feat: first prototype of optimizers + alloca optimizer

.gitignore

@@ -1 +1,2 @@
 build/
+.cache/

CMakeLists.txt

@@ -1,6 +1,8 @@
 cmake_minimum_required(VERSION 3.16.0 FATAL_ERROR)
 project(pl VERSION 0.0.1 LANGUAGES C CXX)
 
+set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
+
 set(CMAKE_CXX_STANDARD 23)
 set(SOURCES
     src/main.cpp

include/ir/ir.hpp

@@ -1,7 +1,7 @@
 #pragma once
 #include <unordered_map>
+#include "parser/nodes.hpp"
 #include "prelude/string.hpp"
-#include "parser/ast.hpp"
 #include "ir/value.hpp"
 #include "ir/ops.hpp"
 
@@ -26,7 +26,7 @@ namespace IR
         {
             auto value = ParseExpression(varDecl->value());
             // TODO: gather type information from var decl signature, aka local <int> v = 0;
-            auto dst = AllocateNamed<Pointer>();
+            auto dst = AllocateNamed<Pointer>(value->GetType());
             m_ops->Push(new AllocateOp(dst, value->GetType()));
             m_ops->Push(new StoreOp(value, reinterpret_cast<Pointer *>(dst)));
             m_locals.insert(std::make_pair(varDecl->name(), reinterpret_cast<Pointer *>(dst)));
@@ -34,14 +34,14 @@ namespace IR
 
         ValueHandle *ParseVariable(const VariableNode *var)
         {
-            // auto dst = AllocateValue();
-            // m_ops->Push(new LoadOp(dst, var->name()));
             if (m_locals.find(var->name()) == m_locals.end())
             {
                 // TODO: throw proper error
                 assert(0 && "ERROR: variable does not exist");
             }
-            return reinterpret_cast<ValueHandle *>(m_locals[var->name()]);
+            auto dst = AllocateNamed<Instruction>(m_locals[var->name()]->GetValueType());
+            m_ops->Push(new LoadOp(dst, m_locals[var->name()]));
+            return reinterpret_cast<ValueHandle *>(dst);
         }
 
         ValueHandle *ParseFnCall(const FnCallNode *fn)
@@ -54,7 +54,7 @@ namespace IR
                 argRegs.Push(arg);
             }
             // TODO: gather return type of the function
-            auto dst = AllocateUnnamed<Void>();
+            auto dst = VoidValue();
             m_ops->Push(new CallOp(dst, fn->name(), argRegs.view()));
             return dst;
         }
@@ -75,7 +75,7 @@ namespace IR
             }
 
             assert(0 && "unreachable");
-            return reinterpret_cast<ValueHandle *>(new Void(0));
+            return reinterpret_cast<ValueHandle *>(new Void());
         }
 
         ValueHandle *ParseExpression(const Node *expression)
@@ -187,12 +187,18 @@ namespace IR
             return new V(ValueHandle::kNoId, std::forward<Args>(args)...);
         }
 
+        ValueHandle *VoidValue()
+        {
+            return new Void();
+        }
+
     private:
         const Node *m_root = nullptr;
 
         OpBuilder *m_ops = nullptr;
         unsigned int m_value_counter = 0;
         unsigned int m_block_counter = 0;
+
         std::unordered_map<StringView, Pointer *> m_locals;
 
         Builder<OpBuilder *> m_containers;

include/ir/ops.hpp

@@ -1,6 +1,5 @@
 #pragma once
 #include "prelude/string.hpp"
-#include "parser/nodes.hpp"
 #include "ir/value.hpp"
 #include "ir/op.hpp"
 #include "ir/block.hpp"
@@ -21,7 +20,7 @@ namespace IR
         {
             StringBuilder sb;
             sb.AppendIndent(indent);
-            sb << "EXTRN " << m_symbol.c_str();
+            sb << "extern " << m_symbol.c_str();
             return sb.view();
         }
 
