/***************************************************************************** Copyright (c) 1995, 2015, Oracle and/or its affiliates. All Rights Reserved. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; version 2 of the License. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA *****************************************************************************/ /**************************************************//** @file os/os0sync.cc The interface to the operating system synchronization primitives. Created 9/6/1995 Heikki Tuuri *******************************************************/ #include "os0sync.h" #ifdef UNIV_NONINL #include "os0sync.ic" #endif #ifdef __WIN__ #include #endif #include "ut0mem.h" #include "srv0start.h" #include "srv0srv.h" /* Type definition for an operating system mutex struct */ struct os_mutex_t{ os_event_t event; /*!< Used by sync0arr.cc for queing threads */ void* handle; /*!< OS handle to mutex */ ulint count; /*!< we use this counter to check that the same thread does not recursively lock the mutex: we do not assume that the OS mutex supports recursive locking, though NT seems to do that */ }; // All the os_*_count variables are accessed atomically /** This is incremented by 1 in os_thread_create and decremented by 1 in os_thread_exit. */ UNIV_INTERN ulint os_thread_count = 0; UNIV_INTERN ulint os_event_count = 0; UNIV_INTERN ulint os_mutex_count = 0; UNIV_INTERN ulint os_fast_mutex_count = 0; /* The number of microsecnds in a second. */ static const ulint MICROSECS_IN_A_SECOND = 1000000; #ifdef UNIV_PFS_MUTEX UNIV_INTERN mysql_pfs_key_t event_os_mutex_key; UNIV_INTERN mysql_pfs_key_t os_mutex_key; #endif /* On Windows (Vista and later), load function pointers for condition variable handling. Those functions are not available in prior versions, so we have to use them via runtime loading, as long as we support XP. */ static void os_cond_module_init(void); #ifdef __WIN__ /* Prototypes and function pointers for condition variable functions */ typedef VOID (WINAPI* InitializeConditionVariableProc) (PCONDITION_VARIABLE ConditionVariable); static InitializeConditionVariableProc initialize_condition_variable; typedef BOOL (WINAPI* SleepConditionVariableCSProc) (PCONDITION_VARIABLE ConditionVariable, PCRITICAL_SECTION CriticalSection, DWORD dwMilliseconds); static SleepConditionVariableCSProc sleep_condition_variable; typedef VOID (WINAPI* WakeAllConditionVariableProc) (PCONDITION_VARIABLE ConditionVariable); static WakeAllConditionVariableProc wake_all_condition_variable; typedef VOID (WINAPI* WakeConditionVariableProc) (PCONDITION_VARIABLE ConditionVariable); static WakeConditionVariableProc wake_condition_variable; #endif /*********************************************************//** Initialitze condition variable */ UNIV_INLINE void os_cond_init( /*=========*/ os_cond_t* cond) /*!< in: condition variable. */ { ut_a(cond); #ifdef __WIN__ ut_a(initialize_condition_variable != NULL); initialize_condition_variable(cond); #else ut_a(pthread_cond_init(cond, NULL) == 0); #endif } /*********************************************************//** Do a timed wait on condition variable. @return TRUE if timed out, FALSE otherwise */ UNIV_INLINE ibool os_cond_wait_timed( /*===============*/ os_cond_t* cond, /*!< in: condition variable. */ os_fast_mutex_t* fast_mutex, /*!< in: fast mutex */ #ifndef __WIN__ const struct timespec* abstime /*!< in: timeout */ #else DWORD time_in_ms /*!< in: timeout in milliseconds*/ #endif /* !