Support watchpoints

src/emulator/address_set.rs

@@ -0,0 +1,231 @@
+use std::{
+    collections::{BTreeMap, HashSet},
+    ops::Bound,
+};
+
+#[derive(Debug, Default)]
+pub struct AddressSet {
+    ranges: HashSet<(u32, usize)>,
+    bounds: BTreeMap<u32, usize>,
+}
+
+impl AddressSet {
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    pub fn add(&mut self, address: u32, length: usize) {
+        if length == 0 || !self.ranges.insert((address, length)) {
+            return;
+        }
+        let end = (address as usize)
+            .checked_add(length)
+            .and_then(|e| u32::try_from(e).ok());
+
+        if let Some(end) = end {
+            let val_before = self.bounds.range(..=end).next_back().map_or(0, |(_, &v)| v);
+            self.bounds.insert(end, val_before);
+        }
+
+        let val_before = self
+            .bounds
+            .range(..address)
+            .next_back()
+            .map_or(0, |(_, &v)| v);
+        if let Some(&val_at) = self.bounds.get(&address) {
+            if val_before == val_at + 1 {
+                self.bounds.remove(&address);
+            }
+        } else {
+            self.bounds.insert(address, val_before);
+        }
+
+        let start_bound = Bound::Included(address);
+        let end_bound = match end {
+            Some(e) => Bound::Excluded(e),
+            None => Bound::Unbounded,
+        };
+
+        for (_, val) in self.bounds.range_mut((start_bound, end_bound)) {
+            *val += 1;
+        }
+    }
+
+    pub fn remove(&mut self, address: u32, length: usize) {
+        if !self.ranges.remove(&(address, length)) {
+            return;
+        }
+        let end = (address as usize)
+            .checked_add(length)
+            .and_then(|e| u32::try_from(e).ok());
+
+        if let Some(end) = end {
+            let val_before = self.bounds.range(..end).next_back().map_or(0, |(_, &v)| v);
+            if let Some(&val_at) = self.bounds.get(&end) {
+                if val_at + 1 == val_before {
+                    self.bounds.remove(&end);
+                }
+            } else {
+                self.bounds.insert(end, val_before);
+            }
+        }
+
+        let val_before = self
+            .bounds
+            .range(..address)
+            .next_back()
+            .map_or(0, |(_, &v)| v);
+        if let Some(&val_at) = self.bounds.get(&address) {
+            if val_before + 1 == val_at {
+                self.bounds.remove(&address);
+            }
+        } else {
+            self.bounds.insert(address, val_before);
+        }
+
+        let start_bound = Bound::Included(address);
+        let end_bound = match end {
+            Some(e) => Bound::Excluded(e),
+            None => Bound::Unbounded,
+        };
+
+        for (_, val) in self.bounds.range_mut((start_bound, end_bound)) {
+            *val -= 1;
+        }
+    }
+
+    pub fn clear(&mut self) {
+        self.ranges.clear();
+        self.bounds.clear();
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.bounds.is_empty()
+    }
+
+    pub fn contains(&self, address: u32) -> bool {
+        self.bounds
+            .range(..=address)
+            .next_back()
+            .is_some_and(|(_, &val)| val > 0)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::AddressSet;
+
+    #[test]
