From 75ad04cf9535bf4852de7b8c8c8d9326255fbb54 Mon Sep 17 00:00:00 2001
From: zxq5 <zxq5@zxq5.dev>
Date: Tue, 10 Feb 2026 16:37:33 +0000
Subject: [PATCH] forgot to commit these

---
 compiler/src/backend/dsa/instruction.rs | 610 ++++++++++++++++++++++++
 compiler/src/backend/dsa/scope.rs       |   7 +
 compiler/src/backend/dsa/variable.rs    | 186 ++++++++
 3 files changed, 803 insertions(+)
 create mode 100644 compiler/src/backend/dsa/instruction.rs
 create mode 100644 compiler/src/backend/dsa/scope.rs
 create mode 100644 compiler/src/backend/dsa/variable.rs

diff --git a/compiler/src/backend/dsa/instruction.rs b/compiler/src/backend/dsa/instruction.rs
new file mode 100644
index 0000000..e419b62
--- /dev/null
+++ b/compiler/src/backend/dsa/instruction.rs
@@ -0,0 +1,610 @@
+use std::fmt;
+
+use crate::{
+    backend::dsa::registers::Register,
+    model::{CompilerError, Expression},
+};
+
+pub struct CodeGen {
+    // For building the final program
+    program: InsBlock,
+
+    // For generating temporary blocks
+    label_counter: usize,
+    stack_offset: i32,
+}
+
+impl CodeGen {
+    pub fn new() -> Self {
+        Self {
+            program: InsBlock::new(),
+            label_counter: 0,
+            stack_offset: 0,
+        }
+    }
+
+    /// Emit directly to program (for top-level constructs)
+    pub fn emit(&mut self, instr: Instruction) {
+        self.program.push(instr);
+    }
+
+    /// Emit a block to program
+    pub fn emit_block(&mut self, block: InsBlock) {
+        self.program.append(block);
+    }
+
+    /// Build expression (returns block for composition)
+    pub fn build_expr(&mut self, expr: &Expression) -> Result<InsBlock, CompilerError> {
+        // ... returns InstrBlock
+        todo!()
+    }
+
+    /// Get final output
+    pub fn finish(mut self) -> String {
+        // Optimize before final output
+        // self.program.remove_dead_code();
+        // self.program.optimize_peephole();
+
+        self.program
+            .instructions
+            .iter()
+            .map(|i| i.to_string())
+            .collect::<Vec<_>>()
+            .join("\n")
+    }
+}
+
+pub struct InsBlock {
+    instructions: Vec<Instruction>,
+}
+
+impl InsBlock {
+    pub fn new() -> Self {
+        Self {
+            instructions: vec![],
+        }
+    }
+
+    pub fn push(&mut self, instr: Instruction) {
+        self.instructions.push(instr);
+    }
+
+    pub fn append(&mut self, mut other: Self) {
+        self.instructions.append(&mut other.instructions);
+    }
+
+    pub fn extend(&mut self, instrs: impl IntoIterator<Item = Instruction>) {
+        self.instructions.extend(instrs);
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.instructions.is_empty()
+    }
+
+    pub fn len(&self) -> usize {
+        self.instructions.len()
+    }
+
+    pub fn iter(&self) -> impl Iterator<Item = &Instruction> {
+        self.instructions.iter()
+    }
+}
+
+pub enum Instruction {
+    // Labels and comments
+    Label(Label),
+    Comment(String),
+
+    // Data movement
+    Mov {
+        src: Register,
+        dest: Register,
+    },
+    Movs {
+        src: Register,
+        dest: Register,
+    },
+
+    // Memory operations
+    Ldb {
+        addr: MemOperand,
+        dest: Register,
+    },
+    Ldh {
+        addr: MemOperand,
+        dest: Register,
+    },
+    Ldw {
+        addr: MemOperand,
+        dest: Register,
+    },
+    Stb {
+        src: Register,
+        addr: MemOperand,
+    },
+    Sth {
+        src: Register,
+        addr: MemOperand,
+    },
+    Stw {
+        src: Register,
+        addr: MemOperand,
+    },
+
+    // Immediate loads
+    Lli {
