[MIR] Complete document parsing and printing

Implement the single LLVM-compatible MIR document path with structured\nembedded-IR references, YAML mappings, constants, and deterministic printing.\n\nKeep pointer LLT widths DataLayout-derived, reject layout-dependent output\nwhen embedded IR is omitted, and record the fixed-baseline compatibility\ncontract and regression coverage.

5 files changed, 260 insertions, 3 deletions

diff --git a/backend/include/rcc/MIR/EmbeddedIR.h b/backend/include/rcc/MIR/EmbeddedIR.h
index 8e9e0f6..50567de 100644
--- a/backend/include/rcc/MIR/EmbeddedIR.h
+++ b/backend/include/rcc/MIR/EmbeddedIR.h
@@ -24,6 +24,20 @@ struct EmbeddedIRReference final {
   TextRange Range;
 };
 
+struct EmbeddedIRMetadataDefinition final {
+  std::string Name;
+  TextRange Range;
+  std::string DynamicType;
+};
+
+struct EmbeddedIRTypeDefinition final {
+  std::string Name;
+  TextRange Range;
+  std::string Definition;
+  TextRange DefinitionRange;
+  bool NumericName = false;
+};
+
 class EmbeddedIRFunction final {
 public:
   [[nodiscard]] std::string_view name() const noexcept {
@@ -46,8 +60,13 @@ public:
     return Allocas;
   }
 
+  [[nodiscard]] std::span<const EmbeddedIRReference> values() const noexcept {
+    return Values;
+  }
+
   [[nodiscard]] const EmbeddedIRReference *block(std::string_view QueryName) const noexcept;
   [[nodiscard]] const EmbeddedIRReference *alloca(std::string_view QueryName) const noexcept;
+  [[nodiscard]] const EmbeddedIRReference *value(std::string_view QueryName) const noexcept;
 
 private:
   struct StringHash final {
@@ -63,8 +82,10 @@ private:
   bool IsDefinition = false;
   std::vector<EmbeddedIRReference> Blocks;
   std::vector<EmbeddedIRReference> Allocas;
+  std::vector<EmbeddedIRReference> Values;
   std::unordered_map<std::string, std::size_t, StringHash, std::equal_to<>> BlockNameToIndex;
   std::unordered_map<std::string, std::size_t, StringHash, std::equal_to<>> AllocaNameToIndex;
+  std::unordered_map<std::string, std::size_t, StringHash, std::equal_to<>> ValueNameToIndex;
 
   friend class EmbeddedIRIndexer;
 };
@@ -101,13 +122,19 @@ public:
     return GlobalObjects;
   }
 
-  [[nodiscard]] std::span<const EmbeddedIRReference> metadataDefinitions() const noexcept {
+  [[nodiscard]] std::span<const EmbeddedIRMetadataDefinition>
+  metadataDefinitions() const noexcept {
     return MetadataDefinitions;
   }
 
+  [[nodiscard]] std::span<const EmbeddedIRTypeDefinition> typeDefinitions() const noexcept {
+    return TypeDefinitions;
+  }
+
   [[nodiscard]] const EmbeddedIRFunction *function(std::string_view Name) const noexcept;
   [[nodiscard]] const EmbeddedIRReference *globalObject(std::string_view Name) const noexcept;
-  [[nodiscard]] const EmbeddedIRReference *metadata(std::string_view Name) const noexcept;
+  [[nodiscard]] const EmbeddedIRMetadataDefinition *metadata(std::string_view Name) const noexcept;
+  [[nodiscard]] const EmbeddedIRTypeDefinition *typeDefinition(std::string_view Name) const noexcept;
   [[nodiscard]] bool hasGlobalValue(std::string_view Name) const noexcept {
     return function(Name) != nullptr || globalObject(Name) != nullptr;
   }
@@ -131,10 +158,12 @@ private:
   std::optional<TextRange> DataLayoutRange;
   std::vector<EmbeddedIRFunction> Functions;
   std::vector<EmbeddedIRReference> GlobalObjects;
-  std::vector<EmbeddedIRReference> MetadataDefinitions;
+  std::vector<EmbeddedIRMetadataDefinition> MetadataDefinitions;
+  std::vector<EmbeddedIRTypeDefinition> TypeDefinitions;
   std::unordered_map<std::string, std::size_t, StringHash, std::equal_to<>> FunctionNameToIndex;
   std::unordered_map<std::string, std::size_t, StringHash, std::equal_to<>> GlobalNameToIndex;
   std::unordered_map<std::string, std::size_t, StringHash, std::equal_to<>> MetadataNameToIndex;
+  std::unordered_map<std::string, std::size_t, StringHash, std::equal_to<>> TypeNameToIndex;
 
