improved debug

.cargo/config.toml

@@ -0,0 +1,2 @@
+[build]
+rustflags = ["-Ctarget-cpu=native"]

Cargo.toml

@@ -1,6 +1,7 @@
 [workspace]
 resolver = "3"
-members = ["simu","asm","bitmap_to_asm"]
+members = ["simu","asm"]
 
 [profile.release]
-debug = "line-tables-only"
+opt-level = 2
+# panic = "abort"

README.md

@@ -31,9 +31,10 @@ syntaxe supplémentaire supportée:
 ## Features
 rajouter --features=[liste séparé par des virgules]
 - div_mul: Support des instruction de multiplication / division
-- rgba: Écran au format RGBA plutot que 0BGR
+- rgba: Écran au format RGBA plutot que 0BGR (alpha ignoré mais stocké)
 - rich_keyboard: rajoute trois champs de mmio pour le clavier. Voir MMIO
 - debug: repl similaire a la version python du simulateur
+- futex: accélère le programme. ne marche pas sur les macs. actif par default, utiliser --no-default-features pour le desactiver
 
 ### instruction spéciale:
 halt (jump 0) met le programme en pause, mais on peut se reveiller par des interuptions
@@ -43,7 +44,7 @@ call 0 termine l'exécution du simulateur proprement
 ## mmio:
 - 0x01000000 à 0x0112c000 : Écran en lecture/écriture. un pixel par 32bits, de gauche a droite puis de bas en haut.
 Format de pixel en 0BGR, 4 bits par couleur.
-Passage au format RGBA (a pour alpha) avec la feature rgba
+Passage au format RGBA (alpha ignoré, mais stocké) avec la feature rgba
 - 0x01200000 : Clavier (scancode) (lecture seule)
 - 0x01200004 : Horloge (millisecondes écoulé depuis le début de la simulation, wrappe tout les 49 jours) (lecture seule)
 - 0x01200008 : Boutons de la souris (OR des boutons préssés)  Gauche = 1, Droit=2, Clic Molette = 3, Autres non testé (lecture seule)
@@ -78,6 +79,7 @@ de même, `reti` ne peut pas être appellé par une sous fonction appellée depu
 Les intéruptions et leurs arguments, par priorité croissante:
 ### 0: Point d'entrée
   C'est ici que le programme commence. Seul cas d'utilisation
+  (ce n'est pas vraiment une interruption, juste la première entrée dans la jump table)
 ### 1: MMIO
   Doivent être activé par le MMIO a l'adresse 0x01201000, en y écrivant le OR des (1<<identifiant)
   On toujours pour argument (r0) leur identifiant

asm/src/main.rs

@@ -207,7 +207,9 @@ impl Instruction {
                     }
                 };
                 if jump_distance > 15 {
-                    println!("Error, cannot skip more than 15 instructions");
+                    println!(
+                        "Error, cannot skip {jump_distance} which is more than 15 instructions"
+                    );
                     return Err(());
                 }
                 encode_op2(11, cond.into(), jump_distance, reg.into(), *op2)

bitmap_to_asm/Cargo.toml

@@ -1,8 +0,0 @@
-[package]
-name = "bitmap_to_asm"
-version = "0.1.0"
-edition = "2024"
-
-[dependencies]
-image = "0.25.10"
-regex = "1.12.3"

bitmap_to_asm/src/main.rs

@@ -1,53 +0,0 @@
-use regex::Regex;
-use std::env::args;
-
-fn main() {
-    let path = args().nth(1).expect("usage: 1 image file argument");
-    let (data, name, width, height) = path_to_img(path.as_str());
-
-    println!("{name}:");
-    println!("    D {width}");
-    println!("    D {height}");
-    for d in data {
-        println!("    D 0x{d:08x}");
-    }
-}
-fn remove_non_alphanumeric(input: &str) -> String {
-    let re = Regex::new(r"[^a-zA-Z0-9_]+").unwrap();
-    re.replace_all(input, "").to_string()
-}
-
-fn path_to_img(path: &str) -> (Vec<u32>, String, u32, u32) {
-    let img = match image::open(path) {
-        Ok(img) => img.to_luma8(),
-        Err(e) => panic!("failed to open image {path}: {e}"),
-    };
-
-    let width = img.width();
-    let height = img.height();
-
-    let mut bytes = Vec::new();
-    let mut bit = 0;
-    let mut byte = 0;
-    for y in 0..height {
-        for x in 0..width {
-            let pix = img.get_pixel(x, y)[0];
-            if pix >= 127 {
-                byte |= 1 << bit
-            }
-            bit += 1;
-            if bit == 32 {
-                bytes.push(byte);
-                byte = 0;
-                bit = 0;
-            }
-        }
-    }
-    if bit != 0 {
-        bytes.push(byte);
-    }
-    let path = path.split('/').next_back().unwrap();
-    let split: Vec<_> = path.split('.').collect();
-    let name = remove_non_alphanumeric(&split[0..split.len() - 1].join("_")).to_lowercase();
-    (bytes, name, width, height)
-}

simu/Cargo.toml

@@ -8,11 +8,17 @@ pixels = "0.15.0"
 winit = { version = "0.30.13", features = ["x11", "x11-dl", "x11rb", "ahash", "bytemuck", "memmap2", "rwh_06", "sctk", "sctk-adwaita"] }
 winit_input_helper = "0.17.0"
 parse_int = { version = "0.9.0", optional = true }
-wait_on_address = "0.1.4"
+wait_on_address = { version = "0.1.4", optional = true }
+clap-repl = { version = "0.3.2", optional = true }
+clap = { version = "4.6.0", optional = true }
 
