1   
//
1   
//
2   
// Copyright (c) 2026 Steve Gerbino
2   
// Copyright (c) 2026 Steve Gerbino
3   
//
3   
//
4   
// Distributed under the Boost Software License, Version 1.0. (See accompanying
4   
// Distributed under the Boost Software License, Version 1.0. (See accompanying
5   
// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
5   
// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
6   
//
6   
//
7   
// Official repository: https://github.com/cppalliance/corosio
7   
// Official repository: https://github.com/cppalliance/corosio
8   
//
8   
//
9  
9  
10   
#ifndef BOOST_COROSIO_NATIVE_DETAIL_SELECT_SELECT_SCHEDULER_HPP
10   
#ifndef BOOST_COROSIO_NATIVE_DETAIL_SELECT_SELECT_SCHEDULER_HPP
11   
#define BOOST_COROSIO_NATIVE_DETAIL_SELECT_SELECT_SCHEDULER_HPP
11   
#define BOOST_COROSIO_NATIVE_DETAIL_SELECT_SELECT_SCHEDULER_HPP
12  
12  
13   
#include <boost/corosio/detail/platform.hpp>
13   
#include <boost/corosio/detail/platform.hpp>
14  
14  
15   
#if BOOST_COROSIO_HAS_SELECT
15   
#if BOOST_COROSIO_HAS_SELECT
16  
16  
17   
#include <boost/corosio/detail/config.hpp>
17   
#include <boost/corosio/detail/config.hpp>
18   
#include <boost/capy/ex/execution_context.hpp>
18   
#include <boost/capy/ex/execution_context.hpp>
19  
19  
20   
#include <boost/corosio/native/detail/reactor/reactor_scheduler.hpp>
20   
#include <boost/corosio/native/detail/reactor/reactor_scheduler.hpp>
21  
21  
22   
#include <boost/corosio/native/detail/select/select_traits.hpp>
22   
#include <boost/corosio/native/detail/select/select_traits.hpp>
23   
#include <boost/corosio/detail/timer_service.hpp>
23   
#include <boost/corosio/detail/timer_service.hpp>
24   
#include <boost/corosio/native/detail/make_err.hpp>
24   
#include <boost/corosio/native/detail/make_err.hpp>
25   
#include <boost/corosio/native/detail/posix/posix_resolver_service.hpp>
25   
#include <boost/corosio/native/detail/posix/posix_resolver_service.hpp>
26   
#include <boost/corosio/native/detail/posix/posix_signal_service.hpp>
26   
#include <boost/corosio/native/detail/posix/posix_signal_service.hpp>
27   
#include <boost/corosio/native/detail/posix/posix_stream_file_service.hpp>
27   
#include <boost/corosio/native/detail/posix/posix_stream_file_service.hpp>
28   
#include <boost/corosio/native/detail/posix/posix_random_access_file_service.hpp>
28   
#include <boost/corosio/native/detail/posix/posix_random_access_file_service.hpp>
29  
29  
30   
#include <boost/corosio/detail/except.hpp>
30   
#include <boost/corosio/detail/except.hpp>
31  
31  
32   
#include <sys/select.h>
32   
#include <sys/select.h>
33   
#include <unistd.h>
33   
#include <unistd.h>
34   
#include <errno.h>
34   
#include <errno.h>
35   
#include <fcntl.h>
35   
#include <fcntl.h>
36  
36  
37   
#include <atomic>
37   
#include <atomic>
38   
#include <chrono>
38   
#include <chrono>
39   
#include <cstdint>
39   
#include <cstdint>
40   
#include <limits>
40   
#include <limits>
41   
#include <mutex>
41   
#include <mutex>
42   
#include <unordered_map>
42   
#include <unordered_map>
43  
43  
44   
namespace boost::corosio::detail {
44   
namespace boost::corosio::detail {
45  
45  
46   
struct select_op;
46   
struct select_op;
47  
47  
48   
/** POSIX scheduler using select() for I/O multiplexing.
48   
/** POSIX scheduler using select() for I/O multiplexing.