   friend class EmbeddedIRIndexer;
 };
@@ -159,6 +188,10 @@ public:
     return Diagnostics;
   }
 
+  [[nodiscard]] std::optional<EmbeddedIRReferenceIndex> takeIndex() && {
+    return std::move(Index);
+  }
+
 private:
   EmbeddedIRIndexResult() = default;
 
diff --git a/backend/include/rcc/MIR/LLVMConstantSyntax.h b/backend/include/rcc/MIR/LLVMConstantSyntax.h
new file mode 100644
index 0000000..9d7a7d4
--- /dev/null
+++ b/backend/include/rcc/MIR/LLVMConstantSyntax.h
@@ -0,0 +1,52 @@
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#ifndef RCC_MIR_LLVMCONSTANTSYNTAX_H
+#define RCC_MIR_LLVMCONSTANTSYNTAX_H
+
+#include "rcc/CodeGen/GenericConstant.h"
+#include "rcc/Support/Error.h"
+
+#include <cstddef>
+#include <cstdint>
+#include <span>
+#include <string>
+#include <string_view>
+
+namespace rcc::mir {
+
+struct LLVMNamedTypeDefinition final {
+  std::string_view Name;
+  std::string_view Definition;
+  bool NumericName = false;
+};
+
+struct LLVMConstantSyntaxOptions final {
+  std::uint32_t DefaultPointerSizeInBits = 0;
+  std::string_view DataLayout;
+  std::span<const LLVMNamedTypeDefinition> NamedTypes;
+  std::size_t MaximumDepth = 1024;
+  std::size_t MaximumNodes = std::size_t{1} << 24U;
+  std::size_t MaximumTokenLength = std::size_t{1} << 24U;
+};
+
+[[nodiscard]] Expected<GenericConstantData>
+parseLLVMConstant(std::string_view Text, LLVMConstantSyntaxOptions Options = {});
+
+[[nodiscard]] Expected<GenericConstantType> parseLLVMType(std::string_view Text,
+                                                          LLVMConstantSyntaxOptions Options = {});
+
+[[nodiscard]] Expected<std::string> printLLVMConstant(const GenericConstantData &Constant);
+
+[[nodiscard]] Expected<std::uint64_t>
+llvmConstantStorageSize(const GenericConstantType &Type, LLVMConstantSyntaxOptions Options = {});
+
+[[nodiscard]] Expected<Alignment>
+llvmConstantPreferredAlignment(const GenericConstantType &Type,
+                               LLVMConstantSyntaxOptions Options = {});
+
+[[nodiscard]] Expected<std::uint32_t> llvmPointerSizeInBits(std::uint32_t AddressSpace,
+                                                            LLVMConstantSyntaxOptions Options = {});
+
+} // namespace rcc::mir
+
+#endif // RCC_MIR_LLVMCONSTANTSYNTAX_H
diff --git a/backend/include/rcc/MIR/MIDebugSyntax.h b/backend/include/rcc/MIR/MIDebugSyntax.h
new file mode 100644
index 0000000..a8427dc
--- /dev/null
+++ b/backend/include/rcc/MIR/MIDebugSyntax.h
@@ -0,0 +1,17 @@
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#ifndef RCC_MIR_MIDEBUGSYNTAX_H
+#define RCC_MIR_MIDEBUGSYNTAX_H
+
+#include <cstdint>
+#include <optional>
+#include <string_view>
+
+namespace rcc::mir {
+
+[[nodiscard]] std::optional<std::uint64_t>
+parseDIExpressionOpcodeName(std::string_view Name) noexcept;
+
+} // namespace rcc::mir
+
+#endif // RCC_MIR_MIDEBUGSYNTAX_H
diff --git a/backend/include/rcc/MIR/MIParser.h b/backend/include/rcc/MIR/MIParser.h
index 978e928..116a97f 100644
--- a/backend/include/rcc/MIR/MIParser.h
+++ b/backend/include/rcc/MIR/MIParser.h
@@ -5,15 +5,18 @@
 