 
 [features]
+default = ["futex"]
 div_mul = []
 rgba = []
 rich_keyboard = []
-debug = ["dep:parse_int"]
+debug = ["dep:parse_int","dep:clap-repl","dep:clap"]
+futex = ["dep:wait_on_address"]
+clap-repl = ["dep:clap-repl"]
+clap = ["dep:clap"]

simu/src/cpu.rs

@@ -1,15 +1,17 @@
+use crate::wait::WaitOnAtomic;
 #[cfg(feature = "debug")]
 use std::collections::HashMap;
 use std::{
-    hint::{likely, unlikely, unreachable_unchecked},
+    hint::{cold_path, likely, unlikely, unreachable_unchecked},
     io::Read,
     mem::transmute,
     ops::{Index, IndexMut},
-    process::exit,
     sync::atomic::AtomicU32,
     time::{self, Instant},
 };
-use wait_on_address::AtomicWait;
+
+#[cfg(not(feature = "debug"))]
+use std::process::exit;
 pub(crate) struct SharedState {
     pub(crate) keyboard: [AtomicU32; 4],
     pub(crate) screen_buf: [AtomicU32; 480 * 640],
@@ -32,17 +34,26 @@ enum Op2 {
 }
 
 struct Reg(u8);
+
+#[allow(unused)] //constructed by transmute
+#[repr(u8)]
 enum Cond {
-    Ifeq,
+    Ifeq = 0b000,
-    Ifne,
+    Ifne = 0b0001,
-    Iflt,
+    UK1 = 2,
-    Ifge,
+    UK2 = 3,
-    Ifgt,
+    UK3 = 4,
-    Ifle,
+    UK4 = 5,
-    Ifult,
+    UK5 = 6,
-    Ifuge,
+    UK6 = 7,
-    Ifugt,
+    Iflt = 0b1000,
-    Ifule,
+    Ifge = 0b1001,
+    Ifgt = 0b1010,
+    Ifle = 0b1011,
+    Ifult = 0b1100,
+    Ifuge = 0b1101,
+    Ifugt = 0b1110,
+    Ifule = 0b1111,
 }
 
 #[derive(Debug)]
@@ -73,20 +84,10 @@ pub enum MMIOInterupt {
 }
 
 impl From<u8> for Cond {
+    #[inline(always)]
     fn from(value: u8) -> Self {
-        match value {
+        //unsafe if called with value >= 15 buuut only ever called on actually an u4 so it's ok
-            0b0000 => Cond::Ifeq,
+        unsafe { transmute(value) }
-            0b0001 => Cond::Ifne,
-            0b1000 => Cond::Iflt,
-            0b1001 => Cond::Ifge,
-            0b1010 => Cond::Ifgt,
-            0b1011 => Cond::Ifle,
-            0b1100 => Cond::Ifult,
-            0b1101 => Cond::Ifuge,
-            0b1110 => Cond::Ifugt,
-            0b1111 => Cond::Ifule,
-            _ => iot(),
-        }
     }
 }
 
@@ -108,8 +109,9 @@ impl From<u32> for InteruptKind {
 }
 
 impl Cond {
-    fn eval(self, a: u32, b: u32) -> bool {
+    #[inline(always)]
-        match self {
+    fn eval(self, a: u32, b: u32) -> Option<bool> {
+        Some(match self {
             Cond::Ifeq => a == b,
             Cond::Ifne => a != b,
             Cond::Iflt => (a as i32) < (b as i32),
@@ -120,7 +122,8 @@ impl Cond {
             Cond::Ifuge => a >= b,
             Cond::Ifugt => a > b,
             Cond::Ifule => a <= b,
-        }
+            _ => return None,
+        })
     }
 }
 
