|
gem5 v24.0.0.0
|
Queue of events sorted in time order. More...
#include <eventq.hh>
Classes | |
| class | ScopedMigration |
| class | ScopedRelease |
Public Member Functions | |
| EventQueue (const std::string &n) | |
| void | schedule (Event *event, Tick when, bool global=false) |
| Schedule the given event on this queue. | |
| void | deschedule (Event *event) |
| Deschedule the specified event. | |
| void | reschedule (Event *event, Tick when, bool always=false) |
| Reschedule the specified event. | |
| Tick | nextTick () const |
| void | setCurTick (Tick newVal) |
| Tick | getCurTick () const |
| While curTick() is useful for any object assigned to this event queue, if an object that is assigned to another event queue (or a non-event object) need to access the current tick of this event queue, this function is used. | |
| Event * | getHead () const |
| Event * | serviceOne () |
| void | serviceEvents (Tick when) |
| process all events up to the given timestamp. | |
| bool | empty () const |
| Returns true if no events are queued. | |
| void | dump () const |
| This is a debugging function which will print everything on the event queue. | |
| bool | debugVerify () const |
| void | handleAsyncInsertions () |
| Function for moving events from the async_queue to the main queue. | |
| virtual void | wakeup (Tick when=(Tick) -1) |
| Function to signal that the event loop should be woken up because an event has been scheduled by an agent outside the gem5 event loop(s) whose event insertion may not have been noticed by gem5. | |
| Event * | replaceHead (Event *s) |
| function for replacing the head of the event queue, so that a different set of events can run without disturbing events that have already been scheduled. | |
| void | checkpointReschedule (Event *event) |
| Reschedule an event after a checkpoint. | |
| virtual | ~EventQueue () |
| virtual const std::string | name () const |
| void | name (const std::string &st) |
| void | lock () |
| Provide an interface for locking/unlocking the event queue. | |
| void | unlock () |
Private Member Functions | |
| void | insert (Event *event) |
| Insert / remove event from the queue. | |
| void | remove (Event *event) |
| void | asyncInsert (Event *event) |
| Function for adding events to the async queue. | |
| EventQueue (const EventQueue &) | |
Private Attributes | |
| std::string | objName |
| Event * | head |
| Tick | _curTick |
| UncontendedMutex | async_queue_mutex |
| Mutex to protect async queue. | |
| std::list< Event * > | async_queue |
| List of events added by other threads to this event queue. | |
| UncontendedMutex | service_mutex |
| Lock protecting event handling. | |
Friends | |
| void | curEventQueue (EventQueue *q) |
Queue of events sorted in time order.
Events are scheduled (inserted into the event queue) using the schedule() method. This method either inserts a synchronous or asynchronous event.
Synchronous events are scheduled using schedule() method with the argument 'global' set to false (default). This should only be done from a thread holding the event queue lock (EventQueue::service_mutex). The lock is always held when an event handler is called, it can therefore always insert events into its own event queue unless it voluntarily releases the lock.
Events can be scheduled across thread (and event queue borders) by either scheduling asynchronous events or taking the target event queue's lock. However, the lock should never be taken directly since this is likely to cause deadlocks. Instead, code that needs to schedule events in other event queues should temporarily release its own queue and lock the new queue. This prevents deadlocks since a single thread never owns more than one event queue lock. This functionality is provided by the ScopedMigration helper class. Note that temporarily migrating between event queues can make the simulation non-deterministic, it should therefore be limited to cases where that can be tolerated (e.g., handling asynchronous IO or fast-forwarding in KVM).
Asynchronous events can also be scheduled using the normal schedule() method with the 'global' parameter set to true. Unlike the previous queue migration strategy, this strategy is fully deterministic. This causes the event to be inserted in a separate queue of asynchronous events (async_queue), which is merged main event queue at the end of each simulation quantum (by calling the handleAsyncInsertions() method). Note that this implies that such events must happen at least one simulation quantum into the future, otherwise they risk being scheduled in the past by handleAsyncInsertions().
|
private |
|
inlinevirtual |
Definition at line 964 of file eventq.hh.
References deschedule(), empty(), and getHead().
|
private |
Function for adding events to the async queue.
The added events are added to main event queue later. Threads, other than the owning thread, should call this function instead of insert().
Definition at line 450 of file eventq.cc.
References async_queue, async_queue_mutex, gem5::MipsISA::event, gem5::UncontendedMutex::lock(), and gem5::UncontendedMutex::unlock().
Referenced by schedule().
| void gem5::EventQueue::checkpointReschedule | ( | Event * | event | ) |
Reschedule an event after a checkpoint.
Since events don't know which event queue they belong to, parent objects need to reschedule events themselves. This method conditionally schedules an event that has the Scheduled flag set. It should be called by parent objects after unserializing an object.
@warn Only use this method after unserializing an Event.
Definition at line 306 of file eventq.cc.
References gem5::MipsISA::event, insert(), and gem5::EventBase::Scheduled.
| bool gem5::EventQueue::debugVerify | ( | ) | const |
Definition at line 340 of file eventq.cc.