+        imm: Imm,
+        dest: Register,
+    },
+    Lui {
+        imm: Imm,
+        dest: Register,
+    },
+
+    // Arithmetic
+    Add {
+        src1: Register,
+        src2: Register,
+        dest: Register,
+    },
+    Sub {
+        src1: Register,
+        src2: Register,
+        dest: Register,
+    },
+    IAdd {
+        src: Register,
+        imm: Imm,
+        dest: Option<Register>,
+    },
+    ISub {
+        src: Register,
+        imm: Imm,
+        dest: Option<Register>,
+    },
+    Inc {
+        reg: Register,
+    },
+    Dec {
+        reg: Register,
+    },
+
+    // Bitwise
+    And {
+        src1: Register,
+        src2: Register,
+        dest: Register,
+    },
+    Or {
+        src1: Register,
+        src2: Register,
+        dest: Register,
+    },
+    Xor {
+        src1: Register,
+        src2: Register,
+        dest: Register,
+    },
+    Not {
+        src: Register,
+        dest: Register,
+    },
+    Nand {
+        src1: Register,
+        src2: Register,
+        dest: Register,
+    },
+    Nor {
+        src1: Register,
+        src2: Register,
+        dest: Register,
+    },
+    Xnor {
+        src1: Register,
+        src2: Register,
+        dest: Register,
+    },
+
+    // Shifts
+    Shl {
+        src1: Register,
+        r_shamt: Register,
+        i_shamt: u16,
+        dest: Register,
+    },
+    Shr {
+        src1: Register,
+        rsh: Register,
+        ish: u16,
+        dest: Register,
+    },
+
+    // Comparison
+    Cmp {
+        reg1: Register,
+        reg2: Register,
+    },
+
+    // Jumps
+    Jmp {
+        target: Label,
+    },
+    Jeq {
+        target: Label,
+    },
+    Jne {
+        target: Label,
+    },
+    Jgt {
+        target: Label,
+    },
+    Jge {
+        target: Label,
+    },
+    Jlt {
+        target: Label,
+    },
+    Jle {
+        target: Label,
+    },
+
+    // Stack
+    Push {
+        reg: Register,
+    },
+    Pop {
+        reg: Register,
+    },
+
+    // Function calls
+    Call {
+        target: String,
+    }, // namespace::function
+    Return,
+
+    // System
+    Hlt,
+    Nop,
+    Int {
+        code: u8,
+    },
+}
+
+impl fmt::Display for Instruction {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        match self {
+            Self::Label(l) => write!(f, "{}:", l),
+            Self::Comment(c) => write!(f, "; {}", c),
+
+            Self::Mov { src, dest } => write!(f, "    mov {}, {}", src, dest),
+            Self::Movs { src, dest } => write!(f, "    movs {}, {}", src, dest),
+
+            Self::Ldb { addr, dest } => {
+                write!(f, "    ldb {}, {}", format_mem_operand(addr), dest)
+            }
+            Self::Ldh { addr, dest } => {
+                write!(f, "    ldh {}, {}", format_mem_operand(addr), dest)
+            }
+            Self::Ldw { addr, dest } => {
+                write!(f, "    ldw {}, {}", format_mem_operand(addr), dest)
+            }
+            // Self::Ldbs { addr, dest } => {
+            //     write!(f, "    ldbs {}, {}", format_mem_operand(addr), dest)
+            // }
+            // Self::Ldhs { addr, dest } => {
+            //     write!(f, "    ldhs {}, {}", format_mem_operand(addr), dest)
+            // }
+            // Self::Ldws { addr, dest } => {
+            //     write!(f, "    ldws {}, {}", format_mem_operand(addr), dest)
+            // }
+            Self::Stb { src, addr } => {
+                write!(f, "    stb {}, {}", src, format_mem_operand(addr))
+            }
+            Self::Sth { src, addr } => {
+                write!(f, "    sth {}, {}", src, format_mem_operand(addr))
