[AMDGPU] Refactor HSAMetadataStream::emitKernel (NFC)

[scott.linder]

Summary

Move all metadata construction into AMDGPUHSAMetadataStreamer.

This patch is part of a series required for AMDGPU to support it's new MessagePack metadata.

[arsenm]

LGTM

[bjope]

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

31 ↗

(On Diff #154841)

I see some build errors. Seems to be due to this commit:

[1276/3790] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUHSAMetadataStreamer.cpp.o
FAILED: lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUHSAMetadataStreamer.cpp.o 
/app/clang/3.6/bin/clang++  -march=corei7  -DGTEST_HAS_RTTI=0 -D_DEBUG -D_GNU_SOURCE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -Ilib/Target/AMDGPU -I../lib/Target/AMDGPU -I/usr/include/libxml2 -Iinclude -I../include -fPIC -fvisibility-inlines-hidden -Werror -Werror=date-time -std=c++11 -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wmissing-field-initializers -pedantic -Wno-long-long -Wcovered-switch-default -Wnon-virtual-dtor -Wdelete-non-virtual-dtor -Wstring-conversion -fdiagnostics-color -ffunction-sections -fdata-sections -O3    -UNDEBUG  -fno-exceptions -fno-rtti -MMD -MT lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUHSAMetadataStreamer.cpp.o -MF lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUHSAMetadataStreamer.cpp.o.d -o lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUHSAMetadataStreamer.cpp.o -c ../lib/Target/AMDGPU/AMDGPUHSAMetadataStreamer.cpp
In file included from ../lib/Target/AMDGPU/AMDGPUHSAMetadataStreamer.cpp:20:
../lib/Target/AMDGPU/SIProgramInfo.h:22:1: error: 'SIProgramInfo' defined as a struct here but previously declared as a class [-Werror,-Wmismatched-tags]
struct SIProgramInfo {
^
../lib/Target/AMDGPU/AMDGPUHSAMetadataStreamer.h:31:1: note: did you mean struct here?
class SIProgramInfo;
^~~~~
struct
1 error generated.
[1277/3790] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUFrameLowering.cpp.o
[1278/3790] Building CXX object lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUAsmPrinter.cpp.o
FAILED: lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUAsmPrinter.cpp.o 
/app/clang/3.6/bin/clang++  -march=corei7  -DGTEST_HAS_RTTI=0 -D_DEBUG -D_GNU_SOURCE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -Ilib/Target/AMDGPU -I../lib/Target/AMDGPU -I/usr/include/libxml2 -Iinclude -I../include -fPIC -fvisibility-inlines-hidden -Werror -Werror=date-time -std=c++11 -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wmissing-field-initializers -pedantic -Wno-long-long -Wcovered-switch-default -Wnon-virtual-dtor -Wdelete-non-virtual-dtor -Wstring-conversion -fdiagnostics-color -ffunction-sections -fdata-sections -O3    -UNDEBUG  -fno-exceptions -fno-rtti -MMD -MT lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUAsmPrinter.cpp.o -MF lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUAsmPrinter.cpp.o.d -o lib/Target/AMDGPU/CMakeFiles/LLVMAMDGPUCodeGen.dir/AMDGPUAsmPrinter.cpp.o -c ../lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
In file included from ../lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp:19:
In file included from ../lib/Target/AMDGPU/AMDGPUAsmPrinter.h:21:
../lib/Target/AMDGPU/SIProgramInfo.h:22:1: error: 'SIProgramInfo' defined as a struct here but previously declared as a class [-Werror,-Wmismatched-tags]
struct SIProgramInfo {
^
../lib/Target/AMDGPU/AMDGPUHSAMetadataStreamer.h:31:1: note: did you mean struct here?
class SIProgramInfo;
^~~~~
struct
1 error generated.

lib/Target/AMDGPU/AMDGPUAsmPrinter.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_LIB_TARGET_AMDGPU_AMDGPUASMPRINTER_H
 #define LLVM_LIB_TARGET_AMDGPU_AMDGPUASMPRINTER_H
 