References gem5::cprintf(), gem5::Event::dump(), head, gem5::Event::nextBin, gem5::Event::nextInBin, gem5::ArmISA::priority, gem5::Event::priority(), and gem5::Event::when().
|
inline |
Definition at line 851 of file eventq.hh.
References head.
Referenced by gem5::ruby::RubySystem::memWriteback(), and ~EventQueue().
| void gem5::EventQueue::handleAsyncInsertions | ( | ) |
Function for moving events from the async_queue to the main queue.
Definition at line 458 of file eventq.cc.
References async_queue, async_queue_mutex, curEventQueue, insert(), gem5::UncontendedMutex::lock(), and gem5::UncontendedMutex::unlock().
Referenced by gem5::doSimLoop(), and gem5::GlobalSyncEvent::BarrierEvent::process().
|
private |
Insert / remove event from the queue.
Should only be called by thread operating this queue.
Definition at line 138 of file eventq.cc.
References gem5::MipsISA::event, head, gem5::Event::insertBefore(), and gem5::Event::nextBin.
Referenced by checkpointReschedule(), handleAsyncInsertions(), reschedule(), and schedule().
|
inline |
Provide an interface for locking/unlocking the event queue.
@warn Do NOT use these methods directly unless you really know what you are doing. Incorrect use can easily lead to simulator deadlocks.
Definition at line 947 of file eventq.hh.
References gem5::UncontendedMutex::lock(), and service_mutex.
Referenced by gem5::DistIface::RecvScheduler::pushPacket(), gem5::DistIface::recvThreadFunc(), gem5::EventQueue::ScopedMigration::ScopedMigration(), serviceOne(), gem5::EventQueue::ScopedMigration::~ScopedMigration(), and gem5::EventQueue::ScopedRelease::~ScopedRelease().
|
inline |
Definition at line 836 of file eventq.hh.
References head, and gem5::Event::when().
Referenced by gem5::doSimLoop(), gem5::Iris::ThreadContext::maintainStepping(), serviceEvents(), and gem5::BaseKvmCPU::setupInstStop().
|
private |
Definition at line 192 of file eventq.cc.
References gem5::MipsISA::event, head, gem5::Event::nextBin, panic, and gem5::Event::removeItem().
Referenced by deschedule(), and reschedule().
function for replacing the head of the event queue, so that a different set of events can run without disturbing events that have already been scheduled.
Already scheduled events can be processed by replacing the original head back. USING THIS FUNCTION CAN BE DANGEROUS TO THE HEALTH OF THE SIMULATOR. NOT RECOMMENDED FOR USE.
Definition at line 382 of file eventq.cc.
References head, gem5::ArmISA::s, and gem5::ArmISA::t.
Referenced by gem5::ruby::RubySystem::startup().
| Event * gem5::EventQueue::serviceOne | ( | ) |
Definition at line 224 of file eventq.cc.
References gem5::Flags< T >::clear(), gem5::MipsISA::event, gem5::Event::flags, head, gem5::EventBase::IsMainQueue, lock(), gem5::EventBase::Managed, gem5::Event::nextBin, gem5::Event::nextInBin, gem5::EventBase::Scheduled, setCurTick(), and gem5::EventBase::Squashed.
Referenced by gem5::doSimLoop(), and serviceEvents().
|
inline |
Definition at line 837 of file eventq.hh.
References _curTick.
Referenced by serviceEvents(), serviceOne(), and gem5::EventManager::setCurTick().
|
inline |
Definition at line 948 of file eventq.hh.
References service_mutex, and gem5::UncontendedMutex::unlock().
Referenced by gem5::DistIface::RecvScheduler::pushPacket(), gem5::DistIface::recvThreadFunc(), gem5::EventQueue::ScopedMigration::ScopedMigration(), gem5::EventQueue::ScopedRelease::ScopedRelease(), and gem5::EventQueue::ScopedMigration::~ScopedMigration().
|
friend |
Definition at line 972 of file eventq.hh.
Referenced by deschedule(), handleAsyncInsertions(), reschedule(), schedule(), gem5::EventQueue::ScopedMigration::ScopedMigration(), and gem5::EventQueue::ScopedMigration::~ScopedMigration().
|
private |
Definition at line 622 of file eventq.hh.
Referenced by getCurTick(), and setCurTick().
List of events added by other threads to this event queue.
Definition at line 628 of file eventq.hh.
Referenced by asyncInsert(), and handleAsyncInsertions().
|
private |
Mutex to protect async queue.
Definition at line 625 of file eventq.hh.
Referenced by asyncInsert(), and handleAsyncInsertions().
|
private |
Definition at line 621 of file eventq.hh.
Referenced by debugVerify(), dump(), empty(), getHead(), insert(), nextTick(), remove(), replaceHead(), and serviceOne().
|
private |
|
private |
Lock protecting event handling.
This lock is always taken when servicing events. It is assumed that the thread scheduling new events (not asynchronous events though) have taken this lock. This is normally done by serviceOne() since new events are typically scheduled as a response to an earlier event.
This lock is intended to be used to temporarily steal an event queue to support inter-thread communication when some deterministic timing can be sacrificed for speed. For example, the KVM CPU can use this support to access devices running in a different thread.