questionable refactoring

src/compiler/arch/riscv/asm.rs

@@ -1,6 +1,5 @@
 use crate::{
-    compiler::program::{Addr, Instr, SymTable},
+    compiler::program::{Addr, Instr, SymTable}, ir::Symbol, util::LabeledFmt
-    ir::Symbol,
 };
 
 use super::*;
@@ -169,14 +168,29 @@ impl LinkerInstruction {
     }
 }
 
+// this is not even remotely worth it but technically it doesn't use the heap I think xdddddddddd
 impl<R: std::fmt::Debug, S: std::fmt::Debug> std::fmt::Debug for LinkerInstruction<R, S> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        self.fmt_label(f, &|f, s| write!(f, "{s:?}"))
+    }
+}
+
+pub struct DebugInstr<'a, R, S, L: Fn(&mut std::fmt::Formatter<'_>, &S) -> std::fmt::Result> {
+    instr: &'a LinkerInstruction<R, S>,
+    label: &'a L,
+}
+
+impl<R: std::fmt::Debug, S: std::fmt::Debug> LabeledFmt<S> for LinkerInstruction<R, S> {
+    fn fmt_label(&self, f: &mut std::fmt::Formatter<'_>, label: &dyn crate::util::Labeler<S>) -> std::fmt::Result {
         match self {
             Self::ECall => write!(f, "ecall"),
             Self::EBreak => write!(f, "ebreak"),
             Self::Li { dest, imm } => write!(f, "li {dest:?}, {imm:?}"),
             Self::Mv { dest, src } => write!(f, "mv {dest:?}, {src:?}"),
-            Self::La { dest, src } => write!(f, "la {dest:?}, {src:?}"),
+            Self::La { dest, src } => {
+                write!(f, "la {dest:?}, @")?;
+                label(f, src)
+            },
             Self::Load {
                 width,
                 dest,
@@ -207,8 +221,14 @@ impl<R: std::fmt::Debug, S: std::fmt::Debug> std::fmt::Debug for LinkerInstructi
                 imm,
             } => write!(f, "{}i {dest:?}, {src:?}, {imm}", opstr(*op, *funct)),
             Self::Jal { dest, offset } => write!(f, "jal {dest:?}, {offset:?}"),
-            Self::Call(s) => write!(f, "call {s:?}"),
+            Self::Call(s) => {
-            Self::J(s) => write!(f, "j {s:?}"),
+                write!(f, "call ")?;
+                label(f, s)
+            }
+            Self::J(s) => {
+                write!(f, "j ")?;
+                label(f, s)
+            }
             Self::Ret => write!(f, "ret"),
         }
     }

src/compiler/arch/riscv/compile.rs

@@ -1,7 +1,7 @@
 use std::collections::HashMap;
 
 use crate::{
-    compiler::{arch::riscv::Reg, create_program, Addr},
+    compiler::{arch::riscv::Reg, debug::DebugInfo, UnlinkedProgram},
     ir::{
         arch::riscv64::{RV64Instruction as AI, RegRef},
         IRLInstruction as IRI, IRLProgram, Len, Size,
@@ -47,9 +47,10 @@ fn mov_mem(
     }
 }
 
-pub fn compile(program: IRLProgram) -> (Vec<u8>, Option<Addr>) {
+pub fn compile(program: &IRLProgram) -> UnlinkedProgram<LI> {
     let mut fns = Vec::new();
     let mut data = Vec::new();
+    let mut dbg = DebugInfo::new(program.labels().to_vec());
     for (sym, d) in program.ro_data() {
         data.push((d.clone(), *sym));
     }
@@ -83,7 +84,9 @@ pub fn compile(program: IRLProgram) -> (Vec<u8>, Option<Addr>) {
                 v.push(LI::sd(ra, stack_ra, sp));
             }
         }
+        let mut irli = Vec::new();
         for i in &f.instructions {
+            irli.push((v.len(), format!("{i:?}")));
             match i {
                 IRI::Mv { dest, src } => todo!(),
                 IRI::Ref { dest, src } => todo!(),
@@ -212,6 +215,7 @@ pub fn compile(program: IRLProgram) -> (Vec<u8>, Option<Addr>) {
                 }
             }
         }
+        dbg.push_fn(irli);
         if has_stack {
             if let Some(stack_ra) = stack_ra {
                 v.push(LI::ld(ra, stack_ra, sp));
@@ -221,5 +225,10 @@ pub fn compile(program: IRLProgram) -> (Vec<u8>, Option<Addr>) {
         v.push(LI::Ret);
         fns.push((v, *sym));
     }
-    create_program(fns, data, Some(program.entry()), &program)
+    UnlinkedProgram {
+        fns: fns.into_iter().map(|(v, s, ..)| (v, s)).collect(),
+        ro_data: data,
+        start: Some(program.entry()),
+        dbg,
+    }
 }