 #include "AMDGPU.h"
 #include "AMDKernelCodeT.h"
 #include "MCTargetDesc/AMDGPUHSAMetadataStreamer.h"
+#include "SIProgramInfo.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/CodeGen/AsmPrinter.h"
 #include "llvm/Support/AMDHSAKernelDescriptor.h"
 #include <cstddef>
 #include <cstdint>
 #include <limits>
 #include <memory>
 #include <string>
struct SIFunctionResourceInfo {
     bool UsesVCC = false;
     bool UsesFlatScratch = false;
     bool HasDynamicallySizedStack = false;
     bool HasRecursion = false;
 
     int32_t getTotalNumSGPRs(const SISubtarget &ST) const;
   };
 
-  // Track resource usage for kernels / entry functions.
-  struct SIProgramInfo {
-    // Fields set in PGM_RSRC1 pm4 packet.
-    uint32_t VGPRBlocks = 0;
-    uint32_t SGPRBlocks = 0;
-    uint32_t Priority = 0;
-    uint32_t FloatMode = 0;
-    uint32_t Priv = 0;
-    uint32_t DX10Clamp = 0;
-    uint32_t DebugMode = 0;
-    uint32_t IEEEMode = 0;
-    uint64_t ScratchSize = 0;
-
-    uint64_t ComputePGMRSrc1 = 0;
-
-    // Fields set in PGM_RSRC2 pm4 packet.
-    uint32_t LDSBlocks = 0;
-    uint32_t ScratchBlocks = 0;
-
-    uint64_t ComputePGMRSrc2 = 0;
-
-    uint32_t NumVGPR = 0;
-    uint32_t NumSGPR = 0;
-    uint32_t LDSSize = 0;
-    bool FlatUsed = false;
-
-    // Number of SGPRs that meets number of waves per execution unit request.
-    uint32_t NumSGPRsForWavesPerEU = 0;
-
-    // Number of VGPRs that meets number of waves per execution unit request.
-    uint32_t NumVGPRsForWavesPerEU = 0;
-
-    // If ReservedVGPRCount is 0 then must be 0. Otherwise, this is the first
-    // fixed VGPR number reserved.
-    uint16_t ReservedVGPRFirst = 0;
-
-    // The number of consecutive VGPRs reserved.
-    uint16_t ReservedVGPRCount = 0;
-
-    // Fixed SGPR number used to hold wave scratch offset for entire kernel
-    // execution, or std::numeric_limits<uint16_t>::max() if the register is not
-    // used or not known.
-    uint16_t DebuggerWavefrontPrivateSegmentOffsetSGPR =
-        std::numeric_limits<uint16_t>::max();
-
-    // Fixed SGPR number of the first 4 SGPRs used to hold scratch V# for entire
-    // kernel execution, or std::numeric_limits<uint16_t>::max() if the register
-    // is not used or not known.
-    uint16_t DebuggerPrivateSegmentBufferSGPR =
-        std::numeric_limits<uint16_t>::max();
-
-    // Whether there is recursion, dynamic allocas, indirect calls or some other
-    // reason there may be statically unknown stack usage.
-    bool DynamicCallStack = false;
-
-    // Bonus information for debugging.
-    bool VCCUsed = false;
-
-    SIProgramInfo() = default;
-  };
-
   SIProgramInfo CurrentProgramInfo;
   DenseMap<const Function *, SIFunctionResourceInfo> CallGraphResourceInfo;
 
   AMDGPU::HSAMD::MetadataStreamer HSAMetadataStream;
   std::map<uint32_t, uint32_t> PALMetadataMap;
 
   uint64_t getFunctionCodeSize(const MachineFunction &MF) const;
   SIFunctionResourceInfo analyzeResourceUsage(const MachineFunction &MF) const;
 
   void readPALMetadata(Module &M);
   void getSIProgramInfo(SIProgramInfo &Out, const MachineFunction &MF);
   void getAmdKernelCode(amd_kernel_code_t &Out, const SIProgramInfo &KernelInfo,
                         const MachineFunction &MF) const;
   void findNumUsedRegistersSI(const MachineFunction &MF,
                               unsigned &NumSGPR,
                               unsigned &NumVGPR) const;
 
