CSY/reowolf Changeset - 4da5e57a9834 · Centrum Wiskunde & Informatica (CWI)

Changeset - 4da5e57a9834

Parent rev.

Child rev.

[Not reviewed]

0 7 0

MH - 4 years ago 2021-12-02 17:01:59
contact@maxhenger.nl

Add error checking to sending component messages

7 files changed with 195 insertions and 200 deletions:

src/runtime2/connector.rs

src/runtime2/consensus.rs

src/runtime2/inbox.rs

src/runtime2/mod.rs

145

src/runtime2/native.rs

src/runtime2/scheduler.rs

148

src/runtime2/tests/mod.rs

0 comments (0 inline, 0 general)

src/runtime2/connector.rs

➞

Show inline comments

@@ @@ -23,23 +23,21 @@ @@
 //     a sync round (for now, that means the port mapping in each branch).
 //     Hence once more we have properties conceptually associated with branches
 //     in two places.
 // - TODO: Write about handling messages, consensus wrapping data
 // - TODO: Write about way information is exchanged between PDL/component and scheduler through ctx
 use std::collections::HashMap;
 use std::sync::atomic::AtomicBool;
 use crate::{PortId, ProtocolDescription};
 use crate::common::ComponentState;
 use crate::protocol::eval::{EvalContinuation, EvalError, Prompt, Value, ValueGroup};
-use crate::protocol::{RunContext, RunResult};
 use crate::protocol::RunContext;
 use super::branch::{BranchId, ExecTree, QueueKind, SpeculativeState, PreparedStatement};
 use super::consensus::{Consensus, Consistency, RoundConclusion, find_ports_in_value_group};
-use super::inbox::{DataMessage, Message, SyncCompMessage, SyncCompContent, SyncPortMessage, SyncControlMessage, PublicInbox};
 use super::inbox::{DataMessage, Message, SyncCompMessage, SyncPortMessage, SyncControlMessage, PublicInbox};
 use super::native::Connector;
 use super::port::{PortKind, PortIdLocal};
 use super::scheduler::{ComponentCtx, SchedulerCtx, MessageTicket};
 pub(crate) struct ConnectorPublic {
     pub inbox: PublicInbox,

src/runtime2/consensus.rs

➞

Show inline comments

@@ @@ -1196,13 +1196,13 @@ impl SolutionCombiner { @@
+        }
         return self.check_for_global_solution(component_index, solution_index);
+    }
     fn add_presence_and_check_for_global_failure(&mut self, component_id: ConnectorId, channels: &[LocalChannelPresence]) -> bool {
-        'new_report_loop: for entry in channels {
         for entry in channels {
             let mut found = false;
             for existing in &mut self.presence {
                 if existing.id == entry.channel_id {
                     // Same entry. We only update if we have the second
                     // component coming in it owns one end of the channel, or if

src/runtime2/inbox.rs

➞

Show inline comments

@@ @@ -3,13 +3,12 @@ use std::collections::VecDeque; @@
 use crate::protocol::eval::ValueGroup;
 use crate::runtime2::consensus::{ComponentPresence, SolutionCombiner};
 use crate::runtime2::port::ChannelId;
 use super::ConnectorId;
 use super::branch::BranchId;
 use super::consensus::{GlobalSolution, LocalSolution};
 use super::port::PortIdLocal;
 // TODO: Remove Debug derive from all types
 #[derive(Debug, Copy, Clone)]
@@ @@ -229,7 +228,12 @@ impl PublicInbox { @@
+    }
     pub fn is_empty(&self) -> bool {
         let lock = self.messages.lock().unwrap();
         return lock.is_empty();
+    }
     pub fn clear(&self) {
         let mut lock = self.messages.lock().unwrap();
         lock.clear();
+    }
+}
@@ \ No newline at end of file @@