src/compiler/debug.rs

@@ -0,0 +1,23 @@
+use crate::ir::Symbol;
+
+pub struct DebugInfo {
+    pub sym_labels: Vec<Option<String>>,
+    pub ir_lower: Vec<Vec<(usize, String)>>,
+}
+
+impl DebugInfo {
+    pub fn new(sym_labels: Vec<Option<String>>) -> Self {
+        Self {
+            ir_lower: Vec::new(),
+            sym_labels,
+        }
+    }
+
+    pub fn push_fn(&mut self, instrs: Vec<(usize, String)>) {
+        self.ir_lower.push(instrs);
+    }
+
+    pub fn sym_label(&self, s: Symbol) -> Option<&String> {
+        self.sym_labels[*s].as_ref()
+    }
+}

src/compiler/elf.rs

@@ -1,4 +1,4 @@
-use super::program::Addr;
+use super::{program::Addr, LinkedProgram};
 
 #[repr(C)]
 pub struct ELF64Header {
@@ -102,3 +102,9 @@ pub fn create(program: Vec<u8>, start_offset: Addr) -> Vec<u8> {
 unsafe fn as_u8_slice<T: Sized>(p: &T) -> &[u8] {
     core::slice::from_raw_parts((p as *const T) as *const u8, size_of::<T>())
 }
+
+impl LinkedProgram {
+    pub fn to_elf(self) -> Vec<u8> {
+        create(self.code, self.start.expect("no start found"))
+    }
+}

src/compiler/mod.rs

@@ -1,13 +1,14 @@
 pub mod arch;
+mod debug;
 mod elf;
 mod program;
 mod target;
 
+use arch::riscv;
 pub use program::*;
 
 use crate::ir::IRLProgram;
 
-pub fn compile(program: IRLProgram) -> Vec<u8> {
+pub fn compile(program: &IRLProgram) -> UnlinkedProgram<riscv::LinkerInstruction> {
-    let (compiled, start) = arch::riscv::compile(program);
+    arch::riscv::compile(program)
-    elf::create(compiled, start.expect("no start method found"))
 }

src/compiler/program.rs

@@ -1,57 +1,63 @@
 use std::collections::HashMap;
 
-use crate::ir::{IRLProgram, Symbol};
+use crate::{
+    ir::Symbol,
+    util::{Labelable, LabeledFmt, Labeler},
+};
 
