CSY/reowolf Changeset - 7d01f1245b7c · Centrum Wiskunde & Informatica (CWI)

Changeset - 7d01f1245b7c

Parent rev.

Child rev.

[Not reviewed]

0 12 0

mh - 4 years ago 2021-10-20 20:26:23
contact@maxhenger.nl

Everything compiles again, pending restructuring of shared runtime objects

12 files changed with 302 insertions and 305 deletions:

src/collections/raw_vec.rs

src/protocol/parser/type_table.rs

src/runtime2/connector.rs

src/runtime2/global_store.rs

src/runtime2/inbox.rs

src/runtime2/messages.rs

src/runtime2/mod.rs

114

src/runtime2/native.rs

src/runtime2/port.rs

src/runtime2/runtime.rs

src/runtime2/scheduler.rs

src/runtime2/tests/mod.rs

117

0 comments (0 inline, 0 general)

src/collections/raw_vec.rs

➞

Show inline comments

@@ @@ -13,112 +13,112 @@ enum AllocError { @@
 /// free the backing memory, but will not run any destructors.
 pub struct RawVec<T: Sized> {
     base: *mut T,
     cap: usize,
     len: usize,
+}
 impl<T: Sized> RawVec<T> {
     const T_ALIGNMENT: usize = mem::align_of::<T>();
     const T_SIZE: usize = mem::size_of::<T>();
     const GROWTH_RATE: usize = 2;
     pub fn new() -> Self {
         Self{
             base: ptr::null_mut(),
             cap: 0,
             len: 0,
+        }
+    }
     pub fn with_capacity(capacity: usize) -> Self {
         // Could be done a bit more efficiently
         let mut result = Self::new();
         result.ensure_space(capacity);
+        result.ensure_space(capacity).unwrap();
         return result;
+    }
     pub unsafe fn get(&self, idx: usize) -> *const T {
         debug_assert!(idx < self.len);
         return self.base.add(idx);
+    }
     pub unsafe fn get_mut(&self, idx: usize) -> *mut T {
         debug_assert!(idx < self.len);
         return self.base.add(idx);
+    }
     pub fn push(&mut self, item: T) {
         self.ensure_space(1);
+        self.ensure_space(1).unwrap();
         unsafe {
             let target = self.base.add(self.len);
             std::ptr::write(target, item);
             self.len += 1;
+        }
+    }
     pub fn len(&self) -> usize {
         return self.len;
+    }
     fn ensure_space(&mut self, additional: usize) -> Result<(), AllocError>{
         debug_assert!(Self::T_SIZE != 0);
         debug_assert!(self.cap >= self.len);
         if self.cap - self.len < additional {
             // Need to resize. Note that due to all checked conditions we have
             // that new_cap >= 1.
             debug_assert!(additional > 0);
             let new_cap = self.len.checked_add(additional).unwrap();
             let new_cap = cmp::max(new_cap, self.cap * Self::GROWTH_RATE);
             let layout = Layout::array::<T>(new_cap)
                 .map_err(|_| AllocError::CapacityOverflow)?;
             debug_assert_eq!(new_cap * Self::T_SIZE, layout.size());
             unsafe {
                 // Allocate new storage, transfer bits, deallocate old store
                 let new_base = alloc(layout);
                 if self.cap > 0 {
                     let old_base = self.base as *mut u8;
                     let (old_size, old_layout) = self.current_layout();
                     ptr::copy_nonoverlapping(new_base, old_base, old_size);
                     dealloc(old_base, old_layout);
+                }
                 self.base = new_base;
+                self.base = new_base as *mut T;
                 self.cap = new_cap;
+            }
         } // else: still enough space
         return Ok(());
+    }
     #[inline]
     fn current_layout(&self) -> (usize, Layout) {
         debug_assert!(Self::T_SIZE > 0);
         let old_size = self.cap * Self::T_SIZE;
         unsafe {
             return (
                 old_size,
                 Layout::from_size_align_unchecked(old_size, Self::T_ALIGNMENT)
             );
+        }
+    }
+}
 impl<T: Sized> Drop for RawVec<T> {
     fn drop(&mut self) {
         if self.cap > 0 {
             debug_assert!(!self.base.is_null());
             let (_, layout) = self.current_layout();
             unsafe {
                 dealloc(self.base, layout);
+                dealloc(self.base as *mut u8, layout);
                 if cfg!(debug_assertions) {
                     self.base = ptr::null_mut();
+                }
+            }
+        }
+    }
+}
@@ \ No newline at end of file @@

src/protocol/parser/type_table.rs

➞

Show inline comments

@@ @@ -1098,49 +1098,48 @@ impl TypeTable { @@
     /// resolvable. If so then the appropriate union variants will be marked as
     /// "living on heap". If not then a `ParseError` will be returned
     fn detect_and_resolve_type_loops_for(&mut self, modules: &[Module], heap: &Heap, concrete_type: ConcreteType) -> Result<(), ParseError> {
         use DefinedTypeVariant as DTV;
         debug_assert!(self.type_loop_breadcrumbs.is_empty());
         debug_assert!(self.type_loops.is_empty());
         debug_assert!(self.encountered_types.is_empty());
         // Push the initial breadcrumb
         let initial_breadcrumb = self.check_member_for_type_loops(&concrete_type);
         if let TypeLoopResult::PushBreadcrumb(definition_id, concrete_type) = initial_breadcrumb {
             self.handle_new_breadcrumb_for_type_loops(definition_id, concrete_type);
         } else {
             unreachable!();
+        }
         // Enter into the main resolving loop
         while !self.type_loop_breadcrumbs.is_empty() {
             // Because we might be modifying the breadcrumb array we need to
             let breadcrumb_idx = self.type_loop_breadcrumbs.len() - 1;
             let mut breadcrumb = self.type_loop_breadcrumbs[breadcrumb_idx].clone();
             let poly_type = self.lookup.get(&breadcrumb.definition_id).unwrap();
             let poly_type_definition_id = poly_type.ast_definition;
             let resolve_result = match &poly_type.definition {
                 DTV::Enum(_) => {
                     TypeLoopResult::TypeExists
                 },
                 DTV::Union(definition) => {
                     let monomorph = &definition.monomorphs[breadcrumb.monomorph_idx];
                     let num_variants = monomorph.variants.len();
                     let mut union_result = TypeLoopResult::TypeExists;
                     'member_loop: while breadcrumb.next_member < num_variants {
                         let mono_variant = &monomorph.variants[breadcrumb.next_member];
                         let num_embedded = mono_variant.embedded.len();
                         while breadcrumb.next_embedded < num_embedded {
                             let mono_embedded = &mono_variant.embedded[breadcrumb.next_embedded];
                             union_result = self.check_member_for_type_loops(&mono_embedded.concrete_type);
                             if union_result != TypeLoopResult::TypeExists {
                                 // In type loop or new breadcrumb pushed, so
                                 // break out of the resolving loop
                                 break 'member_loop;
+                            }