 #include "rcc/CodeGen/MachineIRTypes.h"
 #include "rcc/CodeGen/MachineMetadata.h"
+#include "rcc/CodeGen/Register.h"
 #include "rcc/Support/Error.h"
 #include "rcc/Support/TextDiagnostic.h"
 
 #include <cstddef>
 #include <cstdint>
+#include <functional>
 #include <memory>
 #include <optional>
 #include <span>
 #include <string>
+#include <string_view>
 #include <vector>
 
 namespace rcc {
@@ -23,6 +26,11 @@ class TargetDescription;
 
 namespace mir {
 
+struct MIVirtualRegisterSlot final {
+  std::uint32_t TextID = 0;
+  Register Reg;
+};
+
 struct MIFrameObjectSlot final {
   std::uint32_t TextID = 0;
   FrameIndex Index;
@@ -50,12 +58,15 @@ struct MISyncScopeSlot final {
 };
 
 struct MIParserContext final {
+  std::vector<MIVirtualRegisterSlot> VirtualRegisters;
   std::vector<MIFrameObjectSlot> StackObjects;
   std::vector<MIFrameObjectSlot> FixedStackObjects;
   std::vector<MIConstantPoolSlot> ConstantPoolItems;
   std::vector<MIJumpTableSlot> JumpTables;
   std::vector<MIMetadataSlot> Metadata;
   std::vector<MISyncScopeSlot> SyncScopes;
+  std::function<Expected<SyncScopeID>(std::string_view)> InternSyncScope;
+  std::function<Expected<std::uint32_t>(std::uint32_t)> PointerSizeInBits;
   std::uint32_t DefaultPointerSizeInBits = 0;
 };
 
@@ -66,6 +77,7 @@ struct MIParserOptions final {
   std::size_t MaximumInstructions = std::size_t{1} << 24U;
   std::size_t MaximumOperandsPerInstruction = std::size_t{1} << 20U;
   std::size_t MaximumMemoryOperandsPerInstruction = std::size_t{1} << 20U;
+  bool VerifyAfterParse = true;
 };
 