@@ -163,7 +166,6 @@ impl TryFrom<u32> for Instruction {
         Ok(match value >> 30 {
             0b00 => {
                 let t = value & (1 << 29); // 3rd bit set
-                let value = value - t;
                 let t = t != 0;
                 if t {
                     Self::Call(value)
@@ -275,6 +277,7 @@ impl Display for Cond {
             Cond::Ifuge => write!(f, "ifuge"),
             Cond::Ifugt => write!(f, "ifugt"),
             Cond::Ifule => write!(f, "ifule"),
+            _ => write!(f, "unknown"),
         }
     }
 }
@@ -313,7 +316,7 @@ pub(crate) fn instr_to_text(i: u32, a: u32, book: &HashMap<u32, String>) -> Stri
                 format!("skip {} {cond} {reg} {op2}", addr(d as u32))
             }
             Instruction::Jump(a) => format!("jump {}", addr(a)),
-            Instruction::Call(a) => format!("call {}", addr(a)),
+            Instruction::Call(a) => format!("call {}", addr(a % 0x1000_0000)),
             Instruction::Ret() => format!("ret"),
             Instruction::Reti() => format!("reti"),
             Instruction::Swi() => format!("swi"),
@@ -338,7 +341,7 @@ pub struct Computer {
     #[cfg(feature = "debug")]
     pub(crate) error: bool,
     #[cfg(feature = "debug")]
-    pub(crate) book: HashMap<u32, String>,
+    pub(crate) book: (HashMap<u32, String>, HashMap<String, u32>),
 }
 
 impl Index<Reg> for Computer {
@@ -355,10 +358,6 @@ impl IndexMut<Reg> for Computer {
     }
 }
 
-fn iot() -> ! {
-    exit(1);
-}
-
 impl Computer {
     pub fn new(filename: String) -> Self {
         let mut new = Self {
@@ -371,7 +370,7 @@ impl Computer {
             #[cfg(feature = "debug")]
             error: false,
             #[cfg(feature = "debug")]
-            book: HashMap::new(),
+            book: (HashMap::new(), HashMap::new()),
         };
         let mut buf = String::new();
         std::fs::File::open(filename)
@@ -389,14 +388,16 @@ impl Computer {
         while let Some((_, line)) = lines.next() {
             if let Some([addr, s]) = line.split_ascii_whitespace().collect::<Vec<_>>().as_array() {
                 if let Ok(i) = u32::from_str_radix(addr, 16) {
-                    new.book.insert(i, s.to_string());
+                    new.book.0.insert(i, s.to_string());
+                    new.book.1.insert(s.to_string(), i);
                 }
             }
         }
         new
     }
+
     #[inline(always)]
-    pub fn step(&mut self) {
+    pub fn step(&mut self, s: usize) {
         match self.interupts {
             InteruptState::Disabled => {}
             InteruptState::Enabled => {
@@ -415,7 +416,7 @@ impl Computer {
             }
             InteruptState::Serving(..) => {}
         }
-
+        for _ in 0..s {
             //potentially just changed by interupt.
             let next_opcode = self.ram[self.pc];
 
