[IR] Split LL parsing, assembly writing, and data layout

8 files changed, 128 insertions, 61 deletions

diff --git a/backend/include/rcc/AsmParser/LLParser.h b/backend/include/rcc/AsmParser/LLParser.h
new file mode 100644
index 0000000..e15a006
--- /dev/null
+++ b/backend/include/rcc/AsmParser/LLParser.h
@@ -0,0 +1,36 @@
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#ifndef RCC_ASMPARSER_LLPARSER_H
+#define RCC_ASMPARSER_LLPARSER_H
+
+#include "rcc/IR/Constants.h"
+#include "rcc/Support/Error.h"
+
+#include <cstddef>
+#include <span>
+#include <string_view>
+
+namespace rcc {
+
+struct LLNamedTypeDefinition final {
+  std::string_view Name;
+  std::string_view Definition;
+  bool NumericName = false;
+};
+
+struct LLParserOptions final {
+  std::span<const LLNamedTypeDefinition> 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,
+                                                              LLParserOptions Options = {});
+
+[[nodiscard]] Expected<GenericConstantType> parseLLVMType(std::string_view Text,
+                                                          LLParserOptions Options = {});
+
+} // namespace rcc
+
+#endif // RCC_ASMPARSER_LLPARSER_H
diff --git a/backend/include/rcc/CodeGen/GlobalObject.h b/backend/include/rcc/CodeGen/GlobalObject.h
index 798dcd8..1ddc34a 100644
--- a/backend/include/rcc/CodeGen/GlobalObject.h
+++ b/backend/include/rcc/CodeGen/GlobalObject.h
@@ -3,9 +3,9 @@
 #ifndef RCC_CODEGEN_GLOBALOBJECT_H
 #define RCC_CODEGEN_GLOBALOBJECT_H
 
-#include "rcc/CodeGen/GenericConstant.h"
 #include "rcc/CodeGen/MachineIRTypes.h"
 #include "rcc/CodeGen/TargetConstant.h"
+#include "rcc/IR/Constants.h"
 #include "rcc/Support/Error.h"
 
 #include <cstddef>
diff --git a/backend/include/rcc/CodeGen/MIRYamlMapping.h b/backend/include/rcc/CodeGen/MIRYamlMapping.h
index 82880ca..c25559b 100644
--- a/backend/include/rcc/CodeGen/MIRYamlMapping.h
+++ b/backend/include/rcc/CodeGen/MIRYamlMapping.h
@@ -9,7 +9,7 @@
 #include "rcc/CodeGen/MachineJumpTableInfo.h"
 #include "rcc/CodeGen/MachineMetadata.h"
 #include "rcc/CodeGen/TargetMIRTextInfo.h"
-#include "rcc/MIR/LLVMConstantSyntax.h"
+#include "rcc/IR/DataLayout.h"
 #include "rcc/Support/Error.h"
 
 #include <cstdint>
@@ -106,7 +106,7 @@ private:
   [[nodiscard]] bool convertTargetNode(const yaml::Node &Source, MIRTargetNode &Destination);
   [[nodiscard]] bool validateDocumentReferences(const MachineFunction &Function);
   [[nodiscard]] std::string_view effectiveDataLayout() const noexcept;
-  [[nodiscard]] LLVMConstantSyntaxOptions pointerLayoutOptions() const noexcept;
+  [[nodiscard]] Expected<DataLayout> dataLayout() const;
 
   [[nodiscard]] MachineBasicBlock *blockAt(MachineFunction &Function,
                                            std::uint32_t Number) const noexcept;
diff --git a/backend/include/rcc/CodeGen/MachineConstantPool.h b/backend/include/rcc/CodeGen/MachineConstantPool.h
index a0c269b..104e559 100644
--- a/backend/include/rcc/CodeGen/MachineConstantPool.h
+++ b/backend/include/rcc/CodeGen/MachineConstantPool.h
@@ -3,10 +3,10 @@
 #ifndef RCC_CODEGEN_MACHINECONSTANTPOOL_H
 #define RCC_CODEGEN_MACHINECONSTANTPOOL_H
 