src/runtime2/mod.rs

➞

Show inline comments

 use inbox::Message;
 use port::{ChannelId, Port, PortState};
 /// 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.
 #[derive(Debug)]
 pub(crate) struct ConnectorKey {
     pub index: u32, // of connector
     pub generation: u32,
+}
 impl ConnectorKey {
@@ @@ -270,16 +271,104 @@ impl RuntimeInner { @@
             peer_connector: creating_connector,
         };
         return (getter_port, putter_port);
+    }
     /// Sends a message to a particular connector. If the connector happened to
     /// be sleeping then it will be scheduled for execution.
     pub(crate) fn send_message(&self, target_id: ConnectorId, message: Message) {
         let target = self.get_component_public(target_id);
     /// Sends a message directly (without going through the port) to a
     /// component. This is slightly less efficient then sending over a port, but
     /// might be preferable for some algorithms. If the component was sleeping
     /// then it is scheduled for execution.
     pub(crate) fn send_message_maybe_destroyed(&self, target_id: ConnectorId, message: Message) -> bool {
         let target = {
             let mut lock = self.connectors.read().unwrap();
             lock.get(target_id.index)
         };
         // Do a CAS on the number of users. Most common case the component is
         // alive and we're the only one sending the message. Note that if we
         // finish this block, we're sure that no-one has set the `num_users`
         // value to 0. This is essential! When at 0, the component is added to
         // the freelist and the generation counter will be incremented.
         let mut cur_num_users = 1;
         while let Err(old_num_users) = target.num_users.compare_exchange(cur_num_users, cur_num_users + 1, Ordering::SeqCst, Ordering::Acquire) {
             if old_num_users == 0 {
                 // Cannot send message. Whatever the component state is
                 // (destroyed, at a different generation number, busy being
                 // destroyed, etc.) we cannot send the message and will not
                 // modify the component
                 return false;
+            }
             cur_num_users = old_num_users;
+        }
         // We incremented the counter. But we might still be at the wrong
         // generation number. The generation number is a monotonically
         // increasing value. Since it only increases when someone gets the
         // `num_users` counter to 0, we can simply load the generation number.
         let generation = target.generation.load(Ordering::Acquire);
         if generation != target_id.generation {
             // We're at the wrong generation, so we cannot send the message.
             // However, since we incremented the `num_users` counter, the moment
             // we decrement it we might be the one that are supposed to handle
             // the destruction of the component. Note that all users of the
             // component do an increment-followed-by-decrement, we can simply
             // do a `fetch_sub`.
             let old_num_users = target.num_users.fetch_sub(1, Ordering::SeqCst);
             if old_num_users == 1 {
                 // We're the one that got the counter to 0, so we're the ones
                 // that are supposed to handle component exit
                 self.finish_component_destruction(target_id);
+            }
             return false;
+        }
         // The generation is correct, and since we incremented the `num_users`
         // counter we're now sure that we can send the message and it will be
         // handled by the receiver
         target.connector.public.inbox.insert_message(message);
         // Finally, do the same as above: decrement number of users, if at gets
         // to 0 we're the ones who should handle the exit condition.
         let old_num_users = target.num_users.fetch_sub(1, Ordering::SeqCst);
         if old_num_users == 1 {
             // We're allowed to destroy the component.
             self.finish_component_destruction(target_id);
         } else {
             // Message is sent. If the component is sleeping, then we're sure
             // it is not scheduled and it has not initiated the destruction of
             // the component (because of the way
             // `initiate_component_destruction` does not set sleeping to true).
             // So we can safely schedule it.
             let should_wake_up = target.connector.public.sleeping
                 .compare_exchange(true, false, Ordering::SeqCst, Ordering::Acquire)
                 .is_ok();
             if should_wake_up {
                 let key = unsafe{ ConnectorKey::from_id(target_id) };
                 self.push_work(key);
+            }
+        }
         return true
+    }
     /// Sends a message to a particular component, assumed to occur over a port.
     /// If the component happened to be sleeping then it will be scheduled for
     /// execution. Because of the port management system we may assumed that
     /// we're always accessing the component at the right generation number.
     pub(crate) fn send_message_assumed_alive(&self, target_id: ConnectorId, message: Message) {
         let target = {
             let lock = self.connectors.read().unwrap();
             let entry = lock.get(target_id.index);
             debug_assert_eq!(entry.generation.load(Ordering::Acquire), target_id.generation);
             &mut entry.connector.public
         };
         target.inbox.insert_message(message);
         let should_wake_up = target.sleeping
             .compare_exchange(true, false, Ordering::SeqCst, Ordering::Acquire)
             .is_ok();
@@ @@ -313,29 +402,64 @@ impl RuntimeInner { @@
         };
         self.increment_active_components();
         return key;
+    }
     /// Retrieve private access to the component through its key.
     #[inline]
     pub(crate) fn get_component_private(&self, connector_key: &ConnectorKey) -> &'static mut ScheduledConnector {
         let entry = {
             let lock = self.connectors.read().unwrap();
         return lock.get_private(connector_key);
+    }
