-/* Copyright (c) 2016. The SimGrid Team.
+/* Copyright (c) 2016-2018. The SimGrid Team.
* All rights reserved. */
/* This program is free software; you can redistribute it and/or modify it
#ifndef SIMGRID_SIMIX_BLOCKING_SIMCALL_HPP
#define SIMGRID_SIMIX_BLOCKING_SIMCALL_HPP
-#include <iostream>
-
#include <exception>
+#include <functional>
+#include <future>
+#include <utility>
#include <xbt/sysdep.h>
-#include <future>
-
#include <xbt/future.hpp>
#include <simgrid/kernel/future.hpp>
#include <simgrid/simix.h>
+#include <simgrid/simix.hpp>
-XBT_PUBLIC(void) simcall_run_blocking(std::function<void()> const& code);
+XBT_PUBLIC void simcall_run_blocking(std::function<void()> const& code);
namespace simgrid {
namespace simix {
-XBT_PUBLIC(void) unblock(smx_process_t process);
+XBT_PUBLIC void unblock(smx_actor_t process);
-/** Execute some code in kernel mode and wakes up the process when
+/** Execute some code in kernel mode and wakes up the actor when
* the result is available.
*
- * The code given is executed in SimGrid kernel and expected to return
- * a `simgrid::kernel::Future`. The current process is resumed whenever
- * the Future becomes ready and gets the value or exception of the future:
+ * It is given a callback which is executed in the SimGrid kernel and
+ * returns a `simgrid::kernel::Future<T>`. The kernel blocks the actor
+ * until the Future is ready and:
+ *
+ * - either returns the value wrapped in the future to the actor;
+ *
+ * - or raises the exception stored in the future in the actor.
*
- * This can be used to implement blocking calls in without adding new simcalls.
- * One downside of this approach is that we don't have any semantic on what
- * the process is waiting. This might be a problem for the model-checker and
- * we'll have to device a way to make it work.
+ * This can be used to implement blocking calls without adding new simcalls.
+ * One downside of this approach is that we don't have any semantic on what
+ * the actor is waiting. This might be a problem for the model-checker and
+ * we'll have to devise a way to make it work.
*
- * @param code Kernel code returning a `simgrid::kernel::Future<T>`
- * @return Value of the kernel future
- * @exception Exception from the kernel future
+ * @param code Kernel code returning a `simgrid::kernel::Future<T>`
+ * @return Value of the kernel future
+ * @exception Exception from the kernel future
*/
-template<class F>
-auto blocking_simcall(F code) -> decltype(code().get())
+template <class F> auto kernel_sync(F code) -> decltype(code().get())
{
typedef decltype(code().get()) T;
if (SIMIX_is_maestro())
xbt_die("Can't execute blocking call in kernel mode");
- smx_process_t self = SIMIX_process_self();
+ smx_actor_t self = SIMIX_process_self();
simgrid::xbt::Result<T> result;
simcall_run_blocking([&result, self, &code]{
try {
auto future = code();
- future.then([&result, self](simgrid::kernel::Future<T> value) {
- simgrid::xbt::setPromise(result, value);
+ future.then_([&result, self](std::shared_ptr<simgrid::kernel::FutureState<T>>&& value) {
+ simgrid::xbt::set_promise(result, simgrid::kernel::Future<T>(value));
simgrid::simix::unblock(self);
});
}
});
return result.get();
}
+template <class F>
+XBT_ATTRIB_DEPRECATED_v323("Please use simix::kernel_sync()") auto kernelSync(F code) -> decltype(code().get())
+{
+ return kernel_sync(code);
+}
+
+/** A blocking (`wait()`-based) future for SIMIX processes */
+// TODO, .wait_for()
+// TODO, .wait_until()
+// TODO, SharedFuture
+// TODO, simgrid::simix::when_all - wait for all future to be ready (this one is simple!)
+// TODO, simgrid::simix::when_any - wait for any future to be ready
+template <class T>
+class Future {
+public:
+ Future() { /* Nothing to do*/}
+ explicit Future(simgrid::kernel::Future<T> future) : future_(std::move(future)) {}
+ bool valid() const { return future_.valid(); }
+ T get()
+ {
+ if (not valid())
+ throw std::future_error(std::future_errc::no_state);
+ smx_actor_t self = SIMIX_process_self();
+ simgrid::xbt::Result<T> result;
+ simcall_run_blocking([this, &result, self]{
+ try {
+ // When the kernel future is ready...
+ this->future_.then_([&result, self](std::shared_ptr<simgrid::kernel::FutureState<T>>&& value) {
+ // ... wake up the process with the result of the kernel future.
+ simgrid::xbt::set_promise(result, simgrid::kernel::Future<T>(value));
+ simgrid::simix::unblock(self);
+ });
+ }
+ catch (...) {
+ result.set_exception(std::current_exception());
+ simgrid::simix::unblock(self);
+ }
+ });
+ return result.get();
+ }
+ bool is_ready() const
+ {
+ if (not valid())
+ throw std::future_error(std::future_errc::no_state);
+ return future_.is_ready();
+ }
+ void wait()
+ {
+ // The future is ready! We don't have to wait:
+ if (this->is_ready())
+ return;
+ // The future is not ready. We have to delegate to the SimGrid kernel:
+ std::exception_ptr exception;
+ smx_actor_t self = SIMIX_process_self();
+ simcall_run_blocking([this, &exception, self]{
+ try {
+ // When the kernel future is ready...
+ this->future_.then_([this, self](std::shared_ptr<simgrid::kernel::FutureState<T>>&& value) {
+ // ...store it the simix kernel and wake up.
+ this->future_ = std::move(simgrid::kernel::Future<T>(value));
+ simgrid::simix::unblock(self);
+ });
+ }
+ catch (...) {
+ exception = std::current_exception();
+ simgrid::simix::unblock(self);
+ }
+ });
+ }
+private:
+ // We wrap an event-based kernel future:
+ simgrid::kernel::Future<T> future_;
+};
+
+/** Start some asynchronous work
+ *
+ * @param code SimGrid kernel code which returns a simgrid::kernel::Future
+ * @return Actor future
+ */
+template <class F> auto kernel_async(F code) -> Future<decltype(code().get())>
+{
+ typedef decltype(code().get()) T;
+
+ // Execute the code in the kernel and get the kernel future:
+ simgrid::kernel::Future<T> future = simgrid::simix::simcall(std::move(code));
+
+ // Wrap the kernel future in a actor future:
+ return simgrid::simix::Future<T>(std::move(future));
+}
+template <class F>
+XBT_ATTRIB_DEPRECATED_v323("Please use simix::kernel_sync()") auto kernelAsync(F code) -> Future<decltype(code().get())>
+{
+ return kernel_async(code);
+}
}
}
