bin: compile lttng-sessiond as C++
[lttng-tools.git] / src / bin / lttng-sessiond / timer.cpp
diff --git a/src/bin/lttng-sessiond/timer.cpp b/src/bin/lttng-sessiond/timer.cpp
new file mode 100644 (file)
index 0000000..5a1f35d
--- /dev/null
@@ -0,0 +1,431 @@
+/*
+ * Copyright (C) 2017 Julien Desfossez <jdesfossez@efficios.com>
+ * Copyright (C) 2018 Jérémie Galarneau <jeremie.galarneau@efficios.com>
+ *
+ * SPDX-License-Identifier: GPL-2.0-only
+ *
+ */
+
+#define _LGPL_SOURCE
+#include <inttypes.h>
+#include <signal.h>
+
+#include "timer.h"
+#include "health-sessiond.h"
+#include "rotation-thread.h"
+#include "thread.h"
+
+#define LTTNG_SESSIOND_SIG_QS                          SIGRTMIN + 10
+#define LTTNG_SESSIOND_SIG_EXIT                                SIGRTMIN + 11
+#define LTTNG_SESSIOND_SIG_PENDING_ROTATION_CHECK      SIGRTMIN + 12
+#define LTTNG_SESSIOND_SIG_SCHEDULED_ROTATION          SIGRTMIN + 13
+
+#define UINT_TO_PTR(value)                             \
+       ({                                              \
+               LTTNG_ASSERT(value <= UINTPTR_MAX);             \
+               (void *) (uintptr_t) value;             \
+       })
+#define PTR_TO_UINT(ptr) ((uintptr_t) ptr)
+
+/*
+ * Handle timer teardown race wrt memory free of private data by sessiond
+ * signals are handled by a single thread, which permits a synchronization
+ * point between handling of each signal. Internal lock ensures mutual
+ * exclusion.
+ */
+static
+struct timer_signal_data {
+       /* Thread managing signals. */
+       pthread_t tid;
+       int qs_done;
+       pthread_mutex_t lock;
+} timer_signal = {
+       .tid = 0,
+       .qs_done = 0,
+       .lock = PTHREAD_MUTEX_INITIALIZER,
+};
+
+/*
+ * Set custom signal mask to current thread.
+ */
+static
+void setmask(sigset_t *mask)
+{
+       int ret;
+
+       ret = sigemptyset(mask);
+       if (ret) {
+               PERROR("sigemptyset");
+       }
+       ret = sigaddset(mask, LTTNG_SESSIOND_SIG_QS);
+       if (ret) {
+               PERROR("sigaddset teardown");
+       }
+       ret = sigaddset(mask, LTTNG_SESSIOND_SIG_EXIT);
+       if (ret) {
+               PERROR("sigaddset exit");
+       }
+       ret = sigaddset(mask, LTTNG_SESSIOND_SIG_PENDING_ROTATION_CHECK);
+       if (ret) {
+               PERROR("sigaddset pending rotation check");
+       }
+       ret = sigaddset(mask, LTTNG_SESSIOND_SIG_SCHEDULED_ROTATION);
+       if (ret) {
+               PERROR("sigaddset scheduled rotation");
+       }
+}
+
+/*
+ * This is the same function as timer_signal_thread_qs, when it
+ * returns, it means that no timer signr is currently pending or being handled
+ * by the timer thread. This cannot be called from the timer thread.
+ */
+static
+void timer_signal_thread_qs(unsigned int signr)
+{
+       sigset_t pending_set;
+       int ret;
+
+       /*
+        * We need to be the only thread interacting with the thread
+        * that manages signals for teardown synchronization.
+        */
+       pthread_mutex_lock(&timer_signal.lock);
+
+       /* Ensure we don't have any signal queued for this session. */
+       for (;;) {
+               ret = sigemptyset(&pending_set);
+               if (ret == -1) {
+                       PERROR("sigemptyset");
+               }
+               ret = sigpending(&pending_set);
+               if (ret == -1) {
+                       PERROR("sigpending");
+               }
+               if (!sigismember(&pending_set, signr)) {
+                       break;
+               }
+               caa_cpu_relax();
+       }
+
+       /*
+        * From this point, no new signal handler will be fired that would try to
+        * access "session". However, we still need to wait for any currently
+        * executing handler to complete.
+        */
+       cmm_smp_mb();
+       CMM_STORE_SHARED(timer_signal.qs_done, 0);
+       cmm_smp_mb();
+
+       /*
+        * Kill with LTTNG_SESSIOND_SIG_QS, so signal management thread
+        * wakes up.
+        */
+       kill(getpid(), LTTNG_SESSIOND_SIG_QS);
+
+       while (!CMM_LOAD_SHARED(timer_signal.qs_done)) {
+               caa_cpu_relax();
+       }
+       cmm_smp_mb();
+
+       pthread_mutex_unlock(&timer_signal.lock);
+}
+
+/*
+ * Start a timer on a session that will fire at a given interval
+ * (timer_interval_us) and fire a given signal (signal).