49  
49  
50   
    This scheduler implements the scheduler interface using the POSIX select()
50   
    This scheduler implements the scheduler interface using the POSIX select()
51   
    call for I/O event notification. It inherits the shared reactor threading
51   
    call for I/O event notification. It inherits the shared reactor threading
52   
    model from reactor_scheduler: signal state machine, inline completion
52   
    model from reactor_scheduler: signal state machine, inline completion
53   
    budget, work counting, and the do_one event loop.
53   
    budget, work counting, and the do_one event loop.
54  
54  
55   
    The design mirrors epoll_scheduler for behavioral consistency:
55   
    The design mirrors epoll_scheduler for behavioral consistency:
56   
    - Same single-reactor thread coordination model
56   
    - Same single-reactor thread coordination model
57   
    - Same deferred I/O pattern (reactor marks ready; workers do I/O)
57   
    - Same deferred I/O pattern (reactor marks ready; workers do I/O)
58   
    - Same timer integration pattern
58   
    - Same timer integration pattern
59  
59  
60   
    Known Limitations:
60   
    Known Limitations:
61   
    - FD_SETSIZE (~1024) limits maximum concurrent connections
61   
    - FD_SETSIZE (~1024) limits maximum concurrent connections
62   
    - O(n) scanning: rebuilds fd_sets each iteration
62   
    - O(n) scanning: rebuilds fd_sets each iteration
63   
    - Level-triggered only (no edge-triggered mode)
63   
    - Level-triggered only (no edge-triggered mode)
64  
64  
65   
    @par Thread Safety
65   
    @par Thread Safety
66   
    All public member functions are thread-safe.
66   
    All public member functions are thread-safe.
67   
*/
67   
*/
68   
class BOOST_COROSIO_DECL select_scheduler final : public reactor_scheduler
68   
class BOOST_COROSIO_DECL select_scheduler final : public reactor_scheduler
69   
{
69   
{
70   
public:
70   
public:
71   
    /** Construct the scheduler.
71   
    /** Construct the scheduler.
72  
72  
73   
        Creates a self-pipe for reactor interruption.
73   
        Creates a self-pipe for reactor interruption.
74  
74  
75   
        @param ctx Reference to the owning execution_context.
75   
        @param ctx Reference to the owning execution_context.
76   
        @param concurrency_hint Hint for expected thread count (unused).
76   
        @param concurrency_hint Hint for expected thread count (unused).
77   
    */
77   
    */
78   
    select_scheduler(capy::execution_context& ctx, int concurrency_hint = -1);
78   
    select_scheduler(capy::execution_context& ctx, int concurrency_hint = -1);
79  
79  
80   
    /// Destroy the scheduler.
80   
    /// Destroy the scheduler.
81   
    ~select_scheduler() override;
81   
    ~select_scheduler() override;
82  
82  
83   
    select_scheduler(select_scheduler const&)            = delete;
83   
    select_scheduler(select_scheduler const&)            = delete;
84   
    select_scheduler& operator=(select_scheduler const&) = delete;
84   
    select_scheduler& operator=(select_scheduler const&) = delete;
85  
85  
86   
    /// Shut down the scheduler, draining pending operations.
86   
    /// Shut down the scheduler, draining pending operations.
87   
    void shutdown() override;
87   
    void shutdown() override;
88  
88  
89   
    /** Return the maximum file descriptor value supported.
89   
    /** Return the maximum file descriptor value supported.
90  
90  
91   
        Returns FD_SETSIZE - 1, the maximum fd value that can be
91   
        Returns FD_SETSIZE - 1, the maximum fd value that can be
92   
        monitored by select(). Operations with fd >= FD_SETSIZE
92   
        monitored by select(). Operations with fd >= FD_SETSIZE
93   
        will fail with EINVAL.
93   
        will fail with EINVAL.
94  
94  
95   
        @return The maximum supported file descriptor value.
95   
        @return The maximum supported file descriptor value.
96   
    */
96   
    */
97   
    static constexpr int max_fd() noexcept
97   
    static constexpr int max_fd() noexcept
98   
    {
98   
    {
99   
        return FD_SETSIZE - 1;
99   
        return FD_SETSIZE - 1;
100   
    }
100   
    }
101  
101  
102   
    /** Register a descriptor for persistent monitoring.