__WIN__ */ ) { fast_mutex_t* mutex = &fast_mutex->mutex; #ifdef __WIN__ BOOL ret; DWORD err; ut_a(sleep_condition_variable != NULL); ret = sleep_condition_variable(cond, mutex, time_in_ms); if (!ret) { err = GetLastError(); /* From http://msdn.microsoft.com/en-us/library/ms686301%28VS.85%29.aspx, "Condition variables are subject to spurious wakeups (those not associated with an explicit wake) and stolen wakeups (another thread manages to run before the woken thread)." Check for both types of timeouts. Conditions are checked by the caller.*/ if ((err == WAIT_TIMEOUT) || (err == ERROR_TIMEOUT)) { return(TRUE); } } ut_a(ret); return(FALSE); #else int ret; ret = pthread_cond_timedwait(cond, mutex, abstime); switch (ret) { case 0: case ETIMEDOUT: /* We play it safe by checking for EINTR even though according to the POSIX documentation it can't return EINTR. */ case EINTR: break; default: fprintf(stderr, " InnoDB: pthread_cond_timedwait() returned: " "%d: abstime={%lu,%lu}\n", ret, (ulong) abstime->tv_sec, (ulong) abstime->tv_nsec); ut_error; } return(ret == ETIMEDOUT); #endif } /*********************************************************//** Wait on condition variable */ UNIV_INLINE void os_cond_wait( /*=========*/ os_cond_t* cond, /*!< in: condition variable. */ os_fast_mutex_t* fast_mutex)/*!< in: fast mutex */ { fast_mutex_t* mutex = &fast_mutex->mutex; ut_a(cond); ut_a(mutex); #ifdef __WIN__ ut_a(sleep_condition_variable != NULL); ut_a(sleep_condition_variable(cond, mutex, INFINITE)); #else ut_a(pthread_cond_wait(cond, mutex) == 0); #endif } /*********************************************************//** Wakes all threads waiting for condition variable */ UNIV_INLINE void os_cond_broadcast( /*==============*/ os_cond_t* cond) /*!< in: condition variable. */ { ut_a(cond); #ifdef __WIN__ ut_a(wake_all_condition_variable != NULL); wake_all_condition_variable(cond); #else ut_a(pthread_cond_broadcast(cond) == 0); #endif } /*********************************************************//** Destroys condition variable */ UNIV_INLINE void os_cond_destroy( /*============*/ os_cond_t* cond) /*!< in: condition variable. */ { #ifdef __WIN__ /* Do nothing */ #else ut_a(pthread_cond_destroy(cond) == 0); #endif } /*********************************************************//** On Windows (Vista and later), load function pointers for condition variable handling. Those functions are not available in prior versions, so we have to use them via runtime loading, as long as we support XP. */ static void os_cond_module_init(void) /*=====================*/ { #ifdef __WIN__ HMODULE h_dll; if (!srv_use_native_conditions) return; h_dll = GetModuleHandle("kernel32"); initialize_condition_variable = (InitializeConditionVariableProc) GetProcAddress(h_dll, "InitializeConditionVariable"); sleep_condition_variable = (SleepConditionVariableCSProc) GetProcAddress(h_dll, "SleepConditionVariableCS"); wake_all_condition_variable = (WakeAllConditionVariableProc) GetProcAddress(h_dll, "WakeAllConditionVariable"); wake_condition_variable = (WakeConditionVariableProc) GetProcAddress(h_dll, "WakeConditionVariable"); /* When using native condition variables, check function pointers */ ut_a(initialize_condition_variable); ut_a(sleep_condition_variable); ut_a(wake_all_condition_variable); ut_a(wake_condition_variable); #endif } /*********************************************************//** Initializes global event and OS 'slow' mutex lists. */ UNIV_INTERN void os_sync_init(void) /*==============*/ { /* Now for Windows only */ os_cond_module_init(); } /** Create an event semaphore, i.e., a semaphore which may just have two states: signaled and nonsignaled. The created event is manual reset: it must be reset explicitly by calling sync_os_reset_event. @param[in,out] event memory block where to create the