add API call to display general information about the module (#2298)

Partially load SPIR-V from the stream and decode only selected for the report instructions, needed to retrieve general information about the module: capabilities, extensions, version, memory model and addressing model.

In addition to immediately helpful for back-ends lists of capabilities and extensions declared in SPIR-V module, a general intent also is to extend report details in future by feedbacks about further potentially useful analysis, statistics, etc.

Author: VyacheslavLevytskyy

include/LLVMSPIRVLib.h

@@ -106,6 +106,35 @@ std::unique_ptr<SPIRVModule> readSpirvModule(std::istream &IS,
                                              const SPIRV::TranslatorOpts &Opts,
                                              std::string &ErrMsg);
 
+struct SPIRVModuleReport {
+  SPIRV::VersionNumber Version;
+  uint32_t MemoryModel;
+  uint32_t AddrModel;
+  std::vector<std::string> Extensions;
+  std::vector<std::string> ExtendedInstructionSets;
+  std::vector<uint32_t> Capabilities;
+};
+/// \brief Partially load SPIR-V from the stream and decode only selected
+/// instructions that are needed to retrieve general information
+/// about the module. If this call fails, readSPIRVModule is
+/// expected to fail as well.
+/// \returns nullopt on failure.
+std::optional<SPIRVModuleReport> getSpirvReport(std::istream &IS);
+std::optional<SPIRVModuleReport> getSpirvReport(std::istream &IS, int &ErrCode);
+
+struct SPIRVModuleTextReport {
+  std::string Version;
+  std::string MemoryModel;
+  std::string AddrModel;
+  std::vector<std::string> Extensions;
+  std::vector<std::string> ExtendedInstructionSets;
+  std::vector<std::string> Capabilities;
+};
+/// \brief Create a human-readable form of the report returned by a call to
+/// getSpirvReport by decoding its binary fields.
+/// \returns String with the human-readable report.
+SPIRVModuleTextReport formatSpirvReport(const SPIRVModuleReport &Report);
+
 } // End namespace SPIRV
 
 namespace llvm {

include/LLVMSPIRVOpts.h

@@ -68,6 +68,27 @@ enum class VersionNumber : uint32_t {
   MaximumVersion = SPIRV_1_4
 };
 
+inline constexpr std::string_view formatVersionNumber(uint32_t Version) {
+  switch (Version) {
+  case static_cast<uint32_t>(VersionNumber::SPIRV_1_0):
+    return "1.0";
+  case static_cast<uint32_t>(VersionNumber::SPIRV_1_1):
+    return "1.1";
+  case static_cast<uint32_t>(VersionNumber::SPIRV_1_2):
+    return "1.2";
+  case static_cast<uint32_t>(VersionNumber::SPIRV_1_3):
+    return "1.3";
+  case static_cast<uint32_t>(VersionNumber::SPIRV_1_4):
+    return "1.4";
+  }
+  return "unknown";
+}
+
+inline bool isSPIRVVersionKnown(uint32_t Ver) {
+  return Ver >= static_cast<uint32_t>(VersionNumber::MinimumVersion) &&
+         Ver <= static_cast<uint32_t>(VersionNumber::MaximumVersion);
+}
+
 enum class ExtensionID : uint32_t {
   First,
 #define EXT(X) X,

lib/SPIRV/SPIRVReader.cpp

@@ -4860,6 +4860,137 @@ Instruction *SPIRVToLLVM::transRelational(SPIRVInstruction *I, BasicBlock *BB) {
   return cast<Instruction>(Mutator.getMutated());
 }
 
