summaryrefslogtreecommitdiff
path: root/Userland/Libraries/LibCore/EventLoop.h
blob: 7a0b10b5cc7c009d566b2c28261500cd048b4fdf (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
/*
 * Copyright (c) 2018-2023, Andreas Kling <kling@serenityos.org>
 * Copyright (c) 2022, kleines Filmröllchen <filmroellchen@serenityos.org>
 * Copyright (c) 2022, the SerenityOS developers.
 *
 * SPDX-License-Identifier: BSD-2-Clause
 */

#pragma once

#include <AK/Forward.h>
#include <AK/Function.h>
#include <AK/Noncopyable.h>
#include <AK/NonnullOwnPtr.h>
#include <AK/Time.h>
#include <LibCore/Event.h>
#include <LibCore/Forward.h>

namespace Core {

// The event loop enables asynchronous (not parallel or multi-threaded) computing by efficiently handling events from various sources.
// Event loops are most important for GUI programs, where the various GUI updates and action callbacks run on the EventLoop,
// as well as services, where asynchronous remote procedure calls of multiple clients are handled.
// Event loops, through select(), allow programs to "go to sleep" for most of their runtime until some event happens.
// EventLoop is too expensive to use in realtime scenarios (read: audio) where even the time required by a single select() system call is too large and unpredictable.
//
// There is at most one running event loop per thread.
// Another event loop can be started while another event loop is already running; that new event loop will take over for the other event loop.
// This is mainly used in LibGUI, where each modal window stacks another event loop until it is closed.
// However, that means you need to be careful with storing the current event loop, as it might already be gone at the time of use.
// Event loops currently handle these kinds of events:
// - Deferred invocations caused by various objects. These are just a generic way of telling the EventLoop to run some function as soon as possible at a later point.
// - Timers, which repeatedly (or once after a delay) run a function on the EventLoop. Note that timers are not super accurate.
// - Filesystem notifications, i.e. whenever a file is read from, written to, etc.
// - POSIX signals, which allow the event loop to act as a signal handler and dispatch those signals in a more user-friendly way.
// - Fork events, because the child process event loop needs to clear its events and handlers.
// - Quit events, i.e. the event loop should exit.
// Any event that the event loop needs to wait on or needs to repeatedly handle is stored in a handle, e.g. s_timers.
//
// EventLoop has one final responsibility: Handling the InspectorServer connection and processing requests to the Object hierarchy.
class EventLoop {
    friend struct EventLoopPusher;

public:
    enum class MakeInspectable {
        No,
        Yes,
    };

    enum class WaitMode {
        WaitForEvents,
        PollForEvents,
    };

    explicit EventLoop(MakeInspectable = MakeInspectable::No);
    ~EventLoop();

    static void initialize_wake_pipes();
    static bool has_been_instantiated();

    // Pump the event loop until its exit is requested.
    int exec();

    // Process events, generally called by exec() in a loop.
    // This should really only be used for integrating with other event loops.
    // The wait mode determines whether pump() uses select() to wait for the next event.
    size_t pump(WaitMode = WaitMode::WaitForEvents);

    // Pump the event loop until some condition is met.
    void spin_until(Function<bool()>);

    // Post an event to this event loop.
    void post_event(Object& receiver, NonnullOwnPtr<Event>&&);

    void add_job(NonnullRefPtr<Promise<NonnullRefPtr<Object>>> job_promise);

    void deferred_invoke(Function<void()>);

    void wake();

    void quit(int);
    void unquit();
    bool was_exit_requested() const { return m_exit_requested; }

    // The registration functions act upon the current loop of the current thread.
    static int register_timer(Object&, int milliseconds, bool should_reload, TimerShouldFireWhenNotVisible);
    static bool unregister_timer(int timer_id);

    static void register_notifier(Badge<Notifier>, Notifier&);
    static void unregister_notifier(Badge<Notifier>, Notifier&);

    static int register_signal(int signo, Function<void(int)> handler);
    static void unregister_signal(int handler_id);

    // Note: Boost uses Parent/Child/Prepare, but we don't really have anything
    //       interesting to do in the parent or before forking.
    enum class ForkEvent {
        Child,
    };
    static void notify_forked(ForkEvent);

    static EventLoop& current();

private:
    void wait_for_event(WaitMode);
    Optional<Time> get_next_timer_expiration();
    static void dispatch_signal(int);
    static void handle_signal(int);

    static pid_t s_pid;

    bool m_exit_requested { false };
    int m_exit_code { 0 };

    static thread_local int s_wake_pipe_fds[2];
    static thread_local bool s_wake_pipe_initialized;

    // The wake pipe of this event loop needs to be accessible from other threads.
    int (*m_wake_pipe_fds)[2];

    struct Private;
    NonnullOwnPtr<Private> m_private;
};

void deferred_invoke(Function<void()>);

}