src/runtime2/connector.rs

➞

Show inline comments

 use std::collections::HashMap;
 use std::sync::atomic::AtomicBool;
 use crate::{PortId, ProtocolDescription};
 use crate::protocol::{ComponentState, RunContext, RunResult};
 use crate::protocol::eval::{Prompt, Value, ValueGroup};
 use crate::runtime2::inbox::{MessageContents, SolutionMessage};
+use crate::runtime2::inbox::{Message, MessageContents, SolutionMessage};
 use crate::runtime2::native::Connector;
 use crate::runtime2::port::{Port, PortKind};
 use crate::runtime2::scheduler::ConnectorCtx;
 use super::global_store::ConnectorId;
 use super::inbox::{
     PrivateInbox, PublicInbox, DataMessage, SyncMessage,
     SyncBranchConstraint, SyncConnectorSolution
 };
 use super::port::PortIdLocal;
 /// Represents the identifier of a branch (the index within its container). An
 /// ID of `0` generally means "no branch" (e.g. no parent, or a port did not
 /// yet receive anything from any branch).
 #[derive(Clone, Copy, PartialEq, Eq)]
-pub(crate) struct BranchId {
 pub struct BranchId {
     pub index: u32,
+}
 impl BranchId {
     fn new_invalid() -> Self {
         Self{ index: 0 }
+    }
     fn new(index: u32) -> Self {
         debug_assert!(index != 0);
         Self{ index }
+    }
     #[inline]
     pub(crate) fn is_valid(&self) -> bool {
         return self.index != 0;
+    }
+}
 #[derive(Debug, PartialEq, Eq)]
 pub(crate) enum SpeculativeState {
     // Non-synchronous variants
     RunningNonSync,         // regular execution of code
     Error,                  // encountered a runtime error
@@ @@ -124,92 +124,94 @@ impl PortAssignment { @@
+        }
+    }
     #[inline]
     fn mark_speculative(&mut self, num_times_fired: u32) {
         debug_assert!(!self.last_registered_branch_id.is_valid());
         self.is_assigned = true;
         self.num_times_fired = num_times_fired;
+    }
     #[inline]
     fn mark_definitive(&mut self, branch_id: BranchId, num_times_fired: u32) {
         self.is_assigned = true;
         self.last_registered_branch_id = branch_id;
         self.num_times_fired = num_times_fired;
+    }
+}
 #[derive(Clone)]
 struct PortOwnershipDelta {
     acquired: bool, // if false, then released ownership
     port_id: PortIdLocal,
+}
 #[derive(Debug)]
 enum PortOwnershipError {
     UsedInInteraction(PortIdLocal),
     AlreadyGivenAway(PortIdLocal)
+}
 /// Contains a description of the port mapping during a particular sync session.
 /// TODO: Extend documentation
 pub(crate) struct ConnectorPorts {
     // Essentially a mapping from `port_index` to `port_id`.
     pub owned_ports: Vec<PortIdLocal>,
     // Contains P*B entries, where P is the number of ports and B is the number
     // of branches. One can find the appropriate mapping of port p at branch b
     // at linear index `b*P+p`.
-    pub port_mapping: Vec<PortAssignment>
     port_mapping: Vec<PortAssignment>
+}
 impl ConnectorPorts {
     /// Constructs the initial ports object. Assumes the presence of the
     /// non-sync branch at index 0. Will initialize all entries for the non-sync
     /// branch.
     fn new(owned_ports: Vec<PortIdLocal>) -> Self {
         let num_ports = owned_ports.len();
         let mut port_mapping = Vec::with_capacity(num_ports);
         for _ in 0..num_ports {
             port_mapping.push(PortAssignment::new_unassigned());
+        }
         Self{ owned_ports, port_mapping }
+    }
     /// Prepares the port mapping for a new branch. Assumes that there is no
     /// intermediate branch index that we have skipped.
     fn prepare_sync_branch(&mut self, parent_branch_idx: u32, new_branch_idx: u32) {
         let num_ports = self.owned_ports.len();
         let parent_base_idx = parent_branch_idx as usize * num_ports;
         let new_base_idx = new_branch_idx as usize * num_ports;
         debug_assert!(parent_branch_idx < new_branch_idx);
         debug_assert!(new_base_idx == self.port_mapping.len());
         self.port_mapping.reserve(num_ports);
         for offset in 0..num_ports {
             let parent_port = &self.port_mapping[parent_base_idx + offset];
             self.port_mapping.push(parent_port.clone());
             let parent_port = parent_port.clone();
             self.port_mapping.push(parent_port);
+        }
+    }
     /// Adds a new port. Caller must make sure that the connector is not in the
     /// sync phase.
     fn add_port(&mut self, port_id: PortIdLocal) {
         debug_assert!(self.port_mapping.len() == self.owned_ports.len());
         debug_assert!(!self.owned_ports.contains(&port_id));
         self.owned_ports.push(port_id);
         self.port_mapping.push(PortAssignment::new_unassigned());
+    }
     /// Commits to a particular branch. Essentially just removes the port
     /// mapping information generated during the sync phase.
     fn commit_to_sync(&mut self) {
         self.port_mapping.truncate(self.owned_ports.len());
         debug_assert!(self.port_mapping.iter().all(|v| {
             !v.is_assigned && !v.last_registered_branch_id.is_valid()
         }));
+    }
     /// Removes a particular port from the connector. May only be done if the
     /// connector is in non-sync mode
     fn remove_port(&mut self, port_id: PortIdLocal) {
@@ @@ -282,102 +284,102 @@ impl BranchQueue { @@
         Self{ first: 0, last: 0 }
+    }
     #[inline]
     fn is_empty(&self) -> bool {
         debug_assert!((self.first == 0) == (self.last == 0));
         return self.first == 0;
+    }
     #[inline]
     fn clear(&mut self) {
         self.first = 0;
         self.last = 0;
+    }
+}
 /// Public fields of the connector that can be freely shared between multiple
 /// threads.
 pub(crate) struct ConnectorPublic {
     pub inbox: PublicInbox,
     pub sleeping: AtomicBool,
+}
 impl ConnectorPublic {
     pub fn new() -> Self {
+    pub fn new(initialize_as_sleeping: bool) -> Self {
         ConnectorPublic{
             inbox: PublicInbox::new(),
-            sleeping: AtomicBool::new(false),
+            sleeping: AtomicBool::new(initialize_as_sleeping),
+        }
+    }
+}
 // TODO: Maybe prevent false sharing by aligning `public` to next cache line.
 // TODO: Do this outside of the connector, create a wrapping struct
 pub(crate) struct ConnectorPDL {
     // State and properties of connector itself
     in_sync: bool,
     // Branch management
     branches: Vec<Branch>, // first branch is always non-speculative one
     sync_active: BranchQueue,
     sync_pending_get: BranchQueue,
     sync_finished: BranchQueue,
     sync_finished_last_handled: u32, // TODO: Change to BranchId?
     cur_round: u32,
     // Port/message management
     pub committed_to: Option<(ConnectorId, u64)>,
     pub inbox: PrivateInbox,
     pub ports: ConnectorPorts,
+}
 struct TempCtx {}
 impl RunContext for TempCtx {
     fn did_put(&mut self, port: PortId) -> bool {
         todo!()
+    }
     fn get(&mut self, port: PortId) -> Option<ValueGroup> {
         todo!()
+    }
     fn fires(&mut self, port: PortId) -> Option<Value> {
         todo!()
+    }
     fn get_channel(&mut self) -> Option<(Value, Value)> {
         todo!()
+    }
+}
 impl Connector for ConnectorPDL {
-    fn handle_message(&mut self, message: MessageContents, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) {
+    fn handle_message(&mut self, message: Message, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) {
         use MessageContents as MC;
         match message {
             MC::Data(message) => self.handle_data_message(message),
             MC::Sync(message) => self.handle_sync_message(message, ctx, delta_state),
             MC::RequestCommit(message) => self.handle_request_commit_message(message, ctx, delta_state),
             MC::ConfirmCommit(message) => self.handle_confirm_commit_message(message, ctx, delta_state),
         match message.contents {
             MC::Data(content) => self.handle_data_message(message.receiving_port, content),
             MC::Sync(content) => self.handle_sync_message(content, ctx, delta_state),
             MC::RequestCommit(content) => self.handle_request_commit_message(content, ctx, delta_state),
             MC::ConfirmCommit(content) => self.handle_confirm_commit_message(content, ctx, delta_state),
             MC::Control(_) | MC::Ping => {},
+        }
+    }
     fn run(&mut self, pd: &ProtocolDescription, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) -> ConnectorScheduling {
         if self.in_sync {
             let scheduling = self.run_in_speculative_mode(pd, ctx, delta_state);
             // When in speculative mode we might have generated new sync
             // solutions, we need to turn them into proposed solutions here.
             if self.sync_finished_last_handled != self.sync_finished.last {
                 // Retrieve first element in queue
                 let mut next_id;
                 if self.sync_finished_last_handled == 0 {
                     next_id = self.sync_finished.first;
                 } else {
                     let last_handled = &self.branches[self.sync_finished_last_handled as usize];
                     debug_assert!(last_handled.next_branch_in_queue.is_some()); // because "last handled" != "last in queue"
                     next_id = last_handled.next_branch_in_queue.unwrap();
+                }
                 loop {
                     let branch_id = BranchId::new(next_id);
                     let branch = &self.branches[next_id as usize];
@@ @@ -425,50 +427,50 @@ impl ConnectorPDL { @@
     pub fn new(initial_branch: Branch, owned_ports: Vec<PortIdLocal>) -> Self {
         Self{
             in_sync: false,
             branches: vec![initial_branch],
             sync_active: BranchQueue::new(),
             sync_pending_get: BranchQueue::new(),
             sync_finished: BranchQueue::new(),
             sync_finished_last_handled: 0, // none at all
             cur_round: 0,
             committed_to: None,
             inbox: PrivateInbox::new(),
             ports: ConnectorPorts::new(owned_ports),
+        }
+    }
     pub fn is_in_sync_mode(&self) -> bool {
         return self.in_sync;
+    }
     // -------------------------------------------------------------------------
     // Handling connector messages
     // -------------------------------------------------------------------------
     #[inline]
     pub fn handle_data_message(&mut self, message: DataMessage) {
         self.inbox.insert_message(message);
     pub fn handle_data_message(&mut self, target_port: PortIdLocal, message: DataMessage) {
         self.inbox.insert_message(target_port, message);
+    }
     /// Accepts a synchronous message and combines it with the locally stored
     /// solution(s). Then queue new `Sync`/`Solution` messages when appropriate.
     pub fn handle_sync_message(&mut self, message: SyncMessage, ctx: &ConnectorCtx, results: &mut RunDeltaState) {
         debug_assert!(!message.to_visit.contains(&ctx.id)); // own ID already removed
         debug_assert!(message.constraints.iter().any(|v| v.connector_id == ctx.id)); // we have constraints
         // TODO: Optimize, use some kind of temp workspace vector
         let mut execution_path_branch_ids = Vec::new();
         if self.sync_finished_last_handled != 0 {
             // We have some solutions to match against
             let constraints_index = message.constraints
                 .iter()
                 .position(|v| v.connector_id == ctx.id)
                 .unwrap();
             let constraints = &message.constraints[constraints_index].constraints;
             debug_assert!(!constraints.is_empty());
             // Note that we only iterate over the solutions we've already
             // handled ourselves, not necessarily
             let mut branch_index = self.sync_finished.first;
             'branch_loop: loop {
@@ @@ -556,101 +558,102 @@ impl ConnectorPDL { @@
                         let port = ctx.get_port(port_id);
                         (port.peer_connector, port.peer_id, port.kind == PortKind::Putter)
                     };
                     let mapping = self.ports.get_port(branch_index, port_index);
                     let constraint = if mapping.num_times_fired == 0 {
                         SyncBranchConstraint::SilentPort(peer_port_id)
                     } else {
                         if peer_is_getter {
                             SyncBranchConstraint::PortMapping(peer_port_id, mapping.last_registered_branch_id)
                         } else {
                             SyncBranchConstraint::BranchNumber(mapping.last_registered_branch_id)
+                        }
                     };