+std::optional<SPIRVModuleReport> getSpirvReport(std::istream &IS) {
+  int IgnoreErrCode;
+  return getSpirvReport(IS, IgnoreErrCode);
+}
+
+std::optional<SPIRVModuleReport> getSpirvReport(std::istream &IS,
+                                                int &ErrCode) {
+  SPIRVWord Word;
+  std::string Name;
+  std::unique_ptr<SPIRVModule> BM(SPIRVModule::createSPIRVModule());
+  SPIRVDecoder D(IS, *BM);
+  D >> Word;
+  if (Word != MagicNumber) {
+    ErrCode = SPIRVEC_InvalidMagicNumber;
+    return {};
+  }
+  D >> Word;
+  if (!isSPIRVVersionKnown(Word)) {
+    ErrCode = SPIRVEC_InvalidVersionNumber;
+    return {};
+  }
+  SPIRVModuleReport Report;
+  Report.Version = static_cast<SPIRV::VersionNumber>(Word);
+  // Skip: Generator’s magic number, Bound and Reserved word
+  D.ignore(3);
+
+  bool IsReportGenCompleted = false, IsMemoryModelDefined = false;
+  while (!IS.bad() && !IsReportGenCompleted && D.getWordCountAndOpCode()) {
+    switch (D.OpCode) {
+    case OpCapability:
+      D >> Word;
+      Report.Capabilities.push_back(Word);
+      break;
+    case OpExtension:
+      Name.clear();
+      D >> Name;
+      Report.Extensions.push_back(Name);
+      break;
+    case OpExtInstImport:
+      Name.clear();
+      D >> Word >> Name;
+      Report.ExtendedInstructionSets.push_back(Name);
+      break;
+    case OpMemoryModel:
+      if (IsMemoryModelDefined) {
+        ErrCode = SPIRVEC_RepeatedMemoryModel;
+        return {};
+      }
+      SPIRVAddressingModelKind AddrModel;
+      SPIRVMemoryModelKind MemoryModel;
+      D >> AddrModel >> MemoryModel;
+      if (!isValid(AddrModel)) {
+        ErrCode = SPIRVEC_InvalidAddressingModel;
+        return {};
+      }
+      if (!isValid(MemoryModel)) {
+        ErrCode = SPIRVEC_InvalidMemoryModel;
+        return {};
+      }
+      Report.MemoryModel = MemoryModel;
+      Report.AddrModel = AddrModel;
+      IsMemoryModelDefined = true;
+      // In this report we don't analyze instructions after OpMemoryModel
+      IsReportGenCompleted = true;
+      break;
+    default:
+      // No more instructions to gather information about
+      IsReportGenCompleted = true;
+    }
+  }
+  if (IS.bad()) {
+    ErrCode = SPIRVEC_InvalidModule;
+    return {};
+  }
+  if (!IsMemoryModelDefined) {
+    ErrCode = SPIRVEC_UnspecifiedMemoryModel;
+    return {};
+  }
+  ErrCode = SPIRVEC_Success;
+  return std::make_optional(std::move(Report));
+}
+
+constexpr std::string_view formatAddressingModel(uint32_t AddrModel) {
+  switch (AddrModel) {
+  case AddressingModelLogical:
+    return "Logical";
+  case AddressingModelPhysical32:
+    return "Physical32";
+  case AddressingModelPhysical64:
+    return "Physical64";
+  case AddressingModelPhysicalStorageBuffer64:
+    return "PhysicalStorageBuffer64";
+  default:
+    return "Unknown";
+  }
+}
+
+constexpr std::string_view formatMemoryModel(uint32_t MemoryModel) {
+  switch (MemoryModel) {
+  case MemoryModelSimple:
+    return "Simple";
+  case MemoryModelGLSL450:
+    return "GLSL450";
+  case MemoryModelOpenCL:
+    return "OpenCL";
+  case MemoryModelVulkan:
+    return "Vulkan";
+  default:
+    return "Unknown";
+  }
+}
+
+SPIRVModuleTextReport formatSpirvReport(const SPIRVModuleReport &Report) {
+  SPIRVModuleTextReport TextReport;
+  TextReport.Version =
+      formatVersionNumber(static_cast<uint32_t>(Report.Version));
+  TextReport.AddrModel = formatAddressingModel(Report.AddrModel);
+  TextReport.MemoryModel = formatMemoryModel(Report.MemoryModel);
+  // format capability codes as strings
+  std::string Name;
+  for (auto Capability : Report.Capabilities) {
+    bool Found = SPIRVCapabilityNameMap::find(
+        static_cast<SPIRVCapabilityKind>(Capability), &Name);
+    TextReport.Capabilities.push_back(Found ? Name : "Unknown");
+  }
+  // other fields with string content can be copied as is
+  TextReport.Extensions = Report.Extensions;
+  TextReport.ExtendedInstructionSets = Report.ExtendedInstructionSets;
+  return TextReport;
+}
+
 std::unique_ptr<SPIRVModule> readSpirvModule(std::istream &IS,
                                              const SPIRV::TranslatorOpts &Opts,
                                              std::string &ErrMsg) {

lib/SPIRV/libSPIRV/SPIRVErrorEnum.h

@@ -22,3 +22,9 @@ _SPIRV_OP(InvalidWordCount,
 _SPIRV_OP(Requires1_1, "Feature requires SPIR-V 1.1 or greater:")
 _SPIRV_OP(RequiresExtension,
           "Feature requires the following SPIR-V extension:\n")
+_SPIRV_OP(InvalidMagicNumber,
+          "Invalid Magic Number.")
+_SPIRV_OP(InvalidVersionNumber,
+          "Invalid Version Number.")
+_SPIRV_OP(UnspecifiedMemoryModel, "Unspecified Memory Model.")
+_SPIRV_OP(RepeatedMemoryModel, "Expects a single OpMemoryModel instruction.")