event */ UNIV_INTERN void os_event_create(os_event_t event) { #ifdef __WIN__ if(!srv_use_native_conditions) { event->handle = CreateEvent(NULL, TRUE, FALSE, NULL); if (!event->handle) { fprintf(stderr, "InnoDB: Could not create a Windows event" " semaphore; Windows error %lu\n", (ulong) GetLastError()); } } else /* Windows with condition variables */ #endif { #ifndef PFS_SKIP_EVENT_MUTEX os_fast_mutex_init(event_os_mutex_key, &event->os_mutex); #else os_fast_mutex_init(PFS_NOT_INSTRUMENTED, &event->os_mutex); #endif os_cond_init(&(event->cond_var)); event->init_count_and_set(); } os_atomic_increment_ulint(&os_event_count, 1); } /*********************************************************//** Creates an event semaphore, i.e., a semaphore which may just have two states: signaled and nonsignaled. The created event is manual reset: it must be reset explicitly by calling sync_os_reset_event. @return the event handle */ UNIV_INTERN os_event_t os_event_create(void) /*==================*/ { os_event_t event = static_cast(ut_malloc(sizeof(*event))); os_event_create(event); return(event); } /**********************************************************//** Sets an event semaphore to the signaled state: lets waiting threads proceed. */ UNIV_INTERN void os_event_set( /*=========*/ os_event_t event) /*!< in: event to set */ { ut_a(event); #ifdef __WIN__ if (!srv_use_native_conditions) { ut_a(SetEvent(event->handle)); return; } #endif os_fast_mutex_lock(&(event->os_mutex)); if (UNIV_UNLIKELY(event->is_set())) { /* Do nothing */ } else { event->set(); event->inc_signal_count(); os_cond_broadcast(&(event->cond_var)); } os_fast_mutex_unlock(&(event->os_mutex)); } /**********************************************************//** Resets an event semaphore to the nonsignaled state. Waiting threads will stop to wait for the event. The return value should be passed to os_even_wait_low() if it is desired that this thread should not wait in case of an intervening call to os_event_set() between this os_event_reset() and the os_event_wait_low() call. See comments for os_event_wait_low(). @return current signal_count. */ UNIV_INTERN ib_int64_t os_event_reset( /*===========*/ os_event_t event) /*!< in: event to reset */ { ib_int64_t ret = 0; ut_a(event); #ifdef __WIN__ if(!srv_use_native_conditions) { ut_a(ResetEvent(event->handle)); return(0); } #endif os_fast_mutex_lock(&(event->os_mutex)); if (UNIV_UNLIKELY(!event->is_set())) { /* Do nothing */ } else { event->reset(); } ret = event->signal_count(); os_fast_mutex_unlock(&(event->os_mutex)); return(ret); } /**********************************************************//** Frees an event object. */ UNIV_INTERN void os_event_free( /*==========*/ os_event_t event, /*!< in: event to free */ bool free_memory)/*!< in: if true, deallocate the event memory block too */ { ut_a(event); #ifdef __WIN__ if(!srv_use_native_conditions){ ut_a(CloseHandle(event->handle)); } else /*Windows with condition variables */ #endif { os_fast_mutex_free(&(event->os_mutex)); os_cond_destroy(&(event->cond_var)); } os_atomic_decrement_ulint(&os_event_count, 1); if (free_memory) ut_free(event); } /**********************************************************//** Waits for an event object until it is in the signaled state. Typically, if the event has been signalled after the os_event_reset() we'll return immediately because event->is_set == TRUE. There are, however, situations (e.g.: sync_array code) where we may lose this information. For example: thread A calls os_event_reset() thread B calls os_event_set() [event->is_set == TRUE] thread C calls os_event_reset() [event->is_set == FALSE] thread A calls os_event_wait() [infinite wait!] thread C calls os_event_wait() [infinite wait!] Where such a scenario is possible, to avoid infinite wait, the value returned by os_event_reset() should be passed in as reset_sig_count. */ UNIV_INTERN void os_event_wait_low( /*==============*/ os_event_t event, /*!