102   
    /** Register a descriptor for persistent monitoring.
103  
103  
104   
        The fd is added to the registered_descs_ map and will be
104   
        The fd is added to the registered_descs_ map and will be
105   
        included in subsequent select() calls. The reactor is
105   
        included in subsequent select() calls. The reactor is
106   
        interrupted so a blocked select() rebuilds its fd_sets.
106   
        interrupted so a blocked select() rebuilds its fd_sets.
107  
107  
108   
        @param fd The file descriptor to register.
108   
        @param fd The file descriptor to register.
109   
        @param desc Pointer to descriptor state for this fd.
109   
        @param desc Pointer to descriptor state for this fd.
110   
    */
110   
    */
111   
    void register_descriptor(int fd, reactor_descriptor_state* desc) const;
111   
    void register_descriptor(int fd, reactor_descriptor_state* desc) const;
112  
112  
113   
    /** Deregister a persistently registered descriptor.
113   
    /** Deregister a persistently registered descriptor.
114  
114  
115   
        @param fd The file descriptor to deregister.
115   
        @param fd The file descriptor to deregister.
116   
    */
116   
    */
117   
    void deregister_descriptor(int fd) const;
117   
    void deregister_descriptor(int fd) const;
118  
118  
119   
    /** Interrupt the reactor so it rebuilds its fd_sets.
119   
    /** Interrupt the reactor so it rebuilds its fd_sets.
120  
120  
121   
        Called when a write or connect op is registered after
121   
        Called when a write or connect op is registered after
122   
        the reactor's snapshot was taken. Without this, select()
122   
        the reactor's snapshot was taken. Without this, select()
123   
        may block not watching for writability on the fd.
123   
        may block not watching for writability on the fd.
124   
    */
124   
    */
125   
    void notify_reactor() const;
125   
    void notify_reactor() const;
126  
126  
127   
private:
127   
private:
128   
    void
128   
    void
129   
    run_task(lock_type& lock, context_type* ctx,
129   
    run_task(lock_type& lock, context_type* ctx,
130   
        long timeout_us) override;
130   
        long timeout_us) override;
131   
    void interrupt_reactor() const override;
131   
    void interrupt_reactor() const override;
132   
    long calculate_timeout(long requested_timeout_us) const;
132   
    long calculate_timeout(long requested_timeout_us) const;
133  
133  
134   
    // Self-pipe for interrupting select()
134   
    // Self-pipe for interrupting select()
135   
    int pipe_fds_[2]; // [0]=read, [1]=write
135   
    int pipe_fds_[2]; // [0]=read, [1]=write
136  
136  
137   
    // Per-fd tracking for fd_set building
137   
    // Per-fd tracking for fd_set building
138   
    mutable std::unordered_map<int, reactor_descriptor_state*> registered_descs_;
138   
    mutable std::unordered_map<int, reactor_descriptor_state*> registered_descs_;
139   
    mutable int max_fd_ = -1;
139   
    mutable int max_fd_ = -1;
140   
};
140   
};
141  
141  
142   
inline select_scheduler::select_scheduler(capy::execution_context& ctx, int)
142   
inline select_scheduler::select_scheduler(capy::execution_context& ctx, int)
143   
    : pipe_fds_{-1, -1}
143   
    : pipe_fds_{-1, -1}
144   
    , max_fd_(-1)
144   
    , max_fd_(-1)
145   
{
145   
{
146   
    if (::pipe(pipe_fds_) < 0)
146   
    if (::pipe(pipe_fds_) < 0)
147   
        detail::throw_system_error(make_err(errno), "pipe");
147   
        detail::throw_system_error(make_err(errno), "pipe");
148  
148  
149   
    for (int i = 0; i < 2; ++i)
149   
    for (int i = 0; i < 2; ++i)
150   
    {
150   
    {