lib/SPIRV/libSPIRV/SPIRVModule.cpp

@@ -2094,27 +2094,7 @@ void SPIRVModuleImpl::addUnknownStructField(SPIRVTypeStruct *Struct, unsigned I,
 }
 
 static std::string to_string(uint32_t Version) {
-  std::string Res;
-  switch (Version) {
-  case static_cast<uint32_t>(VersionNumber::SPIRV_1_0):
-    Res = "1.0";
-    break;
-  case static_cast<uint32_t>(VersionNumber::SPIRV_1_1):
-    Res = "1.1";
-    break;
-  case static_cast<uint32_t>(VersionNumber::SPIRV_1_2):
-    Res = "1.2";
-    break;
-  case static_cast<uint32_t>(VersionNumber::SPIRV_1_3):
-    Res = "1.3";
-    break;
-  case static_cast<uint32_t>(VersionNumber::SPIRV_1_4):
-    Res = "1.4";
-    break;
-  default:
-    Res = "unknown";
-  }
-
+  std::string Res(formatVersionNumber(Version));
   Res += " (" + std::to_string(Version) + ")";
   return Res;
 }
@@ -2139,9 +2119,7 @@ std::istream &operator>>(std::istream &I, SPIRVModule &M) {
   }
 
   Decoder >> MI.SPIRVVersion;
-  bool SPIRVVersionIsKnown =
-      static_cast<uint32_t>(VersionNumber::MinimumVersion) <= MI.SPIRVVersion &&
-      MI.SPIRVVersion <= static_cast<uint32_t>(VersionNumber::MaximumVersion);
+  bool SPIRVVersionIsKnown = isSPIRVVersionKnown(MI.SPIRVVersion);
   if (!M.getErrorLog().checkError(
           SPIRVVersionIsKnown, SPIRVEC_InvalidModule,
           "unsupported SPIR-V version number '" + to_string(MI.SPIRVVersion) +

test/negative/spirv_report_bad_input.spt

@@ -0,0 +1,33 @@
+; RUN: llvm-spirv %s -to-binary -o %t.spv
+; The next line is to corrupt the binary file by changing its Magic Number
+; RUN: echo "0" > %t_corrupted.spv && cat %t.spv >> %t_corrupted.spv
+; RUN: not llvm-spirv --spirv-print-report %t_corrupted.spv 2>&1 | FileCheck %s --check-prefix=CHECK-ERROR
+;
+; CHECK-ERROR: Invalid SPIR-V binary
+
+119734787 65536 393230 10 0
+2 Capability Addresses
+2 Capability Kernel
+2 Capability LoopFuseINTEL
+2 Capability BitInstructions
+6 Extension "SPV_INTEL_loop_fuse"
+8 Extension "SPV_KHR_bit_instructions"
+5 ExtInstImport 1 "OpenCL.std"
+3 MemoryModel 1 2
+7 EntryPoint 6 5 "TestSatPacked"
+3 Source 3 102000
+
+5 Decorate 5 FuseLoopsInFunctionINTEL 3 1
+4 TypeInt 3 32 0
+2 TypeVoid 2
+5 TypeFunction 4 2 3 3
+
+5 Function 2 5 0 4
+3 FunctionParameter 3 6
+3 FunctionParameter 3 7
+
+2 Label 8
+4 BitReverse 3 9 6
+1 Return
+
+1 FunctionEnd