+    fn should_not_include_addresses_when_empty() {
+        let set = AddressSet::new();
+        assert!(set.is_empty());
+        assert!(!set.contains(0x13374200));
+    }
+
+    #[test]
+    fn should_include_addresses_when_full() {
+        let mut set = AddressSet::new();
+        set.add(0x00000000, 0x100000000);
+        assert!(set.contains(0x13374200));
+    }
+
+    #[test]
+    fn should_ignore_empty_address_ranges() {
+        let mut set = AddressSet::new();
+        set.add(0x13374200, 0);
+        assert!(set.is_empty());
+        assert!(!set.contains(0x13374200));
+    }
+
+    #[test]
+    fn should_add_addresses_idempotently() {
+        let mut set = AddressSet::new();
+        set.add(0x13374200, 1);
+        set.add(0x13374200, 1);
+        set.remove(0x13374200, 1);
+        assert!(set.is_empty());
+        assert!(!set.contains(0x13374200));
+    }
+
+    #[test]
+    fn should_remove_addresses_idempotently() {
+        let mut set = AddressSet::new();
+        set.add(0x13374200, 1);
+        set.remove(0x13374200, 1);
+        set.remove(0x13374200, 1);
+        assert!(set.is_empty());
+        assert!(!set.contains(0x13374200));
+    }
+
+    #[test]
+    fn should_report_address_in_range() {
+        let mut set = AddressSet::new();
+        set.add(0x13374200, 4);
+        assert!(!set.contains(0x133741ff));
+        for address in 0x13374200..0x13374204 {
+            assert!(set.contains(address));
+        }
+        assert!(!set.contains(0x13374204));
+    }
+
+    #[test]
+    fn should_allow_overlapping_addresses() {
+        let mut set = AddressSet::new();
+        set.add(0x13374200, 4);
+        set.add(0x13374201, 1);
+        set.add(0x13374202, 2);
+
+        assert!(!set.contains(0x133741ff));
+        for address in 0x13374200..0x13374204 {
+            assert!(set.contains(address));
+        }
+        assert!(!set.contains(0x13374204));
+
+        set.remove(0x13374200, 4);
+        assert!(!set.contains(0x13374200));
+        for address in 0x13374201..0x13374204 {
+            assert!(set.contains(address));
+        }
+        assert!(!set.contains(0x13374204));
+    }
+
+    #[test]
+    fn should_allow_removing_overlapped_address_ranges() {
+        let mut set = AddressSet::new();
+        set.add(0x13374200, 8);
+        set.add(0x13374204, 8);
+        set.remove(0x13374204, 8);
+
+        for address in 0x13374200..0x13374208 {
+            assert!(set.contains(address));
+        }
+        for address in 0x13374208..0x1337420c {
+            assert!(!set.contains(address));
+        }
+    }
+
+    #[test]
+    fn should_merge_adjacent_ranges() {
+        let mut set = AddressSet::new();
+        set.add(0x13374200, 4);
+        set.add(0x13374204, 4);
+        set.add(0x13374208, 4);
+        set.add(0x1337420c, 4);
+
+        assert!(!set.contains(0x133741ff));
+        for address in 0x13374200..0x13374210 {
+            assert!(set.contains(address));
+        }
+        assert!(!set.contains(0x13374210));
+
+        set.remove(0x13374200, 4);
+        set.remove(0x13374204, 4);
+        set.remove(0x13374208, 4);
+        set.remove(0x1337420c, 4);
+
+        assert!(set.is_empty());
+        for address in 0x133741ff..=0x13374210 {
+            assert!(!set.contains(address));
+        }
+    }
+}