-#include "rcc/CodeGen/GenericConstant.h"
 #include "rcc/CodeGen/MachineIRTypes.h"
 #include "rcc/CodeGen/MachineMemOperand.h"
 #include "rcc/CodeGen/TargetConstant.h"
+#include "rcc/IR/Constants.h"
 #include "rcc/Support/Error.h"
 
 #include <cstddef>
diff --git a/backend/include/rcc/CodeGen/MachineMemOperand.h b/backend/include/rcc/CodeGen/MachineMemOperand.h
index c5c1357..d236fd3 100644
--- a/backend/include/rcc/CodeGen/MachineMemOperand.h
+++ b/backend/include/rcc/CodeGen/MachineMemOperand.h
@@ -3,10 +3,10 @@
 #ifndef RCC_CODEGEN_MACHINEMEMOPERAND_H
 #define RCC_CODEGEN_MACHINEMEMOPERAND_H
 
-#include "rcc/CodeGen/GenericConstant.h"
 #include "rcc/CodeGen/LowLevelType.h"
 #include "rcc/CodeGen/MachineIRTypes.h"
 #include "rcc/CodeGen/MachineMetadata.h"
+#include "rcc/IR/Constants.h"
 
 #include <cstdint>
 #include <initializer_list>
diff --git a/backend/include/rcc/CodeGen/GenericConstant.h b/backend/include/rcc/IR/Constants.h
index 4219c24..9f894f2 100644
--- a/backend/include/rcc/CodeGen/GenericConstant.h
+++ b/backend/include/rcc/IR/Constants.h
@@ -1,11 +1,10 @@
 // SPDX-License-Identifier: GPL-3.0-or-later
 
-#ifndef RCC_CODEGEN_GENERICCONSTANT_H
-#define RCC_CODEGEN_GENERICCONSTANT_H
+#ifndef RCC_IR_CONSTANTS_H
+#define RCC_IR_CONSTANTS_H
 
 #include "rcc/ADT/APFloat.h"
 #include "rcc/ADT/APInt.h"
-#include "rcc/CodeGen/LowLevelType.h"
 #include "rcc/Support/Error.h"
 
 #include <cstddef>
@@ -63,6 +62,7 @@ struct GenericConstantType final {
   [[nodiscard]] static Expected<GenericConstantType> named(std::string Name,
                                                            bool NumericName = false);
 
+  [[nodiscard]] const GenericConstantType &resolved() const noexcept;
   [[nodiscard]] Expected<void> verify() const;
 
   friend bool operator==(const GenericConstantType &Left,
@@ -164,6 +164,7 @@ struct GenericConstantData final {
   [[nodiscard]] static Expected<GenericConstantData> byteString(GenericConstantType Type,
                                                                 std::vector<std::byte> Bytes);
 
+  [[nodiscard]] Expected<std::string> toLLVMString() const;
   [[nodiscard]] Expected<void> verify() const;
   [[nodiscard]] bool referencesSymbol() const noexcept;
 