                     match new_solution.add_or_check_constraint(peer_connector_id, constraint) {
                         Err(_) => continue 'branch_loop,
                         Ok(false) => continue 'branch_loop,
                         Ok(true) => {},
+                    }
+                }
                 // If here, then the newly generated solution is completely
                 // compatible.
                 let next_branch = branch.next_branch_in_queue;
                 self.submit_sync_solution(new_solution, ctx, results);
                 // Consider the next branch
                 if branch_index == self.sync_finished_last_handled {
                     // At the end of the previously handled solutions
                     break;
+                }
                 debug_assert!(branch.next_branch_in_queue.is_some()); // because we cannot be at the end of the queue
                 branch_index = branch.next_branch_in_queue.unwrap();
                 debug_assert!(next_branch.is_some()); // because we cannot be at the end of the queue
                 branch_index = next_branch.unwrap();
+            }
+        }
+    }
     fn handle_request_commit_message(&mut self, mut message: SolutionMessage, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) {
         let should_propagate_message = match &self.committed_to {
             Some((previous_origin, previous_comparison)) => {
                 // Already committed to something. So will commit to this if it
                 // takes precedence over the current solution
                 message.comparison_number > *previous_comparison ||
                     (message.comparison_number == *previous_comparison && message.connector_origin.0 > previous_origin.0)
             },
             None => {
                 // Not yet committed to a solution, so commit to this one
                 true
+            }
         };
         if should_propagate_message {
             self.committed_to = Some((message.connector_origin, message.comparison_number));
             if message.to_visit.is_empty() {
                 // Visited all of the connectors, so every connector can now
                 // apply the solution
                 // TODO: Use temporary workspace
                 let mut to_visit = Vec::with_capacity(message.local_solutions.len() - 1);
                 for (connector_id, _) in &message.local_solutions {
                     if *connector_id != ctx.id {
                         to_visit.push(*connector_id);
+                    }
+                }
                 message.to_visit = to_visit;
                 self.handle_confirm_commit_message(message.clone(), ctx, delta_state);
                 delta_state.outbox.push(MessageContents::ConfirmCommit(message));
             } else {
                 // Not yet visited all of the connectors
                 delta_state.outbox.push(MessageContents::RequestCommit(message));
+            }
+        }
+    }
-    fn handle_confirm_commit_message(&mut self, message: SolutionMessage, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) {
+    fn handle_confirm_commit_message(&mut self, message: SolutionMessage, ctx: &ConnectorCtx, _delta_state: &mut RunDeltaState) {
         // Make sure this is the message we actually committed to. As long as
         // we're running on a single machine this is fine.
         // TODO: Take care of nefarious peers
         let (expected_connector_id, expected_comparison_number) =
             self.committed_to.unwrap();
         assert_eq!(message.connector_origin, expected_connector_id);
         assert_eq!(message.comparison_number, expected_comparison_number);
         // Find the branch we're supposed to commit to
         let (_, branch_id) = message.local_solutions
             .iter()
             .find(|(id, _)| *id == ctx.id)
             .unwrap();
         let branch_id = *branch_id;
         // Commit to the branch. That is: move the solution branch to the first
         // of the connector's branches
         self.in_sync = false;
         self.branches.swap(0, branch_id.index as usize);
         self.branches.truncate(1); // TODO: Or drain and do not deallocate?
         let solution = &mut self.branches[0];
         // Clear all of the other sync-related variables
         self.sync_active.clear();
@@ @@ -662,264 +665,280 @@ impl ConnectorPDL { @@
         self.committed_to = None;
         self.inbox.clear();
         self.ports.commit_to_sync();
         // Add/remove any of the ports we lost during the sync phase
         for port_delta in &solution.ports_delta {
             if port_delta.acquired {
                 self.ports.add_port(port_delta.port_id);
             } else {
                 self.ports.remove_port(port_delta.port_id);
+            }
+        }
         solution.commit_to_sync();
+    }
     // -------------------------------------------------------------------------
     // Executing connector code
     // -------------------------------------------------------------------------
     /// Runs the connector in synchronous mode. Potential changes to the global
     /// system's state are added to the `RunDeltaState` object by the connector,
     /// where it is the caller's responsibility to immediately take care of
     /// those changes. The return value indicates when (and if) the connector
     /// needs to be scheduled again.
-    pub fn run_in_speculative_mode(&mut self, pd: &ProtocolDescription, context: &ConnectorCtx, results: &mut RunDeltaState) -> ConnectorScheduling {
+    pub fn run_in_speculative_mode(&mut self, pd: &ProtocolDescription, _context: &ConnectorCtx, results: &mut RunDeltaState) -> ConnectorScheduling {
         debug_assert!(self.in_sync);
         debug_assert!(!self.sync_active.is_empty());
         let branch = Self::pop_branch_from_queue(&mut self.branches, &mut self.sync_active);
         // Run the branch to the next blocking point
         let mut run_context = TempCtx{};
         let run_result = branch.code_state.run(&mut run_context, pd);
         // Match statement contains `return` statements only if the particular
         // run result behind handled requires an immediate re-run of the
         // connector.
         match run_result {
             RunResult::BranchInconsistent => {
                 // Speculative branch became inconsistent
                 branch.sync_state = SpeculativeState::Inconsistent;
             },
             RunResult::BranchMissingPortState(port_id) => {
                 // Branch called `fires()` on a port that does not yet have an
                 // assigned speculative value. So we need to create those
                 // branches
                 let local_port_id = PortIdLocal::new(port_id.0.u32_suffix);
                 let local_port_index = self.ports.get_port_index(local_port_id).unwrap();
                 debug_assert!(self.ports.owned_ports.contains(&local_port_id));
                 let silent_branch = &*branch;
                 // Create a copied branch who will have the port set to firing
                 let firing_index = self.branches.len() as u32;
                 let mut firing_branch = Branch::new_sync_branching_from(firing_index, silent_branch);
                 self.ports.prepare_sync_branch(branch.index.index, firing_index);
                 // Create two copied branches, one silent and one firing
                 branch.sync_state = SpeculativeState::HaltedAtBranchPoint;
                 let parent_branch_id = branch.index;
                 let parent_branch = &self.branches[parent_branch_id.index as usize];
                 let firing_port = self.ports.get_port_mut(firing_index, local_port_index);
                 firing_port.mark_speculative(1);
                 let silent_index = self.branches.len() as u32;
                 let firing_index = silent_index + 1;
                 let silent_branch = Branch::new_sync_branching_from(silent_index, parent_branch);
                 self.ports.prepare_sync_branch(parent_branch.index.index, silent_index);
                 // Assign the old branch a silent value
                 let silent_port = self.ports.get_port_mut(silent_branch.index.index, local_port_index);
                 let firing_branch = Branch::new_sync_branching_from(firing_index, parent_branch);
                 self.ports.prepare_sync_branch(parent_branch.index.index, firing_index);
                 // Assign the port values of the two new branches
                 let silent_port = self.ports.get_port_mut(silent_index, local_port_index);
                 silent_port.mark_speculative(0);
                 let firing_port = self.ports.get_port_mut(firing_index, local_port_index);
                 firing_port.mark_speculative(1);
                 // Run both branches again
                 Self::push_branch_into_queue(&mut self.branches, &mut self.sync_active, silent_branch.index);
                 Self::push_branch_into_queue(&mut self.branches, &mut self.sync_active, firing_branch.index);
                 let silent_branch_id = silent_branch.index;
                 self.branches.push(silent_branch);
                 let firing_branch_id = firing_branch.index;
                 self.branches.push(firing_branch);
                 Self::push_branch_into_queue(&mut self.branches, &mut self.sync_active, silent_branch_id);
                 Self::push_branch_into_queue(&mut self.branches, &mut self.sync_active, firing_branch_id);
                 return ConnectorScheduling::Immediate;
             },
             RunResult::BranchMissingPortValue(port_id) => {
                 // Branch performed a `get` on a port that has not yet received
                 // a value in its inbox.
                 let local_port_id = PortIdLocal::new(port_id.0.u32_suffix);
                 let local_port_index = self.ports.get_port_index(local_port_id);
                 if local_port_index.is_none() {
                     todo!("deal with the case where the port is acquired");
+                }
                 let local_port_index = local_port_index.unwrap();
                 let port_mapping = self.ports.get_port_mut(branch.index.index, local_port_index);
                 // Check for port mapping assignment and, if present, if it is
                 // consistent
                 let is_valid_get = if port_mapping.is_assigned {
                     assert!(port_mapping.num_times_fired <= 1); // temporary, until we get rid of `fires`
                     port_mapping.num_times_fired == 1
                 } else {
                     // Not yet assigned
                     port_mapping.mark_speculative(1);
                     true
                 };
                 if is_valid_get {
                     // Mark as a branching point for future messages
                     branch.sync_state = SpeculativeState::HaltedAtBranchPoint;
                     Self::push_branch_into_queue(&mut self.branches, &mut self.sync_pending_get, branch.index);
                     let branch_id = branch.index;
                     Self::push_branch_into_queue(&mut self.branches, &mut self.sync_pending_get, branch_id);
                     // But if some messages can be immediately applied, do so
                     // now.
                     let messages = self.inbox.get_messages(local_port_id, port_mapping.last_registered_branch_id);
                     let mut did_have_messages = false;
                     for message in messages {
                         did_have_messages = true;
                         // For each message prepare a new branch to execute
                         let parent_branch = &self.branches[branch_id.index as usize];
                         let new_branch_index = self.branches.len() as u32;
                         let mut new_branch = Branch::new_sync_branching_from(new_branch_index, branch);
                         self.ports.prepare_sync_branch(branch.index.index, new_branch_index);
                         let mut new_branch = Branch::new_sync_branching_from(new_branch_index, parent_branch);
                         self.ports.prepare_sync_branch(branch_id.index, new_branch_index);
                         let port_mapping = self.ports.get_port_mut(new_branch_index, local_port_index);
                         port_mapping.last_registered_branch_id = message.sender_cur_branch_id;
                         debug_assert!(port_mapping.is_assigned && port_mapping.num_times_fired == 1);
                         new_branch.received.insert(local_port_id, message.clone());
                         // If the message contains any ports then they will now