-pub fn create_program<I: Instr>(
+use super::debug::DebugInfo;
-    fns: Vec<(Vec<I>, Symbol)>,
+
-    ro_data: Vec<(Vec<u8>, Symbol)>,
+pub struct LinkedProgram {
-    start: Option<Symbol>,
+    pub code: Vec<u8>,
-    program: &IRLProgram,
+    pub start: Option<Addr>,
-) -> (Vec<u8>, Option<Addr>) {
+}
-    let mut data = Vec::new();
+
-    let mut sym_table = SymTable::new(fns.len() + ro_data.len());
+pub struct UnlinkedProgram<I: Instr> {
-    let mut missing = HashMap::<Symbol, Vec<(Addr, I)>>::new();
+    pub fns: Vec<(Vec<I>, Symbol)>,
-    for (val, id) in ro_data {
+    pub ro_data: Vec<(Vec<u8>, Symbol)>,
-        sym_table.insert(id, Addr(data.len() as u64));
+    pub start: Option<Symbol>,
-        data.extend(val);
+    pub dbg: DebugInfo,
-    }
+}
-    data.resize(data.len() + (4 - data.len() % 4), 0);
+
-    for (fun, id) in fns {
+impl<I: Instr> UnlinkedProgram<I> {
-        sym_table.insert(id, Addr(data.len() as u64));
+    pub fn link(self) -> LinkedProgram {
-        for i in fun {
+        let mut data = Vec::new();
-            let i_pos = Addr(data.len() as u64);
+        let mut sym_table = SymTable::new(self.fns.len() + self.ro_data.len());
-            if let Some(sym) = i.push(&mut data, &sym_table, i_pos, false) {
+        let mut missing = HashMap::<Symbol, Vec<(Addr, I)>>::new();
-                if let Some(vec) = missing.get_mut(&sym) {
+        for (val, id) in self.ro_data {
-                    vec.push((i_pos, i));
+            sym_table.insert(id, Addr(data.len() as u64));
-                } else {
+            data.extend(val);
-                    missing.insert(sym, vec![(i_pos, i)]);
+        }
+        data.resize(data.len() + (4 - data.len() % 4), 0);
+        for (fun, id) in self.fns {
+            sym_table.insert(id, Addr(data.len() as u64));
+            for i in fun {
+                let i_pos = Addr(data.len() as u64);
+                if let Some(sym) = i.push(&mut data, &sym_table, i_pos, false) {
+                    if let Some(vec) = missing.get_mut(&sym) {
+                        vec.push((i_pos, i));
+                    } else {
+                        missing.insert(sym, vec![(i_pos, i)]);
+                    }
+                }
+            }
+            if let Some(vec) = missing.remove(&id) {
+                for (addr, i) in vec {
+                    let mut replace = Vec::new();
+                    i.push(&mut replace, &sym_table, addr, true);
+                    let pos = addr.val() as usize;
+                    data[pos..pos + replace.len()].copy_from_slice(&replace);
                 }
             }
         }
-        if let Some(vec) = missing.remove(&id) {
+        assert!(missing.is_empty());
-            for (addr, i) in vec {
+        LinkedProgram {
-                let mut replace = Vec::new();
+            code: data,
-                i.push(&mut replace, &sym_table, addr, true);
+            start: self
-                let pos = addr.val() as usize;
+                .start
-                data[pos..pos + replace.len()].copy_from_slice(&replace);
+                .map(|s| sym_table.get(s).expect("start symbol doesn't exist")),
-            }
         }
     }
-    for (s, f) in program.fns() {
-        println!(
-            "{}: {:?}",
-            f.name,
-            sym_table.get(*s).map(|a| {
-                let pos = a.0 + 0x1000 + 0x40 + 0x38;
-                format!("0x{:x}", pos)
-            })
-        );
-    }
-    assert!(missing.is_empty());
-    (
-        data,
-        start.map(|s| sym_table.get(s).expect("start symbol doesn't exist")),
-    )
 }
 
 pub trait Instr {
@@ -83,3 +89,31 @@ impl SymTable {
         }
     }
 }
+
+impl<I: Instr + Labelable<Symbol> + LabeledFmt<Symbol>> std::fmt::Debug for UnlinkedProgram<I> {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        for ((v, s), irli) in self.fns.iter().zip(&self.dbg.ir_lower) {
+            writeln!(f, "{}:", self.dbg.sym_label(*s).unwrap())?;
+            let mut liter = irli.iter();
+            let mut cur = liter.next();
+            for (i, instr) in v.iter().enumerate() {
+                if let Some(c) = cur {
+                    if i == c.0 {
+                        writeln!(f, "   {}:", c.1)?;
+                        cur = liter.next();
+                    }
+                }
+                writeln!(
+                    f,
+                    "      {:?}",
+                    instr.labeled(&|f: &mut std::fmt::Formatter, s: &Symbol| write!(
+                        f,
+                        "{}",
+                        self.dbg.sym_label(*s).unwrap_or(&format!("{:?}", *s))
+                    ))
+                )?;
+            }
+        }
+        Ok(())
+    }
+}

src/ir/lower/func.rs

@@ -5,7 +5,6 @@ use std::collections::HashMap;
 
 #[derive(Debug)]
 pub struct IRLFunction {
-    pub name: String,
     pub instructions: Vec<IRLInstruction>,
     pub stack: HashMap<VarID, Size>,
     pub args: Vec<(VarID, Size)>,

src/ir/lower/program.rs

@@ -58,7 +58,8 @@ impl IRLProgram {
                             continue;
                         }
                         let data = &p.data[src.0];
-                        let sym = builder.ro_data(src, data);
+                        let ddef = p.get_data(*src);
+                        let sym = builder.ro_data(src, data, Some(ddef.label.clone()));
                         instrs.push(IRLInstruction::LoadData {
                             dest: dest.id,
                             offset: 0,
@@ -75,14 +76,14 @@ impl IRLProgram {
                         let Type::Array(ty, len) = &def.ty else {
                             return Err(format!("tried to load {} as slice", p.type_name(&def.ty)));
                         };
-                        let sym = builder.ro_data(src, data);
+                        let sym = builder.ro_data(src, data, Some(def.label.clone()));
                         instrs.push(IRLInstruction::LoadAddr {
                             dest: dest.id,
                             offset: 0,
                             src: sym,
                         });
 