@@ -449,15 +450,15 @@ impl Computer {
                             self.pc += 1;
                         }
                         Instruction::Lsl(reg, reg1, op2) => {
-                        self[reg] = self[reg1] << self.resolve(op2);
+                            self[reg] = (self[reg1] as u64).wrapping_shl(self.resolve(op2)) as u32;
                             self.pc += 1;
                         }
                         Instruction::Lsr(reg, reg1, op2) => {
-                        self[reg] = self[reg1] >> self.resolve(op2);
+                            self[reg] = (self[reg1] as u64).wrapping_shr(self.resolve(op2)) as u32;
                             self.pc += 1;
                         }
                         Instruction::Asr(reg, reg1, op2) => {
-                        self[reg] = (self[reg1] as i32 >> self.resolve(op2)) as u32;
+                            self[reg] = (self[reg1] as i64).wrapping_shr(self.resolve(op2)) as u32;
                             self.pc += 1;
                         }
                         Instruction::Umull(reg, reg1, op2) => {
@@ -508,13 +509,18 @@ impl Computer {
                                 self.ram[addr / 4] = self[reg1];
                             } else if addr <= 0x00ff_ffff + 480 * 640 * 4 {
                                 let buf_addr = (addr - 0x0100_0000) / 4;
-                            let dat = self[reg1] & 0x00FF_FFFF;
+                                let dat = if cfg!(feature = "rgba") {
+                                    self[reg1]
+                                } else {
+                                    self[reg1] & 0x00FF_FFFF
+                                };
                                 (&SHARED.screen_buf[buf_addr])
                                     .store(dat, std::sync::atomic::Ordering::Relaxed);
                             } else if addr == 0x0120_1000 {
                                 (&SHARED.external_enabled_interupts)
                                     .store(self[reg1], std::sync::atomic::Ordering::Relaxed);
                             } else {
+                                cold_path();
                                 self.serve_interupt(InteruptKind::IllegalOpcode, [next_opcode]);
                             }
                         }
@@ -537,23 +543,34 @@ impl Computer {
                             } else {
                                 match addr as isize - 0x0120_0000 {
                                     #[cfg(feature = "rich_keyboard")]
-                                -12 => {
+                                    -12 => SHARED.keyboard[0]
-                                    SHARED.keyboard[0].load(std::sync::atomic::Ordering::Relaxed)
+                                        .load(std::sync::atomic::Ordering::Relaxed),
-                                }
                                     #[cfg(feature = "rich_keyboard")]
-                                -8 => SHARED.keyboard[1].load(std::sync::atomic::Ordering::Relaxed),
+                                    -8 => SHARED.keyboard[1]
+                                        .load(std::sync::atomic::Ordering::Relaxed),
                                     #[cfg(feature = "rich_keyboard")]
-                                -4 => SHARED.keyboard[2].load(std::sync::atomic::Ordering::Relaxed),
+                                    -4 => SHARED.keyboard[2]
-                                0 => SHARED.keyboard[3].load(std::sync::atomic::Ordering::Relaxed),
+                                        .load(std::sync::atomic::Ordering::Relaxed),
+                                    0 => SHARED.keyboard[3]
+                                        .load(std::sync::atomic::Ordering::Relaxed),
                                     4 => time::Instant::now()
                                         .duration_since(self.creation)
-                                    .as_millis() as u32,
+                                        .as_millis()
+                                        as u32,
                                     8 => SHARED.mouse[0].load(std::sync::atomic::Ordering::Relaxed),
-                                12 => SHARED.mouse[1].load(std::sync::atomic::Ordering::Relaxed),
+                                    12 => {
-                                16 => SHARED.mouse[2].load(std::sync::atomic::Ordering::Relaxed),
+                                        SHARED.mouse[1].load(std::sync::atomic::Ordering::Relaxed)
+                                    }
+                                    16 => {
+                                        SHARED.mouse[2].load(std::sync::atomic::Ordering::Relaxed)
+                                    }
                                     //guaranted by the inequality and is multiple of 4
                                     _ => {
-                                    self.serve_interupt(InteruptKind::IllegalOpcode, [next_opcode]);
+                                        cold_path();
+                                        self.serve_interupt(
+                                            InteruptKind::IllegalOpcode,
+                                            [next_opcode],
+                                        );
                                         return;
                                     }
                                 }
@@ -571,14 +588,17 @@ impl Computer {
                         }
                         Instruction::Skip(d, cond, reg, op2) => {
                             self.pc += 1;
-                        if cond.eval(self[reg], self.resolve(op2)) {
+                            match cond.eval(self[reg], self.resolve(op2)) {
-                            self.pc += d as usize
+                                Some(false) => { /*Nothing*/ }
+                                Some(true) => self.pc += d as usize,
+                                None => {
+                                    cold_path();
+                                    self.serve_interupt(InteruptKind::IllegalOpcode, [next_opcode])
                                 }
                             }
-                    Instruction::Jump(mut addr) => {
+                        }
-                        if addr & (1 << 28) != 0 {
+                        Instruction::Jump(addr) => {
-                            addr += 7 << 29;
+                            if unlikely(addr == 0) {
-                        } else if addr == 0 {
                                 #[cfg(feature = "debug")]
                                 {
                                     match self.interupts {
@@ -601,16 +621,16 @@ impl Computer {
                                     println!("awaiting interupt...");
                                 }
                                 SHARED.external_interupts.wait(0);
+                                return;
                             }
-                        self.pc = (addr + self.pc as u32) as usize;
+                            self.pc = ((addr + self.pc as u32) & 0x1FFF_FFFF) as usize;
                         }
-                    Instruction::Call(mut addr) => {
+                        Instruction::Call(addr) => {
+                            //WARNING! addr still has the type bit set at this point!
                             self.sp -= 1;
                             self.ram[self.sp] = ((self.pc << 2) + 4) as u32;
 
-                        if addr & (1 << 28) != 0 {
+                            if unlikely(addr == 0 + (1 << 29) /*t bit*/) {
-                            addr += 7 << 29;
-                        } else if unlikely(addr == 0) {
                                 #[cfg(feature = "debug")]
                                 {
                                     println!("program terminated");
@@ -620,7 +640,8 @@ impl Computer {
                                 #[cfg(not(feature = "debug"))]
                                 exit(0);
                             }
-                        self.pc = (addr + self.pc as u32) as usize;
+                            //The mask take care of both wrapping and shedding the t bit
+                            self.pc = ((addr + self.pc as u32) & 0x1FFF_FFFF) as usize;
                         }
                         Instruction::Ret() => {
                             self.pc = (self.ram[self.sp] >> 2) as usize;
@@ -643,12 +664,13 @@ impl Computer {
                                         InteruptKind::MMIO => {
                                             (&SHARED.external_interupts)
                                                 .store(0, std::sync::atomic::Ordering::Release);
-                                        SHARED.external_interupts.notify_one();
+                                            SHARED.external_interupts.signal();
                                             //no need to check prev because MMIO is the lowest priority
                                             self.interupts = InteruptState::Enabled
                                         }
                                         InteruptKind::Swi => {}
-                                    InteruptKind::DivByZero | InteruptKind::UnsupportedOpcode => {
+                                        InteruptKind::DivByZero
+                                        | InteruptKind::UnsupportedOpcode => {
                                             ret_index = Some(self.regs[0]);
                                             ret_value = self.regs[1];
                                         }
@@ -687,8 +709,9 @@ impl Computer {
 