             lock.get(connector_key.index)
         };
     #[inline]
     pub(crate) fn get_component_public(&self, connector_id: ConnectorId) -> &'static ConnectorPublic {
         let lock = self.connectors.read().unwrap();
         return lock.get_public(connector_id);
         debug_assert_eq!(entry.generation.load(Ordering::Acquire), connector_key.generation, "private access to {:?}", connector_key);
         return &mut entry.connector;
+    }
     pub(crate) fn destroy_component(&self, connector_key: ConnectorKey) {
     // --- Managing component destruction
     /// Start component destruction, may only be done by the scheduler that is
     /// executing the component. This might not actually destroy the component,
     /// since other components might be sending it messages.
     fn initiate_component_destruction(&self, connector_key: ConnectorKey) {
         // Most of the time no-one will be sending messages, so try
         // immediate destruction
         let mut lock = self.connectors.write().unwrap();
         let entry = lock.get(connector_key.index);
         debug_assert_eq!(entry.generation.load(Ordering::Acquire), connector_key.generation);
         debug_assert_eq!(entry.connector.public.sleeping.load(Ordering::Acquire), false); // not sleeping: caller is executing this component
         let old_num_users = entry.num_users.fetch_sub(1, Ordering::SeqCst);
         if old_num_users == 1 {
             // We just brought the number of users down to 0. Destroy the
             // component
             entry.connector.public.inbox.clear();
             entry.generation.fetch_add(1, Ordering::SeqCst);
             lock.destroy(connector_key);
             self.decrement_active_components();
+        }
+    }
     fn finish_component_destruction(&self, connector_id: ConnectorId) {
         let mut lock = self.connectors.write().unwrap();
         let entry = lock.get(connector_id.index);
         debug_assert_eq!(entry.num_users.load(Ordering::Acquire), 0);
         let _old_generation = entry.generation.fetch_add(1, Ordering::SeqCst);
         debug_assert_eq!(_old_generation, connector_id.generation);
         // TODO: In the future we should not only clear out the inbox, but send
         //  messages back to the senders indicating the messages did not arrive.
         entry.connector.public.inbox.clear();
         // Invariant of only one thread being able to handle the internals of
         // component is preserved by the fact that only one thread can decrement
         // `num_users` to 0.
         lock.destroy(unsafe{ ConnectorKey::from_id(connector_id) });
         self.decrement_active_components();
+    }
     // --- Managing exit condition
     #[inline]
     pub(crate) fn increment_active_interfaces(&self) {
         let _old_num = self.active_interfaces.fetch_add(1, Ordering::SeqCst);
@@ @@ -393,12 +517,13 @@ unsafe impl Sync for RuntimeInner {} @@
 // ConnectorStore
 // -----------------------------------------------------------------------------
 struct StoreEntry {
     connector: ScheduledConnector,
     generation: std::sync::atomic::AtomicU32,
     num_users: std::sync::atomic::AtomicU32,
+}
 struct ConnectorStore {
     // Freelist storage of connectors. Storage should be pointer-stable as
     // someone might be mutating the vector while we're executing one of the
     // connectors.
@@ @@ -411,32 +536,19 @@ impl ConnectorStore { @@
         Self {
             entries: RawVec::with_capacity(capacity),
             free: Vec::with_capacity(capacity),
+        }
+    }
     /// Retrieves public part of connector - accessible by many threads at once.
     fn get_public(&self, id: ConnectorId) -> &'static ConnectorPublic {
         unsafe {
             let entry = self.entries.get(id.index as usize);
             debug_assert!(!entry.is_null());
             let cur_generation = (**entry).generation.load(Ordering::Acquire);
             assert_eq!(cur_generation, id.generation, "accessing {}", id.index);
             return &(**entry).connector.public;
+        }
+    }
     /// Retrieves private part of connector - accessible by one thread at a
     /// time.
     fn get_private(&self, key: &ConnectorKey) -> &'static mut ScheduledConnector {
     /// Directly retrieves an entry. There be dragons here. The `connector`
     /// might have its destructor already executed. Accessing it might then lead
     /// to memory corruption.
     fn get(&self, index: u32) -> &'static mut StoreEntry {
         unsafe {
             let entry = self.entries.get_mut(key.index as usize);
             debug_assert!(!entry.is_null());
             let cur_generation = (**entry).generation.load(Ordering::Acquire);
             assert_eq!(cur_generation, key.generation, "accessing {}", key.index);
             return &mut (**entry).connector;
             let entry = self.entries.get_mut(index as usize);
             return &mut **entry;
+        }
+    }
     /// Creates a new connector. Caller should ensure ports are set up correctly
     /// and the connector is queued for execution if needed.
     fn create(&mut self, connector: ConnectorVariant, initially_sleeping: bool) -> ConnectorKey {
@@ @@ -459,21 +571,24 @@ impl ConnectorStore { @@
             };
             connector.ctx.id = key.downcast();
             let connector = Box::into_raw(Box::new(StoreEntry{
                 connector,
                 generation: AtomicU32::new(0),
                 num_users: AtomicU32::new(1),
             }));
             self.entries.push(connector);
         } else {
             // Free spot available
             index = self.free.pop().unwrap();
             unsafe {
                 let target = &mut **self.entries.get_mut(index);
                 std::ptr::write(&mut target.connector as *mut _, connector);
                 let _old_num_users = target.num_users.fetch_add(1, Ordering::SeqCst);
                 debug_assert_eq!(_old_num_users, 0);
                 let generation = target.generation.load(Ordering::Acquire);
                 key = ConnectorKey{ index: index as u32, generation };
                 target.connector.ctx.id = key.downcast();
+            }
+        }