151   
        int flags = ::fcntl(pipe_fds_[i], F_GETFL, 0);
151   
        int flags = ::fcntl(pipe_fds_[i], F_GETFL, 0);
152   
        if (flags == -1)
152   
        if (flags == -1)
153   
        {
153   
        {
154   
            int errn = errno;
154   
            int errn = errno;
155   
            ::close(pipe_fds_[0]);
155   
            ::close(pipe_fds_[0]);
156   
            ::close(pipe_fds_[1]);
156   
            ::close(pipe_fds_[1]);
157   
            detail::throw_system_error(make_err(errn), "fcntl F_GETFL");
157   
            detail::throw_system_error(make_err(errn), "fcntl F_GETFL");
158   
        }
158   
        }
159   
        if (::fcntl(pipe_fds_[i], F_SETFL, flags | O_NONBLOCK) == -1)
159   
        if (::fcntl(pipe_fds_[i], F_SETFL, flags | O_NONBLOCK) == -1)
160   
        {
160   
        {
161   
            int errn = errno;
161   
            int errn = errno;
162   
            ::close(pipe_fds_[0]);
162   
            ::close(pipe_fds_[0]);
163   
            ::close(pipe_fds_[1]);
163   
            ::close(pipe_fds_[1]);
164   
            detail::throw_system_error(make_err(errn), "fcntl F_SETFL");
164   
            detail::throw_system_error(make_err(errn), "fcntl F_SETFL");
165   
        }
165   
        }
166   
        if (::fcntl(pipe_fds_[i], F_SETFD, FD_CLOEXEC) == -1)
166   
        if (::fcntl(pipe_fds_[i], F_SETFD, FD_CLOEXEC) == -1)
167   
        {
167   
        {
168   
            int errn = errno;
168   
            int errn = errno;
169   
            ::close(pipe_fds_[0]);
169   
            ::close(pipe_fds_[0]);
170   
            ::close(pipe_fds_[1]);
170   
            ::close(pipe_fds_[1]);
171   
            detail::throw_system_error(make_err(errn), "fcntl F_SETFD");
171   
            detail::throw_system_error(make_err(errn), "fcntl F_SETFD");
172   
        }
172   
        }
173   
    }
173   
    }
174  
174  
175   
    timer_svc_ = &get_timer_service(ctx, *this);
175   
    timer_svc_ = &get_timer_service(ctx, *this);
176   
    timer_svc_->set_on_earliest_changed(
176   
    timer_svc_->set_on_earliest_changed(
177   
        timer_service::callback(this, [](void* p) {
177   
        timer_service::callback(this, [](void* p) {
178   
            static_cast<select_scheduler*>(p)->interrupt_reactor();
178   
            static_cast<select_scheduler*>(p)->interrupt_reactor();
179   
        }));
179   
        }));
180  
180  
181   
    get_resolver_service(ctx, *this);
181   
    get_resolver_service(ctx, *this);
182   
    get_signal_service(ctx, *this);
182   
    get_signal_service(ctx, *this);
183   
    get_stream_file_service(ctx, *this);
183   
    get_stream_file_service(ctx, *this);
184   
    get_random_access_file_service(ctx, *this);
184   
    get_random_access_file_service(ctx, *this);
185  
185  
186   
    completed_ops_.push(&task_op_);
186   
    completed_ops_.push(&task_op_);
187   
}
187   
}
188  
188  
189   
inline select_scheduler::~select_scheduler()
189   
inline select_scheduler::~select_scheduler()
190   
{
190   
{
191   
    if (pipe_fds_[0] >= 0)
191   
    if (pipe_fds_[0] >= 0)
192   
        ::close(pipe_fds_[0]);
192   
        ::close(pipe_fds_[0]);
193   
    if (pipe_fds_[1] >= 0)
193   
    if (pipe_fds_[1] >= 0)
194   
        ::close(pipe_fds_[1]);
194   
        ::close(pipe_fds_[1]);
195   
}
195   
}
196  
196  
197   
inline void
197   
inline void
198   
select_scheduler::shutdown()
198   
select_scheduler::shutdown()
199   
{
199   
{
200   
    shutdown_drain();
200   
    shutdown_drain();
201  
201  
202   
    if (pipe_fds_[1] >= 0)
202   
    if (pipe_fds_[1] >= 0)
203   
        interrupt_reactor();
203   
        interrupt_reactor();
204   
}
204   
}
205  
205  
206   
inline void