src/emulator/shrooms_vb_core.rs

@@ -5,6 +5,8 @@ use bitflags::bitflags;
 use num_derive::{FromPrimitive, ToPrimitive};
 use serde::{Deserialize, Serialize};
 
+use super::address_set::AddressSet;
+
 #[repr(C)]
 struct VB {
     _data: [u8; 0],
@@ -62,9 +64,31 @@ pub enum VBRegister {
     PC,
 }
 
-type OnFrame = extern "C" fn(sim: *mut VB) -> c_int;
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+pub enum VBWatchpointType {
+    Read,
+    Write,
+    Access,
+}
+
 type OnExecute =
     extern "C" fn(sim: *mut VB, address: u32, code: *const u16, length: c_int) -> c_int;
+type OnFrame = extern "C" fn(sim: *mut VB) -> c_int;
+type OnRead = extern "C" fn(
+    sim: *mut VB,
+    address: u32,
+    type_: VBDataType,
+    value: *mut i32,
+    cycles: *mut u32,
+) -> c_int;
+type OnWrite = extern "C" fn(
+    sim: *mut VB,
+    address: u32,
+    type_: VBDataType,
+    value: *mut i32,
+    cycles: *mut u32,
+    cancel: *mut c_int,
+) -> c_int;
 
 #[link(name = "vb")]
 extern "C" {
@@ -112,15 +136,17 @@ extern "C" {
     #[link_name = "vbSetCartROM"]
     fn vb_set_cart_rom(sim: *mut VB, rom: *mut c_void, size: u32) -> c_int;
     #[link_name = "vbSetExecuteCallback"]
-    fn vb_set_execute_callback(sim: *mut VB, callback: Option<OnExecute>);
+    fn vb_set_execute_callback(sim: *mut VB, callback: Option<OnExecute>) -> Option<OnExecute>;
     #[link_name = "vbSetFrameCallback"]
-    fn vb_set_frame_callback(sim: *mut VB, callback: Option<OnFrame>);
+    fn vb_set_frame_callback(sim: *mut VB, callback: Option<OnFrame>) -> Option<OnFrame>;
     #[link_name = "vbSetKeys"]
     fn vb_set_keys(sim: *mut VB, keys: u16) -> u16;
     #[link_name = "vbSetOption"]
     fn vb_set_option(sim: *mut VB, key: VBOption, value: c_int);
     #[link_name = "vbSetPeer"]
     fn vb_set_peer(sim: *mut VB, peer: *mut VB);
+    #[link_name = "vbSetReadCallback"]
+    fn vb_set_read_callback(sim: *mut VB, callback: Option<OnRead>) -> Option<OnRead>;
     #[link_name = "vbSetSamples"]
     fn vb_set_samples(
         sim: *mut VB,
@@ -130,6 +156,8 @@ extern "C" {
     ) -> c_int;
     #[link_name = "vbSetUserData"]
     fn vb_set_user_data(sim: *mut VB, tag: *mut c_void);
+    #[link_name = "vbSetWriteCallback"]
+    fn vb_set_write_callback(sim: *mut VB, callback: Option<OnWrite>) -> Option<OnWrite>;
     #[link_name = "vbSizeOf"]
     fn vb_size_of() -> usize;
 }
@@ -147,18 +175,73 @@ extern "C" fn on_execute(sim: *mut VB, address: u32, _code: *const u16, _length:
     // There is no way for the userdata to be null or otherwise invalid.
     let data: &mut VBState = unsafe { &mut *vb_get_user_data(sim).cast() };
 
-    let mut stopped = 0;
+    let mut stopped = data.stop_reason.is_some();
     if data.step_from.is_some_and(|s| s != address) {
         data.step_from = None;
         data.stop_reason = Some(StopReason::Stepped);
-        stopped = 1;
+        stopped = true;
     }
     if data.breakpoints.binary_search(&address).is_ok() {
         data.stop_reason = Some(StopReason::Breakpoint);
-        stopped = 1;
+        stopped = true;
     }
 
-    stopped
+    if stopped {
+        1
+    } else {
+        0
+    }
+}
+
+extern "C" fn on_read(
+    sim: *mut VB,
+    address: u32,
+    _type: VBDataType,
+    _value: *mut i32,
+    _cycles: *mut u32,
+) -> c_int {
+    // SAFETY: the *mut VB owns its userdata.
+    // There is no way for the userdata to be null or otherwise invalid.
+    let data: &mut VBState = unsafe { &mut *vb_get_user_data(sim).cast() };
+
+    if data.read_watchpoints.contains(address) {
+        let watch = if data.write_watchpoints.contains(address) {
+            VBWatchpointType::Access
+        } else {
+            VBWatchpointType::Read
+        };
+        data.stop_reason = Some(StopReason::Watchpoint(watch, address));
+    }
+
+    // Don't stop here, the debugger expects us to break after the memory access.
+    // We'll stop in on_execute instead.
+    0
+}
+
+extern "C" fn on_write(
+    sim: *mut VB,
+    address: u32,
+    _type: VBDataType,
+    _value: *mut i32,
+    _cycles: *mut u32,
+    _cancel: *mut c_int,
+) -> c_int {
+    // SAFETY: the *mut VB owns its userdata.
+    // There is no way for the userdata to be null or otherwise invalid.
+    let data: &mut VBState = unsafe { &mut *vb_get_user_data(sim).cast() };
+
+    if data.write_watchpoints.contains(address) {
+        let watch = if data.read_watchpoints.contains(address) {
+            VBWatchpointType::Access
+        } else {
+            VBWatchpointType::Write
+        };
+        data.stop_reason = Some(StopReason::Watchpoint(watch, address));
+    }
+
+    // Don't stop here, the debugger expects us to break after the memory access.
+    // We'll stop in on_execute instead.
+    0
 }
 
 const AUDIO_CAPACITY_SAMPLES: usize = 834 * 4;
@@ -170,6 +253,23 @@ struct VBState {
     stop_reason: Option<StopReason>,
     step_from: Option<u32>,
     breakpoints: Vec<u32>,
+    read_watchpoints: AddressSet,
+    write_watchpoints: AddressSet,
+}
+
+impl VBState {
+    fn needs_execute_callback(&self) -> bool {
+        self.step_from.is_some()
+            || !self.breakpoints.is_empty()
+            || !self.read_watchpoints.is_empty()
+            || !self.write_watchpoints.is_empty()
+    }
+}
+
+pub enum StopReason {
+    Breakpoint,
+    Watchpoint(VBWatchpointType, u32),
+    Stepped,
 }
 