src/runtime2/native.rs

➞

Show inline comments

 use std::collections::VecDeque;
 use std::sync::{Arc, Mutex, Condvar};
 use std::sync::atomic::Ordering;
 use std::collections::HashMap;
 use crate::protocol::ComponentCreationError;
 use crate::protocol::eval::ValueGroup;
 use crate::runtime2::consensus::RoundConclusion;
-use super::{ConnectorKey, ConnectorId, RuntimeInner};
 use super::{ConnectorId, RuntimeInner};
 use super::branch::{BranchId, FakeTree, QueueKind, SpeculativeState};
 use super::scheduler::{SchedulerCtx, ComponentCtx, MessageTicket};
 use super::port::{Port, PortIdLocal, Channel, PortKind};
 use super::consensus::{Consensus, Consistency, find_ports_in_value_group};
 use super::connector::{ConnectorScheduling, ConnectorPDL};
 use super::inbox::{
@@ @@ -531,27 +529,18 @@ impl ApplicationInterface { @@
     /// Called by runtime to set associated connector's ID.
     pub(crate) fn set_connector_id(&mut self, id: ConnectorId) {
         self.connector_id = id;
+    }
     fn wake_up_connector_with_ping(&self) {
         let connector = self.runtime.get_component_public(self.connector_id);
         connector.inbox.insert_message(Message::Control(ControlMessage {
         let message = ControlMessage {
             id: 0,
             sending_component_id: self.connector_id,
             content: ControlContent::Ping,
         }));
         let should_wake_up = connector.sleeping
             .compare_exchange(true, false, Ordering::SeqCst, Ordering::Acquire)
             .is_ok();
         if should_wake_up {
             let key = unsafe{ ConnectorKey::from_id(self.connector_id) };
             self.runtime.push_work(key);
+        }
         };
         self.runtime.send_message_maybe_destroyed(self.connector_id, Message::Control(message));
+    }
     fn find_port_by_id(&self, port_id: PortIdLocal) -> Option<PortKind> {
         return self.owned_ports.iter()
             .find(|(_, owned_id)| *owned_id == port_id)
             .map(|(port_kind, _)| *port_kind);