206   
inline void
207   
select_scheduler::register_descriptor(
207   
select_scheduler::register_descriptor(
208   
    int fd, reactor_descriptor_state* desc) const
208   
    int fd, reactor_descriptor_state* desc) const
209   
{
209   
{
210   
    if (fd < 0 || fd >= FD_SETSIZE)
210   
    if (fd < 0 || fd >= FD_SETSIZE)
211   
        detail::throw_system_error(make_err(EINVAL), "select: fd out of range");
211   
        detail::throw_system_error(make_err(EINVAL), "select: fd out of range");
212  
212  
213   
    desc->registered_events = reactor_event_read | reactor_event_write;
213   
    desc->registered_events = reactor_event_read | reactor_event_write;
214   
    desc->fd                = fd;
214   
    desc->fd                = fd;
215   
    desc->scheduler_        = this;
215   
    desc->scheduler_        = this;
216   
    desc->mutex.set_enabled(!single_threaded_);
216   
    desc->mutex.set_enabled(!single_threaded_);
217   
    desc->ready_events_.store(0, std::memory_order_relaxed);
217   
    desc->ready_events_.store(0, std::memory_order_relaxed);
218  
218  
219   
    {
219   
    {
220   
        conditionally_enabled_mutex::scoped_lock lock(desc->mutex);
220   
        conditionally_enabled_mutex::scoped_lock lock(desc->mutex);
221   
        desc->impl_ref_.reset();
221   
        desc->impl_ref_.reset();
222   
        desc->read_ready  = false;
222   
        desc->read_ready  = false;
223   
        desc->write_ready = false;
223   
        desc->write_ready = false;
224   
    }
224   
    }
225  
225  
226   
    {
226   
    {
227   
        mutex_type::scoped_lock lock(mutex_);
227   
        mutex_type::scoped_lock lock(mutex_);
228   
        registered_descs_[fd] = desc;
228   
        registered_descs_[fd] = desc;
229   
        if (fd > max_fd_)
229   
        if (fd > max_fd_)
230   
            max_fd_ = fd;
230   
            max_fd_ = fd;
231   
    }
231   
    }
232  
232  
233   
    interrupt_reactor();
233   
    interrupt_reactor();
234   
}
234   
}
235  
235  
236   
inline void
236   
inline void
237   
select_scheduler::deregister_descriptor(int fd) const
237   
select_scheduler::deregister_descriptor(int fd) const
238   
{
238   
{
239   
    mutex_type::scoped_lock lock(mutex_);
239   
    mutex_type::scoped_lock lock(mutex_);
240  
240  
241   
    auto it = registered_descs_.find(fd);
241   
    auto it = registered_descs_.find(fd);
242   
    if (it == registered_descs_.end())
242   
    if (it == registered_descs_.end())
243   
        return;
243   
        return;
244  
244  
245   
    registered_descs_.erase(it);
245   
    registered_descs_.erase(it);
246  
246  
247   
    if (fd == max_fd_)
247   
    if (fd == max_fd_)
248   
    {
248   
    {
249   
        max_fd_ = pipe_fds_[0];
249   
        max_fd_ = pipe_fds_[0];
250   
        for (auto& [registered_fd, state] : registered_descs_)
250   
        for (auto& [registered_fd, state] : registered_descs_)
251   
        {
251   
        {
252   
            if (registered_fd > max_fd_)
252   
            if (registered_fd > max_fd_)
253   
                max_fd_ = registered_fd;
253   
                max_fd_ = registered_fd;
254   
        }
254   
        }
255   
    }
255   
    }
256   
}
256   
}
257  
257  
258   
inline void
258   
inline void
259   
select_scheduler::notify_reactor() const
259   
select_scheduler::notify_reactor() const
260   
{
260   
{
261   
    interrupt_reactor();
261   
    interrupt_reactor();
262   
}
262   
}
263  
263  
264   
inline void
264   
inline void
265   
select_scheduler::interrupt_reactor() const
265   
select_scheduler::interrupt_reactor() const
266   
{
266   
{
267   
    char byte               = 1;
267   
    char byte               = 1;
268   