@@ -173,4 +174,4 @@ struct GenericConstantData final {
 
 } // namespace rcc
 
-#endif // RCC_CODEGEN_GENERICCONSTANT_H
+#endif // RCC_IR_CONSTANTS_H
diff --git a/backend/include/rcc/IR/DataLayout.h b/backend/include/rcc/IR/DataLayout.h
new file mode 100644
index 0000000..0ca3f07
--- /dev/null
+++ b/backend/include/rcc/IR/DataLayout.h
@@ -0,0 +1,82 @@
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#ifndef RCC_IR_DATALAYOUT_H
+#define RCC_IR_DATALAYOUT_H
+
+#include "rcc/IR/Constants.h"
+#include "rcc/Support/Error.h"
+
+#include <cstddef>
+#include <cstdint>
+#include <string>
+#include <string_view>
+#include <vector>
+
+namespace rcc {
+
+class DataLayout final {
+public:
+  [[nodiscard]] static Expected<DataLayout> parse(std::string_view LayoutString);
+
+  [[nodiscard]] Expected<std::uint64_t> getTypeAllocSize(const GenericConstantType &Type) const;
+  [[nodiscard]] Expected<std::uint64_t> getABITypeAlign(const GenericConstantType &Type) const;
+  [[nodiscard]] Expected<std::uint64_t> getPrefTypeAlign(const GenericConstantType &Type) const;
+  [[nodiscard]] Expected<std::uint32_t> getPointerSizeInBits(std::uint32_t AddressSpace = 0) const;
+
+private:
+  struct AlignmentSpec final {
+    std::uint64_t ABI = 1;
+    std::uint64_t Preferred = 1;
+  };
+
+  struct PrimitiveSpec final {
+    char Kind = '\0';
+    std::uint32_t BitWidth = 0;
+    AlignmentSpec Alignment;
+  };
+
+  struct PointerSpec final {
+    std::uint32_t AddressSpace = 0;
+    std::uint32_t SizeInBits = 0;
+    AlignmentSpec Alignment;
+    std::string AddressSpaceName;
+  };
+
+  struct TypeLayout final {
+    std::uint64_t StoreSize = 0;
+    std::uint64_t AllocationSize = 0;
+    std::uint64_t ABIAlignment = 1;
+    std::uint64_t PreferredAlignment = 1;
+  };
+
+  [[nodiscard]] static Expected<std::uint64_t> parseUnsignedField(std::string_view Text,
+                                                                  std::string_view Description);
+  [[nodiscard]] static Expected<std::uint64_t> alignmentBytes(std::string_view Text, bool AllowZero,
+                                                              std::string_view Description);
+  [[nodiscard]] static std::vector<std::string_view> split(std::string_view Text, char Separator);
+
+  void setPrimitiveSpec(PrimitiveSpec Entry);
+  void setPointerSpec(PointerSpec Entry);
+  [[nodiscard]] AlignmentSpec naturalAlignment(std::uint64_t StoreSize) const noexcept;
+  [[nodiscard]] AlignmentSpec scalarAlignment(char Kind, std::uint32_t BitWidth,
+                                              std::uint64_t StoreSize) const noexcept;
+  [[nodiscard]] AlignmentSpec integerAlignment(std::uint32_t BitWidth,
+                                               std::uint64_t StoreSize) const noexcept;
+  [[nodiscard]] AlignmentSpec floatingAlignment(std::uint32_t BitWidth,
+                                                std::uint64_t StoreSize) const noexcept;
+  [[nodiscard]] AlignmentSpec vectorAlignment(std::uint32_t BitWidth,
+                                              std::uint64_t StoreSize) const noexcept;
+  [[nodiscard]] Expected<PointerSpec> getPointerSpec(std::uint32_t AddressSpace) const;
+  [[nodiscard]] Expected<TypeLayout> getTypeLayout(const GenericConstantType &Type,
+                                                   std::vector<std::string> &NamedTypeStack,
+                                                   std::size_t Depth) const;
+
+  bool Explicit = false;
+  std::vector<PrimitiveSpec> PrimitiveSpecs;
+  std::vector<PointerSpec> PointerSpecs;
+  AlignmentSpec AggregateAlignment{1, 8};
+};
+
+} // namespace rcc
+
+#endif // RCC_IR_DATALAYOUT_H
diff --git a/backend/include/rcc/MIR/LLVMConstantSyntax.h b/backend/include/rcc/MIR/LLVMConstantSyntax.h
deleted file mode 100644
index 7e18d60..0000000
--- a/backend/include/rcc/MIR/LLVMConstantSyntax.h
+++ /dev/null
@@ -1,52 +0,0 @@
-// 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/CodeGen/MachineIRTypes.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::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