                         // be owned by the new branch
                         debug_assert!(results.ports.is_empty());
                         find_ports_in_value_group(&message.message, &mut results.ports);
                         Self::acquire_ports_during_sync(&mut self.ports, &mut new_branch, &results.ports);
                         results.ports.clear();
                         // Schedule the new branch
                         debug_assert!(new_branch.sync_state == SpeculativeState::RunningInSync);
-                        Self::push_branch_into_queue(&mut self.branches, &mut self.sync_active, new_branch.index);
+                        let new_branch_id = new_branch.index;
                         self.branches.push(new_branch);
                         Self::push_branch_into_queue(&mut self.branches, &mut self.sync_active, new_branch_id);
+                    }
                     if did_have_messages {
                         // If we did create any new branches, then we can run
                         // them immediately.
                         return ConnectorScheduling::Immediate;
+                    }
                 } else {
                     branch.sync_state = SpeculativeState::Inconsistent;
+                }
             },
             RunResult::BranchAtSyncEnd => {
                 // Branch is done, go through all of the ports that are not yet
                 // assigned and map them to non-firing.
                 for port_idx in 0..self.ports.num_ports() {
                     let port_mapping = self.ports.get_port_mut(branch.index.index, port_idx);
                     if !port_mapping.is_assigned {
                         port_mapping.mark_speculative(0);
+                    }
+                }
                 let branch_id = branch.index;
                 branch.sync_state = SpeculativeState::ReachedSyncEnd;
-                Self::push_branch_into_queue(&mut self.branches, &mut self.sync_finished, branch.index);
+                Self::push_branch_into_queue(&mut self.branches, &mut self.sync_finished, branch_id);
             },
             RunResult::BranchPut(port_id, value_group) => {
                 // Branch performed a `put` on a particualar port.
                 let local_port_id = PortIdLocal{ index: port_id.0.u32_suffix };
                 let local_port_index = self.ports.get_port_index(local_port_id);
                 if local_port_index.is_none() {
                     todo!("handle case where port was received before (i.e. in ports_delta)")
+                }
                 let local_port_index = local_port_index.unwrap();
                 // Check the port mapping for consistency
                 // TODO: For now we can only put once, so that simplifies stuff
                 let port_mapping = self.ports.get_port_mut(branch.index.index, local_port_index);
                 let is_valid_put = if port_mapping.is_assigned {
                     // Already assigned, so must be speculative and one time
                     // firing, otherwise we are `put`ing multiple times.
                     if port_mapping.last_registered_branch_id.is_valid() {
                         // Already did a `put`
                         todo!("handle error through RunDeltaState");
                     } else {
                         // Valid if speculatively firing
                         port_mapping.num_times_fired == 1
+                    }
                 } else {
                     // Not yet assigned, do so now
                     true
                 };
                 if is_valid_put {
                     // Put in run results for thread to pick up and transfer to
                     // the correct connector inbox.
                     port_mapping.mark_definitive(branch.index, 1);
                     let message = DataMessage{
                         sending_port: local_port_id,
                         sender_prev_branch_id: BranchId::new_invalid(),
                         sender_cur_branch_id: branch.index,
                         message: value_group,
                     };
                     // If the message contains any ports then we release our
                     // ownership over them in this branch
                     debug_assert!(results.ports.is_empty());
                     find_ports_in_value_group(&message.message, &mut results.ports);
                     Self::release_ports_during_sync(&mut self.ports, branch, &results.ports);
+                    Self::release_ports_during_sync(&mut self.ports, branch, &results.ports).unwrap();
                     results.ports.clear();
                     results.outbox.push(MessageContents::Data(message));
                     return ConnectorScheduling::Immediate
                 } else {
                     branch.sync_state = SpeculativeState::Inconsistent;
+                }
             },
             _ => unreachable!("unexpected run result '{:?}' while running in sync mode", run_result),
+        }
         // Not immediately scheduling, so schedule again if there are more
         // branches to run
         if self.sync_active.is_empty() {
             return ConnectorScheduling::NotNow;
         } else {
             return ConnectorScheduling::Later;
+        }
+    }
     /// Runs the connector in non-synchronous mode.
-    pub fn run_in_deterministic_mode(&mut self, pd: &ProtocolDescription, context: &ConnectorCtx, results: &mut RunDeltaState) -> ConnectorScheduling {
+    pub fn run_in_deterministic_mode(&mut self, pd: &ProtocolDescription, _context: &ConnectorCtx, results: &mut RunDeltaState) -> ConnectorScheduling {
         debug_assert!(!self.in_sync);
         debug_assert!(self.sync_active.is_empty() && self.sync_pending_get.is_empty() && self.sync_finished.is_empty());
         debug_assert!(self.branches.len() == 1);
         let branch = &mut self.branches[0];
         debug_assert!(branch.sync_state == SpeculativeState::RunningNonSync);
         let mut run_context = TempCtx{};
         let run_result = branch.code_state.run(&mut run_context, pd);
         match run_result {
             RunResult::ComponentTerminated => {
                 // Need to wait until all children are terminated
                 // TODO: Think about how to do this?
                 branch.sync_state = SpeculativeState::Finished;
                 return ConnectorScheduling::NotNow;
             },
             RunResult::ComponentAtSyncStart => {
                 // Prepare for sync execution and reschedule immediately
                 self.in_sync = true;
                 let first_sync_branch = Branch::new_sync_branching_from(1, branch);
-                Self::push_branch_into_queue(&mut self.branches, &mut self.sync_active, first_sync_branch.index);
+                let first_sync_branch_id = first_sync_branch.index;
                 self.branches.push(first_sync_branch);
                 Self::push_branch_into_queue(&mut self.branches, &mut self.sync_active, first_sync_branch_id);
                 return ConnectorScheduling::Later;
             },
             RunResult::NewComponent(definition_id, monomorph_idx, arguments) => {
                 // Construction of a new component. Find all references to ports
                 // inside of the arguments
                 debug_assert!(results.ports.is_empty());
                 find_ports_in_value_group(&arguments, &mut results.ports);
                 if !results.ports.is_empty() {
                     // Ports changing ownership
                     if let Err(_) = Self::release_ports_during_non_sync(&mut self.ports, branch, &results.ports) {
                         todo!("fatal error handling");
+                    }
+                }
                 // Add connector for later execution
                 let new_connector_state = ComponentState {
                     prompt: Prompt::new(&pd.types, &pd.heap, definition_id, monomorph_idx, arguments)
                 };
                 let new_connector_ports = results.ports.clone(); // TODO: Do something with this
                 let new_connector_branch = Branch::new_initial_branch(new_connector_state);
                 let new_connector = ConnectorPDL::new(new_connector_branch, new_connector_ports);
@@ @@ -999,48 +1018,50 @@ impl ConnectorPDL { @@
         // Walk through all elements in queue to find branch to delete
         let mut prev_index = 0;
         let mut next_index = queue.first;
         while next_index != 0 {
             if next_index == to_delete.index {
                 // Found the element we're going to delete
                 // - check if at the first element or not
                 if prev_index == 0 {
                     queue.first = branch_next_index_unwrapped;
                 } else {
                     let prev_branch = &mut branches[prev_index as usize];
                     prev_branch.next_branch_in_queue = branch_next_index_option;
+                }
                 // - check if at last element or not (also takes care of "no elements left in queue")
                 if branch_next_index_option.is_none() {
                     queue.last = prev_index;
+                }
                 return;
+            }
             prev_index = next_index;
             let entry = &branches[next_index as usize];
             next_index = entry.next_branch_in_queue.unwrap_or(0);
+        }
         // If here, then we didn't find the element
         panic!("branch does not exist in provided queue");
+    }
     // Helpers for local port management. Specifically for adopting/losing
     // ownership over ports, and for checking if specific ports can be sent
     // over another port.
     /// Releasing ownership of ports while in non-sync mode. This only occurs
     /// while instantiating new connectors
     fn release_ports_during_non_sync(ports: &mut ConnectorPorts, branch: &mut Branch, port_ids: &[PortIdLocal]) -> Result<(), PortOwnershipError> {
         debug_assert!(!branch.index.is_valid()); // branch in non-sync mode
         for port_id in port_ids {
             // We must own the port, or something is wrong with our code
             todo!("Set up some kind of message router");
             debug_assert!(ports.get_port_index(*port_id).is_some());
             ports.remove_port(*port_id);
+        }
         return Ok(())
+    }
@@ @@ -1195,49 +1216,49 @@ impl ConnectorPDL { @@
             let constraint = if port_mapping.is_assigned {
                 if port.kind == PortKind::Getter {
                     SyncBranchConstraint::BranchNumber(port_mapping.last_registered_branch_id)
                 } else {
                     SyncBranchConstraint::PortMapping(port.peer_id, port_mapping.last_registered_branch_id)
+                }
             } else {
                 SyncBranchConstraint::SilentPort(port.peer_id)
             };
             if !sync_message.add_or_check_constraint(port.peer_connector, constraint).unwrap() {
                 return None;
+            }
+        }
         return Some(sync_message);
+    }
     fn submit_sync_solution(&mut self, partial_solution: SyncMessage, ctx: &ConnectorCtx, results: &mut RunDeltaState) {
         if partial_solution.to_visit.is_empty() {
             // Solution is completely consistent. So ask everyone to commit
             // TODO: Maybe another package for random?
             let comparison_number: u64 = unsafe {
                 let mut random_array = [0u8; 8];
                 getrandom::getrandom(&mut random_array);
+                getrandom::getrandom(&mut random_array).unwrap();
                 std::mem::transmute(random_array)
             };
             let num_local = partial_solution.local_solutions.len();
             let mut full_solution = SolutionMessage{
                 comparison_number,
                 connector_origin: ctx.id,
                 local_solutions: Vec::with_capacity(num_local),
                 to_visit: Vec::with_capacity(num_local - 1),
             };
             for local_solution in &partial_solution.local_solutions {
                 full_solution.local_solutions.push((local_solution.connector_id, local_solution.terminating_branch_id));
                 if local_solution.connector_id != ctx.id {
                     full_solution.to_visit.push(local_solution.connector_id);
+                }
+            }
             debug_assert!(self.committed_to.is_none());
             self.committed_to = Some((full_solution.connector_origin, full_solution.comparison_number));
             results.outbox.push(MessageContents::RequestCommit(full_solution));
         } else {
             // Still have connectors to visit