@@ -46,7 +45,7 @@ namespace IR
         {
             StringBuilder sb;
             sb.AppendIndent(indent);
-            sb << "LABEL " << m_name.c_str() << ':' << '\n';
+            sb << "label " << m_name.c_str() << ':' << '\n';
             sb << m_body.Format(indent);
             return sb.view();
         }
@@ -79,7 +78,7 @@ namespace IR
         {
             StringBuilder sb;
             sb.AppendIndent(indent);
-            sb << result()->Format() << " = ALLOCATE " << m_typ->Format();
+            sb << result()->Format() << " = allocate " << m_typ->Format();
             return sb.view();
         }
 
@@ -90,8 +89,8 @@ namespace IR
     class LoadOp : public OpValued
     {
     public:
-        LoadOp(ValueHandle *dest, StringView addr)
+        LoadOp(ValueHandle *dest, Pointer *ptr)
-            : OpValued(dest), m_addr(addr) {}
+            : OpValued(dest), m_ptr(ptr) {}
         ~LoadOp() {}
 
         OP_TYPE(LOAD)
@@ -100,15 +99,12 @@ namespace IR
         {
             StringBuilder sb;
             sb.AppendIndent(indent);
-            sb << result()->Format() << " = LOAD \"" << m_addr.c_str() << "\"";
+            sb << result()->Format() << " = load " << m_ptr->Format();
             return sb.view();
         }
 
-    public:
-        const StringView &addr() const { return m_addr; }
-
     private:
-        StringView m_addr;
+        Pointer* m_ptr;
     };
 
     class StoreOp : public Op
@@ -124,7 +120,7 @@ namespace IR
         {
             StringBuilder sb;
             sb.AppendIndent(indent);
-            sb << "STORE " << src()->Format() << ", " << dst()->Format();
+            sb << "store " << src()->Format() << ", " << dst()->Format();
             return sb.view();
         }
 
@@ -152,13 +148,14 @@ namespace IR
             switch (m_typ)
             {
             case OpType::ADD:
-                return StringView("ADD");
+                return StringView("add");
             case OpType::MUL:
-                return StringView("MUL");
+                return StringView("mul");
             case OpType::SUB:
-                return StringView("SUB");
+                return StringView("sub");
             case OpType::DIV:
-                return StringView("DIV");
+                return StringView("div");
+            default: break;
             }
 
             assert(false && "unreachable");
@@ -198,10 +195,10 @@ namespace IR
             for (size_t i = 0; i < m_args.size; ++i)
             {
                 sb.AppendIndent(indent);
-                sb << "PARAM " << m_args.data[i]->Format() << '\n';
+                sb << "param " << m_args.data[i]->Format() << '\n';
             }
             sb.AppendIndent(indent);
-            sb << result()->Format() << " = CALL " << m_callee.c_str();
+            sb << result()->Format() << " = call " << m_callee.c_str();
             return sb.view();
         }
 

include/ir/optimize.hpp

@@ -0,0 +1,34 @@
+#pragma once
+#include "ir/op.hpp"
+#include "prelude/linkedlist.hpp"
+
+namespace IR {
+
+class Optimizer
+{
+public:
+    virtual bool Apply(DoubleLinkedList<Op*>& ops) = 0;
+};
+
+class AllocOptimizer : public Optimizer
+{
+public:
+    bool Apply(DoubleLinkedList<Op*>& ops) override
+    {
+        for (ListNode<Op*>* cur = ops.Begin(); cur != nullptr; )
+        {
+            auto op = cur->value;
+            if (op->GetType() == OpType::ALLOCATE)
+            {
+                auto tmp = cur;
+                cur = cur->next;
+                ops.SpliceFront(tmp);
+            } else {
+                cur = cur->next;
+            }
+        }
+        return true;
+    }
+};
+
+}

include/ir/value.hpp

@@ -1,4 +1,5 @@
 #pragma once
+#include <cstdint>
 #include "prelude/string.hpp"
 
 namespace IR
@@ -104,8 +105,8 @@ namespace IR
     class Pointer : public ValueHandle
     {
     public:
-        Pointer(unsigned int id)
+        Pointer(unsigned int id, const Type* value_typ)
-            : ValueHandle(id)
+            : ValueHandle(id), m_value_typ(new Type(value_typ))
         {
             m_type = new Type{Type::Kind::Ptr};
         }
@@ -120,16 +121,19 @@ namespace IR
             return sb.view();
         }
 