src/runtime2/scheduler.rs

➞

Show inline comments

 use std::collections::VecDeque;
 use std::mem::MaybeUninit;
 use std::sync::Arc;
 use std::sync::atomic::Ordering;
-use crate::collections::RawVec;
 use crate::protocol::eval::EvalError;
 use crate::runtime2::port::ChannelId;
 use super::{ScheduledConnector, RuntimeInner, ConnectorId, ConnectorKey};
 use super::port::{Port, PortState, PortIdLocal};
 use super::native::Connector;
 use super::branch::{BranchId};
 use super::connector::{ConnectorPDL, ConnectorScheduling};
 use super::inbox::{
-    Message, DataMessage, SyncHeader,
     Message, DataMessage,
     ControlMessage, ControlContent,
     SyncControlMessage, SyncControlContent,
 };
 // Because it contains pointers we're going to do a copy by value on this one
 #[derive(Clone, Copy)]
@@ @@ -65,13 +64,13 @@ impl Scheduler { @@
                     // Nothing to do. But we're stil waiting for all our pending
                     // control messages to be answered.
                     self.debug_conn(connector_id, &format!("Shutting down, {} Acks remaining", scheduled.router.num_pending_acks()));
                     if scheduled.router.num_pending_acks() == 0 {
                         // We're actually done, we can safely destroy the
                         // currently running connector
-                        self.runtime.destroy_component(connector_key);
+                        self.runtime.initiate_component_destruction(connector_key);
                         continue 'thread_loop;
                     } else {
                         cur_schedule = ConnectorScheduling::NotNow;
+                    }
                 } else {
                     self.debug_conn(connector_id, "Running ...");
@@ @@ -110,26 +109,26 @@ impl Scheduler { @@
                         if port.state != PortState::Closed {
                             let message = scheduled.router.prepare_closing_channel(
                                 port.self_id, port.peer_id,
                                 connector_id
                             );
                             self.debug_conn(connector_id, &format!("Sending message to {:?} [ exit ] \n --- {:?}", port.peer_connector, message));
-                            self.runtime.send_message(port.peer_connector, Message::Control(message));
+                            self.runtime.send_message_assumed_alive(port.peer_connector, Message::Control(message));
+                        }
+                    }
                     // Any messages still in the public inbox should be handled
                     scheduled.ctx.inbox.clear_read_messages();
                     while let Some(ticket) = scheduled.ctx.get_next_message_ticket_even_if_not_in_sync() {
                         let message = scheduled.ctx.take_message_using_ticket(ticket);
                         self.handle_message_while_shutting_down(message, scheduled);
+                    }
                     if scheduled.router.num_pending_acks() == 0 {
                         // All ports (if any) already closed
-                        self.runtime.destroy_component(connector_key);
+                        self.runtime.initiate_component_destruction(connector_key);
                         continue 'thread_loop;
+                    }
                     self.try_go_to_sleep(connector_key, scheduled);
                 },
+            }
@@ @@ -149,13 +148,13 @@ impl Scheduler { @@
                 if let Some(other_component_id) = scheduled.router.should_reroute(target_port) {
                     self.debug_conn(connector_id, " ... Rerouting the message");
                     // We insert directly into the private inbox. Since we have
                     // a reroute entry the component can not yet be running.
                     if let Message::Control(_) = &message {
-                        self.runtime.send_message(other_component_id, message);
+                        self.runtime.send_message_assumed_alive(other_component_id, message);
                     } else {
                         let key = unsafe { ConnectorKey::from_id(other_component_id) };
                         let component = self.runtime.get_component_private(&key);
                         component.ctx.inbox.insert_new(message);
+                    }
@@ @@ -196,13 +195,13 @@ impl Scheduler { @@
                         let ack_message = Message::Control(ControlMessage {
                             id: message.id,
                             sending_component_id: connector_id,
                             content: ControlContent::Ack,
                         });
                         self.debug_conn(connector_id, &format!("Sending message to {:?} [pp ack]\n --- {:?}", message.sending_component_id, ack_message));
-                        self.runtime.send_message(message.sending_component_id, ack_message);
+                        self.runtime.send_message_assumed_alive(message.sending_component_id, ack_message);
                     },
                     ControlContent::CloseChannel(port_id) => {
                         // Mark the port as being closed