src/runtime2/global_store.rs

➞

Show inline comments

 use std::ptr;
 use std::sync::{Arc, RwLock};
 use std::sync::atomic::{AtomicBool, AtomicU32};
 use crate::collections::{MpmcQueue, RawVec};
 use super::connector::{ConnectorPDL, ConnectorPublic};
 use super::scheduler::Router;
 use crate::ProtocolDescription;
 use crate::runtime2::connector::{ConnectorScheduling, RunDeltaState};
 use crate::runtime2::inbox::MessageContents;
 use crate::runtime2::native::{Connector, ConnectorApplication};
 use crate::runtime2::scheduler::ConnectorCtx;
 /// A kind of token that, once obtained, allows mutable access to a connector.
 /// We're trying to use move semantics as much as possible: the owner of this
 /// key is the only one that may execute the connector's code.
 pub(crate) struct ConnectorKey {
     pub index: u32, // of connector
+}
 impl ConnectorKey {
     /// Downcasts the `ConnectorKey` type, which can be used to obtain mutable
     /// access, to a "regular ID" which can be used to obtain immutable access.
     #[inline]
     pub fn downcast(&self) -> ConnectorId {
         return ConnectorId(self.index);
+    }
     /// Turns the `ConnectorId` into a `ConnectorKey`, marked as unsafe as it
     /// bypasses the type-enforced `ConnectorKey`/`ConnectorId` system
     #[inline]
     pub unsafe fn from_id(id: ConnectorId) -> ConnectorKey {
         return ConnectorKey{ index: id.0 };
+    }
+}
 /// A kind of token that allows shared access to a connector. Multiple threads
 /// may hold this
 #[derive(Debug, Copy, Clone, PartialEq, Eq)]
 pub(crate) struct ConnectorId(pub u32);
 use super::scheduler::{Router, ConnectorCtx};
 use super::connector::{ConnectorPDL, ConnectorPublic, ConnectorScheduling, RunDeltaState};
 use super::inbox::Message;
 use super::native::{Connector, ConnectorApplication};
 impl ConnectorId {
     // TODO: Like the other `new_invalid`, maybe remove
     #[inline]
     pub fn new_invalid() -> ConnectorId {
         return ConnectorId(u32::MAX);
+    }
     #[inline]
     pub(crate) fn is_valid(&self) -> bool {
         return self.0 != u32::MAX;
+    }
+}
 // TODO: Change this, I hate this. But I also don't want to put `public` and
 //  `router` of `ScheduledConnector` back into `Connector`. The reason I don't
 //  want `Box<dyn Connector>` everywhere is because of the v-table overhead. But
 //  to truly design this properly I need some benchmarks.
 pub enum ConnectorVariant {
     UserDefined(ConnectorPDL),
     Native(Box<dyn Connector>),
+}
 impl Connector for ConnectorVariant {
     fn handle_message(&mut self, message: MessageContents, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) {
         match self {
             ConnectorVariant::UserDefined(c) => c.handle_message(message, ctx, delta_state),
             ConnectorVariant::Native(c) => c.handle_message(message, ctx, delta_state),
+        }
+    }
     fn run(&mut self, protocol_description: &ProtocolDescription, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) -> ConnectorScheduling {
         match self {
             ConnectorVariant::UserDefined(c) => c.run(protocol_description, ctx, delta_state),
             ConnectorVariant::Native(c) => c.run(protocol_description, ctx, delta_state),
+        }
+    }
+}
 pub struct ScheduledConnector {
     pub connector: ConnectorVariant, // access by connector
     pub context: ConnectorCtx, // mutable access by scheduler, immutable by connector
     pub public: ConnectorPublic, // accessible by all schedulers and connectors
     pub router: Router,
+}
 /// The registry containing all connectors. The idea here is that when someone
 /// owns a `ConnectorKey`, then one has unique access to that connector.
 /// Otherwise one has shared access.
 ///
 /// This datastructure is built to be wrapped in a RwLock.
 pub(crate) struct ConnectorStore {
     pub(crate) port_counter: Arc<AtomicU32>,
     inner: RwLock<ConnectorStoreInner>,
+}
 struct ConnectorStoreInner {
     connectors: RawVec<*mut ScheduledConnector>,
     free: Vec<usize>,
+}
 impl ConnectorStore {
     fn with_capacity(capacity: usize) -> Self {
         return Self{
             port_counter: Arc::new(AtomicU32::new(0)),
             inner: RwLock::new(ConnectorStoreInner {
                 connectors: RawVec::with_capacity(capacity),
                 free: Vec::with_capacity(capacity),
             }),
@@ @@ -117,140 +42,138 @@ impl ConnectorStore { @@
         let lock = self.inner.read().unwrap();
         unsafe {
             let connector = lock.connectors.get(connector_id.0 as usize);
             debug_assert!(!connector.is_null());
             return &(**connector).public;
+        }
+    }
     /// Retrieves a particular connector. Only the thread that pulled the
     /// associated key out of the execution queue should (be able to) call this.
     pub(crate) fn get_mut(&self, key: &ConnectorKey) -> &'static mut ScheduledConnector {
         let lock = self.inner.read().unwrap();
         unsafe {
             let connector = lock.connectors.get_mut(key.index as usize);
             debug_assert!(!connector.is_null());
             return &mut (**connector);
+        }
+    }
     pub(crate) fn create_interface(&self, connector: ConnectorApplication) -> ConnectorKey {
         // Connector interface does not own any initial ports, and cannot be
         // created by another connector
         let key = self.create_connector_raw(ConnectorVariant::Native(Box::new(connector)));
+        let key = self.create_connector_raw(ConnectorVariant::Native(Box::new(connector)), true);
         return key;
+    }
     /// Create a new connector, returning the key that can be used to retrieve
     /// and/or queue it. The caller must make sure that the constructed
     /// connector's code is initialized with the same ports as the ports in the
     /// `initial_ports` array.
     /// `initial_ports` array. Furthermore the connector is initialized as not
     /// sleeping, so MUST be put on the connector queue by the caller.
     pub(crate) fn create_pdl(&self, created_by: &mut ScheduledConnector, connector: ConnectorPDL) -> ConnectorKey {
         let key = self.create_connector_raw(ConnectorVariant::UserDefined(connector));
+        let key = self.create_connector_raw(ConnectorVariant::UserDefined(connector), false);
         let new_connector = self.get_mut(&key);
         // Transferring ownership of ports (and crashing if there is a
         // programmer's mistake in port management)
         match &new_connector.connector {
             ConnectorVariant::UserDefined(connector) => {
                 for port_id in &connector.ports.owned_ports {
                     let mut port = created_by.context.remove_port(*port_id);
                     new_connector.context.add_port(port);
+                }
             },
             ConnectorVariant::Native(_) => unreachable!(),
+        }
         return key;
+    }
     pub(crate) fn destroy(&self, key: ConnectorKey) {
         let lock = self.inner.write().unwrap();
+        let mut lock = self.inner.write().unwrap();
         unsafe {
             let connector = lock.connectors.get_mut(key.index as usize);
             ptr::drop_in_place(*connector);
             // Note: but not deallocating!
+        }
         lock.free.push(key.index as usize);
+    }
     /// Creates a connector but does not set its initial ports
     fn create_connector_raw(&self, connector: ConnectorVariant) -> ConnectorKey {
+    fn create_connector_raw(&self, connector: ConnectorVariant, initialize_as_sleeping: bool) -> ConnectorKey {
         // Creation of the connector in the global store, requires a lock
         let index;
+        {
             let lock = self.inner.write().unwrap();
+            let mut lock = self.inner.write().unwrap();
             let connector = ScheduledConnector {
                 connector,
                 context: ConnectorCtx::new(self.port_counter.clone()),
                 public: ConnectorPublic::new(),
+                public: ConnectorPublic::new(initialize_as_sleeping),
                 router: Router::new(),
             };
             if lock.free.is_empty() {
                 let connector = Box::into_raw(Box::new(connector));
                 unsafe {
                     // Cheating a bit here. Anyway, move to heap, store in list
                     index = lock.connectors.len();
                     lock.connectors.push(connector);
+                }
                 index = lock.connectors.len();
                 lock.connectors.push(connector);
             } else {
                 index = lock.free.pop().unwrap();
                 unsafe {
                     let target = lock.connectors.get_mut(index);
                     debug_assert!(!target.is_null());
                     ptr::write(*target, connector);
+                }
+            }
+        }
         // Generate key and retrieve the connector to set its ID
         let key = ConnectorKey{ index: index as u32 };
         let new_connector = self.get_mut(&key);
         new_connector.context.id = key.downcast();
         // Return the connector key
         return key;
+    }
+}
 impl Drop for ConnectorStore {
     fn drop(&mut self) {
         let lock = self.inner.write().unwrap();
         for idx in 0..lock.connectors.len() {
             unsafe {
                 let memory = *lock.connectors.get_mut(idx);
                 let _ = Box::from_raw(memory); // takes care of deallocation
+            }
+        }
+    }
+}
 /// Global store of connectors, ports and queues that are used by the sceduler
 /// threads. The global store has the appearance of a thread-safe datatype, but
 /// one needs to be careful using it.
 ///
 /// TODO: @docs
 /// TODO: @Optimize, very lazy implementation of concurrent datastructures.
 ///     This includes the `should_exit` and `did_exit` pair!
 pub struct GlobalStore {
+pub(crate) struct GlobalStore {
     pub connector_queue: MpmcQueue<ConnectorKey>,
     pub connectors: ConnectorStore,
     pub should_exit: AtomicBool,    // signal threads to exit
+}
 impl GlobalStore {
     pub fn new() -> Self {
+    pub(crate) fn new() -> Self {
         Self{
             connector_queue: MpmcQueue::with_capacity(256),
             connectors: ConnectorStore::with_capacity(256),
             should_exit: AtomicBool::new(false),
+        }
+    }
+}
@@ \ No newline at end of file @@

src/runtime2/inbox.rs

➞

Show inline comments

@@ @@ -217,126 +217,126 @@ pub struct PublicInbox { @@
 impl PublicInbox {
     pub fn new() -> Self {
         Self{
             messages: Mutex::new(VecDeque::new()),
+        }
+    }
     pub fn insert_message(&self, message: Message) {
         let mut lock = self.messages.lock().unwrap();
         lock.push_back(message);
+    }
     pub fn take_message(&self) -> Option<Message> {
         let mut lock = self.messages.lock().unwrap();
         return lock.pop_front();
+    }
     pub fn is_empty(&self) -> bool {
         let lock = self.messages.lock().unwrap();
         return lock.is_empty();
+    }
+}
 pub struct PrivateInbox {
+pub(crate) struct PrivateInbox {
     // "Normal" messages, intended for a PDL protocol. These need to stick
     // around during an entire sync-block (to handle `put`s for which the
     // corresponding `get`s have not yet been reached).
     messages: Vec<DataMessage>,
+    messages: Vec<(PortIdLocal, DataMessage)>,
     len_read: usize,
+}
 impl PrivateInbox {
     pub fn new() -> Self {
         Self{
             messages: Vec::new(),
             len_read: 0,
+        }
+    }
     /// Will insert the message into the inbox. Only exception is when the tuple
     /// (prev_branch_id, cur_branch_id, receiving_port_id) already exists, then
     /// nothing is inserted..
     pub fn insert_message(&mut self, message: DataMessage) {
         for existing in self.messages.iter() {
     pub(crate) fn insert_message(&mut self, target_port: PortIdLocal, message: DataMessage) {
         for (existing_target_port, existing) in self.messages.iter() {
             if existing.sender_prev_branch_id == message.sender_prev_branch_id &&
                     existing.sender_cur_branch_id == message.sender_cur_branch_id &&
-                    existing.sending_port == message.sending_port {
+                    *existing_target_port == target_port {
                 // Message was already received
                 return;
+            }
+        }
         self.messages.push(message);
+        self.messages.push((target_port, message));
+    }
     /// Retrieves all previously read messages that satisfy the provided
     /// speculative conditions. Note that the inbox remains read-locked until
     /// the returned iterator is dropped. Should only be called by the
     /// inbox-reader (i.e. the thread executing a connector's PDL code).
     ///
     /// This function should only be used to check if already-received messages
     /// could be received by a newly encountered `get` call in a connector's
     /// PDL code.
     pub fn get_messages(&self, port_id: PortIdLocal, prev_branch_id: BranchId) -> InboxMessageIter {
+    pub(crate) fn get_messages(&self, port_id: PortIdLocal, prev_branch_id: BranchId) -> InboxMessageIter {
         return InboxMessageIter{
             messages: &self.messages,
             next_index: 0,
             max_index: self.len_read,
             match_port_id: port_id,
             match_prev_branch_id: prev_branch_id,
         };
+    }
     /// Retrieves the next unread message. Should only be called by the
     /// inbox-reader.
     pub fn next_message(&mut self) -> Option<&DataMessage> {
+    pub(crate) fn next_message(&mut self) -> Option<&DataMessage> {
         if self.len_read == self.messages.len() {
             return None;
+        }
         let to_return = &self.messages[self.len_read];
+        let (_, to_return) = &self.messages[self.len_read];
         self.len_read += 1;
         return Some(to_return);
+    }
     /// Simply empties the inbox
     pub fn clear(&mut self) {
+    pub(crate) fn clear(&mut self) {
         self.messages.clear();
         self.len_read = 0;
+    }
+}
 /// Iterator over previously received messages in the inbox.
 pub struct InboxMessageIter<'i> {
     messages: &'i Vec<DataMessage>,
 pub(crate) struct InboxMessageIter<'i> {
     messages: &'i Vec<(PortIdLocal, DataMessage)>,
     next_index: usize,
     max_index: usize,
     match_port_id: PortIdLocal,
     match_prev_branch_id: BranchId,
+}
 impl<'i> Iterator for InboxMessageIter<'i> {
     type Item = &'i DataMessage;
     fn next(&mut self) -> Option<Self::Item> {
         // Loop until match is found or at end of messages
         while self.next_index < self.max_index {
             let cur_message = &self.messages[self.next_index];
             if cur_message.receiving_port == self.match_port_id && cur_message.sender_prev_branch_id == self.match_prev_branch_id {
             let (target_port, cur_message) = &self.messages[self.next_index];
             if *target_port == self.match_port_id && cur_message.sender_prev_branch_id == self.match_prev_branch_id {
                 // Found a match
                 break;
+            }
             self.next_index += 1;
+        }
         if self.next_index == self.max_index {
             return None;
+        }
         let message = &self.messages[self.next_index];
+        let (_, message) = &self.messages[self.next_index];
         self.next_index += 1;
         return Some(message);
+    }
+}
@@ \ No newline at end of file @@