    [[maybe_unused]] auto r = ::write(pipe_fds_[1], &byte, 1);
268   
    [[maybe_unused]] auto r = ::write(pipe_fds_[1], &byte, 1);
269   
}
269   
}
270  
270  
271   
inline long
271   
inline long
272   
select_scheduler::calculate_timeout(long requested_timeout_us) const
272   
select_scheduler::calculate_timeout(long requested_timeout_us) const
273   
{
273   
{
274   
    if (requested_timeout_us == 0)
274   
    if (requested_timeout_us == 0)
275   
        return 0;
275   
        return 0;
276  
276  
277   
    auto nearest = timer_svc_->nearest_expiry();
277   
    auto nearest = timer_svc_->nearest_expiry();
278   
    if (nearest == timer_service::time_point::max())
278   
    if (nearest == timer_service::time_point::max())
279   
        return requested_timeout_us;
279   
        return requested_timeout_us;
280  
280  
281   
    auto now = std::chrono::steady_clock::now();
281   
    auto now = std::chrono::steady_clock::now();
282   
    if (nearest <= now)
282   
    if (nearest <= now)
283   
        return 0;
283   
        return 0;
284  
284  
285   
    auto timer_timeout_us =
285   
    auto timer_timeout_us =
286   
        std::chrono::duration_cast<std::chrono::microseconds>(nearest - now)
286   
        std::chrono::duration_cast<std::chrono::microseconds>(nearest - now)
287   
            .count();
287   
            .count();
288  
288  
289   
    constexpr auto long_max =
289   
    constexpr auto long_max =
290   
        static_cast<long long>((std::numeric_limits<long>::max)());
290   
        static_cast<long long>((std::numeric_limits<long>::max)());
291   
    auto capped_timer_us =
291   
    auto capped_timer_us =
292   
        (std::min)((std::max)(static_cast<long long>(timer_timeout_us),
292   
        (std::min)((std::max)(static_cast<long long>(timer_timeout_us),
293   
                              static_cast<long long>(0)),
293   
                              static_cast<long long>(0)),
294   
                   long_max);
294   
                   long_max);
295  
295  
296   
    if (requested_timeout_us < 0)
296   
    if (requested_timeout_us < 0)
297   
        return static_cast<long>(capped_timer_us);
297   
        return static_cast<long>(capped_timer_us);
298  
298  
299   
    return static_cast<long>(
299   
    return static_cast<long>(
300   
        (std::min)(static_cast<long long>(requested_timeout_us),
300   
        (std::min)(static_cast<long long>(requested_timeout_us),
301   
                   capped_timer_us));
301   
                   capped_timer_us));
302   
}
302   
}
303  
303  
304   
inline void
304   
inline void
305   
select_scheduler::run_task(
305   
select_scheduler::run_task(
306   
    lock_type& lock, context_type* ctx, long timeout_us)
306   
    lock_type& lock, context_type* ctx, long timeout_us)
307   
{
307   
{
308   
    long effective_timeout_us =
308   
    long effective_timeout_us =
309   
        task_interrupted_ ? 0 : calculate_timeout(timeout_us);
309   
        task_interrupted_ ? 0 : calculate_timeout(timeout_us);
310  
310  
311   
    // Snapshot registered descriptors while holding lock.
311   
    // Snapshot registered descriptors while holding lock.
312   
    // Record which fds need write monitoring to avoid a hot loop:
312   
    // Record which fds need write monitoring to avoid a hot loop:
313   
    // select is level-triggered so writable sockets (nearly always
313   
    // select is level-triggered so writable sockets (nearly always
314   
    // writable) would cause select() to return immediately every
314   
    // writable) would cause select() to return immediately every
315   
    // iteration if unconditionally added to write_fds.
315   
    // iteration if unconditionally added to write_fds.