 class MIParser final {
diff --git a/backend/include/rcc/MIR/MIRDocument.h b/backend/include/rcc/MIR/MIRDocument.h
new file mode 100644
index 0000000..7980ca1
--- /dev/null
+++ b/backend/include/rcc/MIR/MIRDocument.h
@@ -0,0 +1,143 @@
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#ifndef RCC_MIR_MIRDOCUMENT_H
+#define RCC_MIR_MIRDOCUMENT_H
+
+#include "rcc/CodeGen/CompilationUnit.h"
+#include "rcc/CodeGen/StableID.h"
+#include "rcc/MIR/EmbeddedIR.h"
+#include "rcc/MIR/MIParser.h"
+#include "rcc/MIR/MIPrinter.h"
+#include "rcc/Support/TextDiagnostic.h"
+#include "rcc/Support/YAML.h"
+
+#include <cstdint>
+#include <optional>
+#include <span>
+#include <string>
+#include <utility>
+#include <vector>
+
+namespace rcc {
+
+class TargetDescription;
+
+namespace mir {
+
+struct MIRFunctionDocumentInfo final {
+  MachineFunctionStableID Function;
+  std::optional<bool> NoPHIs;
+  std::optional<bool> IsSSA;
+  std::optional<bool> NoVRegs;
+  std::optional<bool> HasFakeUses;
+
+  friend bool operator==(const MIRFunctionDocumentInfo &Left,
+                         const MIRFunctionDocumentInfo &Right) = default;
+};
+
+class MIRDocument final {
+public:
+  MIRDocument();
+  explicit MIRDocument(CompilationUnit Unit);
+  MIRDocument(const MIRDocument &) = delete;
+  MIRDocument &operator=(const MIRDocument &) = delete;
+  MIRDocument(MIRDocument &&) noexcept;
+  MIRDocument &operator=(MIRDocument &&) noexcept;
+  ~MIRDocument();
+
+  [[nodiscard]] CompilationUnit &unit() noexcept {
+    return Unit;
+  }
+
+  [[nodiscard]] const CompilationUnit &unit() const noexcept {
+    return Unit;
+  }
+
+  [[nodiscard]] const EmbeddedIRReferenceIndex *embeddedIR() const noexcept {
+    return EmbeddedIR ? &*EmbeddedIR : nullptr;
+  }
+
+  [[nodiscard]] std::span<const MIRFunctionDocumentInfo> functionDocumentInfo() const noexcept {
+    return FunctionInfo;
+  }
+
+  [[nodiscard]] std::span<const MISyncScopeSlot> syncScopes() const noexcept {
+    return SyncScopes;
+  }
+
+private:
+  CompilationUnit Unit;
+  std::optional<EmbeddedIRReferenceIndex> EmbeddedIR;
+  std::vector<MIRFunctionDocumentInfo> FunctionInfo;
+  std::vector<MISyncScopeSlot> SyncScopes;
+
+  friend class MIRDocumentParser;
+};
+
+struct MIRDocumentParserOptions final {
+  yaml::ParseOptions YAML;
+  MIParserOptions MachineInstructions;
+  std::string DataLayout;
+  std::uint32_t DefaultPointerSizeInBits = 0;
+  bool VerifyMachineFunctions = true;
+};
+
+class MIRDocumentParseResult final {
+public:
+  MIRDocumentParseResult(const MIRDocumentParseResult &) = delete;
+  MIRDocumentParseResult &operator=(const MIRDocumentParseResult &) = delete;
+  MIRDocumentParseResult(MIRDocumentParseResult &&) noexcept;
+  MIRDocumentParseResult &operator=(MIRDocumentParseResult &&) noexcept;
+  ~MIRDocumentParseResult();
+
+  [[nodiscard]] bool succeeded() const noexcept {
+    return Document.has_value() && Diagnostics.empty();
+  }
+
+  [[nodiscard]] MIRDocument *document() noexcept {
+    return Document ? &*Document : nullptr;
+  }
+
+  [[nodiscard]] const MIRDocument *document() const noexcept {
+    return Document ? &*Document : nullptr;
+  }
+
+  [[nodiscard]] const TextSource &source() const noexcept {
+    return Source;
+  }
+
+  [[nodiscard]] std::span<const TextDiagnostic> diagnostics() const noexcept {
+    return Diagnostics;
+  }
+
+private:
+  MIRDocumentParseResult(std::string SourceName, std::string SourceText)
+      : Source(std::move(SourceName), std::move(SourceText)) {}
+
+  TextSource Source;
+  std::optional<MIRDocument> Document;
+  std::vector<TextDiagnostic> Diagnostics;
+
+  friend class MIRDocumentParser;
+};
+
+[[nodiscard]] MIRDocumentParseResult parseMIRDocument(const TargetDescription &Target,
+                                                      std::string SourceName,
+                                                      std::string SourceText,
+                                                      MIRDocumentParserOptions Options = {});
+
+struct MIRDocumentPrinterOptions final {
+  MIPrintMode Mode = MIPrintMode::Complete;
+  bool EmitEmbeddedIR = true;
+  std::string DataLayout;
+  std::uint32_t DefaultPointerSizeInBits = 0;
+};
+
+[[nodiscard]] Expected<std::string> printMIRDocument(const MIRDocument &Document,
+                                                     const TargetDescription &Target,
+                                                     MIRDocumentPrinterOptions Options = {});
+
+} // namespace mir
+} // namespace rcc
+
+#endif // RCC_MIR_MIRDOCUMENT_H