                         let port = scheduled.ctx.get_port_mut_by_id(port_id).unwrap();
                         port.state = PortState::Closed;
@@ @@ -210,13 +209,13 @@ impl Scheduler { @@
                         let ack_message = Message::Control(ControlMessage {
                             id: message.id,
                             sending_component_id: connector_id,
                             content: ControlContent::Ack,
                         });
                         self.debug_conn(connector_id, &format!("Sending message to {:?} [cc ack] \n --- {:?}", message.sending_component_id, ack_message));
-                        self.runtime.send_message(message.sending_component_id, ack_message);
+                        self.runtime.send_message_assumed_alive(message.sending_component_id, ack_message);
                     },
                     ControlContent::Ack => {
                         if let Some(component_key) = scheduled.router.handle_ack(message.id) {
                             self.runtime.push_work(component_key);
                         };
                     },
@@ @@ -255,13 +254,13 @@ impl Scheduler { @@
                 let message = SyncControlMessage {
                     in_response_to_sync_round: sync_round,
                     target_component_id: port.peer_connector,
                     content: SyncControlContent::ChannelIsClosed(port.peer_id),
                 };
                 self.debug_conn(scheduled.ctx.id, &format!("Sending message to {:?} [shutdown]\n --- {:?}", port.peer_connector, message));
-                self.runtime.send_message(port.peer_connector, Message::SyncControl(message));
+                self.runtime.send_message_assumed_alive(port.peer_connector, Message::SyncControl(message));
+            }
+        }
+    }
     /// Handles changes to the context that were made by the component. This is
     /// the way (due to Rust's borrowing rules) that we bubble up changes in the
@@ @@ -269,47 +268,51 @@ impl Scheduler { @@
     /// that the component wants to send, a port that has been added).
     fn handle_changes_in_context(&mut self, scheduled: &mut ScheduledConnector) {
         let connector_id = scheduled.ctx.id;
         // Handling any messages that were sent
         while let Some(message) = scheduled.ctx.outbox.pop_front() {
             let target_component_id = match &message {
+            let (target_component_id, over_port) = match &message {
                 Message::Data(content) => {
                     // Data messages are always sent to a particular port, and
                     // may end up being rerouted.
                     let port_desc = scheduled.ctx.get_port_by_id(content.data_header.sending_port).unwrap();
                     debug_assert_eq!(port_desc.peer_id, content.data_header.target_port);
                     if port_desc.state == PortState::Closed {
                         todo!("handle sending over a closed port")
+                    }
                     port_desc.peer_connector
+                    (port_desc.peer_connector, true)
                 },
                 Message::SyncComp(content) => {
                     // Sync messages are always sent to a particular component,
                     // the sender must make sure it actually wants to send to
                     // the specified component (and is not using an inconsistent
                     // component ID associated with a port).
                     content.target_component_id
+                    (content.target_component_id, false)
                 },
                 Message::SyncPort(content) => {
                     let port_desc = scheduled.ctx.get_port_by_id(content.source_port).unwrap();
                     debug_assert_eq!(port_desc.peer_id, content.target_port);
                     if port_desc.state == PortState::Closed {
                         todo!("handle sending over a closed port")
+                    }
                     port_desc.peer_connector
+                    (port_desc.peer_connector, true)
                 },
                 Message::SyncControl(_) => unreachable!("component sending 'SyncControl' messages directly"),
                 Message::Control(_) => unreachable!("component sending 'Control' messages directly"),
             };
             self.debug_conn(connector_id, &format!("Sending message to {:?} [outbox] \n --- {:#?}", target_component_id, message));
             self.runtime.send_message(target_component_id, message);
             self.debug_conn(connector_id, &format!("Sending message to {:?} [outbox, over port: {}] \n --- {:#?}", target_component_id, over_port, message));
             if over_port {
                 self.runtime.send_message_assumed_alive(target_component_id, message);
             } else {
                 self.runtime.send_message_maybe_destroyed(target_component_id, message);
+            }
+        }
         while let Some(state_change) = scheduled.ctx.state_changes.pop_front() {
             match state_change {
                 ComponentStateChange::CreatedComponent(component, initial_ports) => {