src/runtime2/messages.rs

➞

Show inline comments

 use std::collections::hash_map::Entry;
 use std::collections::HashMap;
 use crate::PortId;
 use crate::protocol::*;
 use crate::protocol::eval::*;
 /// A message residing in a connector's inbox (waiting to be put into some kind
 /// of speculative branch), or a message waiting to be sent.
 #[derive(Clone)]
 pub struct BufferedMessage {
     pub(crate) sending_port: PortId,
     pub(crate) receiving_port: PortId,
     pub(crate) peer_prev_branch_id: Option<u32>,
     pub(crate) peer_cur_branch_id: u32,
     pub(crate) message: ValueGroup,
+}
 /// A connector's global inbox. Any received message ends up here. This is
 /// because a message might be received before a branch arrives at the
 /// corresponding `get()` that is supposed to receive that message. Hence we
 /// need to store it for all future branches that might be able to receive it.
 pub struct ConnectorInbox {
     // TODO: @optimize, HashMap + Vec is a bit stupid.
     messages: HashMap<PortAction, Vec<BufferedMessage>>
+}
 /// An action performed on a port. Unsure about this

src/runtime2/mod.rs

➞

Show inline comments

 // Structure of module
 mod runtime;
 mod messages;
 mod connector;
 mod native;
 mod port;
 mod global_store;
 mod scheduler;
 mod inbox;
 #[cfg(test)] mod tests;
 // Imports
 use std::sync::{Arc, Mutex};
 use std::sync::atomic::Ordering;
+use std::sync::atomic::{AtomicU32, Ordering};
 use std::thread::{self, JoinHandle};
 use crate::ProtocolDescription;
 use global_store::{ConnectorVariant, GlobalStore};
 use scheduler::Scheduler;
 use native::{ConnectorApplication, ApplicationInterface};
 // Runtime API
 // TODO: Exit condition is very dirty. Take into account:
 //  - Connector hack with &'static references. May only destroy (unforced) if all connectors are done working
 //  - Running schedulers: schedulers need to be signaled that they should exit, then wait until all are done
 //  - User-owned interfaces: As long as these are owned user may still decide to create new connectors.
 /// A kind of token that, once obtained, allows mutable access to a connector.
 /// We're trying to use move semantics as much as possible: the owner of this
 /// key is the only one that may execute the connector's code.
 pub(crate) struct ConnectorKey {
     pub index: u32, // of connector
+}
 impl ConnectorKey {
     /// Downcasts the `ConnectorKey` type, which can be used to obtain mutable
     /// access, to a "regular ID" which can be used to obtain immutable access.
     #[inline]
     pub fn downcast(&self) -> ConnectorId {
         return ConnectorId(self.index);
+    }
     /// Turns the `ConnectorId` into a `ConnectorKey`, marked as unsafe as it
     /// bypasses the type-enforced `ConnectorKey`/`ConnectorId` system
     #[inline]
     pub unsafe fn from_id(id: ConnectorId) -> ConnectorKey {
         return ConnectorKey{ index: id.0 };
+    }
+}
 /// A kind of token that allows shared access to a connector. Multiple threads
 /// may hold this
 #[derive(Debug, Copy, Clone, PartialEq, Eq)]
 pub struct ConnectorId(pub u32);
 impl ConnectorId {
     // TODO: Like the other `new_invalid`, maybe remove
     #[inline]
     pub fn new_invalid() -> ConnectorId {
         return ConnectorId(u32::MAX);
+    }
     #[inline]
     pub(crate) fn is_valid(&self) -> bool {
         return self.0 != u32::MAX;
+    }
+}
 // TODO: Change this, I hate this. But I also don't want to put `public` and
 //  `router` of `ScheduledConnector` back into `Connector`. The reason I don't
 //  want `Box<dyn Connector>` everywhere is because of the v-table overhead. But
 //  to truly design this properly I need some benchmarks.
 pub(crate) enum ConnectorVariant {
     UserDefined(ConnectorPDL),
     Native(Box<dyn Connector>),
+}
 impl Connector for ConnectorVariant {
     fn handle_message(&mut self, message: Message, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) {
         match self {
             ConnectorVariant::UserDefined(c) => c.handle_message(message, ctx, delta_state),
             ConnectorVariant::Native(c) => c.handle_message(message, ctx, delta_state),
+        }
+    }
     fn run(&mut self, protocol_description: &ProtocolDescription, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) -> ConnectorScheduling {
         match self {
             ConnectorVariant::UserDefined(c) => c.run(protocol_description, ctx, delta_state),
             ConnectorVariant::Native(c) => c.run(protocol_description, ctx, delta_state),
+        }
+    }
+}
 pub(crate) struct ScheduledConnector {
     pub connector: ConnectorVariant, // access by connector
     pub context: ConnectorCtx, // mutable access by scheduler, immutable by connector
     pub public: ConnectorPublic, // accessible by all schedulers and connectors
     pub router: Router,
+}
 /// Externally facing runtime.
 pub struct Runtime {
     inner: Arc<RuntimeInner>,
+}
 pub(crate) struct RuntimeInner {
-    pub(crate) global_store: GlobalStore,
+    // Protocol
     pub(crate) protocol_description: ProtocolDescription,
     schedulers: Mutex<Vec<JoinHandle<()>>>, // TODO: Revise, make exit condition something like: all interfaces dropped
     // Storage of connectors in a kind of freelist. Note the vector of points to
     // ensure pointer stability: the vector can be changed but the entries
     // themselves remain valid.
     pub connectors_list: RawVec<*mut ScheduledConnector>,
     pub connectors_free: Vec<usize>,
     pub(crate) global_store: GlobalStore,
     schedulers: Mutex<Vec<JoinHandle<()>>>,
     active_interfaces: AtomicU32, // active API interfaces that can add connectors/channels
+}
 impl RuntimeInner {
     #[inline]
     pub(crate) fn increment_active_interfaces(&self) {
         let _old_num = self.active_interfaces.fetch_add(1, Ordering::SeqCst);
         debug_assert_ne!(_old_num, 1); // once it hits 0, it stays zero
+    }
     pub(crate) fn decrement_active_interfaces(&self) {
         let old_num = self.active_interfaces.fetch_sub(1, Ordering::SeqCst);
         debug_assert!(old_num > 0);
         if old_num == 1 {
             // Became 0
             // TODO: Check num connectors, if 0, then set exit flag
+        }
+    }
+}
 // TODO: Come back to this at some point
 unsafe impl Send for RuntimeInner {}
 unsafe impl Sync for RuntimeInner {}
 impl Runtime {
-    pub fn new(num_threads: usize, protocol_description: ProtocolDescription) -> Runtime {
+    pub fn new(num_threads: u32, protocol_description: ProtocolDescription) -> Runtime {
         // Setup global state
         assert!(num_threads > 0, "need a thread to run connectors");
         let runtime_inner = Arc::new(RuntimeInner{
             global_store: GlobalStore::new(),
             protocol_description,
             schedulers: Mutex::new(Vec::new()),
             active_interfaces: AtomicU32::new(1), // we are the active interface
         });
         // Launch threads
+        {
             let mut schedulers = Vec::with_capacity(num_threads);
             for _ in 0..num_threads {
             let mut schedulers = Vec::with_capacity(num_threads as usize);
             for thread_index in 0..num_threads {
                 let cloned_runtime_inner = runtime_inner.clone();
                 let thread = thread::spawn(move || {
                     let mut scheduler = Scheduler::new(cloned_runtime_inner);
                     scheduler.run();
                 });
                 let thread = thread::Builder::new()
                     .name(format!("thread-{}", thread_index))
                     .spawn(move || {
                         let mut scheduler = Scheduler::new(cloned_runtime_inner, thread_index);
                         scheduler.run();
                     })
                     .unwrap();
                 schedulers.push(thread);
+            }
             let mut lock = runtime_inner.schedulers.lock().unwrap();
             *lock = schedulers;
+        }
         // Return runtime
         return Runtime{ inner: runtime_inner };
+    }
     /// Returns a new interface through which channels and connectors can be
     /// created.
     pub fn create_interface(&self) -> ApplicationInterface {
         let (connector, mut interface) = ConnectorApplication::new(self.inner.clone());
         let connector_key = self.inner.global_store.connectors.create_interface(connector);
         interface.set_connector_id(connector_key.downcast());
         // Note that we're not scheduling. That is done by the interface in case
         // it is actually needed.
         return interface;
+    }
+}
 impl Drop for Runtime {
     fn drop(&mut self) {
         self.inner.global_store.should_exit.store(true, Ordering::Release);
         let mut schedulers = self.inner.schedulers.lock().unwrap();
         for scheduler in schedulers.drain(..) {
             scheduler.join();
+            scheduler.join().unwrap();
+        }
+    }
+}
@@ \ No newline at end of file @@