-                        let sym = builder.anon_ro_data(&(*len as u64).to_le_bytes());
+                        let sym = builder.anon_ro_data(&(*len as u64).to_le_bytes(), Some(format!("len: {}", len)));
                         instrs.push(IRLInstruction::LoadData {
                             dest: dest.id,
                             offset: 8,
@@ -133,7 +134,6 @@ impl IRLProgram {
             builder.write_fn(
                 sym,
                 IRLFunction {
-                    name: f.name.clone(),
                     instructions: instrs,
                     makes_call,
                     args: f
@@ -144,14 +144,10 @@ impl IRLProgram {
                     ret_size: p.size_of_type(&f.ret).expect("unsized type"),
                     stack,
                 },
+                Some(f.name.clone()),
             );
         }
         let sym_space = builder.finish().expect("we failed the mission");
-        // println!("fns:");
-        // for (a, f) in sym_space.fns() {
-        //     println!("    {:?}: {}", a, f.name);
-        // }
-        // println!("datas: {}", sym_space.ro_data().len());
         Ok(Self { sym_space, entry })
     }
 

src/ir/lower/symbol.rs

@@ -18,6 +18,7 @@ impl std::ops::Deref for WritableSymbol {
 pub struct SymbolSpace {
     ro_data: Vec<(Symbol, Vec<u8>)>,
     fns: Vec<(Symbol, IRLFunction)>,
+    labels: Vec<Option<String>>,
 }
 
 pub struct SymbolSpaceBuilder {
@@ -27,6 +28,7 @@ pub struct SymbolSpaceBuilder {
     data_map: HashMap<DataID, Symbol>,
     ro_data: Vec<(Symbol, Vec<u8>)>,
     fns: Vec<(Symbol, IRLFunction)>,
+    labels: Vec<Option<String>>,
 }
 
 impl SymbolSpace {
@@ -38,6 +40,7 @@ impl SymbolSpace {
             data_map: HashMap::new(),
             ro_data: Vec::new(),
             fns: Vec::new(),
+            labels: Vec::new(),
         };
         for e in entries {
             s.func(e);
@@ -50,23 +53,26 @@ impl SymbolSpace {
     pub fn fns(&self) -> &[(Symbol, IRLFunction)] {
         &self.fns
     }
+    pub fn labels(&self) -> &[Option<String>] {
+        &self.labels
+    }
 }
 
 impl SymbolSpaceBuilder {
     pub fn pop_fn(&mut self) -> Option<(WritableSymbol, FnID)> {
         self.unwritten_fns.pop()
     }
-    pub fn anon_ro_data(&mut self, data: &[u8]) -> Symbol {
+    pub fn anon_ro_data(&mut self, data: &[u8], label: Option<String>) -> Symbol {
         let sym = self.reserve();
-        self.write_ro_data(sym, data.to_vec())
+        self.write_ro_data(sym, data.to_vec(), label)
     }
-    pub fn ro_data(&mut self, id: &DataID, data: &[u8]) -> Symbol {
+    pub fn ro_data(&mut self, id: &DataID, data: &[u8], label: Option<String>) -> Symbol {
         match self.data_map.get(id) {
             Some(s) => *s,
             None => {
                 let sym = self.reserve();
                 self.data_map.insert(*id, *sym);
-                self.write_ro_data(sym, data.to_vec())
+                self.write_ro_data(sym, data.to_vec(), label)
             }
         }
     }
@@ -82,18 +88,31 @@ impl SymbolSpaceBuilder {
             }
         }
     }
-    pub fn write_ro_data(&mut self, sym: WritableSymbol, data: Vec<u8>) -> Symbol {
+    pub fn write_ro_data(
+        &mut self,
+        sym: WritableSymbol,
+        data: Vec<u8>,
+        name: Option<String>,
+    ) -> Symbol {
         let data = data.into();
         self.ro_data.push((*sym, data));
+        self.labels[sym.0 .0] = name;
         *sym
     }
-    pub fn write_fn(&mut self, sym: WritableSymbol, func: IRLFunction) -> Symbol {
+    pub fn write_fn(
+        &mut self,
+        sym: WritableSymbol,
+        func: IRLFunction,
+        name: Option<String>,
+    ) -> Symbol {
         self.fns.push((*sym, func));
+        self.labels[sym.0 .0] = name;
         *sym
     }
     pub fn reserve(&mut self) -> WritableSymbol {
         let val = self.symbols;
         self.symbols += 1;
+        self.labels.push(None);
         WritableSymbol(Symbol(val))
     }
     pub fn len(&self) -> usize {
@@ -104,6 +123,7 @@ impl SymbolSpaceBuilder {
             Some(SymbolSpace {
                 fns: self.fns,
                 ro_data: self.ro_data,
+                labels: self.labels,
             })
         } else {
             None

src/ir/upper/def.rs

@@ -29,6 +29,7 @@ pub struct VarDef {
 pub struct DataDef {
     pub ty: Type,
     pub origin: Origin,
+    pub label: String,
 }
 