+            }
+            Self::Stw { src, addr } => {
+                write!(f, "    stw {}, {}", src, format_mem_operand(addr))
+            }
+
+            Self::Lli { imm, dest } => write!(f, "    lli {}, {}", imm, dest),
+            Self::Lui { imm, dest } => write!(f, "    lui {}, {}", imm, dest),
+
+            // arithmetic
+            Self::Add { src1, src2, dest } => {
+                write!(f, "    add {}, {}, {}", src1, src2, dest)
+            }
+            Self::Sub { src1, src2, dest } => {
+                write!(f, "    sub {}, {}, {}", src1, src2, dest)
+            }
+            Self::And { src1, src2, dest } => {
+                write!(f, "    and {}, {}, {}", src1, src2, dest)
+            }
+            Self::Or { src1, src2, dest } => {
+                write!(f, "    or {}, {}, {}", src1, src2, dest)
+            }
+            Self::Nand { src1, src2, dest } => {
+                write!(f, "    nand {}, {}, {}", src1, src2, dest)
+            }
+            Self::Xor { src1, src2, dest } => {
+                write!(f, "    xor {}, {}, {}", src1, src2, dest)
+            }
+            Self::Nor { src1, src2, dest } => {
+                write!(f, "    nor {}, {}, {}", src1, src2, dest)
+            }
+            Self::Not { src, dest } => {
+                write!(f, "    not {} {}", src, dest)
+            }
+            Self::Xnor { src1, src2, dest } => {
+                write!(f, "    xnor {}, {}, {}", src1, src2, dest)
+            }
+            Self::IAdd { src, imm, dest } => {
+                if let Some(d) = dest {
+                    write!(f, "    iadd {}, {}, {}", src, imm, d)
+                } else {
+                    write!(f, "    iadd {}, {}", src, imm)
+                }
+            }
+            Self::ISub { src, imm, dest } => {
+                if let Some(d) = dest {
+                    write!(f, "    isub {}, {}, {}", src, imm, d)
+                } else {
+                    write!(f, "    isub {}, {}", src, imm)
+                }
+            }
+
+            // shift instructions
+            Self::Shl {
+                src1,
+                r_shamt,
+                i_shamt,
+                dest,
+            } => {
+                write!(f, "    shl {}, {}, {}, {}", src1, r_shamt, i_shamt, dest)
+            }
+            Self::Shr {
+                src1,
+                rsh: r_shamt,
+                ish: i_shamt,
+                dest,
+            } => {
+                write!(f, "    shl {}, {}, {}, {}", src1, r_shamt, i_shamt, dest)
+            }
+
+            // increment instructions
+            Self::Inc { reg } => write!(f, "    inc {}", reg),
+            Self::Dec { reg } => write!(f, "    dec {}", reg),
+
+            Self::Cmp { reg1, reg2 } => write!(f, "    cmp {}, {}", reg1, reg2),
+
+            // jump instructions
+            Self::Jmp { target } => write!(f, "    jmp {}", target),
+            Self::Jeq { target } => write!(f, "    jeq {}", target),
+            Self::Jne { target } => write!(f, "    jne {}", target),
+            Self::Jgt { target } => write!(f, "    jgt {}", target),
+            Self::Jge { target } => write!(f, "    jge {}", target),
+            Self::Jlt { target } => write!(f, "    jlt {}", target),
+            Self::Jle { target } => write!(f, "    jle {}", target),
+
+            // stack pseudoinstructions
+            Self::Push { reg } => write!(f, "    push {}", reg),
+            Self::Pop { reg } => write!(f, "    pop {}", reg),
+
+            // call & return pseudoinstructions
+            Self::Call { target } => write!(f, "    call {}", target),