                     // Creating a new component. Need to relinquish control of
@@ @@ -351,13 +354,13 @@ impl Scheduler { @@
                             let control_message = scheduled.router.prepare_changed_port_peer(
                                 control_id, scheduled.ctx.id,
                                 port.peer_connector, port.peer_id,
                                 new_component_id, port.self_id
                             );
                             self.debug_conn(connector_id, &format!("Sending message to {:?} [newcom]\n --- {:#?}", port.peer_connector, control_message));
-                            self.runtime.send_message(port.peer_connector, Message::Control(control_message));
+                            self.runtime.send_message_assumed_alive(port.peer_connector, Message::Control(control_message));
+                        }
+                    }
                 },
                 ComponentStateChange::CreatedPort(port) => {
                     scheduled.ctx.ports.push(port);
                 },
@@ @@ -644,215 +647,101 @@ impl<'a> Iterator for MessagesIter<'a> { @@
 // -----------------------------------------------------------------------------
 /// A structure that contains inbox messages. Some messages are left inside and
 /// continuously re-read. Others are taken out, but may potentially be put back
 /// for later reading. Later reading in this case implies that they are put back
 /// for reading in the next sync round.
 /// TODO: Again, lazy concurrency, see git history for other implementation
 struct Inbox {
     temp_m: Vec<Message>,
     temp_d: Vec<Message>,
     messages: RawVec<Message>,
     next_delay_idx: u32,
     start_read_idx: u32,
     messages: Vec<Message>,
     delayed: Vec<Message>,
     next_read_idx: u32,
     last_read_idx: u32,
     generation: u32,
+}
 #[derive(Clone, Copy)]
 pub(crate) struct MessageTicket {
     index: u32,
     generation: u32,
+}
 impl Inbox {
     fn new() -> Self {
         return Inbox {
             temp_m: Vec::new(), temp_d: Vec::new(),
             messages: RawVec::new(),
             next_delay_idx: 0,
             start_read_idx: 0,
             messages: Vec::new(),
             delayed: Vec::new(),
             next_read_idx: 0,
             last_read_idx: 0,
             generation: 0,
+        }
+    }
     fn insert_new(&mut self, message: Message) {
         assert!(self.messages.len() < u32::MAX as usize); // TODO: @Size
         self.temp_m.push(message);
         return;
         self.messages.push(message);
+    }
     fn get_next_message_ticket(&mut self) -> Option<MessageTicket> {
-        if self.next_read_idx as usize >= self.temp_m.len() { return None };
+        if self.next_read_idx as usize >= self.messages.len() { return None };
         let idx = self.next_read_idx;
         self.generation += 1;
         self.next_read_idx += 1;
         return Some(MessageTicket{ index: idx, generation: self.generation });
         let cur_read_idx = self.next_read_idx as usize;
         if cur_read_idx >= self.messages.len() {
             return None;
+        }
         self.generation += 1;
         self.next_read_idx += 1;
         return Some(MessageTicket{
             index: cur_read_idx as u32,
             generation: self.generation
         });
+    }
     fn read_message_using_ticket(&self, ticket: MessageTicket) -> &Message {
         debug_assert_eq!(self.generation, ticket.generation);
         return &self.temp_m[ticket.index as usize];
         return unsafe{ &*self.messages.get(ticket.index as usize) }
         return &self.messages[ticket.index as usize];
+    }
     fn take_message_using_ticket(&mut self, ticket: MessageTicket) -> Message {
         debug_assert_eq!(self.generation, ticket.generation);
         debug_assert!(ticket.index < self.next_read_idx);
         self.next_read_idx -= 1;
         return self.temp_m.remove(ticket.index as usize);
         unsafe {
             let take_idx = ticket.index as usize;
             let val = std::ptr::read(self.messages.get(take_idx));
             // Move messages to the right, clearing up space in the
             // front.
             let num_move_right = take_idx - self.start_read_idx as usize;
             self.messages.move_range(
                 self.start_read_idx as usize,
                 self.start_read_idx as usize + 1,
                 num_move_right
             );
             self.start_read_idx += 1;
             return val;
+        }
         return self.messages.remove(ticket.index as usize);
+    }
     fn put_back_message(&mut self, message: Message) {
         // We have space in front of the array because we've taken out a message
         // before.
         self.temp_d.push(message);
         return;
         debug_assert!(self.next_delay_idx < self.start_read_idx);
         unsafe {
             // Write to front of the array
             std::ptr::write(self.messages.get_mut(self.next_delay_idx as usize), message);
             self.next_delay_idx += 1;
+        }
         self.delayed.push(message);
+    }
     fn get_read_data_messages(&self, match_port_id: PortIdLocal) -> MessagesIter {