316   
    struct fd_entry
316   
    struct fd_entry
317   
    {
317   
    {
318   
        int fd;
318   
        int fd;
319   
        reactor_descriptor_state* desc;
319   
        reactor_descriptor_state* desc;
320   
        bool needs_write;
320   
        bool needs_write;
321   
    };
321   
    };
322   
    fd_entry snapshot[FD_SETSIZE];
322   
    fd_entry snapshot[FD_SETSIZE];
323   
    int snapshot_count = 0;
323   
    int snapshot_count = 0;
324  
324  
325   
    for (auto& [fd, desc] : registered_descs_)
325   
    for (auto& [fd, desc] : registered_descs_)
326   
    {
326   
    {
327   
        if (snapshot_count < FD_SETSIZE)
327   
        if (snapshot_count < FD_SETSIZE)
328   
        {
328   
        {
329   
            conditionally_enabled_mutex::scoped_lock desc_lock(desc->mutex);
329   
            conditionally_enabled_mutex::scoped_lock desc_lock(desc->mutex);
330   
            snapshot[snapshot_count].fd   = fd;
330   
            snapshot[snapshot_count].fd   = fd;
331   
            snapshot[snapshot_count].desc = desc;
331   
            snapshot[snapshot_count].desc = desc;
332   
            snapshot[snapshot_count].needs_write =
332   
            snapshot[snapshot_count].needs_write =
333   
                (desc->write_op || desc->connect_op);
333   
                (desc->write_op || desc->connect_op);
334   
            ++snapshot_count;
334   
            ++snapshot_count;
335   
        }
335   
        }
336   
    }
336   
    }
337  
337  
338   
    if (lock.owns_lock())
338   
    if (lock.owns_lock())
339   
        lock.unlock();
339   
        lock.unlock();
340  
340  
341   
    task_cleanup on_exit{this, &lock, ctx};
341   
    task_cleanup on_exit{this, &lock, ctx};
342  
342  
343   
    fd_set read_fds, write_fds, except_fds;
343   
    fd_set read_fds, write_fds, except_fds;
344   
    FD_ZERO(&read_fds);
344   
    FD_ZERO(&read_fds);
345   
    FD_ZERO(&write_fds);
345   
    FD_ZERO(&write_fds);
346   
    FD_ZERO(&except_fds);
346   
    FD_ZERO(&except_fds);
347  
347  
348   
    FD_SET(pipe_fds_[0], &read_fds);
348   
    FD_SET(pipe_fds_[0], &read_fds);
349   
    int nfds = pipe_fds_[0];
349   
    int nfds = pipe_fds_[0];
350  
350  
351   
    for (int i = 0; i < snapshot_count; ++i)
351   
    for (int i = 0; i < snapshot_count; ++i)
352   
    {
352   
    {
353   
        int fd = snapshot[i].fd;
353   
        int fd = snapshot[i].fd;
354   
        FD_SET(fd, &read_fds);
354   
        FD_SET(fd, &read_fds);
355   
        if (snapshot[i].needs_write)
355   
        if (snapshot[i].needs_write)
356   
            FD_SET(fd, &write_fds);
356   
            FD_SET(fd, &write_fds);
357   
        FD_SET(fd, &except_fds);
357   
        FD_SET(fd, &except_fds);
358   
        if (fd > nfds)
358   
        if (fd > nfds)
359   
            nfds = fd;
359   
            nfds = fd;
360   
    }
360   
    }
361  
361  
362   
    struct timeval tv;
362   
    struct timeval tv;
363   
    struct timeval* tv_ptr = nullptr;
363   
    struct timeval* tv_ptr = nullptr;
364   
    if (effective_timeout_us >= 0)
364   
    if (effective_timeout_us >= 0)
365   
    {
365   
    {
366   
        tv.tv_sec  = effective_timeout_us / 1000000;
366   
        tv.tv_sec  = effective_timeout_us / 1000000;
367   
        tv.tv_usec = effective_timeout_us % 1000000;
367   
        tv.tv_usec = effective_timeout_us % 1000000;
368   
        tv_ptr     = &tv;
368   
        tv_ptr     = &tv;
369   
    }
369   
    }
370  
370  
371   
    int ready = ::select(nfds + 1, &read_fds, &write_fds, &except_fds, tv_ptr);
371   
    int ready = ::select(nfds + 1, &read_fds, &write_fds, &except_fds, tv_ptr);
372  
372  
373   
    // EINTR: signal interrupted select(), just retry.
373   
    // EINTR: signal interrupted select(), just retry.
374   
    // EBADF: an fd was closed between snapshot and select(); retry
374   
    // EBADF: an fd was closed between snapshot and select(); retry
375   
    // with a fresh snapshot from registered_descs_.