                             (&SHARED.external_enabled_interupts)
                                 .store(0, std::sync::atomic::Ordering::Relaxed);
-                        (&SHARED.external_interupts).store(0, std::sync::atomic::Ordering::Relaxed);
+                            (&SHARED.external_interupts)
-                        SHARED.external_interupts.notify_all();
+                                .store(0, std::sync::atomic::Ordering::Relaxed);
+                            SHARED.external_interupts.signal();
                         }
                         Instruction::Swi() => {
                             self.pc += 1;
@@ -712,14 +735,26 @@ impl Computer {
                 Err((kind, rx, ry, op2, opcode)) => {
                     self.pc += 1;
                     match kind {
-                    InteruptKind::UnsupportedOpcode => self
+                        InteruptKind::UnsupportedOpcode => self.serve_interupt(
-                        .serve_interupt(kind, [rx.0.into(), self[ry], self.resolve(op2), opcode]),
+                            kind,
-                    InteruptKind::IllegalOpcode => self.serve_interupt(kind, [next_opcode]),
+                            [rx.0.into(), self[ry], self.resolve(op2), opcode],
+                        ),
+                        InteruptKind::IllegalOpcode => {
+                            cold_path();
+
+                            self.serve_interupt(kind, [next_opcode])
+                        }
                         _ => unsafe { unreachable_unchecked() },
                     }
                 }
             }
         }
+    }
+    #[cfg(feature = "debug")]
+    pub fn debug_step(&mut self, s: usize) {
+        self.step(s);
+    }
+    #[inline(always)]
     fn resolve(&self, op2: Op2) -> u32 {
         match op2 {
             Op2::Direct(v) => v,

simu/src/main.rs

@@ -1,9 +1,4 @@
-#![feature(
+#![feature(likely_unlikely, widening_mul, int_lowest_highest_one)]
-    likely_unlikely,
-    widening_mul,
-    sync_unsafe_cell,
-    int_lowest_highest_one
-)]
 #![deny(clippy::all)]
 
 use std::env::args;
@@ -16,17 +11,18 @@ use std::sync::{
 use std::thread::scope;
 use std::time::{Duration, Instant};
 
-use pixels::{Error, Pixels, SurfaceTexture};
+use pixels::wgpu::{BlendState, Color};
-use wait_on_address::AtomicWait;
+use pixels::{Error, Pixels, PixelsBuilder, SurfaceTexture};
 use winit::application::ApplicationHandler;
 use winit::dpi::LogicalSize;
 use winit::event::WindowEvent;
 use winit::event_loop::EventLoop;
 use winit::platform::scancode::PhysicalKeyExtScancode;
 use winit::window::Window;
-// use winit_input_helper::WinitInputHelper;
 
 use crate::cpu::{Computer, MMIOInterupt};
+mod wait;
+use wait::WaitOnAtomic;
 mod cpu;
 use cpu::SHARED;
 
@@ -66,7 +62,12 @@ impl<'a> ApplicationHandler for App<'a> {
         self.w = Some(window.clone());
         let size = window.inner_size();
         let surface_texture = SurfaceTexture::new(size.width, size.height, window);
-        self.pixels = Some(Pixels::new(WIDTH, HEIGHT, surface_texture).unwrap());
+        let pix = PixelsBuilder::new(WIDTH, HEIGHT, surface_texture)
+            .clear_color(Color::BLACK)
+            .blend_state(BlendState::REPLACE)
+            .build();
+
+        self.pixels = Some(pix.unwrap());
     }
 
