elna/backend/riscv.cpp

#include "elna/backend/riscv.hpp"
#include <memory>

namespace elna::riscv
{
    instruction::instruction(base_opcode opcode)
    {
        this->representation = static_cast<std::underlying_type<base_opcode>::type>(opcode);
    }

    instruction& instruction::i(x_register rd, funct3_t funct3, x_register rs1, std::uint32_t immediate)
    {
        this->representation |= (static_cast<std::underlying_type<x_register>::type>(rd) << 7)
            | (static_cast<std::underlying_type<funct3_t>::type>(funct3) << 12)
            | (static_cast<std::underlying_type<x_register>::type>(rs1) << 15)
            | (immediate << 20);

        return *this;
    }

    instruction& instruction::s(std::uint32_t imm, funct3_t funct3, x_register rs1, x_register rs2)
    {
        this->representation |= ((imm & 0x1f) << 7)
            | (static_cast<std::underlying_type<funct3_t>::type>(funct3) << 12)
            | (static_cast<std::underlying_type<x_register>::type>(rs1) << 15)
            | (static_cast<std::underlying_type<x_register>::type>(rs2) << 20)
            | ((imm & 0xfe0) << 20);

        return *this;
    }

    instruction& instruction::b(std::uint32_t imm, funct3_t funct3, x_register rs1, x_register rs2)
    {
        this->representation |= ((imm & 0x800) >> 4) | ((imm & 0x1e) << 7)
            | (static_cast<std::underlying_type<funct3_t>::type>(funct3) << 12)
            | (static_cast<std::underlying_type<x_register>::type>(rs1) << 15)
            | (static_cast<std::underlying_type<x_register>::type>(rs2) << 20)
            | ((imm & 0x7e0) << 20) | ((imm & 0x1000) << 19);

        return *this;
    }

    instruction& instruction::r(x_register rd, funct3_t funct3, x_register rs1, x_register rs2, funct7_t funct7)
    {
        this->representation |= (static_cast<std::underlying_type<x_register>::type>(rd) << 7)
            | (static_cast<std::underlying_type<funct3_t>::type>(funct3) << 12)
            | (static_cast<std::underlying_type<x_register>::type>(rs1) << 15)
            | (static_cast<std::underlying_type<x_register>::type>(rs2) << 20)
            | (static_cast<std::underlying_type<funct7_t>::type>(funct7) << 25);

        return *this;
    }

    instruction& instruction::u(x_register rd, std::uint32_t imm)
    {
        this->representation |= (static_cast<std::underlying_type<x_register>::type>(rd) << 7) | (imm << 12);

        return *this;
    }

    instruction& instruction::j(x_register rd, std::uint32_t imm)
    {
        this->representation |= (static_cast<std::underlying_type<x_register>::type>(rd) << 7)
            | (imm & 0xff000) | ((imm & 0x800) <<  9) | ((imm & 0x7fe) << 20) | ((imm & 0x100000) << 11);

        return *this;
    }

    const std::byte *instruction::cbegin() const
    {
        return reinterpret_cast<const std::byte *>(&this->representation);
    }

    const std::byte *instruction::cend() const
    {
        return reinterpret_cast<const std::byte *>(&this->representation) + sizeof(this->representation);
    }

    visitor::visitor(std::function<void(const std::string&, const std::byte *, std::size_t)> write_text,
            std::function<std::string_view(const std::byte *, std::size_t)> write_read_only)
        : write_text(write_text), write_read_only(write_read_only)
    {
    }

    void visitor::visit(source::declaration *declaration)
    {
    }

    void visitor::visit(source::definition *definition)
    {
    }

    void visitor::visit(source::block *block)
    {
        this->instructions.push_back(instruction(base_opcode::opImm));
        this->instructions.push_back(instruction(base_opcode::store));
        this->instructions.push_back(instruction(base_opcode::store));
        this->instructions.push_back(instruction(base_opcode::opImm));

        table = block->table();
        block->body().accept(this);

        // Set the return value (0).
        this->instructions.push_back(instruction(base_opcode::op)
            .r(x_register::a0, funct3_t::_and, x_register::zero, x_register::zero));