+            Self::Return => write!(f, "    return"),
+
+            // misc instructions
+            Self::Int { code } => write!(f, "    int {}", code),
+            Self::Hlt => write!(f, "    hlt"),
+            Self::Nop => write!(f, "    nop"),
+        }
+    }
+}
+
+impl Instruction {
+    // Movement
+    pub fn mov(src: Register, dest: Register) -> Self {
+        Self::Mov { src, dest }
+    }
+
+    // Memory loads
+    pub fn ldw_reg(base: Register, dest: Register) -> Self {
+        Self::Ldw {
+            addr: MemOperand::RegIndirect(base),
+            dest,
+        }
+    }
+
+    pub fn ldw_reg_offset(base: Register, offset: i32, dest: Register) -> Self {
+        Self::Ldw {
+            addr: MemOperand::RegOffset(base, offset),
+            dest,
+        }
+    }
+
+    pub fn ldw_label(label: impl Into<Label>, dest: Register) -> Self {
+        Self::Ldw {
+            addr: MemOperand::Label(label.into()),
+            dest,
+        }
+    }
+
+    // Memory stores
+    pub fn stw_reg(src: Register, base: Register) -> Self {
+        Self::Stw {
+            src,
+            addr: MemOperand::RegIndirect(base),
+        }
+    }
+
+    pub fn stw_reg_offset(src: Register, base: Register, offset: i32) -> Self {
+        Self::Stw {
+            src,
+            addr: MemOperand::RegOffset(base, offset),
+        }
+    }
+
+    // Arithmetic
+    pub fn add(src1: Register, src2: Register, dest: Register) -> Self {
+        Self::Add { src1, src2, dest }
+    }
+
+    pub fn sub(src1: Register, src2: Register, dest: Register) -> Self {
+        Self::Sub { src1, src2, dest }
+    }
+
+    pub fn iadd(src: Register, imm: i32) -> Self {
+        Self::IAdd {
+            src,
+            imm: Imm(imm),
+            dest: None,
+        }
+    }
+
+    pub fn iadd_dest(src: Register, imm: i32, dest: Register) -> Self {
+        Self::IAdd {
+            src,
+            imm: Imm(imm),
+            dest: Some(dest),
+        }
+    }
+
+    pub fn inc(reg: Register) -> Self {
+        Self::Inc { reg }
+    }
+
+    pub fn dec(reg: Register) -> Self {
+        Self::Dec { reg }
+    }
+
+    // Immediate loads
+    pub fn load_imm32(value: u32, dest: Register) -> Vec<Self> {
+        let lower = (value & 0xFFFF) as i32;
+        let upper = ((value >> 16) & 0xFFFF) as i32;
+
+        vec![
+            Self::Lli {
+                imm: Imm(lower),
+                dest,
+            },
+            Self::Lui {
+                imm: Imm(upper),
+                dest,
+            },
+        ]
+    }
+
+    // Control flow
+    pub fn label(name: impl Into<String>) -> Self {
+        Self::Label(Label(name.into()))
+    }
+
+    pub fn jmp(target: impl Into<Label>) -> Self {
+        Self::Jmp {
+            target: target.into(),
+        }
+    }
+
+    pub fn jeq(target: impl Into<Label>) -> Self {
+        Self::Jeq {
+            target: target.into(),
+        }
+    }
+
+    pub fn cmp(reg1: Register, reg2: Register) -> Self {
+        Self::Cmp { reg1, reg2 }
+    }
+
+    // Stack
+    pub fn push(reg: Register) -> Self {
+        Self::Push { reg }
+    }
+
+    pub fn pop(reg: Register) -> Self {
+        Self::Pop { reg }
+    }
+
+    // Functions
+    pub fn call(target: impl Into<String>) -> Self {
+        Self::Call {
+            target: target.into(),
+        }
+    }
+
+    pub fn int(code: u8) -> Self {
+        Self::Int { code }
+    }
+
+    pub fn ret() -> Self {
+        Self::Return
+    }
+
+    // Utilities
+    pub fn comment(text: impl Into<String>) -> Self {
+        Self::Comment(text.into())
+    }
+}
+
+// Convenience trait for Label conversion
+impl From<String> for Label {