+    public:
+        const Type *GetValueType() const { return m_value_typ; }
+
     private:
         Type *m_type;
+        Type *m_value_typ;
     };
 
-    // TODO: Remove void value and use void type only
     class Void : public ValueHandle
     {
     public:
-        Void(unsigned int id)
+        Void()
-            : ValueHandle(id)
+            : ValueHandle()
         {
             m_type = new Type{Type::Kind::Void};
         }

include/parser/ast.hpp

@@ -1,7 +1,5 @@
 #pragma once
-#include <cstring>
 #include <cassert>
-#include <vector>
 #include "parser/lexer.hpp"
 #include "parser/nodes.hpp"
 

include/parser/lexer.hpp

@@ -1,5 +1,4 @@
 #pragma once
-#include <iostream>
 #include <cstring>
 #include <cctype>
 

include/prelude/file.hpp

@@ -0,0 +1,79 @@
+#pragma once
+#include "prelude/string.hpp"
+#include <cstdio>
+
+class File
+{
+public:
+    enum class Mode
+    {
+        WRITE,
+        READ,
+    };
+
+private:
+    File(FILE* handle, Mode mode)
+        : m_handle(handle), m_mode(mode) {}
+
+public:
+    static File Open(const StringView& filename, Mode mode)
+    {
+        StringBuilder sb;
+        switch(mode)
+        {
+        case Mode::WRITE:
+            sb.Append('w');
+            break;
+        case Mode::READ:
+            sb.Append('r');
+            break;
+        }
+        FILE* handle = fopen(filename.c_str(), sb.c_str());
+        return File(handle, mode);
+    }
+
+private:
+    inline bool Writable() const { return m_mode == File::Mode::WRITE; }
+    inline bool Readable() const { return m_mode == File::Mode::READ; }
+
+public:
+    bool Write(const StringView& content)
+    {
+        assert(Writable() && "attempt to write to a file in read only mode");
+        if (fwrite(content.c_str(), sizeof(char), content.size - 1, m_handle) == 0)
+        {
+            return false;
+        }
+        return true;
+    }
+
+public:
+    StringView ReadAll(bool* status = nullptr)
+    {
+        auto fail = [&]() -> StringView {
+            if (status) *status = false;
+            return {};
+        };
+
+        if (status) *status = true;
+
+        assert(Readable() && "attempt to read from a file in write only mode");
+        if (fseek(m_handle, 0, SEEK_END) != 0) return fail();
+
+        auto size = ftell(m_handle);
+        if (size == -1 && status) return fail();
+
+        StringBuilder sb;
+        sb.ensure_extra(size);
+
+        size_t wrote = fread(sb.data, sizeof(char), size, m_handle);
+        if (wrote == 0) return fail();
+
+        sb.size = wrote;
+        return sb.view();
+    }
+
+private:
+    FILE* m_handle;
+    Mode m_mode;
+};