375   
    // with a fresh snapshot from registered_descs_.
376   
    if (ready < 0)
376   
    if (ready < 0)
377   
    {
377   
    {
378   
        if (errno == EINTR || errno == EBADF)
378   
        if (errno == EINTR || errno == EBADF)
379   
            return;
379   
            return;
380   
        detail::throw_system_error(make_err(errno), "select");
380   
        detail::throw_system_error(make_err(errno), "select");
381   
    }
381   
    }
382  
382  
383   
    // Process timers outside the lock
383   
    // Process timers outside the lock
384   
    timer_svc_->process_expired();
384   
    timer_svc_->process_expired();
385  
385  
386   
    op_queue local_ops;
386   
    op_queue local_ops;
387  
387  
388   
    if (ready > 0)
388   
    if (ready > 0)
389   
    {
389   
    {
390   
        if (FD_ISSET(pipe_fds_[0], &read_fds))
390   
        if (FD_ISSET(pipe_fds_[0], &read_fds))
391   
        {
391   
        {
392   
            char buf[256];
392   
            char buf[256];
393   
            while (::read(pipe_fds_[0], buf, sizeof(buf)) > 0)
393   
            while (::read(pipe_fds_[0], buf, sizeof(buf)) > 0)
394   
            {
394   
            {
395   
            }
395   
            }
396   
        }
396   
        }
397  
397  
398   
        for (int i = 0; i < snapshot_count; ++i)
398   
        for (int i = 0; i < snapshot_count; ++i)
399   
        {
399   
        {
400   
            int fd                        = snapshot[i].fd;
400   
            int fd                        = snapshot[i].fd;
401   
            reactor_descriptor_state* desc = snapshot[i].desc;
401   
            reactor_descriptor_state* desc = snapshot[i].desc;
402  
402  
403   
            std::uint32_t flags = 0;
403   
            std::uint32_t flags = 0;
404   
            if (FD_ISSET(fd, &read_fds))
404   
            if (FD_ISSET(fd, &read_fds))
405   
                flags |= reactor_event_read;
405   
                flags |= reactor_event_read;
406   
            if (FD_ISSET(fd, &write_fds))
406   
            if (FD_ISSET(fd, &write_fds))
407   
                flags |= reactor_event_write;
407   
                flags |= reactor_event_write;
408   
            if (FD_ISSET(fd, &except_fds))
408   
            if (FD_ISSET(fd, &except_fds))
409   
                flags |= reactor_event_error;
409   
                flags |= reactor_event_error;
410  
410  
411   
            if (flags == 0)
411   
            if (flags == 0)
412   
                continue;
412   
                continue;
413  
413  
414   
            desc->add_ready_events(flags);
414   
            desc->add_ready_events(flags);
415  
415  
416   
            bool expected = false;
416   
            bool expected = false;
417   
            if (desc->is_enqueued_.compare_exchange_strong(
417   
            if (desc->is_enqueued_.compare_exchange_strong(
418   
                    expected, true, std::memory_order_release,
418   
                    expected, true, std::memory_order_release,
419   
                    std::memory_order_relaxed))
419   
                    std::memory_order_relaxed))
420   
            {
420   
            {
421   
                local_ops.push(desc);
421   
                local_ops.push(desc);
422   
            }
422   
            }
423   
        }
423   
        }
424   
    }
424   
    }
425  
425  
426   
    lock.lock();
426   
    lock.lock();
427  
427  
428   
    if (!local_ops.empty())
428   
    if (!local_ops.empty())
429   
        completed_ops_.splice(local_ops);
429   
        completed_ops_.splice(local_ops);
430   
}
430   
}
431  
431  
432   
} // namespace boost::corosio::detail
432   
} // namespace boost::corosio::detail
433  
433  
434   
#endif // BOOST_COROSIO_HAS_SELECT
434   
#endif // BOOST_COROSIO_HAS_SELECT
435  
435  
436   
#endif // BOOST_COROSIO_NATIVE_DETAIL_SELECT_SELECT_SCHEDULER_HPP
436   
#endif // BOOST_COROSIO_NATIVE_DETAIL_SELECT_SELECT_SCHEDULER_HPP