         return MessagesIter{
             messages: self.temp_m.as_slice(),
             next_index: self.start_read_idx as usize,
             max_index: self.next_read_idx as usize,
             match_port_id
         };
         return MessagesIter{
             messages: self.messages.as_slice(),
-            next_index: self.start_read_idx as usize,
+            next_index: 0,
             max_index: self.next_read_idx as usize,
             match_port_id
         };
+    }
     fn clear_read_messages(&mut self) {
         self.temp_m.drain(0..self.next_read_idx as usize);
         for (idx, v) in self.temp_d.drain(..).enumerate() {
             self.temp_m.insert(idx, v);
         self.messages.drain(0..self.next_read_idx as usize);
         for (idx, v) in self.delayed.drain(..).enumerate() {
             self.messages.insert(idx, v);
+        }
         self.next_read_idx = 0;
         return;
         // Deallocate everything that was read
         self.destroy_range(self.start_read_idx, self.next_read_idx);
         self.generation += 1;
         // Join up all remaining values with the delayed ones in the front
         let num_to_move = self.messages.len() - self.next_read_idx as usize;
         self.messages.move_range(
             self.next_read_idx as usize,
             self.next_delay_idx as usize,
             num_to_move
         );
         // Set all indices (and the RawVec len) to make sense in this new state
         let new_len = self.next_delay_idx as usize + num_to_move;
         self.next_delay_idx = 0;
         self.start_read_idx = 0;
         self.next_read_idx = 0;
         self.messages.len = new_len;
+    }
     fn transfer_messages_for_port(&mut self, port: PortIdLocal, new_inbox: &mut Inbox) {
-        debug_assert!(self.temp_d.is_empty());
+        debug_assert!(self.delayed.is_empty());
         let mut idx = 0;
         while idx < self.temp_m.len() {
             let msg = &self.temp_m[idx];
         while idx < self.messages.len() {
             let msg = &self.messages[idx];
             if let Some(target) = msg.target_port() {
                 if target == port {
-                    new_inbox.temp_m.push(self.temp_m.remove(idx));
+                    new_inbox.messages.push(self.messages.remove(idx));
                     continue;
+                }
+            }
             idx += 1;
+        }
         return;
         let mut idx = 0;
         while idx < self.messages.len() {
             let message = unsafe{ &*self.messages.get(idx) };
             if let Some(target_port) = message.target_port() {
                 if target_port == port {
                     // Transfer port
                     unsafe {
                         let message = std::ptr::read(message as *const _);
                         let remaining = self.messages.len() - idx - 1; // idx < len, due to loop condition
                         if remaining > 0 {
                             self.messages.move_range(idx + 1, idx, remaining);
+                        }
                         self.messages.len -= 1;
                         new_inbox.insert_new(message);
+                    }
                     continue; // do not increment index
+    }
+}
             idx += 1;
+        }
+    }
     #[inline]
     fn destroy_range(&mut self, start_idx: u32, end_idx: u32) {
         for idx in (start_idx as usize)..(end_idx as usize) {
             unsafe {
                 let msg = self.messages.get_mut(idx);
                 std::ptr::drop_in_place(msg);
+            }
+        }
+    }
+}
 //
 // impl Drop for Inbox {
 //     fn drop(&mut self) {
 //         // Whether in sync or not in sync. We have two ranges of allocated
 //         // messages:
 //         // - delayed messages: from 0 to `next_delay_idx` (which is 0 if in non-sync)
 //         // - readable messages: from `start_read_idx` to `messages.len`
 //         self.destroy_range(0, self.next_delay_idx);
 //         self.destroy_range(self.start_read_idx, self.messages.len as u32);
 //     }
 // }
 // -----------------------------------------------------------------------------
 // Control messages
 // -----------------------------------------------------------------------------
 struct ControlEntry {
     id: u32,

src/runtime2/tests/mod.rs

➞

Show inline comments

@@ @@ -13,13 +13,13 @@ use crate::runtime2::native::{ApplicationSyncAction}; @@
 // Generic testing constants, use when appropriate to simplify stress-testing
 // pub(crate) const NUM_THREADS: u32 =  8;     // number of threads in runtime
 // pub(crate) const NUM_INSTANCES: u32 = 750;  // number of test instances constructed
 // pub(crate) const NUM_LOOPS: u32 = 10;       // number of loops within a single test (not used by all tests)
 pub(crate) const NUM_THREADS: u32 = 6;
-pub(crate) const NUM_INSTANCES: u32 = 1;
+pub(crate) const NUM_INSTANCES: u32 = 2;
 pub(crate) const NUM_LOOPS: u32 = 1;
 fn create_runtime(pdl: &str) -> Runtime {
     let protocol = ProtocolDescription::parse(pdl.as_bytes()).expect("parse pdl");
     let runtime = Runtime::new(NUM_THREADS, protocol);

0 comments (0 inline, 0 general)