_block_invoke_kernel
 ; CHECK-NEXT:   SymbolName:      '__test_block_invoke_kernel@kd'

 ; CHECK-NEXT:       ValueKind:     HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:     I64
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenPrintfBuffer
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
 define amdgpu_kernel void @__test_block_invoke_kernel(
-    <{ i32, i32, i8*, i8 addrspace(1)*, i8 }> %arg) #0
+    <{ i32, i32, i8*, i8 addrspace(1)*, i8 }> %arg) #1
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !110
     !kernel_arg_base_type !110 !kernel_arg_type_qual !4 {
   ret void
 }
 
 ; CHECK:      - Name:            test_enqueue_kernel_caller
 ; CHECK-NEXT:   SymbolName:      'test_enqueue_kernel_caller@kd'
 ; CHECK-NEXT:   Language:        OpenCL C

 ; CHECK-NEXT:       ValueKind:     HiddenDefaultQueue
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
 ; CHECK-NEXT:     - Size:          8
 ; CHECK-NEXT:       Align:         8
 ; CHECK-NEXT:       ValueKind:     HiddenCompletionAction
 ; CHECK-NEXT:       ValueType:     I8
 ; CHECK-NEXT:       AddrSpaceQual: Global
-define amdgpu_kernel void @test_enqueue_kernel_caller(i8 %a) #1
+define amdgpu_kernel void @test_enqueue_kernel_caller(i8 %a) #2
     !kernel_arg_addr_space !1 !kernel_arg_access_qual !2 !kernel_arg_type !9
     !kernel_arg_base_type !9 !kernel_arg_type_qual !4 {
   ret void
 }
 
-attributes #0 = { "runtime-handle"="__test_block_invoke_kernel_runtime_handle" }
-attributes #1 = { "calls-enqueue-kernel" }
+attributes #0 = { "amdgpu-implicitarg-num-bytes"="48" }
+attributes #1 = { "amdgpu-implicitarg-num-bytes"="48" "runtime-handle"="__test_block_invoke_kernel_runtime_handle" }
+attributes #2 = { "amdgpu-implicitarg-num-bytes"="48" "calls-enqueue-kernel" }
 
 !llvm.printf.fmts = !{!100, !101}
 
 !1 = !{i32 0}
 !2 = !{!"none"}
 !3 = !{!"int"}
 !4 = !{!""}
 !5 = !{i32 undef, i32 1}

llvm/trunk/test/CodeGen/AMDGPU/hsa-metadata-hidden-args.ll

 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx700 -filetype=obj -o - < %s | llvm-readobj -elf-output-style=GNU -notes | FileCheck --check-prefix=CHECK --check-prefix=GFX700 --check-prefix=NOTES %s
 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx803 -filetype=obj -o - < %s | llvm-readobj -elf-output-style=GNU -notes | FileCheck --check-prefix=CHECK --check-prefix=GFX803 --check-prefix=NOTES %s
 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -filetype=obj -o - < %s | llvm-readobj -elf-output-style=GNU -notes | FileCheck --check-prefix=CHECK --check-prefix=GFX900 --check-prefix=NOTES %s
 
 ; CHECK: ---
 ; CHECK:  Version: [ 1, 0 ]
 ; CHECK:  Kernels:
 
-; CHECK:      - Name:       test
-; CHECK:        SymbolName: 'test@kd'
+; CHECK:      - Name:       test0
+; CHECK:        SymbolName: 'test0@kd'
+; CHECK:        Args:
+; CHECK-NEXT:     - Name:            r
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            a
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            b
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:   CodeProps:
+define amdgpu_kernel void @test0(
+    half addrspace(1)* %r,
+    half addrspace(1)* %a,
+    half addrspace(1)* %b) {
+entry:
+  %a.val = load half, half addrspace(1)* %a
+  %b.val = load half, half addrspace(1)* %b
+  %r.val = fadd half %a.val, %b.val
+  store half %r.val, half addrspace(1)* %r
+  ret void
+}
+
+; CHECK:      - Name:       test8
+; CHECK:        SymbolName: 'test8@kd'
+; CHECK:        Args:
+; CHECK-NEXT:     - Name:            r
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            a
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            b
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetX
+; CHECK-NEXT:       ValueType:       I64
+; CHECK-NEXT:   CodeProps:
+define amdgpu_kernel void @test8(
+    half addrspace(1)* %r,
+    half addrspace(1)* %a,
+    half addrspace(1)* %b) #0 {
+entry:
+  %a.val = load half, half addrspace(1)* %a
+  %b.val = load half, half addrspace(1)* %b
+  %r.val = fadd half %a.val, %b.val
+  store half %r.val, half addrspace(1)* %r
+  ret void
+}
+
+; CHECK:      - Name:       test16
+; CHECK:        SymbolName: 'test16@kd'
+; CHECK:        Args:
+; CHECK-NEXT:     - Name:            r
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            a
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            b
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetX
+; CHECK-NEXT:       ValueType:       I64
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetY
+; CHECK-NEXT:       ValueType:       I64
+; CHECK-NEXT:   CodeProps:
+define amdgpu_kernel void @test16(
+    half addrspace(1)* %r,
+    half addrspace(1)* %a,
+    half addrspace(1)* %b) #1 {
+entry:
+  %a.val = load half, half addrspace(1)* %a
+  %b.val = load half, half addrspace(1)* %b
+  %r.val = fadd half %a.val, %b.val
+  store half %r.val, half addrspace(1)* %r
+  ret void
+}
+
+; CHECK:      - Name:       test24
+; CHECK:        SymbolName: 'test24@kd'
+; CHECK:        Args:
+; CHECK-NEXT:     - Name:            r
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            a
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            b
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetX
+; CHECK-NEXT:       ValueType:       I64
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetY
+; CHECK-NEXT:       ValueType:       I64
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetZ
+; CHECK-NEXT:       ValueType:       I64
+; CHECK-NEXT:   CodeProps:
+define amdgpu_kernel void @test24(
+    half addrspace(1)* %r,
+    half addrspace(1)* %a,
+    half addrspace(1)* %b) #2 {
+entry:
+  %a.val = load half, half addrspace(1)* %a
+  %b.val = load half, half addrspace(1)* %b
+  %r.val = fadd half %a.val, %b.val
+  store half %r.val, half addrspace(1)* %r
+  ret void
+}
+
+; CHECK:      - Name:       test32
+; CHECK:        SymbolName: 'test32@kd'
 ; CHECK:        Args:
 ; CHECK-NEXT:     - Name:            r
 ; CHECK-NEXT:       Size:            8
 ; CHECK-NEXT:       Align:           8
 ; CHECK-NEXT:       ValueKind:       GlobalBuffer
 ; CHECK-NEXT:       ValueType:       F16
 ; CHECK-NEXT:       AddrSpaceQual:   Global
 ; CHECK-NEXT:     - Name:            a