+ *
+ * Returns a negative value on error, 0 if a timer was created, and
+ * a positive value if no timer was created (not an error).
+ */
+static
+int timer_start(timer_t *timer_id, struct ltt_session *session,
+               unsigned int timer_interval_us, int signal, bool one_shot)
+{
+       int ret = 0, delete_ret;
+       struct sigevent sev = {};
+       struct itimerspec its;
+
+       sev.sigev_notify = SIGEV_SIGNAL;
+       sev.sigev_signo = signal;
+       sev.sigev_value.sival_ptr = session;
+       ret = timer_create(CLOCK_MONOTONIC, &sev, timer_id);
+       if (ret == -1) {
+               PERROR("timer_create");
+               goto end;
+       }
+
+       its.it_value.tv_sec = timer_interval_us / 1000000;
+       its.it_value.tv_nsec = (timer_interval_us % 1000000) * 1000;
+       if (one_shot) {
+               its.it_interval.tv_sec = 0;
+               its.it_interval.tv_nsec = 0;
+       } else {
+               its.it_interval.tv_sec = its.it_value.tv_sec;
+               its.it_interval.tv_nsec = its.it_value.tv_nsec;
+       }
+
+       ret = timer_settime(*timer_id, 0, &its, NULL);
+       if (ret == -1) {
+               PERROR("timer_settime");
+               goto error_destroy_timer;
+       }
+       goto end;
+
+error_destroy_timer:
+       delete_ret = timer_delete(*timer_id);
+       if (delete_ret == -1) {
+               PERROR("timer_delete");
+       }
+
+end:
+       return ret;
+}
+
+static
+int timer_stop(timer_t *timer_id, int signal)
+{
+       int ret = 0;
+
+       ret = timer_delete(*timer_id);
+       if (ret == -1) {
+               PERROR("timer_delete");
+               goto end;
+       }
+
+       timer_signal_thread_qs(signal);
+       *timer_id = 0;
+end:
+       return ret;
+}
+
+int timer_session_rotation_pending_check_start(struct ltt_session *session,
+               unsigned int interval_us)
+{
+       int ret;
+
+       if (!session_get(session)) {
+               ret = -1;
+               goto end;
+       }
+       DBG("Enabling session rotation pending check timer on session %" PRIu64,
+                       session->id);
+       /*
+        * We arm this timer in a one-shot mode so we don't have to disable it
+        * explicitly (which could deadlock if the timer thread is blocked
+        * writing in the rotation_timer_pipe).
+        *
+        * Instead, we re-arm it if needed after the rotation_pending check as
+        * returned. Also, this timer is usually only needed once, so there is
+        * no need to go through the whole signal teardown scheme everytime.
+        */
+       ret = timer_start(&session->rotation_pending_check_timer,
+                       session, interval_us,
+                       LTTNG_SESSIOND_SIG_PENDING_ROTATION_CHECK,
+                       /* one-shot */ true);
+       if (ret == 0) {
+               session->rotation_pending_check_timer_enabled = true;
+       }
+end:
+       return ret;
+}
+
+/*
+ * Call with session and session_list locks held.
+ */
+int timer_session_rotation_pending_check_stop(struct ltt_session *session)
+{
+       int ret;
+
+       LTTNG_ASSERT(session);
+       LTTNG_ASSERT(session->rotation_pending_check_timer_enabled);
+
+       DBG("Disabling session rotation pending check timer on session %" PRIu64,
+                       session->id);
+       ret = timer_stop(&session->rotation_pending_check_timer,
+                       LTTNG_SESSIOND_SIG_PENDING_ROTATION_CHECK);
+       if (ret == -1) {
+               ERR("Failed to stop rotate_pending_check timer");
+       } else {
+               session->rotation_pending_check_timer_enabled = false;
+               /*
+                * The timer's reference to the session can be released safely.
+                */
+               session_put(session);
+       }
+       return ret;
+}
+
+/*
+ * Call with session and session_list locks held.
+ */
+int timer_session_rotation_schedule_timer_start(struct ltt_session *session,
+               unsigned int interval_us)
+{
+       int ret;
+
+       if (!session_get(session)) {
+               ret = -1;
+               goto end;
+       }
+       DBG("Enabling scheduled rotation timer on session \"%s\" (%ui %s)", session->name,
+                       interval_us, USEC_UNIT);
+       ret = timer_start(&session->rotation_schedule_timer, session,
+                       interval_us, LTTNG_SESSIOND_SIG_SCHEDULED_ROTATION,
+                       /* one-shot */ false);
+       if (ret < 0) {
+               goto end;
+       }
+       session->rotation_schedule_timer_enabled = true;
+end:
+       return ret;
+}
+
+/*
+ * Call with session and session_list locks held.