src/runtime2/native.rs

➞

Show inline comments

 use std::collections::VecDeque;
 use std::sync::{Arc, Mutex, Condvar};
 use std::sync::atomic::Ordering;
 use crate::protocol::ComponentCreationError;
 use crate::protocol::ComponentCreationError;
 use crate::protocol::eval::ValueGroup;
 use crate::ProtocolDescription;
 use crate::runtime2::connector::{Branch, find_ports_in_value_group};
 use crate::runtime2::global_store::ConnectorKey;
 use crate::runtime2::inbox::MessageContents;
 use crate::runtime2::port::{Port, PortKind};
 use crate::runtime2::scheduler::ConnectorCtx;
 use super::RuntimeInner;
 use super::global_store::ConnectorId;
 use super::port::{Channel, PortIdLocal};
 use super::connector::{ConnectorPDL, ConnectorScheduling, RunDeltaState};
 use super::inbox::Message;
 /// Generic connector interface from the scheduler's point of view.
 pub trait Connector {
+pub(crate) trait Connector {
     /// Handle a new message (preprocessed by the scheduler). You probably only
     /// want to handle `Data`, `Sync`, and `Solution` messages. The others are
     /// intended for the scheduler itself.
-    fn handle_message(&mut self, message: MessageContents, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState);
+    fn handle_message(&mut self, message: Message, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState);
     /// Should run the connector's behaviour up until the next blocking point.
     fn run(&mut self, protocol_description: &ProtocolDescription, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) -> ConnectorScheduling;
+}
 type SyncDone = Arc<(Mutex<bool>, Condvar)>;
-type JobQueue = Arc<Mutex<Vec<ApplicationJob>>>;
+type JobQueue = Arc<Mutex<VecDeque<ApplicationJob>>>;
 enum ApplicationJob {
     NewChannel((Port, Port)),
     NewConnector(ConnectorPDL),
+}
 /// The connector which an application can directly interface with. Once may set
 /// up the next synchronous round, and retrieve the data afterwards.
 pub struct ConnectorApplication {
     sync_done: SyncDone,
     job_queue: JobQueue,
+}
 impl ConnectorApplication {
     pub(crate) fn new(runtime: Arc<RuntimeInner>) -> (Self, ApplicationInterface) {
         let sync_done = Arc::new(( Mutex::new(false), Condvar::new() ));
-        let job_queue = Arc::new(Mutex::new(Vec::with_capacity(32)));
+        let job_queue = Arc::new(Mutex::new(VecDeque::with_capacity(32)));
         let connector = ConnectorApplication { sync_done: sync_done.clone(), job_queue: job_queue.clone() };
         let interface = ApplicationInterface::new(sync_done, job_queue, runtime);
         return (connector, interface);
+    }
+}
 impl Connector for ConnectorApplication {
     fn handle_message(&mut self, message: MessageContents, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) {
         todo!("handling messages in ConnectorApplication (API for runtime)")
     fn handle_message(&mut self, message: Message, _ctx: &ConnectorCtx, _delta_state: &mut RunDeltaState) {
         use MessageContents as MC;
         match message.contents {
             MC::Data(_) => unreachable!("data message in API connector"),
             MC::Sync(_) | MC::RequestCommit(_) | MC::ConfirmCommit(_) => {
                 // Handling sync in API
             },
             MC::Control(_) => {},
             MC::Ping => {},
+        }
+    }
     fn run(&mut self, protocol_description: &ProtocolDescription, ctx: &ConnectorCtx, delta_state: &mut RunDeltaState) -> ConnectorScheduling {
         let mut queue = self.job_queue.lock().unwrap();
-        while let Some(job) = queue.pop() {
+        while let Some(job) = queue.pop_front() {
             match job {
                 ApplicationJob::NewChannel((endpoint_a, endpoint_b)) => {
                     println!("DEBUG: API adopting ports");
                     delta_state.new_ports.reserve(2);
                     delta_state.new_ports.push(endpoint_a);
                     delta_state.new_ports.push(endpoint_b);
+                }
                 ApplicationJob::NewConnector(connector) => {
                     println!("DEBUG: API creating connector");
                     delta_state.new_connectors.push(connector);
+                }
+            }
+        }
         return ConnectorScheduling::NotNow;
+    }
+}
 /// The interface to a `ApplicationConnector`. This allows setting up the
 /// interactions the `ApplicationConnector` performs within a synchronous round.
 pub struct ApplicationInterface {
     sync_done: SyncDone,
     job_queue: JobQueue,
     runtime: Arc<RuntimeInner>,
     connector_id: ConnectorId,
     owned_ports: Vec<PortIdLocal>,
+}
 impl ApplicationInterface {
     pub(crate) fn new(sync_done: SyncDone, job_queue: JobQueue, runtime: Arc<RuntimeInner>) -> Self {
     fn new(sync_done: SyncDone, job_queue: JobQueue, runtime: Arc<RuntimeInner>) -> Self {
         runtime.active_interfaces += 1;
         return Self{
             sync_done, job_queue, runtime,
             connector_id: ConnectorId::new_invalid(),
             owned_ports: Vec::new(),
+        }
+    }
     /// Creates a new channel.
     pub fn create_channel(&mut self) -> Channel {
         // TODO: Duplicated logic in scheduler
         let getter_id = self.runtime.global_store.connectors.port_counter.fetch_add(2, Ordering::SeqCst);
         let putter_id = PortIdLocal::new(getter_id + 1);
         let getter_id = PortIdLocal::new(getter_id);
         // Create ports and add a job such that they are transferred to the
         // API component. (note that we do not send a ping, this is only
         // necessary once we create a connector)
         let getter_port = Port{
             self_id: getter_id,
             peer_id: putter_id,
             kind: PortKind::Getter,
             peer_connector: self.connector_id,
         };
         let putter_port = Port{
             self_id: putter_id,
             peer_id: getter_id,
             kind: PortKind::Putter,
             peer_connector: self.connector_id,
         };
+        {
             let mut lock = self.job_queue.lock().unwrap();
-            lock.push(ApplicationJob::NewChannel((getter_port, putter_port)));
+            lock.push_back(ApplicationJob::NewChannel((getter_port, putter_port)));
+        }
         // Add to owned ports for error checking while creating a connector
         self.owned_ports.reserve(2);
         self.owned_ports.push(putter_id);
         self.owned_ports.push(getter_id);
         return Channel{ putter_id, getter_id };
+    }
     /// Creates a new connector. Note that it is not scheduled immediately, but
     /// depends on the `ApplicationConnector` to run, followed by the created
     /// connector being scheduled.
     // TODO: Optimize by yanking out scheduler logic for common use.
     pub fn create_connector(&mut self, module: &str, routine: &str, arguments: ValueGroup) -> Result<(), ComponentCreationError> {
         // Retrieve ports and make sure that we own the ones that are currently
         // specified. This is also checked by the scheduler, but that is done
         // asynchronously.
         let mut initial_ports = Vec::new();
         find_ports_in_value_group(&arguments, &mut initial_ports);
         for port_to_remove in &initial_ports {
             match self.owned_ports.iter().position(|v| v == port_to_remove) {
                 Some(index_to_remove) => {
                     // We own the port, so continue
                     self.owned_ports.remove(index_to_remove);
                 },
                 None => {
                     // We don't own the port
                     return Err(ComponentCreationError::UnownedPort);
+                }
+            }
+        }
         let state = self.runtime.protocol_description.new_component_v2(module.as_bytes(), routine.as_bytes(), arguments)?;
         let connector = ConnectorPDL::new(Branch::new_initial_branch(state), initial_ports);
         // Put on job queue
+        {
             let mut queue = self.job_queue.lock().unwrap();
-            queue.push(ApplicationJob::NewConnector(connector));
+            queue.push_back(ApplicationJob::NewConnector(connector));
+        }
         // Send ping message to wake up connector
         let connector = self.runtime.global_store.connectors.get_shared(self.connector_id);
         connector.inbox.insert_message(Message{
             sending_connector: ConnectorId::new_invalid(),
             receiving_port: PortIdLocal::new_invalid(),
             contents: MessageContents::Ping,
         });
         let should_wake_up = connector.sleeping
             .compare_exchange(true, false, Ordering::SeqCst, Ordering::Acquire)
             .is_ok();
         if should_wake_up {
             println!("DEBUG: Waking up connector");
             let key = unsafe{ ConnectorKey::from_id(self.connector_id) };
             self.runtime.global_store.connector_queue.push_back(key);
         } else {
             println!("DEBUG: NOT waking up connector");
+        }
         return Ok(());
+    }
     /// Check if the next sync-round is finished.
     pub fn try_wait(&self) -> bool {
         let (is_done, _) = &*self.sync_done;
         let lock = is_done.lock().unwrap();
         return *lock;
+    }
     /// Wait until the next sync-round is finished
     pub fn wait(&self) {
         let (is_done, condition) = &*self.sync_done;
         let lock = is_done.lock().unwrap();
         condition.wait_while(lock, |v| !*v); // wait while not done
+        condition.wait_while(lock, |v| !*v).unwrap(); // wait while not done
+    }
     /// Called by runtime to set associated connector's ID.
     pub(crate) fn set_connector_id(&mut self, id: ConnectorId) {
         self.connector_id = id;
+    }
+}
 impl Drop for ApplicationInterface {
     fn drop(&mut self) {
+    }
+}
@@ \ No newline at end of file @@

src/runtime2/port.rs

➞

Show inline comments

 use super::global_store::ConnectorId;
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
-pub(crate) struct PortIdLocal {
 pub struct PortIdLocal {
     pub index: u32,
+}
 impl PortIdLocal {
     pub fn new(id: u32) -> Self {
         Self{ index: id }
+    }
     // TODO: Unsure about this, maybe remove, then also remove all struct
     //  instances where I call this
     pub fn new_invalid() -> Self {
         Self{ index: u32::MAX }
+    }
     pub fn is_valid(&self) -> bool {
         return self.index != u32::MAX;
+    }
+}
 #[derive(Eq, PartialEq)]
 pub enum PortKind {
     Putter,
     Getter,
+}

src/runtime2/runtime.rs

➞

Show inline comments

@@ @@ -1041,49 +1041,49 @@ impl Runtime { @@
             result.push(next_branch_index);
             let branch = &desc.branches[next_branch_index as usize];
             match branch.parent_index {
                 Some(index) => next_branch_index = index,
                 None => return,
+            }
+        }
+    }
+}
 /// Context accessible by the code while being executed by the runtime. When the
 /// code is being executed by the runtime it sometimes needs to interact with
 /// the runtime. This is achieved by the "code throwing an error code", after
 /// which the runtime modifies the appropriate variables and continues executing
 /// the code again.
 struct Context<'a> {
     // Temporary references to branch related storage
     inbox: &'a HashMap<(PortId, u32), ValueGroup>,
     port_mapping: &'a HashMap<PortId, BranchPortDesc>,
     branch_ctx: &'a mut BranchContext,
+}
 impl<'a> crate::protocol::RunContext for Context<'a> {
-    fn did_put(&mut self, port: PortId) -> bool {
+    fn did_put(&mut self, _port: PortId) -> bool {
         // Note that we want "did put" to return false if we have fired zero
         // times, because this implies we did a prevous
         let old_value = self.branch_ctx.just_called_did_put;
         self.branch_ctx.just_called_did_put = false;
         return old_value;
+    }
     fn get(&mut self, port: PortId) -> Option<ValueGroup> {
         let inbox_key = (port, 1);
         match self.inbox.get(&inbox_key) {
             None => None,
             Some(value) => Some(value.clone()),
+        }
+    }
     fn fires(&mut self, port: PortId) -> Option<Value> {
         match self.port_mapping.get(&port) {
             None => None,
             Some(port_info) => Some(Value::Bool(port_info.num_times_fired != 0)),
+        }
+    }
     fn get_channel(&mut self) -> Option<(Value, Value)> {
         self.branch_ctx.pending_channel.take()