test/spirv_report.spt

@@ -0,0 +1,43 @@
+; RUN: llvm-spirv %s -to-binary -o %t.spv
+; RUN: llvm-spirv --spirv-print-report %t.spv | FileCheck %s --check-prefix=CHECK-DAG
+
+; CHECK-DAG: Version: 1.0
+; CHECK-DAG: Memory model: OpenCL
+; CHECK-DAG: Addressing model: Physical32
+; CHECK-DAG: Number of capabilities: 4
+; CHECK-DAG: Capability: Addresses
+; CHECK-DAG: Capability: Kernel
+; CHECK-DAG: Capability: LoopFuseINTEL
+; CHECK-DAG: Capability: BitInstructions
+; CHECK-DAG: Number of extensions: 2
+; CHECK-DAG: Extension: SPV_INTEL_loop_fuse
+; CHECK-DAG: Extension: SPV_KHR_bit_instructions
+; CHECK-DAG: Number of extended instruction sets: 1
+; CHECK-DAG: Extended Instruction Set: OpenCL.std
+
+119734787 65536 393230 10 0
+2 Capability Addresses
+2 Capability Kernel
+2 Capability LoopFuseINTEL
+2 Capability BitInstructions
+6 Extension "SPV_INTEL_loop_fuse"
+8 Extension "SPV_KHR_bit_instructions"
+5 ExtInstImport 1 "OpenCL.std"
+3 MemoryModel 1 2
+7 EntryPoint 6 5 "TestSatPacked"
+3 Source 3 102000
+
+5 Decorate 5 FuseLoopsInFunctionINTEL 3 1
+4 TypeInt 3 32 0
+2 TypeVoid 2
+5 TypeFunction 4 2 3 3
+
+5 Function 2 5 0 4
+3 FunctionParameter 3 6
+3 FunctionParameter 3 7
+
+2 Label 8
+4 BitReverse 3 9 6
+1 Return
+
+1 FunctionEnd

tools/llvm-spirv/llvm-spirv.cpp

@@ -196,6 +196,12 @@ static cl::opt<bool> SpecConstInfo(
     cl::desc("Display id of constants available for specializaion and their "
              "size in bytes"));
 
+static cl::opt<bool>
+    SPIRVPrintReport("spirv-print-report", cl::init(false),
+                     cl::desc("Display general information about the module "
+                              "(capabilities, extensions, version, memory model"
+                              " and addressing model)"));
+
 static cl::opt<SPIRV::FPContractMode> FPCMode(
     "spirv-fp-contract", cl::desc("Set FP Contraction mode:"),
     cl::init(SPIRV::FPContractMode::On),
@@ -797,7 +803,7 @@ int main(int Ac, char **Av) {
     return convertSPIRV();
 #endif
 
-  if (!IsReverse && !IsRegularization && !SpecConstInfo)
+  if (!IsReverse && !IsRegularization && !SpecConstInfo && !SPIRVPrintReport)
     return convertLLVMToSPIRV(Opts);
 
   if (IsReverse && IsRegularization) {
@@ -824,5 +830,39 @@ int main(int Ac, char **Av) {
                 << ", size in bytes = " << SpecConst.Size
                 << ", type = " << SpecConst.Type << "\n";
   }
+
+  if (SPIRVPrintReport) {
+    std::ifstream IFS(InputFile, std::ios::binary);
+    int ErrCode = 0;
+    std::optional<SPIRV::SPIRVModuleReport> BinReport =
+        SPIRV::getSpirvReport(IFS, ErrCode);
+    if (!BinReport) {
+      std::cerr << "Invalid SPIR-V binary, error code is " << ErrCode << "\n";
+      return -1;
+    }
+
+    SPIRV::SPIRVModuleTextReport TextReport =
+        SPIRV::formatSpirvReport(BinReport.value());
+
+    std::cout << "SPIR-V module report:"
+              << "\n Version: " << TextReport.Version
+              << "\n Memory model: " << TextReport.MemoryModel
+              << "\n Addressing model: " << TextReport.AddrModel << "\n";
+
+    std::cout << " Number of capabilities: " << TextReport.Capabilities.size()
+              << "\n";
+    for (auto &Capability : TextReport.Capabilities)
+      std::cout << "  Capability: " << Capability << "\n";
+
+    std::cout << " Number of extensions: " << TextReport.Extensions.size()
+              << "\n";
+    for (auto &Extension : TextReport.Extensions)
+      std::cout << "  Extension: " << Extension << "\n";
+
+    std::cout << " Number of extended instruction sets: "
+              << TextReport.ExtendedInstructionSets.size() << "\n";
+    for (auto &ExtendedInstructionSet : TextReport.ExtendedInstructionSets)
+      std::cout << "  Extended Instruction Set: " << ExtendedInstructionSet << "\n";
+  }
   return 0;
 }