 #[derive(Debug, Clone, Copy)]

src/main.rs

@@ -1,5 +1,6 @@
 #![feature(box_patterns)]
 #![feature(try_trait_v2)]
+#![feature(trait_alias)]
 
 use ir::{IRLProgram, IRUProgram};
 use parser::{NodeParsable, PModule, PStatement, ParserCtx};
@@ -21,15 +22,16 @@ use common::*;
 fn main() {
     let file = std::env::args_os().nth(1);
     let gdb = std::env::args().nth(2).is_some_and(|a| a == "--debug");
+    let asm = std::env::args().nth(2).is_some_and(|a| a == "--asm");
     if let Some(path) = file {
         let file = std::fs::read_to_string(path).expect("failed to read file");
-        run_file(&file, gdb);
+        run_file(&file, gdb, asm);
     } else {
         run_stdin();
     }
 }
 
-fn run_file(file: &str, gdb: bool) {
+fn run_file(file: &str, gdb: bool, asm: bool) {
     let mut ctx = ParserCtx::from(file);
     let res = PModule::parse_node(&mut ctx);
     if ctx.output.errs.is_empty() {
@@ -50,9 +52,14 @@ fn run_file(file: &str, gdb: bool) {
                 output.write_for(&mut stdout(), file);
                 if output.errs.is_empty() {
                     let program = IRLProgram::create(&namespace).expect("morir");
-                    let bin = compiler::compile(program);
+                    let unlinked = compiler::compile(&program);
-                    println!("compiled");
+                    if asm {
-                    save_run(&bin, gdb);
+                        println!("{:?}", unlinked);
+                    } else {
+                        let bin = unlinked.link().to_elf();
+                        println!("compiled");
+                        save_run(&bin, gdb);
+                    }
                 }
             }
         }

src/parser/v3/lower/expr.rs

@@ -13,6 +13,7 @@ impl FnLowerable for PExpr {
                         DataDef {
                             ty: Type::Bits(8).arr(data.len() as u32),
                             origin: Origin::File(l.span),
+                            label: format!("string \"{}\"", s.replace("\n", "\\n"))
                         },
                         data,
                     );
@@ -26,6 +27,7 @@ impl FnLowerable for PExpr {
                         DataDef {
                             ty,
                             origin: Origin::File(l.span),
+                            label: format!("char '{c}'"),
                         },
                         c.to_string().as_bytes().to_vec(),
                     );
@@ -40,6 +42,7 @@ impl FnLowerable for PExpr {
                         DataDef {
                             ty,
                             origin: Origin::File(l.span),
+                            label: format!("num {n:?}")
                         },
                         n.whole.parse::<i64>().unwrap().to_le_bytes().to_vec(),
                     );

src/util/label.rs

@@ -0,0 +1,40 @@
+// this is not even remotely worth it but technically it doesn't use the heap I think xdddddddddd
+
+use std::marker::PhantomData;
+pub trait Labeler<S> = Fn(&mut std::fmt::Formatter<'_>, &S) -> std::fmt::Result;
+
+pub trait Labelable<S> {
+    fn labeled<L: Labeler<S>>(&self, l: L) -> Labeled<Self, L, S>
+    where
+        Self: Sized;
+}
+
+pub struct Labeled<'a, T, L: Labeler<S>, S> {
+    data: &'a T,
+    labeler: L,
+    pd: PhantomData<S>,
+}
+
+pub trait LabeledFmt<S> {
+    fn fmt_label(
+        &self,
+        f: &mut std::fmt::Formatter<'_>,
+        label: &dyn Labeler<S>,
+    ) -> std::fmt::Result;
+}
+
+impl<T: LabeledFmt<S>, S> Labelable<S> for T {
+    fn labeled<L: Labeler<S>>(&self, l: L) -> Labeled<Self, L, S> {
+        Labeled {
+            data: self,
+            labeler: l,
+            pd: PhantomData,
+        }
+    }
+}
+
+impl<T: LabeledFmt<S>, L: Labeler<S>, S> std::fmt::Debug for Labeled<'_, T, L, S> {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        self.data.fmt_label(f, &self.labeler)
+    }
+}

src/util/mod.rs

@@ -1,5 +1,7 @@
 mod padder;
 mod bits;
+mod label;
 
 pub use padder::*;
 pub use bits::*;
+pub use label::*;