Merge pull request #622 from scratchcpp/fix_loop_type_prediction

Implement type analysis in loops

Author: adazem009

src/engine/internal/llvm/CMakeLists.txt

@@ -16,6 +16,7 @@ target_sources(scratchcpp
     llvmtypes.cpp
     llvmtypes.h
     llvmfunctions.cpp
+    llvmloopscope.h
     llvmcompilercontext.cpp
     llvmcompilercontext.h
     llvmexecutablecode.cpp

src/engine/internal/llvm/llvmcodebuilder.cpp

@@ -20,6 +20,7 @@ namespace libscratchcpp
 
 class LLVMCompilerContext;
 class LLVMConstantRegister;
+class LLVMLoopScope;
 
 class LLVMCodeBuilder : public ICodeBuilder
 {
@@ -119,22 +120,24 @@ class LLVMCodeBuilder : public ICodeBuilder
         void createListMap();
         void pushScopeLevel();
         void popScopeLevel();
+        void pushLoopScope(bool buildPhase);
+        void popLoopScope();
 
         std::string getMainFunctionName(BlockPrototype *procedurePrototype);
         std::string getResumeFunctionName(BlockPrototype *procedurePrototype);
         llvm::FunctionType *getMainFunctionType(BlockPrototype *procedurePrototype);
         llvm::Function *getOrCreateFunction(const std::string &name, llvm::FunctionType *type);
         void verifyFunction(llvm::Function *func);
 
-        LLVMRegister *addReg(std::shared_ptr<LLVMRegister> reg);
+        LLVMRegister *addReg(std::shared_ptr<LLVMRegister> reg, std::shared_ptr<LLVMInstruction> ins);
 
         llvm::Value *addAlloca(llvm::Type *type);
         void freeLater(llvm::Value *value);
         void freeScopeHeap();
         llvm::Value *castValue(LLVMRegister *reg, Compiler::StaticType targetType);
         llvm::Value *castRawValue(LLVMRegister *reg, Compiler::StaticType targetType);
         llvm::Constant *castConstValue(const Value &value, Compiler::StaticType targetType);
-        Compiler::StaticType optimizeRegisterType(LLVMRegister *reg);
+        Compiler::StaticType optimizeRegisterType(LLVMRegister *reg) const;
         llvm::Type *getType(Compiler::StaticType type);
         Compiler::StaticType getProcedureArgType(BlockPrototype::ArgType type);
         llvm::Value *isNaN(llvm::Value *num);
@@ -146,9 +149,15 @@ class LLVMCodeBuilder : public ICodeBuilder
         void reloadVariables(llvm::Value *targetVariables);
         void reloadLists();
         void updateListDataPtr(const LLVMListPtr &listPtr);
+        bool isVarOrListTypeSafe(std::shared_ptr<LLVMInstruction> ins, Compiler::StaticType expectedType) const;
+        bool isVarOrListTypeSafe(std::shared_ptr<LLVMInstruction> ins, Compiler::StaticType expectedType, std::unordered_set<LLVMInstruction *> &log, int &c) const;
+        bool isVarOrListWriteResultTypeSafe(std::shared_ptr<LLVMInstruction> ins, Compiler::StaticType expectedType, bool ignoreSavedType, std::unordered_set<LLVMInstruction *> &log, int &c) const;
 
+        LLVMRegister *createOp(LLVMInstruction::Type type, Compiler::StaticType retType, Compiler::StaticType argType, const Compiler::Args &args);
+        LLVMRegister *createOp(LLVMInstruction::Type type, Compiler::StaticType retType, const Compiler::ArgTypes &argTypes = {}, const Compiler::Args &args = {});
         LLVMRegister *createOp(const LLVMInstruction &ins, Compiler::StaticType retType, Compiler::StaticType argType, const Compiler::Args &args);
         LLVMRegister *createOp(const LLVMInstruction &ins, Compiler::StaticType retType, const Compiler::ArgTypes &argTypes = {}, const Compiler::Args &args = {});
+        std::shared_ptr<LLVMLoopScope> currentLoopScope() const;
 
         void createValueStore(LLVMRegister *reg, llvm::Value *targetPtr, Compiler::StaticType sourceType, Compiler::StaticType targetType);
         void createReusedValueStore(LLVMRegister *reg, llvm::Value *targetPtr, Compiler::StaticType sourceType, Compiler::StaticType targetType);
@@ -215,14 +224,21 @@ class LLVMCodeBuilder : public ICodeBuilder
         llvm::StructType *m_valueDataType = nullptr;
         llvm::FunctionType *m_resumeFuncType = nullptr;
 
