-/* Copyright (c) 2007-2019. The SimGrid Team.
- * All rights reserved. */
+/* Copyright (c) 2007-2023. The SimGrid Team. All rights reserved. */
/* This program is free software; you can redistribute it and/or modify it
* under the terms of the license (GNU LGPL) which comes with this package. */
#ifndef SIMGRID_SIMIX_HPP
#define SIMGRID_SIMIX_HPP
-#include <simgrid/simix.h>
-#include <xbt/functional.hpp>
-#include <xbt/future.hpp>
+#include <simgrid/s4u/Actor.hpp>
+#include <xbt/promise.hpp>
#include <xbt/signal.hpp>
-#include <boost/heap/fibonacci_heap.hpp>
#include <string>
#include <unordered_map>
-XBT_PUBLIC void simcall_run_kernel(std::function<void()> const& code);
+XBT_PUBLIC void simcall_run_answered(std::function<void()> const& code,
+ simgrid::kernel::actor::SimcallObserver* observer);
+XBT_PUBLIC void simcall_run_blocking(std::function<void()> const& code,
+ simgrid::kernel::actor::SimcallObserver* observer);
+XBT_PUBLIC void simcall_run_object_access(std::function<void()> const& code,
+ simgrid::kernel::actor::ObjectAccessSimcallItem* item);
-/** Execute some code in the kernel and block
+namespace simgrid::kernel::actor {
+
+/** Execute some code in kernel context on behalf of the user code.
*
- * run_blocking() is a generic blocking simcall. It is given a callback
- * which is executed immediately in the SimGrid kernel. The callback is
- * responsible for setting the suitable logic for waking up the process
- * when needed.
+ * Every modification of the environment must be protected this way: every setter, constructor and similar.
+ * Getters don't have to be protected this way, and setters may use the simcall_object_access() variant (see below).
*
- * @ref simix::kernelSync() is a higher level wrapper for this.
- */
-XBT_PUBLIC void simcall_run_blocking(std::function<void()> const& code);
-
-template<class F> inline
-void simcall_run_kernel(F& f)
-{
- simcall_run_kernel(std::function<void()>(std::ref(f)));
-}
-template<class F> inline
-void simcall_run_blocking(F& f)
-{
- simcall_run_blocking(std::function<void()>(std::ref(f)));
-}
-
-namespace simgrid {
-namespace simix {
-
-/** Execute some code in the kernel/maestro
+ * This allows deterministic parallel simulation without any locking, even if almost nobody uses parallel simulation in
+ * SimGrid. More interestingly it makes every modification of the simulated world observable by the model-checker,
+ * allowing the whole MC business.
+ *
+ * It is highly inspired from the syscalls in a regular operating system, allowing the user code to get some specific
+ * code executed in the kernel context. But here, there is almost no security involved. Parameters get checked for
+ * finiteness but that's all. The main goal remain to ensure reproducible ordering of uncomparable events (in
+ * [parallel] simulation) and observability of events (in model-checking).
*
- * This can be used to enforce mutual exclusion with other simcall.
- * More importantly, this enforces a deterministic/reproducible ordering
- * of the operation with respect to other simcalls.
+ * The code passed as argument is supposed to terminate at the exact same simulated timestamp.
+ * Do not use it if your code may block waiting for a subsequent event, e.g. if you lock a mutex,
+ * you may need to wait for that mutex to be unlocked by its current owner.
+ * Potentially blocking simcall must be issued using simcall_blocking(), right below in this file.
*/
-template <class F> typename std::result_of<F()>::type simcall(F&& code)
+template <class F> typename std::result_of_t<F()> simcall_answered(F&& code, SimcallObserver* observer = nullptr)
{
// If we are in the maestro, we take the fast path and execute the
- // code directly without simcall mashalling/unmarshalling/dispatch:
- if (SIMIX_is_maestro())
+ // code directly without simcall marshalling/unmarshalling/dispatch:
+ if (s4u::Actor::is_maestro())
return std::forward<F>(code)();
// If we are in the application, pass the code to the maestro which
// executes it for us and reports the result. We use a std::future which
// conveniently handles the success/failure value for us.
- typedef typename std::result_of<F()>::type R;
+ using R = typename std::result_of_t<F()>;
simgrid::xbt::Result<R> result;
- simcall_run_kernel([&result, &code] { simgrid::xbt::fulfill_promise(result, std::forward<F>(code)); });
+ simcall_run_answered([&result, &code] { simgrid::xbt::fulfill_promise(result, std::forward<F>(code)); }, observer);
return result.get();
}
-XBT_ATTRIB_DEPRECATED_v325("Please manifest if you actually need this function.")