+    fn from(s: String) -> Self {
+        Label(s)
+    }
+}
+
+impl From<&str> for Label {
+    fn from(s: &str) -> Self {
+        Label(s.to_string())
+    }
+}
+
+fn format_mem_operand(op: &MemOperand) -> String {
+    match op {
+        MemOperand::RegIndirect(reg) => format!("{}", reg),
+        MemOperand::RegOffset(reg, offset) => {
+            if *offset >= 0 {
+                format!("{}, {}", reg, offset)
+            } else {
+                format!("{}, {}", reg, offset)
+            }
+        }
+        MemOperand::Label(label) => format!("{}", label),
+        MemOperand::LabelOffset(label, offset) => {
+            format!("{}, {}", label, offset)
+        }
+    }
+}
+
+/// Memory operand for loads/stores
+#[derive(Debug, Clone)]
+pub enum MemOperand {
+    /// Register indirect: [reg]
+    RegIndirect(Register),
+    /// Register with offset: [reg + offset]
+    RegOffset(Register, i32),
+    /// Label: [label]
+    Label(Label),
+    /// Label with offset: [label + offset]
+    LabelOffset(Label, i32),
+}
+
+/// Immediate value (16-bit or 32-bit)
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct Imm(pub i32);
+
+impl fmt::Display for Imm {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "{}", self.0)
+    }
+}
+
+/// Label reference
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct Label(pub String);
+
+impl fmt::Display for Label {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "{}", self.0)
+    }
+}
diff --git a/compiler/src/backend/dsa/scope.rs b/compiler/src/backend/dsa/scope.rs
new file mode 100644
index 0000000..e720664
--- /dev/null
+++ b/compiler/src/backend/dsa/scope.rs
@@ -0,0 +1,7 @@
+pub struct FunctionContext {
+    name: String,
+
+    stack_offset: i32,
+
+    registers: [(Register, bool); 16],
+}
diff --git a/compiler/src/backend/dsa/variable.rs b/compiler/src/backend/dsa/variable.rs
new file mode 100644
index 0000000..cabcf7b
--- /dev/null
+++ b/compiler/src/backend/dsa/variable.rs
@@ -0,0 +1,186 @@
+use std::{collections::HashMap, hash::Hash};
+
+use crate::{
+    backend::dsa::{
+        instruction::{InsBlock, Reg},
+        registers::Register,
+    },
+    model::{CompilerError, TypeId},
+};
+
+pub struct Variable {
+    name: String,
+
+    /// the type of the variable.
+    r#type: TypeId,
+
+    /// size taken up in bytes.
+    /// if size > 4, value must be stored on the stack.
+    size: usize,
+
+    // location
+    /// this must be None if it cannot be stored in a register.
+    allocated_register: Option<Register>,
+
+    /// represents the offset from the base pointer (Bpr) of the stack frame.
+    /// needs to be signed as offset is positive for function args and negative for local
+    /// variables. as we can't access values at negative offsets, we use the following
+    /// formula: addr = Spr + offset - (Spr - Bpr) where we know (Spr-Bpr) at compile
+    /// time.
+    bpr_offset: Option<isize>,
+}
+
+impl Variable {
+    pub fn new_uninit(name: String, r#type: TypeId) -> Self {
+        Self {
+            name,
+            size: r#type.size(),
+            r#type,
+            allocated_register: None,
+            bpr_offset: None,
+        }
+    }
+
+    fn alloc_register(
+        &mut self,
+        scope: &'_ mut Scope,
+    ) -> Result<Register, CompilerError> {
+        if self.size > 4 {
+            return Err(CompilerError::Generic(format!(
+                "Type {} cannot be allocated a register as it has a size of {} bytes",
+                self.r#type, self.size
+            )));
+        }
+
+        todo!("integrate with register alloc logic")
+
+        // self.allocated_register = Some(...)