-  AMDGPU::HSAMD::Kernel::CodeProps::Metadata getHSACodeProps(
-      const MachineFunction &MF,
-      const SIProgramInfo &ProgramInfo) const;
-  AMDGPU::HSAMD::Kernel::DebugProps::Metadata getHSADebugProps(
-      const MachineFunction &MF,
-      const SIProgramInfo &ProgramInfo) const;
-
   /// Emit register usage information so that the GPU driver
   /// can correctly setup the GPU state.
   void EmitProgramInfoSI(const MachineFunction &MF,
                          const SIProgramInfo &KernelInfo);
   void EmitPALMetadata(const MachineFunction &MF,
                        const SIProgramInfo &KernelInfo);
   void emitCommonFunctionComments(uint32_t NumVGPR,
                                   uint32_t NumSGPR,

lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp

void AMDGPUAsmPrinter::EmitFunctionBodyStart() {
   if (STM.isAmdCodeObjectV2(MF->getFunction())) {
     getAmdKernelCode(KernelCode, CurrentProgramInfo, *MF);
     getTargetStreamer()->EmitAMDKernelCodeT(KernelCode);
   }
 
   if (TM.getTargetTriple().getOS() != Triple::AMDHSA)
     return;
 
-  HSAMetadataStream.emitKernel(MF->getFunction(),
-                               getHSACodeProps(*MF, CurrentProgramInfo),
-                               getHSADebugProps(*MF, CurrentProgramInfo));
+  HSAMetadataStream.emitKernel(*MF, CurrentProgramInfo);
 }
 
 void AMDGPUAsmPrinter::EmitFunctionBodyEnd() {
   const SIMachineFunctionInfo &MFI = *MF->getInfo<SIMachineFunctionInfo>();
   if (!MFI.isEntryFunction())
     return;
   if (!IsaInfo::hasCodeObjectV3(getSTI()) ||
       TM.getTargetTriple().getOS() != Triple::AMDHSA)
void AMDGPUAsmPrinter::getAmdKernelCode(amd_kernel_code_t &Out,
   if (STM.debuggerEmitPrologue()) {
     Out.debug_wavefront_private_segment_offset_sgpr =
       CurrentProgramInfo.DebuggerWavefrontPrivateSegmentOffsetSGPR;
     Out.debug_private_segment_buffer_sgpr =
       CurrentProgramInfo.DebuggerPrivateSegmentBufferSGPR;
   }
 }
 