< in: event to wait */ ib_int64_t reset_sig_count)/*!< in: zero or the value returned by previous call of os_event_reset(). */ { #ifdef __WIN__ if(!srv_use_native_conditions) { DWORD err; ut_a(event); UT_NOT_USED(reset_sig_count); /* Specify an infinite wait */ err = WaitForSingleObject(event->handle, INFINITE); ut_a(err == WAIT_OBJECT_0); return; } #endif os_fast_mutex_lock(&event->os_mutex); if (!reset_sig_count) { reset_sig_count = event->signal_count(); } while (!event->is_set() && event->signal_count() == reset_sig_count) { os_cond_wait(&(event->cond_var), &(event->os_mutex)); /* Solaris manual said that spurious wakeups may occur: we have to check if the event really has been signaled after we came here to wait */ } os_fast_mutex_unlock(&event->os_mutex); } /**********************************************************//** Waits for an event object until it is in the signaled state or a timeout is exceeded. @return 0 if success, OS_SYNC_TIME_EXCEEDED if timeout was exceeded */ UNIV_INTERN ulint os_event_wait_time_low( /*===================*/ os_event_t event, /*!< in: event to wait */ ulint time_in_usec, /*!< in: timeout in microseconds, or OS_SYNC_INFINITE_TIME */ ib_int64_t reset_sig_count) /*!< in: zero or the value returned by previous call of os_event_reset(). */ { ibool timed_out = FALSE; #ifdef __WIN__ DWORD time_in_ms; if (!srv_use_native_conditions) { DWORD err; ut_a(event); if (time_in_usec != OS_SYNC_INFINITE_TIME) { time_in_ms = static_cast(time_in_usec / 1000); err = WaitForSingleObject(event->handle, time_in_ms); } else { err = WaitForSingleObject(event->handle, INFINITE); } if (err == WAIT_OBJECT_0) { return(0); } else if ((err == WAIT_TIMEOUT) || (err == ERROR_TIMEOUT)) { return(OS_SYNC_TIME_EXCEEDED); } ut_error; /* Dummy value to eliminate compiler warning. */ return(42); } else { ut_a(sleep_condition_variable != NULL); if (time_in_usec != OS_SYNC_INFINITE_TIME) { time_in_ms = static_cast(time_in_usec / 1000); } else { time_in_ms = INFINITE; } } #else struct timespec abstime; if (time_in_usec != OS_SYNC_INFINITE_TIME) { struct timeval tv; int ret; ulint sec; ulint usec; ret = ut_usectime(&sec, &usec); ut_a(ret == 0); tv.tv_sec = sec; tv.tv_usec = usec; tv.tv_usec += time_in_usec; if ((ulint) tv.tv_usec >= MICROSECS_IN_A_SECOND) { tv.tv_sec += tv.tv_usec / MICROSECS_IN_A_SECOND; tv.tv_usec %= MICROSECS_IN_A_SECOND; } abstime.tv_sec = tv.tv_sec; abstime.tv_nsec = tv.tv_usec * 1000; } else { abstime.tv_nsec = 999999999; abstime.tv_sec = (time_t) ULINT_MAX; } ut_a(abstime.tv_nsec <= 999999999); #endif /* __WIN__ */ os_fast_mutex_lock(&event->os_mutex); if (!reset_sig_count) { reset_sig_count = event->signal_count(); } do { if (event->is_set() || event->signal_count() != reset_sig_count) { break; } timed_out = os_cond_wait_timed( &event->cond_var, &event->os_mutex, #ifndef __WIN__ &abstime #else time_in_ms #endif /* !__WIN__ */ ); } while (!timed_out); os_fast_mutex_unlock(&event->os_mutex); return(timed_out ? OS_SYNC_TIME_EXCEEDED : 0); } /*********************************************************//** Creates an operating system mutex semaphore. Because these are slow, the mutex semaphore of InnoDB itself (ib_mutex_t) should be used where possible. @return the mutex handle */ UNIV_INTERN os_ib_mutex_t os_mutex_create(void) /*=================*/ { os_fast_mutex_t* mutex; os_ib_mutex_t mutex_str; mutex = static_cast( ut_malloc(sizeof(os_fast_mutex_t))); os_fast_mutex_init(os_mutex_key, mutex); mutex_str = static_cast(ut_malloc(sizeof *mutex_str)); mutex_str->handle = mutex; mutex_str->count = 0; mutex_str->event = os_event_create(); os_atomic_increment_ulint(&os_mutex_count, 1); return(mutex_str); } /**********************************************************//** Acquires ownership of a mutex semaphore. */ UNIV_INTERN void os_mutex_enter( /*===========*/ os_ib_mutex_t mutex) /*!