 ; CHECK-NEXT:       Align:           8
 ; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetZ
 ; CHECK-NEXT:       ValueType:       I64
 ; CHECK-NEXT:     - Size:            8
 ; CHECK-NEXT:       Align:           8
 ; CHECK-NEXT:       ValueKind:       HiddenNone
 ; CHECK-NEXT:       ValueType:       I8
 ; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:   CodeProps:
+define amdgpu_kernel void @test32(
+    half addrspace(1)* %r,
+    half addrspace(1)* %a,
+    half addrspace(1)* %b) #3 {
+entry:
+  %a.val = load half, half addrspace(1)* %a
+  %b.val = load half, half addrspace(1)* %b
+  %r.val = fadd half %a.val, %b.val
+  store half %r.val, half addrspace(1)* %r
+  ret void
+}
+
+; CHECK:      - Name:       test48
+; CHECK:        SymbolName: 'test48@kd'
+; CHECK:        Args:
+; CHECK-NEXT:     - Name:            r
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            a
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Name:            b
+; CHECK-NEXT:       Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       GlobalBuffer
+; CHECK-NEXT:       ValueType:       F16
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetX
+; CHECK-NEXT:       ValueType:       I64
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetY
+; CHECK-NEXT:       ValueType:       I64
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenGlobalOffsetZ
+; CHECK-NEXT:       ValueType:       I64
 ; CHECK-NEXT:     - Size:            8
 ; CHECK-NEXT:       Align:           8
 ; CHECK-NEXT:       ValueKind:       HiddenNone
 ; CHECK-NEXT:       ValueType:       I8
 ; CHECK-NEXT:       AddrSpaceQual:   Global
 ; CHECK-NEXT:     - Size:            8
 ; CHECK-NEXT:       Align:           8
 ; CHECK-NEXT:       ValueKind:       HiddenNone
 ; CHECK-NEXT:       ValueType:       I8
 ; CHECK-NEXT:       AddrSpaceQual:   Global
-define amdgpu_kernel void @test(
+; CHECK-NEXT:     - Size:            8
+; CHECK-NEXT:       Align:           8
+; CHECK-NEXT:       ValueKind:       HiddenNone
+; CHECK-NEXT:       ValueType:       I8
+; CHECK-NEXT:       AddrSpaceQual:   Global
+; CHECK-NEXT:   CodeProps:
+define amdgpu_kernel void @test48(
     half addrspace(1)* %r,
     half addrspace(1)* %a,
-    half addrspace(1)* %b) {
+    half addrspace(1)* %b) #4 {
 entry:
   %a.val = load half, half addrspace(1)* %a
   %b.val = load half, half addrspace(1)* %b
   %r.val = fadd half %a.val, %b.val
   store half %r.val, half addrspace(1)* %r
   ret void
 }
 
-!opencl.ocl.version = !{!0}
-!0 = !{i32 2, i32 0}
+attributes #0 = { "amdgpu-implicitarg-num-bytes"="8" }
+attributes #1 = { "amdgpu-implicitarg-num-bytes"="16" }
+attributes #2 = { "amdgpu-implicitarg-num-bytes"="24" }
+attributes #3 = { "amdgpu-implicitarg-num-bytes"="32" }
+attributes #4 = { "amdgpu-implicitarg-num-bytes"="48" }