+ */
+int timer_session_rotation_schedule_timer_stop(struct ltt_session *session)
+{
+       int ret = 0;
+
+       LTTNG_ASSERT(session);
+
+       if (!session->rotation_schedule_timer_enabled) {
+               goto end;
+       }
+
+       DBG("Disabling scheduled rotation timer on session %s", session->name);
+       ret = timer_stop(&session->rotation_schedule_timer,
+                       LTTNG_SESSIOND_SIG_SCHEDULED_ROTATION);
+       if (ret < 0) {
+               ERR("Failed to stop scheduled rotation timer of session \"%s\"",
+                               session->name);
+               goto end;
+       }
+
+       session->rotation_schedule_timer_enabled = false;
+       /* The timer's reference to the session can be released safely. */
+       session_put(session);
+       ret = 0;
+end:
+       return ret;
+}
+
+/*
+ * Block the RT signals for the entire process. It must be called from the
+ * sessiond main before creating the threads
+ */
+int timer_signal_init(void)
+{
+       int ret;
+       sigset_t mask;
+
+       /* Block signal for entire process, so only our thread processes it. */
+       setmask(&mask);
+       ret = pthread_sigmask(SIG_BLOCK, &mask, NULL);
+       if (ret) {
+               errno = ret;
+               PERROR("pthread_sigmask");
+               return -1;
+       }
+       return 0;
+}
+
+/*
+ * This thread is the sighandler for the timer signals.
+ */
+static
+void *thread_timer(void *data)
+{
+       int signr;
+       sigset_t mask;
+       siginfo_t info;
+       struct timer_thread_parameters *ctx = (timer_thread_parameters *) data;
+
+       rcu_register_thread();
+       rcu_thread_online();
+
+       health_register(the_health_sessiond, HEALTH_SESSIOND_TYPE_TIMER);
+       health_code_update();
+
+       /* Only self thread will receive signal mask. */
+       setmask(&mask);
+       CMM_STORE_SHARED(timer_signal.tid, pthread_self());
+
+       while (1) {
+               health_code_update();
+
+               health_poll_entry();
+               signr = sigwaitinfo(&mask, &info);
+               health_poll_exit();
+
+               /*
+                * NOTE: cascading conditions are used instead of a switch case
+                * since the use of SIGRTMIN in the definition of the signals'
+                * values prevents the reduction to an integer constant.
+                */
+               if (signr == -1) {
+                       if (errno != EINTR) {
+                               PERROR("sigwaitinfo");
+                       }
+                       continue;
+               } else if (signr == LTTNG_SESSIOND_SIG_QS) {
+                       cmm_smp_mb();
+                       CMM_STORE_SHARED(timer_signal.qs_done, 1);
+                       cmm_smp_mb();
+               } else if (signr == LTTNG_SESSIOND_SIG_EXIT) {
+                       goto end;
+               } else if (signr == LTTNG_SESSIOND_SIG_PENDING_ROTATION_CHECK) {
+                       struct ltt_session *session =
+                                       (struct ltt_session *) info.si_value.sival_ptr;
+
+                       rotation_thread_enqueue_job(ctx->rotation_thread_job_queue,
+                                       ROTATION_THREAD_JOB_TYPE_CHECK_PENDING_ROTATION,
+                                       session);
+               } else if (signr == LTTNG_SESSIOND_SIG_SCHEDULED_ROTATION) {
+                       rotation_thread_enqueue_job(ctx->rotation_thread_job_queue,
+                                       ROTATION_THREAD_JOB_TYPE_SCHEDULED_ROTATION,
+                                       (struct ltt_session *) info.si_value.sival_ptr);
+                       /*
+                        * The scheduled periodic rotation timer is not in
+                        * "one-shot" mode. The reference to the session is not
+                        * released since the timer is still enabled and can
+                        * still fire.
+                        */
+               } else {
+                       ERR("Unexpected signal %d", info.si_signo);
+               }
+       }
+
+end:
+       DBG("Thread exit");
+       health_unregister(the_health_sessiond);
+       rcu_thread_offline();
+       rcu_unregister_thread();
+       return NULL;
+}
+
+static
+bool shutdown_timer_thread(void *data)
+{
+       return kill(getpid(), LTTNG_SESSIOND_SIG_EXIT) == 0;
+}
+
+bool launch_timer_thread(
+               struct timer_thread_parameters *timer_thread_parameters)
+{
+       struct lttng_thread *thread;
+
+       thread = lttng_thread_create("Timer",
+                       thread_timer,
+                       shutdown_timer_thread,
+                       NULL,
+                       timer_thread_parameters);
+       if (!thread) {
+               goto error;
+       }
+       lttng_thread_put(thread);
+       return true;
+error:
+       return false;
+}
This page took 0.027665 seconds and 4 git commands to generate.