-AMDGPU::HSAMD::Kernel::CodeProps::Metadata AMDGPUAsmPrinter::getHSACodeProps(
-    const MachineFunction &MF,
-    const SIProgramInfo &ProgramInfo) const {
-  const SISubtarget &STM = MF.getSubtarget<SISubtarget>();
-  const SIMachineFunctionInfo &MFI = *MF.getInfo<SIMachineFunctionInfo>();
-  HSAMD::Kernel::CodeProps::Metadata HSACodeProps;
-
-  HSACodeProps.mKernargSegmentSize =
-    STM.getKernArgSegmentSize(MF.getFunction(), MFI.getABIArgOffset());
-  HSACodeProps.mGroupSegmentFixedSize = ProgramInfo.LDSSize;
-  HSACodeProps.mPrivateSegmentFixedSize = ProgramInfo.ScratchSize;
-  HSACodeProps.mKernargSegmentAlign =
-      std::max(uint32_t(4), MFI.getMaxKernArgAlign());
-  HSACodeProps.mWavefrontSize = STM.getWavefrontSize();
-  HSACodeProps.mNumSGPRs = CurrentProgramInfo.NumSGPR;
-  HSACodeProps.mNumVGPRs = CurrentProgramInfo.NumVGPR;
-  HSACodeProps.mMaxFlatWorkGroupSize = MFI.getMaxFlatWorkGroupSize();
-  HSACodeProps.mIsDynamicCallStack = ProgramInfo.DynamicCallStack;
-  HSACodeProps.mIsXNACKEnabled = STM.isXNACKEnabled();
-  HSACodeProps.mNumSpilledSGPRs = MFI.getNumSpilledSGPRs();
-  HSACodeProps.mNumSpilledVGPRs = MFI.getNumSpilledVGPRs();
-
-  return HSACodeProps;
-}
-
-AMDGPU::HSAMD::Kernel::DebugProps::Metadata AMDGPUAsmPrinter::getHSADebugProps(
-    const MachineFunction &MF,
-    const SIProgramInfo &ProgramInfo) const {
-  const SISubtarget &STM = MF.getSubtarget<SISubtarget>();
-  HSAMD::Kernel::DebugProps::Metadata HSADebugProps;
-
-  if (!STM.debuggerSupported())
-    return HSADebugProps;
-
-  HSADebugProps.mDebuggerABIVersion.push_back(1);
-  HSADebugProps.mDebuggerABIVersion.push_back(0);
-  HSADebugProps.mReservedNumVGPRs = ProgramInfo.ReservedVGPRCount;
-  HSADebugProps.mReservedFirstVGPR = ProgramInfo.ReservedVGPRFirst;
-
-  if (STM.debuggerEmitPrologue()) {
-    HSADebugProps.mPrivateSegmentBufferSGPR =
-        ProgramInfo.DebuggerPrivateSegmentBufferSGPR;
-    HSADebugProps.mWavefrontPrivateSegmentOffsetSGPR =
-        ProgramInfo.DebuggerWavefrontPrivateSegmentOffsetSGPR;
-  }
-
-  return HSADebugProps;
-}
-
 bool AMDGPUAsmPrinter::PrintAsmOperand(const MachineInstr *MI, unsigned OpNo,
                                        unsigned AsmVariant,
                                        const char *ExtraCode, raw_ostream &O) {
   // First try the generic code, which knows about modifiers like 'c' and 'n'.
   if (!AsmPrinter::PrintAsmOperand(MI, OpNo, AsmVariant, ExtraCode, O))
     return false;
 
   if (ExtraCode && ExtraCode[0]) {

lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.h

 
 namespace llvm {
 
 class Argument;
 class DataLayout;
 class Function;
 class MDNode;
 class Module;
+class SIProgramInfo;
 class Type;
 
 namespace AMDGPU {
 namespace HSAMD {
 
 class MetadataStreamer final {
 private:
   Metadata HSAMetadata;
ValueKind getValueKind(Type *Ty, StringRef TypeQual,
                          StringRef BaseTypeName) const;
 
   ValueType getValueType(Type *Ty, StringRef TypeName) const;
 
   std::string getTypeName(Type *Ty, bool Signed) const;
 
   std::vector<uint32_t> getWorkGroupDimensions(MDNode *Node) const;
 
+  Kernel::CodeProps::Metadata getHSACodeProps(
+      const MachineFunction &MF,
+      const SIProgramInfo &ProgramInfo) const;
+  Kernel::DebugProps::Metadata getHSADebugProps(
+      const MachineFunction &MF,
+      const SIProgramInfo &ProgramInfo) const;
+
   void emitVersion();
 
   void emitPrintf(const Module &Mod);
 
   void emitKernelLanguage(const Function &Func);
 
   void emitKernelAttrs(const Function &Func);
 
public:
   const Metadata &getHSAMetadata() const {
     return HSAMetadata;
   }
 
   void begin(const Module &Mod);
 
   void end();
 
-  void emitKernel(const Function &Func,
-                  const Kernel::CodeProps::Metadata &CodeProps,
-                  const Kernel::DebugProps::Metadata &DebugProps);
+  void emitKernel(const MachineFunction &MF, const SIProgramInfo &ProgramInfo);
 };
 
 } // end namespace HSAMD
 } // end namespace AMDGPU
 } // end namespace llvm
 
 #endif // LLVM_LIB_TARGET_AMDGPU_MCTARGETDESC_AMDGPUHSAMETADATASTREAMER_H

lib/Target/AMDGPU/MCTargetDesc/AMDGPUHSAMetadataStreamer.cpp

 /// \file
 /// AMDGPU HSA Metadata Streamer.
 ///
 //
 //===----------------------------------------------------------------------===//
 