src/runtime2/scheduler.rs

➞

Show inline comments

@@ @@ -67,129 +67,135 @@ impl ConnectorCtx { @@
     pub(crate) fn get_port(&self, id: PortIdLocal) -> &Port {
         let index = self.port_id_to_index(id);
         return &self.ports[index];
+    }
     pub(crate) fn get_port_mut(&mut self, id: PortIdLocal) -> &mut Port {
         let index = self.port_id_to_index(id);
         return &mut self.ports[index];
+    }
     fn port_id_to_index(&self, id: PortIdLocal) -> usize {
         for (idx, port) in self.ports.iter().enumerate() {
             if port.self_id == id {
                 return idx;
+            }
+        }
         panic!("port {:?}, not owned by connector", id);
+    }
+}
 pub(crate) struct Scheduler {
     runtime: Arc<RuntimeInner>,
     scheduler_id: u32,
+}
 // Thinking aloud: actual ports should be accessible by connector, but managed
 // by the scheduler (to handle rerouting messages). We could just give a read-
 // only context, instead of an extra call on the "Connector" trait.
 impl Scheduler {
     pub fn new(runtime: Arc<RuntimeInner>) -> Self {
         return Self{ runtime };
     pub fn new(runtime: Arc<RuntimeInner>, scheduler_id: u32) -> Self {
         return Self{ runtime, scheduler_id };
+    }
     pub fn run(&mut self) {
         // Setup global storage and workspaces that are reused for every
         // connector that we run
         let scheduler_id = self.scheduler_id;
         let mut delta_state = RunDeltaState::new();
         'thread_loop: loop {
             // Retrieve a unit of work
             let connector_key = self.runtime.global_store.connector_queue.pop_front();
             if connector_key.is_none() {
             let mut connector_key = self.runtime.global_store.connector_queue.pop_front();
             while connector_key.is_none() {
                 // TODO: @Performance, needs condition or something, and most
                 //  def' not sleeping
                 println!("DEBUG [{}]: Nothing to do", scheduler_id);
                 thread::sleep(Duration::new(1, 0));
                 if self.runtime.global_store.should_exit.load(Ordering::Acquire) {
                     // Thread exits!
                     println!("DEBUG [{}]: ... So I am quitting", scheduler_id);
                     break 'thread_loop;
+                }
                 println!("DEBUG [{}]: ... But I'm still running", scheduler_id);
                 continue 'thread_loop;
+            }
             // We have something to do
             let connector_key = connector_key.unwrap();
             println!("DEBUG [{}]: Running connector {}", scheduler_id, connector_key.index);
             let scheduled = self.runtime.global_store.connectors.get_mut(&connector_key);
             // Keep running until we should no longer immediately schedule the
             // connector.
             let mut cur_schedule = ConnectorScheduling::Immediate;
             while cur_schedule == ConnectorScheduling::Immediate {
                 // Check all the message that are in the shared inbox
                 while let Some(message) = scheduled.public.inbox.take_message() {
                     // Check for rerouting
                     if let Some(other_connector_id) = scheduled.router.should_reroute(message.sending_connector, message.receiving_port) {
                         self.send_message_and_wake_up_if_sleeping(other_connector_id, message);
                         continue;
+                    }
                     // Check for messages that requires special action from the
                     // scheduler.
                     if let MessageContents::Control(content) = message.contents {
                         match content.content {
                             ControlMessageVariant::ChangePortPeer(port_id, new_target_connector_id) => {
                                 // Need to change port target
                                 let port = scheduled.context.get_port_mut(port_id);
                                 port.peer_connector = new_target_connector_id;
                                 debug_assert!(delta_state.outbox.is_empty());
                                 // And respond with an Ack
                                 // Note: after this code has been reached, we may not have any
                                 // messages in the outbox that send to the port whose owning
                                 // connector we just changed. This is because the `ack` will
                                 // clear the rerouting entry of the `ack`-receiver.
                                 self.send_message_and_wake_up_if_sleeping(
                                     message.sending_connector,
                                     Message{
                                         sending_connector: connector_key.downcast(),
                                         receiving_port: PortIdLocal::new_invalid(),
                                         contents: MessageContents::Control(ControlMessage{
                                             id: content.id,
                                             content: ControlMessageVariant::Ack,
                                         }),
+                                    }
                                 );
                             },
                             ControlMessageVariant::Ack => {
                                 scheduled.router.handle_ack(content.id);
+                            }
+                        }
                     } else {
                         // Let connector handle message
-                        scheduled.connector.handle_message(message.contents, &scheduled.context, &mut delta_state);
                         scheduled.connector.handle_message(message, &scheduled.context, &mut delta_state);
+                    }
+                }
                 // Actually run the connector
                 let new_schedule = scheduled.connector.run(
                     &self.runtime.protocol_description, &scheduled.context, &mut delta_state
                 );
                 // Handle all of the output from the current run: messages to
                 // send and connectors to instantiate.
                 self.handle_delta_state(&connector_key, &mut scheduled.context, &mut delta_state);
                 cur_schedule = new_schedule;
+            }
             // If here then the connector does not require immediate execution.
             // So enqueue it if requested, and otherwise put it in a sleeping
             // state.
             match cur_schedule {
                 ConnectorScheduling::Immediate => unreachable!(),
                 ConnectorScheduling::Later => {
                     // Simply queue it again later
                     self.runtime.global_store.connector_queue.push_back(connector_key);
                 },
@@ @@ -328,49 +334,50 @@ struct ReroutedTraffic { @@
 pub(crate) struct Router {
     id_counter: u32,
     active: Vec<ReroutedTraffic>,
+}
 impl Router {
     pub fn new() -> Self {
         Router{
             id_counter: 0,
             active: Vec::new(),
+        }
+    }
     /// Prepares rerouting messages due to changed ownership of a port. The
     /// control message returned by this function must be sent to the
     /// transferred port's peer connector.
     pub fn prepare_reroute(
         &mut self,
         port_id: PortIdLocal, peer_port_id: PortIdLocal,
         self_connector_id: ConnectorId, peer_connector_id: ConnectorId,
         new_owner_connector_id: ConnectorId
     ) -> Message {
         let id = self.id_counter;
         self.id_counter.overflowing_add(1);
         let (new_id_counter, _) = self.id_counter.overflowing_add(1);
         self.id_counter = new_id_counter;
         self.active.push(ReroutedTraffic{
             id,
             target_port: port_id,
             source_connector: peer_connector_id,
             target_connector: new_owner_connector_id,
         });
         return Message{
             sending_connector: self_connector_id,
             receiving_port: PortIdLocal::new_invalid(),
             contents: MessageContents::Control(ControlMessage{
                 id,
                 content: ControlMessageVariant::ChangePortPeer(peer_port_id, new_owner_connector_id),
             })
         };
+    }
     /// Returns true if the supplied message should be rerouted. If so then this
     /// function returns the connector that should retrieve this message.
     pub fn should_reroute(&self, sending_connector: ConnectorId, target_port: PortIdLocal) -> Option<ConnectorId> {
         for reroute in &self.active {
             if reroute.source_connector == sending_connector &&
                 reroute.target_port == target_port {

src/runtime2/tests/mod.rs

➞

Show inline comments

 use std::sync::Arc;
 use super::runtime::*;
 use crate::ProtocolDescription;
 use super::*;
 use crate::{PortId, ProtocolDescription};
 use crate::common::Id;
 use crate::protocol::eval::*;
 #[test]
 fn test_single_message() {
     // Simple test were we have a `putter` component, which will simply send a
     // single message (a boolean), and a `getter` component, which will receive
     // that message.
     // We will write this behaviour in the various ways that the language
     // currently allows. We will cheat a bit by peeking into the runtime to make
     // sure that the getter actually received the message.
     // TODO: Expose ports to a "native application"
     fn check_store_bool(value: &Value, expected: bool) {
         if let Value::Bool(value) = value {
             assert_eq!(*value, expected);
         } else {
             assert!(false);
+        }
+    }
     fn run_putter_getter(code: &[u8]) {
         // Compile code
         let pd = ProtocolDescription::parse(code)
             .expect("code successfully compiles");
         let pd = Arc::new(pd);
         // Construct runtime and the appropriate ports and connectors
         let mut rt = Runtime::new(pd);
         let (put_port, get_port) = rt.add_channel();
         let mut put_args = ValueGroup::new_stack(vec![
             put_port,
         ]);
         rt.add_component("", "putter", put_args)
             .expect("'putter' component created");
         let mut get_args = ValueGroup::new_stack(vec![
             get_port,
         ]);
         rt.add_component("", "getter", get_args)
             .expect("'getter' component created");
         // Run until completion
         rt.run();
 fn runtime_for(num_threads: u32, pdl: &str) -> Runtime {
     let protocol = ProtocolDescription::parse(pdl.as_bytes()).expect("parse pdl");
     let runtime = Runtime::new(num_threads, protocol);
         // Check for success (the 'received' and 'did_receive" flags)
         let getter_component = rt.connectors.get(&1).unwrap();
         let branch = &getter_component.branches[0];
         assert_eq!(branch.branch_state, BranchState::Finished);
     return runtime;
+}
         // Note: with the stack structure of the store, the first entry is the
         // "previous stack pos" and the second one is the input port passed to
         // the procedure. Hence the third/fourth entries are the boolean
         // variables on the stack.
         check_store_bool(&branch.code_state.prompt.store.stack[2], true);
         check_store_bool(&branch.code_state.prompt.store.stack[3], true);
 #[test]
 fn test_put_and_get() {
     let rt = runtime_for(4, "
 primitive putter(out<bool> sender, u32 loops) {
     u32 index = 0;
     while (index < loops) {
         synchronous {
             print(\"putting!\");
             put(sender, true);
+        }
         index += 1;
+    }
     // Without `fires()`, just a single valid behaviour
     run_putter_getter(
         b"primitive putter(out<bool> put_here) {
             synchronous {
                 put(put_here, true);
+            }
+}
 primitive getter(in<bool> receiver, u32 loops) {
     u32 index = 0;
     while (index < loops) {
         synchronous {
             print(\"getting!\");
             auto result = get(receiver);
             assert(result);
+        }
         index += 1;
+    }
+}
     ");
         primitive getter(in<bool> get_here) {
             bool received = false;
             bool did_receive = false;
             synchronous {
                 received = get(get_here);
                 if (received) {
                     print(\"value was 'true'\");
                 } else {
                     print(\"value was 'false'\");
+                }
                 did_receive = true;
+            }
         }");
     // With `fires()`, but eliminating on the putter side
     run_putter_getter(
         b"primitive putter(out<bool> put_here) {
             synchronous {
                 if (!fires(put_here)) {
                     assert(false);
                 } else {
                     put(put_here, true);
+                }
+            }
+        }
     let mut api = rt.create_interface();
     let channel = api.create_channel();
     let num_loops = 5;
         primitive getter(in<bool> get_here) {
             bool received = false; bool did_receive = false;
             synchronous {
                 if (fires(get_here)) {
                     received = get(get_here);
                     did_receive = true;
+                }
+            }
         }");
     api.create_connector("", "putter", ValueGroup::new_stack(vec![
         Value::Output(PortId(Id{ connector_id: 0, u32_suffix: channel.putter_id.index })),
         Value::UInt32(num_loops)
     ])).expect("create putter");
     // With `fires()`, but eliminating on the getter side
     run_putter_getter(
         b"primitive putter(out<bool> put_here) {
             synchronous {
                 if (fires(put_here)) {
                     put(put_here, true);
+                }
+            }
+        }
     api.create_connector("", "getter", ValueGroup::new_stack(vec![
         Value::Input(PortId(Id{ connector_id: 0, u32_suffix: channel.getter_id.index })),
         Value::UInt32(num_loops)
     ])).expect("create getter");
         primitive getter(in<bool> get_here) {
             bool received = false; bool did_receive = false;
             synchronous {
                 if (fires(get_here)) {
                     received = get(get_here);
                     did_receive = true;
                 } else {
                     assert(false);
+                }
+            }
         }"
     );
     println!("Am I running?");
+}
@@ \ No newline at end of file @@

0 comments (0 inline, 0 general)