        // Prologue.
        auto main_symbol =
            std::dynamic_pointer_cast<source::procedure_info>(table->lookup("main"));
        const uint stack_size = static_cast<std::uint32_t>(variable_counter * 4 + 8 + main_symbol->stack_size());

        this->instructions[0].i(x_register::sp, funct3_t::addi, x_register::sp, -stack_size);
        this->instructions[1].s(stack_size - 4, funct3_t::sw, x_register::sp, x_register::s0);
        this->instructions[2].s(stack_size - 8, funct3_t::sw, x_register::sp, x_register::ra);
        this->instructions[3].i(x_register::s0, funct3_t::addi, x_register::sp, stack_size);

        // Epilogue.
        this->instructions.push_back(instruction(base_opcode::load)
            .i(x_register::s0, funct3_t::lw, x_register::sp, stack_size - 4));
        this->instructions.push_back(instruction(base_opcode::load)
            .i(x_register::ra, funct3_t::lw, x_register::sp, stack_size - 8));
        this->instructions.push_back(instruction(base_opcode::opImm)
            .i(x_register::sp, funct3_t::addi, x_register::sp, stack_size));
        this->instructions.push_back(instruction(base_opcode::jalr)
            .i(x_register::zero, funct3_t::jalr, x_register::ra, 0));
    }

    void visitor::visit(source::program *program)
    {
        visit(dynamic_cast<source::block *>(program));
        write_text("main", reinterpret_cast<const std::byte *>(this->instructions.data()),
                this->instructions.size() * sizeof(instruction));
    }

    void visitor::visit(source::bang_statement *statement)
    {
        statement->body().accept(this);

        // Print the result.
        this->instructions.push_back(instruction(base_opcode::opImm)
            .i(x_register::a1, funct3_t::addi, x_register::a0, 0));

        auto format_string = write_read_only(reinterpret_cast<const std::byte *>("%d\n\0"), 4);

        this->references.push_back(reference());
        this->references.back().name = format_string;
        this->references.back().offset = instructions.size() * 4;
        this->references.back().target = address_t::high20;
        this->instructions.push_back(instruction(base_opcode::lui).u(x_register::a5, 0));
        this->references.push_back(reference());
        this->references.back().name = format_string;
        this->references.back().offset = instructions.size() * 4;
        this->references.back().target = address_t::lower12i;

        this->instructions.push_back(instruction(base_opcode::opImm)
            .i(x_register::a0, funct3_t::addi, x_register::a5, 0));
        this->references.push_back(reference());
        this->references.back().name = "printf";
        this->references.back().offset = instructions.size() * 4;
        this->references.back().target = address_t::text;
        this->instructions.push_back(instruction(base_opcode::auipc).u(x_register::ra, 0));
        this->instructions.push_back(instruction(base_opcode::jalr)
            .i(x_register::ra, funct3_t::jalr, x_register::ra, 0));
    }

    void visitor::visit(source::question_mark_statement *statement)
    {
        statement->body().accept(this);

        // Print the result.
        this->instructions.push_back(instruction(base_opcode::opImm)
            .i(x_register::a1, funct3_t::addi, x_register::a0, 0));

        auto format_string = write_read_only(reinterpret_cast<const std::byte *>("%d\n\0"), 4);

        this->references.push_back(reference());
        this->references.back().name = format_string;
        this->references.back().offset = instructions.size() * 4;
        this->references.back().target = address_t::high20;
        this->instructions.push_back(instruction(base_opcode::lui).u(x_register::a5, 0));
        this->references.push_back(reference());
        this->references.back().name = format_string;
        this->references.back().offset = instructions.size() * 4;
        this->references.back().target = address_t::lower12i;

        this->instructions.push_back(instruction(base_opcode::opImm)
            .i(x_register::a0, funct3_t::addi, x_register::a5, 0));
        this->references.push_back(reference());
        this->references.back().name = "printf";
        this->references.back().offset = instructions.size() * 4;
        this->references.back().target = address_t::text;
        this->instructions.push_back(instruction(base_opcode::auipc).u(x_register::ra, 0));
        this->instructions.push_back(instruction(base_opcode::jalr)
            .i(x_register::ra, funct3_t::jalr, x_register::ra, 0));
    }

    void visitor::visit(source::compound_statement *statement)
    {
        for (auto& nested_statement : statement->statements())
        {
            nested_statement->accept(this);
        }
    }