include/prelude/linkedlist.hpp

@@ -0,0 +1,85 @@
+#pragma once
+#include <cstddef>
+#include "prelude/string.hpp"
+
+template<typename T>
+struct ListNode
+{
+    T value;
+    ListNode* prev = nullptr;
+    ListNode* next = nullptr;
+};
+
+template<typename T>
+class DoubleLinkedList {
+public:
+    DoubleLinkedList() = default;
+
+public:
+    static DoubleLinkedList<T> FromView(const View<T> &view)
+    {
+        auto list = DoubleLinkedList<T>();
+        for (size_t i = 0; i < view.size; ++i)
+        {
+            list.Append(new ListNode<T>(view.data[i]));
+        }
+        return list;
+    }
+
+public:
+    void Append(ListNode<T>* node)
+    {
+        if (IsEmpty())
+        {
+            node->next = nullptr;
+            node->prev = nullptr;
+            tail = node;
+            head = node;
+        } else {
+            head->next = node;
+            node->prev = head;
+            node->next = nullptr;
+            head = node;
+        }
+        m_size++;
+    }
+
+    void Prepend(ListNode<T>* node)
+    {
+        if (IsEmpty())
+        {
+            Append(node);
+        } else {
+            tail->prev = node;
+            node->next = tail;
+            node->prev = nullptr;
+            tail = node;
+            m_size++;
+        }
+    }
+
+    void SpliceFront(ListNode<T>* node)
+    {
+        if (m_size == 1 && tail == node && head == node) return;
+        assert((node->next || node->prev) && "node should be inside of the linked list");
+        if (node == tail) return;
+        if (node == head) {
+            node->prev->next = nullptr;
+            head = node->prev;
+        } else {
+            node->prev->next = node->next;
+            node->next->prev = node->prev;
+        }
+        m_size--;
+        Prepend(node);
+    }
+
+    bool IsEmpty() const { return m_size == 0; }
+
+    ListNode<T>* Begin() const { return tail; }
+    ListNode<T>* End() const { return head; }
+private:
+    ListNode<T>* head = nullptr;
+    ListNode<T>* tail = nullptr;
+    size_t m_size = 0;
+};

include/prelude/string.hpp

@@ -68,8 +68,6 @@ public:
 };
 
 
-// using StringView = View<char>;
-
 class StringView final : public View<char>
 {
 public:

src/main.cpp

@@ -1,11 +1,14 @@
-#include <iostream>
 #include <fstream>
 #include <cstdio>
 #include <cstdlib>
 #include <print>
+#include "ir/op.hpp"
+#include "ir/optimize.hpp"
 #include "parser/lexer.hpp"
 #include "parser/ast.hpp"
 #include "ir/ir.hpp"
+#include "prelude/string.hpp"
+#include "prelude/linkedlist.hpp"
 
 // #include "codegen/fasm_stack.hpp"
 
@@ -59,8 +62,36 @@ int main(int argc, char **argv)
 
     for (size_t i = 0; i < ops.size; ++i)
     {
-        printf("%s\n", ops.data[i]->Format(0).c_str());
+        auto op = ops.data[i];
+        if (op->GetType() == IR::OpType::FN)
+        {
+            auto fn = reinterpret_cast<IR::FnOp*>(op);
+            auto blockList = DoubleLinkedList<IR::Op*>::FromView(fn->body().ops());
+
+            auto allocaoptimz = IR::AllocOptimizer();
+            printf("fn <%s>: status = %d\n", fn->name().c_str(), allocaoptimz.Apply(blockList));
+            printf("fn %s:\n", fn->name().c_str());
+            for (ListNode<IR::Op*>* cur = blockList.Begin(); cur != nullptr; cur = cur->next)
+            {
+                printf("%s\n", cur->value->Format(2).c_str());
             }
+        }
+    }
+
+    // StringBuilder sb;
+    // for (size_t i = 0; i < ops.size; ++i)
+    // {
+    //     sb.AppendFormat("%s\n", ops.data[i]->Format(0).c_str());
+    // }
+
+    // printf("%s\n", sb.c_str());
+
+    // auto output = File::Open("example.ll", File::Mode::WRITE);
+    // if (!output.Write(sb.view()))
+    // {
+    //     fprintf(stderr, "ERROR: Failed to write IR to a file");
+    // }
+    // std::println("OK");
 
     // StackFasmX86_64Generator gen;