     fn window_event(
@@ -91,6 +92,7 @@ impl<'a> ApplicationHandler for App<'a> {
                 print!("Keyboard event: ");
                 #[cfg(feature = "rich_keyboard")]
                 {
+                    use winit::platform::modifier_supplement::KeyEventExtModifierSupplement;
                     let kb0 = key_event.text_with_all_modifiers().map_or(u32::MAX, |s| {
                         s.as_bytes()
                             .into_iter()
@@ -121,7 +123,7 @@ impl<'a> ApplicationHandler for App<'a> {
                     (&SHARED.external_interupts).store(MMIOInterupt::Keyboard.into(), Release);
                     #[cfg(feature = "debug")]
                     println!("wake due to keyboard event");
-                    SHARED.external_interupts.notify_one();
+                    SHARED.external_interupts.signal();
                 }
             }
             WindowEvent::CursorMoved { position, .. } => {
@@ -150,7 +152,7 @@ impl<'a> ApplicationHandler for App<'a> {
                     (&SHARED.external_interupts).store(MMIOInterupt::MouseMove.into(), Release);
                     #[cfg(feature = "debug")]
                     println!("wake due mouse move");
-                    SHARED.external_interupts.notify_one();
+                    SHARED.external_interupts.signal();
                 }
             }
             // WindowEvent::MouseWheel {
@@ -184,7 +186,7 @@ impl<'a> ApplicationHandler for App<'a> {
                     (&SHARED.external_interupts).store(MMIOInterupt::MouseClick.into(), Release);
                     #[cfg(feature = "debug")]
                     println!("wake mouse click");
-                    SHARED.external_interupts.notify_one();
+                    SHARED.external_interupts.signal();
                 }
             }
             WindowEvent::ScaleFactorChanged { .. } => {
@@ -197,7 +199,7 @@ impl<'a> ApplicationHandler for App<'a> {
                     != 0;
                 if enabled {
                     (&SHARED.external_interupts).store(MMIOInterupt::VSync.into(), Relaxed);
-                    SHARED.external_interupts.notify_one();
+                    SHARED.external_interupts.signal();
                     wait_int();
                 }
                 let pix = self.pixels.as_mut().unwrap();
@@ -206,14 +208,9 @@ impl<'a> ApplicationHandler for App<'a> {
                 for (addr, ubgr) in cpu::SHARED.screen_buf.iter().enumerate() {
                     let raw = ubgr.load(std::sync::atomic::Ordering::Relaxed);
                     #[cfg(not(feature = "rgba"))]
-                    let rgba = [raw as u8, (raw >> 8) as u8, (raw >> 16) as u8, 0xff];
+                    let rgba: [u8; 4] = raw.to_le_bytes();
                     #[cfg(feature = "rgba")]
-                    let rgba = [
+                    let rgba = raw.to_be_bytes();
-                        (raw >> 24) as u8,
-                        (raw >> 16) as u8,
-                        (raw >> 8) as u8,
-                        raw as u8,
-                    ];
                     for i in 0..4 {
                         screen[addr * 4 + i] = rgba[i];
                     }
@@ -264,6 +261,7 @@ impl<'a> ApplicationHandler for App<'a> {
                 let now = Instant::now();
                 if next < now {
                     next = now + Duration::from_secs_f64(1. / 30.);
+                    println!("Warning: rendering is lagging!")
                 }
                 event_loop.set_control_flow(winit::event_loop::ControlFlow::WaitUntil(next));
                 if let Some(w) = self.w.as_ref() {
@@ -301,147 +299,13 @@ fn main() -> Result<(), Error> {
             let mut simulation = Computer::new(program);
             #[cfg(not(feature = "debug"))]
             loop {
-                simulation.step();
+                simulation.step(64);
             }
+
+            //ugly debug code, I should improve that using a real TUI crate
             #[cfg(feature = "debug")]
             {
-                let mut input = stdin().lines();
+                debug_loop(&mut simulation);
-                loop {
-                    {
-                        println!(
-                            "interrupts are {:?}, with mmio interupts flags {}",
-                            simulation.interupts,
-                            SHARED.external_enabled_interupts.load(Relaxed)
-                        );
-                        for i in 0..8 {
-                            println!(
-                                "r{i} 0x{:08x} r{} 0x{:08x}",
-                                simulation.regs[i],
-                                i + 8,
-                                simulation.regs[i + 8]
-                            );
-                        }
-                        println!(
-                            "SP: {:08x}     PC: {:08x}",
-                            simulation.sp * 4,
-                            simulation.pc * 4
-                        );
-                        println!("RAM near SP");
-                        let min_pc = (simulation.pc).min(0x0100_0000 / 4 - 8);
-                        let min_sp = (simulation.sp).min(0x0100_0000 / 4 - 8);
-                        for i in 0..8 {
-                            println!(
-                                "{:8x}: 0x{:08x}",
-                                (min_sp + i) * 4,
-                                simulation.ram[min_sp + i],
-                            );
-                        }
-                        println!("Ram near PC");
-                        for i in 0..8 {
-                            let idx = min_pc + i;
-                            let istr = simulation.ram[idx];
-                            if let Some(s) = simulation.book.get(&(idx as u32 * 4)) {
-                                println!("{s}:")
-                            };
-                            println!(
-                                "{:8x}: 0x{:08x} {}",
-                                idx * 4,
-                                istr,
-                                cpu::instr_to_text(istr, idx as u32 * 4, &simulation.book)
-                            )
-                        }
-                    }
-                    while {
-                        let next = input.next().unwrap().unwrap();
-                        let next: Vec<_> = next.split_ascii_whitespace().collect();
-                        if next.len() > 0 {
-                            match next[0] {
-                                "s" | "step" => {
-                                    let n: usize = {
-                                        if next.len() >= 2 {
-                                            parse_int::parse(next[1]).unwrap_or(1)
-                                        } else {
-                                            1
-                                        }
-                                    };
-                                    for _ in 0..n {
-                                        if simulation.error {
-                                            println!("cannot step, cpu killed");
-                                            break;
-                                        }
-                                        simulation.step();
-                                    }
-                                    false
-                                }
-                                "r" | "run" => {
-                                    while !simulation.error {
-                                        simulation.step();
-                                    }
-                                    false
-                                }
-                                "p" | "print" => {
-                                    if next.len() >= 2 {
-                                        match parse_int::parse::<u32>(next[1]) {
-                                            Ok(i) => {
-                                                let v = simulation.ram[i as usize / 4];
-                                                println!(
-                                                    "0x{:8x} -- {}",
-                                                    v,
-                                                    cpu::instr_to_text(v, i, &simulation.book)
-                                                );
-                                                true
-                                            }
-                                            Err(e) => {
-                                                println!("{e}");
-                                                true
-                                            }
-                                        }
-                                    } else {
-                                        println!("{HELP_MSG}");
-                                        true
-                                    }
-                                }
-                                "c" | "context" => false,
-                                "u" | "up" => {
-                                    while !simulation.error
-                                        && simulation.ram[simulation.pc] != 0x8800_0000
-                                    {
-                                        simulation.step();
-                                    }
-                                    false
-                                }
-                                "t" | "to" => {
-                                    if next.len() >= 2 {
-                                        match parse_int::parse::<u32>(next[1]) {
-                                            Ok(v) => {
-                                                while !simulation.error
-                                                    && simulation.pc != (v as usize / 4)
-                                                {
-                                                    simulation.step();
-                                                }
-                                                false
-                                            }
-                                            Err(e) => {
-                                                println!("{e}");
-                                                true
-                                            }
-                                        }
-                                    } else {
-                                        println!("{HELP_MSG}");
-                                        true
-                                    }
-                                }
-                                _ => {
-                                    println!("{HELP_MSG}");
-                                    true
-                                }
-                            }
-                        } else {
-                            println!("{HELP_MSG}");
-                            true
-                        }
-                    } {}
-                }
             }
         });
 