 #include "AMDGPUHSAMetadataStreamer.h"
 #include "AMDGPU.h"
+#include "AMDGPUSubtarget.h"
+#include "SIMachineFunctionInfo.h"
+#include "SIProgramInfo.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(
std::vector<uint32_t> MetadataStreamer::getWorkGroupDimensions(
   if (Node->getNumOperands() != 3)
     return Dims;
 
   for (auto &Op : Node->operands())
     Dims.push_back(mdconst::extract<ConstantInt>(Op)->getZExtValue());
   return Dims;
 }
 
+Kernel::CodeProps::Metadata MetadataStreamer::getHSACodeProps(
+    const MachineFunction &MF,
+    const SIProgramInfo &ProgramInfo) const {
+  const SISubtarget &STM = MF.getSubtarget<SISubtarget>();
+  const SIMachineFunctionInfo &MFI = *MF.getInfo<SIMachineFunctionInfo>();
+  HSAMD::Kernel::CodeProps::Metadata HSACodeProps;
+
+  HSACodeProps.mKernargSegmentSize =
+    STM.getKernArgSegmentSize(MF.getFunction(), MFI.getABIArgOffset());
+  HSACodeProps.mGroupSegmentFixedSize = ProgramInfo.LDSSize;
+  HSACodeProps.mPrivateSegmentFixedSize = ProgramInfo.ScratchSize;
+  HSACodeProps.mKernargSegmentAlign =
+      std::max(uint32_t(4), MFI.getMaxKernArgAlign());
+  HSACodeProps.mWavefrontSize = STM.getWavefrontSize();
+  HSACodeProps.mNumSGPRs = ProgramInfo.NumSGPR;
+  HSACodeProps.mNumVGPRs = ProgramInfo.NumVGPR;
+  HSACodeProps.mMaxFlatWorkGroupSize = MFI.getMaxFlatWorkGroupSize();
+  HSACodeProps.mIsDynamicCallStack = ProgramInfo.DynamicCallStack;
+  HSACodeProps.mIsXNACKEnabled = STM.isXNACKEnabled();
+  HSACodeProps.mNumSpilledSGPRs = MFI.getNumSpilledSGPRs();
+  HSACodeProps.mNumSpilledVGPRs = MFI.getNumSpilledVGPRs();
+
+  return HSACodeProps;
+}
+
+Kernel::DebugProps::Metadata MetadataStreamer::getHSADebugProps(
+    const MachineFunction &MF,
+    const SIProgramInfo &ProgramInfo) const {
+  const SISubtarget &STM = MF.getSubtarget<SISubtarget>();
+  HSAMD::Kernel::DebugProps::Metadata HSADebugProps;
+
+  if (!STM.debuggerSupported())
+    return HSADebugProps;
+
+  HSADebugProps.mDebuggerABIVersion.push_back(1);
+  HSADebugProps.mDebuggerABIVersion.push_back(0);
+  HSADebugProps.mReservedNumVGPRs = ProgramInfo.ReservedVGPRCount;
+  HSADebugProps.mReservedFirstVGPR = ProgramInfo.ReservedVGPRFirst;
+
+  if (STM.debuggerEmitPrologue()) {
+    HSADebugProps.mPrivateSegmentBufferSGPR =
+        ProgramInfo.DebuggerPrivateSegmentBufferSGPR;
+    HSADebugProps.mWavefrontPrivateSegmentOffsetSGPR =
+        ProgramInfo.DebuggerWavefrontPrivateSegmentOffsetSGPR;
+  }
+
+  return HSADebugProps;
+}
+
+
 void MetadataStreamer::emitVersion() {
   auto &Version = HSAMetadata.mVersion;
 
   Version.push_back(VersionMajor);
   Version.push_back(VersionMinor);
 }
 
 void MetadataStreamer::emitPrintf(const Module &Mod) {
if (toString(HSAMetadata, HSAMetadataString))
     return;
 
   if (DumpHSAMetadata)
     dump(HSAMetadataString);
   if (VerifyHSAMetadata)
     verify(HSAMetadataString);
 }
 