    void visitor::visit(source::assign_statement *statement)
    {
        const auto free_register = this->register_in_use ? x_register::a0 : x_register::t0;
        auto symbol = table->lookup(statement->lvalue());
        auto variable_symbol = std::dynamic_pointer_cast<source::variable_info>(symbol);

        statement->rvalue().accept(this);

        this->instructions.push_back(instruction(base_opcode::store)
            .s(variable_symbol->offset, funct3_t::sw, x_register::s0, x_register::a0));
    }

    void visitor::visit(source::if_statement *statement)
    {
        const auto free_register = this->register_in_use ? x_register::a0 : x_register::t0;

        statement->prerequisite().accept(this);

        auto before_branch = instructions.size();
        instructions.push_back(instruction(base_opcode::branch));
        statement->body().accept(this);
        instructions[before_branch]
            .b((instructions.size() - before_branch) * 4 - 3, funct3_t::beq, x_register::zero, free_register);
    }

    void visitor::visit(source::while_statement *statement)
    {
        statement->prerequisite().accept(this);
        statement->body().accept(this);
    }

    void visitor::visit(source::variable_expression *variable)
    {
        const auto free_register = this->register_in_use ? x_register::a0 : x_register::t0;

        auto symbol = table->lookup(variable->name());
        if (auto constant_symbol = std::dynamic_pointer_cast<source::constant_info>(symbol))
        {
            this->instructions.push_back(
                instruction(base_opcode::opImm) // movl $x, %eax; where $x is a number.
                    .i(free_register, funct3_t::addi, x_register::zero, constant_symbol->value())
            );
        }
        else if (auto variable_symbol = std::dynamic_pointer_cast<source::variable_info>(symbol))
        {
            this->instructions.push_back(
                instruction(base_opcode::store)
                    .i(free_register, funct3_t::lw, x_register::s0, variable_symbol->offset)
            );
        }
    }

    void visitor::visit(source::binary_expression *expression)
    {
        const auto lhs_register = this->register_in_use ? x_register::a0 : x_register::t0;

        this->register_in_use = true;
        expression->lhs().accept(this);

        this->instructions.push_back( // movl %eax, -x(%rbp); where x is a number.
            instruction(base_opcode::store)
                .s(static_cast<std::uint32_t>(this->variable_counter * 4), funct3_t::sw, x_register::sp, x_register::a0)
        );
        auto lhs_stack_position = ++this->variable_counter;

        this->register_in_use = false;
        expression->rhs().accept(this);

         this->instructions.push_back(instruction(base_opcode::load)
            .i(x_register::a0, funct3_t::lw, x_register::sp,
                static_cast<std::int8_t>((lhs_stack_position - 1) * 4))
        );

        // Calculate the result and assign it to a variable on the stack.
        switch (expression->operation())
        {
            case source::binary_operator::sum:
                this->instructions.push_back(instruction(base_opcode::op)
                    .r(lhs_register, funct3_t::add, x_register::a0, x_register::t0));
                break;
            case source::binary_operator::subtraction:
                this->instructions.push_back(instruction(base_opcode::op)
                    .r(lhs_register, funct3_t::sub, x_register::a0, x_register::t0, funct7_t::sub));
                break;
            case source::binary_operator::multiplication:
                this->instructions.push_back(instruction(base_opcode::op)
                    .r(lhs_register, funct3_t::mul, x_register::a0, x_register::t0, funct7_t::muldiv));
                break;
            case source::binary_operator::division:
                this->instructions.push_back(instruction(base_opcode::op)
                    .r(lhs_register, funct3_t::div, x_register::a0, x_register::t0, funct7_t::muldiv));
                break;
        }
    }

    void visitor::visit(source::integer_literal *number)
    {
        const auto free_register = this->register_in_use ? x_register::a0 : x_register::t0;

        this->instructions.push_back(
            instruction(base_opcode::opImm) // movl $x, %eax; where $x is a number.
                .i(free_register, funct3_t::addi, x_register::zero, number->number())
        );
    }

    void visitor::visit(source::boolean_literal *number)
    {
        const auto free_register = this->register_in_use ? x_register::a0 : x_register::t0;

        this->instructions.push_back(
            instruction(base_opcode::opImm) // movl $x, %eax; where $x is a number.
                .i(free_register, funct3_t::addi, x_register::zero, number->boolean())
        );
    }
}