@@ -458,11 +322,210 @@ fn main() -> Result<(), Error> {
 }
 
 #[cfg(feature = "debug")]
-const HELP_MSG: &str = "
+fn debug_loop(com: &mut Computer) {
-step n  - step trough n instructions (alias s)
+    struct Wrap(DefaultCompleter, DefaultCompleter);
-run     - run program until exit / error (alias r)
+
-context - print context (alias c)
+    use clap::Parser;
-print n - print ram content at address n and next 8 (alias p)
+    use clap_repl::ClapEditor;
-up      - run until the nex ret is reached (alias u)
+    use clap_repl::reedline::{
-to n    - tun until PC = n (alias t)
+        Completer, DefaultCompleter, DefaultPrompt, DefaultPromptSegment, FileBackedHistory,
-";
+    };
+
+    use crate::cpu::instr_to_text;
+
+    #[derive(Debug, Parser)]
+    #[command(name = "")] // This name will show up in clap's error messages, so it is important to set it to "".
+    enum Commands {
+        /// Step by single instrcution
+        #[command(alias = "s")]
+        Step {
+            ///number of instruction to step (default one)
+            num: Option<usize>,
+        },
+        /// Run until program halt, or specified instruction is reached
+        #[command(alias = "r")]
+        Run {
+            /// Can be either a label or a address (support hex format)
+            desigantor: Option<String>,
+        },
+        /// run until the current function return.
+        #[command(alias = "u")]
+        Up,
+        /// Print memory at address. Support hexa format
+        #[command(alias = "p")]
+        Print { address: String },
+        /// Print the address associated with a label
+        #[command(alias = "l")]
+        Label { label: String },
+        /// Print context
+        #[command(alias = "c")]
+        Context,
+    }
+    let prompt = DefaultPrompt {
+        left_prompt: DefaultPromptSegment::Basic(">>".to_owned()),
+        right_prompt: DefaultPromptSegment::Empty,
+    };
+
+    let commands_comp = DefaultCompleter::new_with_wordlen(
+        ["step", "run", "up", "print", "label", "help", "context"]
+            .map(|s| s.to_string())
+            .to_vec(),
+        0,
+    );
+    let mut labels_comp =
+        DefaultCompleter::with_inclusions("_0123456789".chars().collect::<Vec<_>>().as_slice())
+            .set_min_word_len(0);
+    labels_comp.insert(com.book.0.values().cloned().collect());
+
+    let editor = ClapEditor::<Commands>::builder()
+        .with_prompt(Box::new(prompt))
+        .with_editor_hook(|reed| {
+            reed.with_history(Box::new(
+                FileBackedHistory::with_file(1000, "debug_cmd.hist".into()).unwrap(),
+            ))
+            .with_completer(Box::new(Wrap(commands_comp, labels_comp)))
+            .with_quick_completions(true)
+            .with_partial_completions(true)
+        })
+        .build();
+
+    debug_context(com);
+    editor.repl(|command| match command {
+        Commands::Step { num } => {
+            let steps = num.unwrap_or(1);
+            com.debug_step(steps);
+            debug_context(com);
+        }
+        Commands::Run { desigantor } => match desigantor {
+            Some(s) => match parse_int::parse::<usize>(s.as_str()) {
+                Ok(addr) => {
+                    while com.pc != (addr / 4) && !com.error {
+                        com.debug_step(1);
+                    }
+                    debug_context(com);
+                }
+                Err(_) => match com.book.1.get(s.as_str()).cloned() {
+                    Some(addr) => {
+                        while com.pc != (addr as usize / 4) && !com.error {
+                            com.debug_step(1);
+                        }
+                        debug_context(com);
+                    }
+                    None => {
+                        println!("Error, {s} cannot be interpreted as addr nor label")
+                    }
+                },
+            },
+            None => {
+                while !com.error {
+                    com.debug_step(64);
+                }
+                debug_context(com);
+            }
+        },
+        Commands::Up => {
+            let curr_sp = com.sp;
+            while (com.sp < curr_sp
+                || (com.ram[com.pc] != (0b10001000 << 24) && com.ram[com.pc] != (0b10101000 << 24)))
+                && !com.error
+            {
+                com.debug_step(1);