- XBT_PUBLIC const std::vector<smx_actor_t>& process_get_runnable();
-
-// What's executed as SIMIX actor code:
-typedef std::function<void()> ActorCode;
-
-// Create an ActorCode based on a std::string
-typedef std::function<ActorCode(std::vector<std::string> args)> ActorCodeFactory;
-
-XBT_PUBLIC void register_function(const std::string& name, ActorCodeFactory factory);
-
-typedef std::pair<double, Timer*> TimerQelt;
-static boost::heap::fibonacci_heap<TimerQelt, boost::heap::compare<xbt::HeapComparator<TimerQelt>>> simix_timers;
-
-/** @brief Timer datatype */
-class Timer {
- double date = 0.0;
-
-public:
- decltype(simix_timers)::handle_type handle_;
-
- Timer(double date, simgrid::xbt::Task<void()> callback) : date(date), callback(std::move(callback)) {}
-
- simgrid::xbt::Task<void()> callback;
- double get_date() { return date; }
- void remove();
-
- template <class F> static inline Timer* set(double date, F callback)
- {
- return set(date, simgrid::xbt::Task<void()>(std::move(callback)));
- }
-
- template <class R, class T> static inline Timer* set(double date, R (*callback)(T*), T* arg)
- {
- return set(date, [callback, arg]() { callback(arg); });
- }
-
- static Timer* set(double date, void (*callback)(void*), void* arg);
- static Timer* set(double date, simgrid::xbt::Task<void()> callback);
- static double next() { return simix_timers.empty() ? -1.0 : simix_timers.top().first; }
-};
+/** Use a setter on the `item` object. That's a simcall only if running in parallel or with MC activated.
+ *
+ * Simulation without MC and without parallelism (contexts/nthreads=1) will not pay the price of a simcall for an
+ * harmless setter. When running in parallel, you want your write access to be done in a mutual exclusion way, while the
+ * getters can still occur out of order.
+ *
+ * When running in MC, you want to make this access visible to the checker. Actually in this case, it's not visible from
+ * the checker (and thus still use a fast track) if the setter is called from the actor that created the object.
+ */
+template <class F> typename std::result_of_t<F()> simcall_object_access(ObjectAccessSimcallItem* item, F&& code)
+{
+ // If we are in the maestro, we take the fast path and execute the code directly
+ if (simgrid::s4u::Actor::is_maestro())
+ return std::forward<F>(code)();
-} // namespace simix
-} // namespace simgrid
+ // If called from another thread, do a real simcall. It will be short-cut on need
+ using R = typename std::result_of_t<F()>;
+ simgrid::xbt::Result<R> result;
+ simcall_run_object_access([&result, &code] { simgrid::xbt::fulfill_promise(result, std::forward<F>(code)); }, item);
-XBT_PUBLIC smx_actor_t simcall_process_create(std::string name, simgrid::simix::ActorCode code, void* data,
- sg_host_t host, std::unordered_map<std::string, std::string>* properties);
+ return result.get();
+}
-XBT_PUBLIC smx_timer_t SIMIX_timer_set(double date, simgrid::xbt::Task<void()> callback);
+/** Execute some code (that does not return immediately) in kernel context
+ *
+ * This is very similar to simcall_answered() above, but the calling actor will not get rescheduled until
+ * actor->simcall_answer() is called explicitly.
+ *
+ * This is meant for blocking actions. For example, locking a mutex is a blocking simcall.
+ * First it's a simcall because that's obviously a modification of the world. Then, that's a blocking simcall because if
+ * the mutex happens not to be free, the actor is added to a queue of actors in the mutex. Every mutex->unlock() takes
+ * the first actor from the queue, mark it as current owner of the mutex and call actor->simcall_answer() to mark that
+ * this mutex is now unblocked and ready to run again. If the mutex is initially free, the calling actor is unblocked
+ * right away with actor->simcall_answer() once the mutex is marked as locked.
+ *
+ * If your code never calls actor->simcall_answer() itself, the actor will never return from its simcall.
+ *
+ * The return value is obtained from observer->get_result() if it exists. Otherwise void is returned.
+ */
+template <class F> void simcall_blocking(F&& code, SimcallObserver* observer = nullptr)
+{
+ xbt_assert(not s4u::Actor::is_maestro(), "Cannot execute blocking call in kernel mode");
+ // Pass the code to the maestro which executes it for us and reports the result. We use a std::future which
+ // conveniently handles the success/failure value for us.
+ simgrid::xbt::Result<void> result;
+ simcall_run_blocking([&result, &code] { simgrid::xbt::fulfill_promise(result, std::forward<F>(code)); }, observer);
+ result.get(); // rethrow stored exception if any
+}
+template <class F, class Observer>
+auto simcall_blocking(F&& code, Observer* observer) -> decltype(observer->get_result())
+{
+ simcall_blocking(std::forward<F>(code), static_cast<SimcallObserver*>(observer));
+ return observer->get_result();
+}
+} // namespace simgrid::kernel::actor
#endif