-        std::vector<LLVMInstruction> m_instructions;
+        std::vector<std::shared_ptr<LLVMInstruction>> m_instructions;
         std::vector<std::shared_ptr<LLVMRegister>> m_regs;
         std::vector<std::shared_ptr<CompilerLocalVariable>> m_localVars;
         BlockPrototype *m_procedurePrototype = nullptr;
         bool m_defaultWarp = false;
         bool m_warp = false;
         int m_defaultArgCount = 0;
 
+        long m_loopScope = -1; // index
+        std::vector<std::shared_ptr<LLVMLoopScope>> m_loopScopes;
+        long m_loopScopeCounter = 0; // replacement for m_loopScopes size in build phase
+        std::vector<long> m_loopScopeTree;
+        bool m_loopCondition = false; // whether we're currently compiling a loop condition
+        std::vector<std::shared_ptr<LLVMInstruction>> m_variableInstructions;
+        std::vector<std::shared_ptr<LLVMInstruction>> m_listInstructions;
         std::vector<std::vector<llvm::Value *>> m_heap; // scopes
 
         std::shared_ptr<ExecutableCode> m_output;

src/engine/internal/llvm/llvmcodebuilder.h

@@ -10,6 +10,7 @@ namespace libscratchcpp
 {
 
 class BlockPrototype;
+class LLVMLoopScope;
 
 struct LLVMInstruction
 {
@@ -77,8 +78,10 @@ struct LLVMInstruction
             ProcedureArg
         };
 
-        LLVMInstruction(Type type) :
-            type(type)
+        LLVMInstruction(Type type, std::shared_ptr<LLVMLoopScope> loopScope, bool loopCondition) :
+            type(type),
+            loopScope(loopScope),
+            loopCondition(loopCondition)
         {
         }
 
@@ -92,6 +95,8 @@ struct LLVMInstruction
         List *workList = nullptr;         // for lists
         BlockPrototype *procedurePrototype = nullptr;
         size_t procedureArgIndex = 0;
+        std::shared_ptr<LLVMLoopScope> loopScope;
+        bool loopCondition = false; // whether the instruction is part of a loop condition
 };
 
 } // namespace libscratchcpp

src/engine/internal/llvm/llvminstruction.h

@@ -3,6 +3,7 @@
 #pragma once
 
 #include <scratchcpp/compiler.h>
+#include <unordered_map>
 
 namespace llvm
 {
@@ -14,6 +15,9 @@ class Value;
 namespace libscratchcpp
 {
 
+class LLVMLoopScope;
+class LLVMInstruction;
+
 struct LLVMListPtr
 {
         llvm::Value *ptr = nullptr;
@@ -22,6 +26,9 @@ struct LLVMListPtr
         llvm::Value *allocatedSizePtr = nullptr;
         llvm::Value *dataPtrDirty = nullptr;
         Compiler::StaticType type = Compiler::StaticType::Unknown;
+
+        // Used in build phase to check the type safety of lists in loops
+        std::unordered_map<std::shared_ptr<LLVMLoopScope>, std::vector<std::shared_ptr<LLVMInstruction>>> loopListWrites; // loop scope, write instructions
 };
 
 } // namespace libscratchcpp

src/engine/internal/llvm/llvmlistptr.h

@@ -0,0 +1,18 @@
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#include <vector>
+#include <memory>
+
+namespace libscratchcpp
+{
+
+struct LLVMLoopScope
+{
+        bool containsYield = false;
+        std::shared_ptr<LLVMLoopScope> parentScope;
+        std::vector<std::shared_ptr<LLVMLoopScope>> childScopes;
+};
+
+} // namespace libscratchcpp

src/engine/internal/llvm/llvmloopscope.h

@@ -15,12 +15,15 @@ class Value;
 namespace libscratchcpp
 {
 
+class LLVMInstruction;
+
 struct LLVMRegisterBase
 {
         virtual const Value &constValue() const = 0;
 
         llvm::Value *value = nullptr;
         bool isRawValue = false;
+        std::shared_ptr<LLVMInstruction> instruction;
 };
 
 } // namespace libscratchcpp

src/engine/internal/llvm/llvmregisterbase.h

@@ -3,6 +3,7 @@
 #pragma once
 
 #include <scratchcpp/compiler.h>
+#include <unordered_map>
 
 namespace llvm
 {
@@ -14,13 +15,19 @@ class Value;
 namespace libscratchcpp
 {
 
+class LLVMLoopScope;
+class LLVMInstruction;
+
 struct LLVMVariablePtr
 {
         llvm::Value *stackPtr = nullptr;
         llvm::Value *heapPtr = nullptr;
         Compiler::StaticType type = Compiler::StaticType::Unknown;
         bool onStack = false;
         bool changed = false;
+
+        // Used in build phase to check the type safety of variables in loops
+        std::unordered_map<std::shared_ptr<LLVMLoopScope>, std::vector<std::shared_ptr<LLVMInstruction>>> loopVariableWrites; // loop scope, write instructions
 };
 
 } // namespace libscratchcpp