+    }
+
+    fn alloc_stack(&mut self, scope: &'_ mut Scope) -> Result<usize, CompilerError> {
+        todo!("integrate with stack alloc logic")
+
+        // self.bpr_offset = Some(...)
+    }
+
+    pub fn load(&mut self, scope: &'_ mut Scope) -> Result<Register, CompilerError> {
+        todo!("load var from stack to reg (if possible)")
+    }
+
+    pub fn new_local(
+        name: String,
+        r#type: TypeId,
+        scope: &'_ mut Scope,
+    ) -> Result<Self, CompilerError> {
+        let mut var = Self::new_uninit(name, r#type);
+        var.alloc_register(scope)?;
+        Ok(var)
+    }
+
+    pub fn new_stack(
+        name: String,
+        r#type: TypeId,
+        scope: &'_ mut Scope,
+    ) -> Result<Self, CompilerError> {
+        let mut init = Self::new_uninit(name, r#type);
+        init.alloc_stack(scope)?;
+        Ok(init)
+    }
+
+    pub fn drop(&mut self, scope: &'_ mut Scope) -> Result<(), CompilerError> {
+        if let Some(reg) = self.allocated_register {
+            todo!("dealloc reg in current function")
+        }
+
+        if let Some(offset) = self.bpr_offset {
+            todo!("free stack slot in current function")
+        }
+
+        Ok(())
+    }
+
+    pub fn spill(&mut self, scope: &'_ mut Scope) -> Result<(), CompilerError> {
+        todo!()
+    }
+}
+
+/// scope object
+pub struct Scope<'a> {
+    /// outer scope, for a function this will be the global scope.
+    parent: Option<&'a mut Scope<'a>>,
+
+    /// is the scope a function body or just a loop?
+    /// depending on the type, ending a scope will have different behaviour
+    r#type: ScopeType,
+
+    /// variables
+    variables: HashMap<String, Variable>,
+
+    /// tells us if a given register is being used or not.
+    /// this can be an array as registers have u8 representation.
+    in_use: [(Register, bool); 16],
+
+    stack_offset: i32,
+}
+
+impl<'a> Scope<'a> {
+    pub fn new(parent: &'a mut Scope<'a>, r#type: ScopeType) -> Scope<'a> {
+        Self {
+            stack_offset: parent.stack_offset,
+            parent: Some(parent),
+            r#type,
+            variables: HashMap::new(),
+            in_use: Register::get_gp().map(|reg| (reg, false)),
+        }
+    }
+
+    pub fn stack_offset(&self) -> i32 {
+        self.stack_offset
+    }
+
+    pub fn stack_offset_mut(&mut self) -> &mut i32 {
+        &mut self.stack_offset
+    }
+
+    pub fn close(&mut self) -> Result<(), CompilerError> {
+        // closing a scope means we need to drop all variables in scope and free registers.
+        for (name, var) in self.variables {
+            if let Some(reg) = var.allocated_register {
+
+            }
+
+            if let Some(off)
+        }
+
+        for reg in self.in_use
+
+        Ok(())
+    }
+
+    pub fn alloc_temp_reg(&mut self) -> Result<(Register, InsBlock), CompilerError> {
+        todo!()
+    }
+
+    pub fn alloc_var_reg(&mut self) -> Result<(Register, InsBlock), CompilerError> {
+        todo!()
+    }
+
+    pub fn alloc_var_stack(&mut self) -> Result<(Register, InsBlock), CompilerError> {
+        todo!()
+    }
+
+    pub fn free_var_stack(&mut self) -> Result<(Register, InsBlock), CompilerError> {
+        todo!()
+    }
+
+    pub fn free_temp_reg(&mut self) -> Result<(Register, InsBlock), CompilerError> {
+        todo!()
+    }
+}
+
+#[derive(PartialEq, Copy, Clone, Debug)]
+enum ScopeType {
+    Function,
+    IfBlock,
+    LoopBlock,
+}