-void MetadataStreamer::emitKernel(
-    const Function &Func,
-    const Kernel::CodeProps::Metadata &CodeProps,
-    const Kernel::DebugProps::Metadata &DebugProps) {
+void MetadataStreamer::emitKernel(const MachineFunction &MF, const SIProgramInfo &ProgramInfo) {
+  auto &Func = MF.getFunction();
+  auto CodeProps = getHSACodeProps(MF, ProgramInfo);
+  auto DebugProps = getHSADebugProps(MF, ProgramInfo);
+
   if (Func.getCallingConv() != CallingConv::AMDGPU_KERNEL)
     return;
 
   HSAMetadata.mKernels.push_back(Kernel::Metadata());
   auto &Kernel = HSAMetadata.mKernels.back();
 
   Kernel.mName = Func.getName();
   Kernel.mSymbolName = (Twine(Func.getName()) + Twine("@kd")).str();

lib/Target/AMDGPU/SIProgramInfo.h

+//===--- SIProgramInfo.h ----------------------------------------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+/// \file
+/// Defines struct to track resource usage for kernels and entry functions.
+///
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIB_TARGET_AMDGPU_SIPROGRAMINFO_H
+#define LLVM_LIB_TARGET_AMDGPU_SIPROGRAMINFO_H
+
+namespace llvm {
+
+/// Track resource usage for kernels / entry functions.
+struct SIProgramInfo {
+    // Fields set in PGM_RSRC1 pm4 packet.
+    uint32_t VGPRBlocks = 0;
+    uint32_t SGPRBlocks = 0;
+    uint32_t Priority = 0;
+    uint32_t FloatMode = 0;
+    uint32_t Priv = 0;
+    uint32_t DX10Clamp = 0;
+    uint32_t DebugMode = 0;
+    uint32_t IEEEMode = 0;
+    uint64_t ScratchSize = 0;
+
+    uint64_t ComputePGMRSrc1 = 0;
+
+    // Fields set in PGM_RSRC2 pm4 packet.
+    uint32_t LDSBlocks = 0;
+    uint32_t ScratchBlocks = 0;
+
+    uint64_t ComputePGMRSrc2 = 0;
+
+    uint32_t NumVGPR = 0;
+    uint32_t NumSGPR = 0;
+    uint32_t LDSSize = 0;
+    bool FlatUsed = false;
+
+    // Number of SGPRs that meets number of waves per execution unit request.
+    uint32_t NumSGPRsForWavesPerEU = 0;
+
+    // Number of VGPRs that meets number of waves per execution unit request.
+    uint32_t NumVGPRsForWavesPerEU = 0;
+
+    // If ReservedVGPRCount is 0 then must be 0. Otherwise, this is the first
+    // fixed VGPR number reserved.
+    uint16_t ReservedVGPRFirst = 0;
+
+    // The number of consecutive VGPRs reserved.
+    uint16_t ReservedVGPRCount = 0;
+
+    // Fixed SGPR number used to hold wave scratch offset for entire kernel
+    // execution, or std::numeric_limits<uint16_t>::max() if the register is not
+    // used or not known.
+    uint16_t DebuggerWavefrontPrivateSegmentOffsetSGPR =
+        std::numeric_limits<uint16_t>::max();
+
+    // Fixed SGPR number of the first 4 SGPRs used to hold scratch V# for entire
+    // kernel execution, or std::numeric_limits<uint16_t>::max() if the register
+    // is not used or not known.
+    uint16_t DebuggerPrivateSegmentBufferSGPR =
+        std::numeric_limits<uint16_t>::max();
+
+    // Whether there is recursion, dynamic allocas, indirect calls or some other
+    // reason there may be statically unknown stack usage.
+    bool DynamicCallStack = false;
+
+    // Bonus information for debugging.
+    bool VCCUsed = false;
+
+    SIProgramInfo() = default;
+};
+
+} // namespace llvm
+
+#endif // LLVM_LIB_TARGET_AMDGPU_SIPROGRAMINFO_H