< in: mutex to acquire */ { os_fast_mutex_lock(static_cast(mutex->handle)); (mutex->count)++; ut_a(mutex->count == 1); } /**********************************************************//** Releases ownership of a mutex. */ UNIV_INTERN void os_mutex_exit( /*==========*/ os_ib_mutex_t mutex) /*!< in: mutex to release */ { ut_a(mutex); ut_a(mutex->count == 1); (mutex->count)--; os_fast_mutex_unlock(static_cast(mutex->handle)); } /**********************************************************//** Frees a mutex object. */ UNIV_INTERN void os_mutex_free( /*==========*/ os_ib_mutex_t mutex) /*!< in: mutex to free */ { ut_a(mutex); os_event_free(mutex->event); os_atomic_decrement_ulint(&os_mutex_count, 1); os_fast_mutex_free(static_cast(mutex->handle)); ut_free(mutex->handle); ut_free(mutex); } /*********************************************************//** Initializes an operating system fast mutex semaphore. */ UNIV_INTERN void os_fast_mutex_init_func( /*====================*/ fast_mutex_t* fast_mutex) /*!< in: fast mutex */ { #ifdef __WIN__ ut_a(fast_mutex); InitializeCriticalSection((LPCRITICAL_SECTION) fast_mutex); #else ut_a(0 == pthread_mutex_init(fast_mutex, MY_MUTEX_INIT_FAST)); #endif os_atomic_increment_ulint(&os_fast_mutex_count, 1); } /**********************************************************//** Acquires ownership of a fast mutex. */ UNIV_INTERN void os_fast_mutex_lock_func( /*====================*/ fast_mutex_t* fast_mutex) /*!< in: mutex to acquire */ { #ifdef __WIN__ EnterCriticalSection((LPCRITICAL_SECTION) fast_mutex); #else pthread_mutex_lock(fast_mutex); #endif } /**********************************************************//** Releases ownership of a fast mutex. */ UNIV_INTERN void os_fast_mutex_unlock_func( /*======================*/ fast_mutex_t* fast_mutex) /*!< in: mutex to release */ { #ifdef __WIN__ LeaveCriticalSection(fast_mutex); #else pthread_mutex_unlock(fast_mutex); #endif } /**********************************************************//** Releases ownership of a fast mutex. Implies a full memory barrier even on platforms such as PowerPC where this is not normally required. */ UNIV_INTERN void os_fast_mutex_unlock_full_barrier( /*=================*/ os_fast_mutex_t* fast_mutex) /*!< in: mutex to release */ { #ifdef __WIN__ LeaveCriticalSection(&fast_mutex->mutex); #else pthread_mutex_unlock(&fast_mutex->mutex); #ifdef __powerpc__ os_mb; #endif #endif } /**********************************************************//** Frees a mutex object. */ UNIV_INTERN void os_fast_mutex_free_func( /*====================*/ fast_mutex_t* fast_mutex) /*!< in: mutex to free */ { #ifdef __WIN__ ut_a(fast_mutex); DeleteCriticalSection((LPCRITICAL_SECTION) fast_mutex); #else int ret; ret = pthread_mutex_destroy(fast_mutex); if (UNIV_UNLIKELY(ret != 0)) { ut_print_timestamp(stderr); fprintf(stderr, " InnoDB: error: return value %lu when calling\n" "InnoDB: pthread_mutex_destroy().\n", (ulint) ret); fprintf(stderr, "InnoDB: Byte contents of the pthread mutex at %p:\n", (void*) fast_mutex); ut_print_buf(stderr, fast_mutex, sizeof(os_fast_mutex_t)); putc('\n', stderr); } #endif os_atomic_decrement_ulint(&os_fast_mutex_count, 1); }