+            }
+            debug_context(com);
+        }
+        Commands::Print { address } => match parse_int::parse::<usize>(address.as_str()) {
+            Ok(addr) => match com.ram.get(addr / 4) {
+                Some(i) => {
+                    println!(
+                        "RAM at {addr:8x}: {:8x} {}",
+                        i,
+                        instr_to_text(*i, u32::MAX, &com.book.0)
+                    )
+                }
+                None => println!("Cannot index RAM at address {addr:8x}"),
+            },
+            Err(_) => match com.book.1.get(address.as_str()).cloned() {
+                Some(addr) => println!(
+                    "RAM at {addr:8x}: {:8x} {}",
+                    com.ram[addr as usize / 4],
+                    instr_to_text(com.ram[addr as usize / 4], addr, &com.book.0)
+                ),
+                None => {
+                    println!("Error, {address} cannot be interpreted as addr nor label")
+                }
+            },
+        },
+        Commands::Label { label } => match com.book.1.get(label.as_str()) {
+            Some(addr) => println!("label is at addr {addr}"),
+            None => println!("error: label not found"),
+        },
+        Commands::Context => debug_context(com),
+    });
+    exit(0);
+
+    impl Completer for Wrap {
+        fn complete(&mut self, line: &str, pos: usize) -> Vec<clap_repl::reedline::Suggestion> {
+            let trimmed = line.trim_start();
+            let line_parts = trimmed.splitn(2, ' ').collect::<Vec<_>>();
+
+            if line_parts.len() <= 1 {
+                self.0.complete(line, pos)
+            } else {
+                match line_parts[0] {
+                    "r" | "run" | "p" | "print" | "l" | "label" => {
+                        let trimmed_2 = line_parts[1].trim_start();
+                        let offset = line.len() - trimmed_2.len();
+                        let mut sub = self.1.complete(trimmed_2, pos - offset);
+                        for sug in sub.iter_mut() {
+                            use clap_repl::reedline::Span;
+
+                            sug.span = Span {
+                                start: sug.span.start + offset,
+                                end: sug.span.end + offset,
+                            }
+                        }
+                        sub
+                    }
+                    _ => Vec::new(),
+                }
+            }
+        }
+    }
+}
+
+#[cfg(feature = "debug")]
+fn debug_context(com: &Computer) {
+    use crate::cpu::instr_to_text;
+    println!("Interupt state: {:?}", com.interupts);
+    for i in 0..8 {
+        println!(
+            "r{i} = {:8x}     r{:<2} = {:8x}",
+            com.regs[i],
+            i + 8,
+            com.regs[i + 8]
+        );
+    }
+    println!("SP={:08x}        PC={:08x}", com.sp * 4, com.pc * 4);
+    println!("RAM at SP | Ram at PC:");
+
+    let mut pc_lines = Vec::new();
+
+    for i in 0..16 {
+        match com.book.0.get(&((com.pc + i) as u32 * 4)) {
+            Some(label) => pc_lines.push(format!("        {label}:")),
+            None => {}
+        };
+        pc_lines.push(format!(
+            "{:08x}  {}",
+            com.ram[com.pc + i],
+            instr_to_text(com.ram[com.pc + i], (com.pc + i) as u32 * 4, &com.book.0)
+        ));
+    }
+    for (i, pc_l) in pc_lines.iter().enumerate() {
+        if com.sp + i < com.ram.len() {
+            println!("{:08x}  |  {pc_l}", com.ram[com.sp + i])
+        } else {
+            println!("   --     |  {pc_l}")
+        }
+    }
+}

simu/src/wait.rs

@@ -0,0 +1,35 @@
+use std::sync::atomic::AtomicU32;
+
+#[cfg(feature = "futex")]
+use wait_on_address::AtomicWait;
+
+pub trait WaitOnAtomic {
+    fn wait(&self, v: u32);
+    fn signal(&self);
+}
+
+#[cfg(feature = "futex")]
+impl WaitOnAtomic for AtomicU32 {
+    fn wait(&self, v: u32) {
+        <AtomicU32 as AtomicWait>::wait(self, v);
+    }
+
+    fn signal(&self) {
+        self.notify_one();
+    }
+}
+
+#[cfg(not(feature = "futex"))]
+impl WaitOnAtomic for AtomicU32 {
+    fn wait(&self, v: u32) {
+        while self.load(std::sync::atomic::Ordering::Acquire) == v {
+            use std::{thread::sleep, time::Duration};
+
+            sleep(Duration::from_micros(500));
+        }
+    }
+
+    fn signal(&self) {
+        //
+    }
+}