 #[repr(transparent)]
@@ -177,11 +277,6 @@ pub struct Sim {
     sim: *mut VB,
 }
 
-pub enum StopReason {
-    Breakpoint,
-    Stepped,
-}
-
 impl Sim {
     pub fn new() -> Self {
         // init the VB instance itself
@@ -201,6 +296,8 @@ impl Sim {
             stop_reason: None,
             step_from: None,
             breakpoints: vec![],
+            read_watchpoints: AddressSet::new(),
+            write_watchpoints: AddressSet::new(),
         };
         unsafe { vb_set_user_data(sim, Box::into_raw(Box::new(state)).cast()) };
         unsafe { vb_set_frame_callback(sim, Some(on_frame)) };
@@ -374,7 +471,71 @@ impl Sim {
         let data = self.get_state();
         if let Ok(index) = data.breakpoints.binary_search(&address) {
             data.breakpoints.remove(index);
-            if data.step_from.is_none() && data.breakpoints.is_empty() {
+            if !data.needs_execute_callback() {
+                unsafe { vb_set_execute_callback(self.sim, None) };
+            }
+        }
+    }
+
+    pub fn add_watchpoint(&mut self, address: u32, length: usize, watch: VBWatchpointType) {
+        match watch {
+            VBWatchpointType::Read => self.add_read_watchpoint(address, length),
+            VBWatchpointType::Write => self.add_write_watchpoint(address, length),
+            VBWatchpointType::Access => {
+                self.add_read_watchpoint(address, length);
+                self.add_write_watchpoint(address, length);
+            }
+        }
+    }
+
+    fn add_read_watchpoint(&mut self, address: u32, length: usize) {
+        let state = self.get_state();
+        state.read_watchpoints.add(address, length);
+        if !state.read_watchpoints.is_empty() {
+            unsafe { vb_set_read_callback(self.sim, Some(on_read)) };
+            unsafe { vb_set_execute_callback(self.sim, Some(on_execute)) };
+        }
+    }
+
+    fn add_write_watchpoint(&mut self, address: u32, length: usize) {
+        let state = self.get_state();
+        state.write_watchpoints.add(address, length);
+        if !state.write_watchpoints.is_empty() {
+            unsafe { vb_set_write_callback(self.sim, Some(on_write)) };
+            unsafe { vb_set_execute_callback(self.sim, Some(on_execute)) };
+        }
+    }
+
+    pub fn remove_watchpoint(&mut self, address: u32, length: usize, watch: VBWatchpointType) {
+        match watch {
+            VBWatchpointType::Read => self.remove_read_watchpoint(address, length),
+            VBWatchpointType::Write => self.remove_write_watchpoint(address, length),
+            VBWatchpointType::Access => {
+                self.remove_read_watchpoint(address, length);
+                self.remove_write_watchpoint(address, length);
+            }
+        }
+    }
+
+    fn remove_read_watchpoint(&mut self, address: u32, length: usize) {
+        let state = self.get_state();
+        state.read_watchpoints.remove(address, length);
+        let needs_execute = state.needs_execute_callback();
+        if state.read_watchpoints.is_empty() {
+            unsafe { vb_set_read_callback(self.sim, None) };
+            if !needs_execute {
+                unsafe { vb_set_execute_callback(self.sim, None) };
+            }
+        }
+    }
+
+    fn remove_write_watchpoint(&mut self, address: u32, length: usize) {
+        let state = self.get_state();
+        state.write_watchpoints.remove(address, length);
+        let needs_execute = state.needs_execute_callback();
+        if state.write_watchpoints.is_empty() {
+            unsafe { vb_set_write_callback(self.sim, None) };
+            if !needs_execute {
                 unsafe { vb_set_execute_callback(self.sim, None) };
             }
         }
@@ -393,13 +554,17 @@ impl Sim {
         let data = self.get_state();
         data.step_from = None;
         data.breakpoints.clear();
+        data.read_watchpoints.clear();
+        data.write_watchpoints.clear();
+        unsafe { vb_set_read_callback(self.sim, None) };
+        unsafe { vb_set_write_callback(self.sim, None) };
         unsafe { vb_set_execute_callback(self.sim, None) };
     }
 
     pub fn stop_reason(&mut self) -> Option<StopReason> {
         let data = self.get_state();
         let reason = data.stop_reason.take();
-        if data.step_from.is_none() && data.breakpoints.is_empty() {
+        if !data.needs_execute_callback() {
             unsafe { vb_set_execute_callback(self.sim, None) };
         }
         reason