/**************************************************************************** * include/nuttx/clock.h * * Licensed to the Apache Software Foundation (ASF) under one or more * contributor license agreements. See the NOTICE file distributed with * this work for additional information regarding copyright ownership. The * ASF licenses this file to you under the Apache License, Version 2.0 (the * "License"); you may not use this file except in compliance with the * License. You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the * License for the specific language governing permissions and limitations * under the License. * ****************************************************************************/ #ifndef __INCLUDE_NUTTX_CLOCK_H #define __INCLUDE_NUTTX_CLOCK_H /**************************************************************************** * Included Files ****************************************************************************/ #include #include #include #include #include /**************************************************************************** * Pre-processor Definitions ****************************************************************************/ /* Configuration ************************************************************/ /* Efficient, direct access to OS global timer variables will be supported * if the execution environment has direct access to kernel global data. * The code in this execution context can access the kernel global data * directly if: * * 1. We are not running tick-less (in which case there is no global timer * data), * 2. This is an un-protected, non-kernel build, * 3. This is a protected build, but this code is being built for execution * within the kernel space. * 4. It we are building with SYSCALLs enabled, but not in a kernel build, * then we can't know a priori whether the code has access to the * global variables or not. In that case we have to assume not. */ #undef __HAVE_KERNEL_GLOBALS #if defined(CONFIG_SCHED_TICKLESS) /* Case 1: There is no global timer data */ #elif defined(__KERNEL__) /* Case 3: Kernel mode of protected/kernel build */ # define __HAVE_KERNEL_GLOBALS 1 #elif defined(CONFIG_LIB_SYSCALL) /* Case 4: Building with SYSCALLs enabled, but not part of a kernel build */ #else /* Case 2: Un-protected, non-kernel build */ # define __HAVE_KERNEL_GLOBALS 1 #endif /* If CONFIG_SYSTEM_TIME64 is selected and the CPU supports long long types, * then a 64-bit system time will be used. */ #ifndef CONFIG_HAVE_LONG_LONG # undef CONFIG_SYSTEM_TIME64 #endif /* Timing constants *********************************************************/ #define NSEC_PER_SEC 1000000000L /* Seconds */ #define USEC_PER_SEC 1000000L #define MSEC_PER_SEC 1000L #define DSEC_PER_SEC 10L #define HSEC_PER_SEC 2L #define NSEC_PER_HSEC 500000000L /* Half seconds */ #define USEC_PER_HSEC 500000L #define MSEC_PER_HSEC 500L #define DSEC_PER_HSEC 5L #define NSEC_PER_DSEC 100000000L /* Deciseconds */ #define USEC_PER_DSEC 100000L #define MSEC_PER_DSEC 100L #define NSEC_PER_MSEC 1000000L /* Milliseconds */ #define USEC_PER_MSEC 1000L #define NSEC_PER_USEC 1000L /* Microseconds */ #define SEC_PER_MIN 60L #define NSEC_PER_MIN (NSEC_PER_SEC * SEC_PER_MIN) #define USEC_PER_MIN (USEC_PER_SEC * SEC_PER_MIN) #define MSEC_PER_MIN (MSEC_PER_SEC * SEC_PER_MIN) #define DSEC_PER_MIN (DSEC_PER_SEC * SEC_PER_MIN) #define HSEC_PER_MIN (HSEC_PER_SEC * SEC_PER_MIN) #define MIN_PER_HOUR 60L #define NSEC_PER_HOUR (NSEC_PER_MIN * MIN_PER_HOUR) #define USEC_PER_HOUR (USEC_PER_MIN * MIN_PER_HOUR) #define MSEC_PER_HOUR (MSEC_PER_MIN * MIN_PER_HOUR) #define DSEC_PER_HOUR (DSEC_PER_MIN * MIN_PER_HOUR) #define HSEC_PER_HOUR (HSEC_PER_MIN * MIN_PER_HOUR) #define SEC_PER_HOUR (SEC_PER_MIN * MIN_PER_HOUR) #define HOURS_PER_DAY 24L #define SEC_PER_DAY (HOURS_PER_DAY * SEC_PER_HOUR) /* If CONFIG_SCHED_TICKLESS is not defined, then the interrupt interval of * the system timer is given by USEC_PER_TICK. This is the expected number * of microseconds between calls from the processor-specific logic to * nxsched_process_timer(). The default value of USEC_PER_TICK is 10000 * microseconds (100 Hz). However, this default setting can be overridden * by defining the interval in microseconds as CONFIG_USEC_PER_TICK in the * NuttX configuration file. * * The following calculations are only accurate when (1) there is no * truncation involved and (2) the underlying system timer is an even * multiple of microseconds. If (2) is not true, you will probably want * to redefine all of the following. */ #ifdef CONFIG_USEC_PER_TICK # define USEC_PER_TICK (CONFIG_USEC_PER_TICK) #else # define USEC_PER_TICK (10000) #endif /* MSEC_PER_TICK can be very inaccurate if CONFIG_USEC_PER_TICK is not an * even multiple of milliseconds. Calculations using USEC_PER_TICK are * preferred for that reason (at the risk of overflow) */ #define TICK_PER_HOUR (USEC_PER_HOUR / USEC_PER_TICK) /* Truncates! */ #define TICK_PER_MIN (USEC_PER_MIN / USEC_PER_TICK) /* Truncates! */ #define TICK_PER_SEC (USEC_PER_SEC / USEC_PER_TICK) /* Truncates! */ #define TICK_PER_MSEC (USEC_PER_MSEC / USEC_PER_TICK) /* Truncates! */ #define TICK_PER_DSEC (USEC_PER_DSEC / USEC_PER_TICK) /* Truncates! */ #define TICK_PER_HSEC (USEC_PER_HSEC / USEC_PER_TICK) /* Truncates! */ #define MSEC_PER_TICK (USEC_PER_TICK / USEC_PER_MSEC) /* Truncates! */ #define NSEC_PER_TICK (USEC_PER_TICK * NSEC_PER_USEC) /* Exact */ #define NSEC2TICK(nsec) (((nsec)+(NSEC_PER_TICK/2))/NSEC_PER_TICK) /* Rounds */ #define USEC2TICK(usec) (((usec)+(USEC_PER_TICK/2))/USEC_PER_TICK) /* Rounds */ #if (MSEC_PER_TICK * USEC_PER_MSEC) == USEC_PER_TICK # define MSEC2TICK(msec) (((msec)+(MSEC_PER_TICK/2))/MSEC_PER_TICK) /* Rounds */ #else # define MSEC2TICK(msec) USEC2TICK((msec) * USEC_PER_MSEC) /* Rounds */ #endif #define DSEC2TICK(dsec) MSEC2TICK((dsec) * MSEC_PER_DSEC) /* Rounds */ #define HSEC2TICK(dsec) MSEC2TICK((dsec) * MSEC_PER_HSEC) /* Rounds */ #define SEC2TICK(sec) MSEC2TICK((sec) * MSEC_PER_SEC) /* Rounds */ #define TICK2NSEC(tick) ((tick) * NSEC_PER_TICK) /* Exact */ #define TICK2USEC(tick) ((tick) * USEC_PER_TICK) /* Exact */ #if (MSEC_PER_TICK * USEC_PER_MSEC) == USEC_PER_TICK # define TICK2MSEC(tick) ((tick)*MSEC_PER_TICK) /* Exact */ #else # define TICK2MSEC(tick) (((tick)*USEC_PER_TICK)/USEC_PER_MSEC) /* Rounds */ #endif #define TICK2DSEC(tick) (((tick)+(TICK_PER_DSEC/2))/TICK_PER_DSEC) /* Rounds */ #define TICK2HSEC(tick) (((tick)+(TICK_PER_HSEC/2))/TICK_PER_HSEC) /* Rounds */ #define TICK2SEC(tick) (((tick)+(TICK_PER_SEC/2))/TICK_PER_SEC) /* Rounds */ #if defined(CONFIG_DEBUG_FEATURES) && defined(CONFIG_SYSTEM_TIME64) && \ defined(CONFIG_CLOCK_MONOTONIC) && !defined(CONFIG_SCHED_TICKLESS) /* Initial system timer ticks value close to maximum 32-bit value, to test * 64-bit system-timer after going over 32-bit value. This is to make errors * of casting 64-bit system-timer to 32-bit variables more visible. */ # define INITIAL_SYSTEM_TIMER_TICKS \ ((uint64_t)(UINT32_MAX - (TICK_PER_SEC * 5))) #else # define INITIAL_SYSTEM_TIMER_TICKS 0 #endif /**************************************************************************** * Public Types ****************************************************************************/ /* This structure is used to report CPU usage for a particular thread */ #ifdef CONFIG_SCHED_CPULOAD struct cpuload_s { volatile uint32_t total; /* Total number of clock ticks */ volatile uint32_t active; /* Number of ticks while this thread was active */ }; #endif /* This non-standard type used to hold relative clock ticks that may take * negative values. Because of its non-portable nature the type sclock_t * should be used only within the OS proper and not by portable applications. */ #ifdef CONFIG_SYSTEM_TIME64 typedef int64_t sclock_t; #else typedef int32_t sclock_t; #endif /**************************************************************************** * Public Data ****************************************************************************/ #ifdef __cplusplus #define EXTERN extern "C" extern "C" { #else #define EXTERN extern #endif /* Access to raw system clock ***********************************************/ /* Direct access to the system timer/counter is supported only if (1) the * system timer counter is available (i.e., we are not configured to use * a hardware periodic timer), and (2) the execution environment has direct * access to kernel global data */ #ifdef __HAVE_KERNEL_GLOBALS EXTERN volatile clock_t g_system_timer; #ifndef CONFIG_SYSTEM_TIME64 # define clock_systime_ticks() g_system_timer #endif #endif /**************************************************************************** * Public Function Prototypes ****************************************************************************/ /**************************************************************************** * Name: clock_timespec_compare * * Description: * Return < 0 if time ts1 is before time ts2 * Return > 0 if time ts2 is before time ts1 * Return 0 if time ts1 is the same as time ts2 * ****************************************************************************/ int clock_timespec_compare(FAR const struct timespec *ts1, FAR const struct timespec *ts2); /**************************************************************************** * Name: clock_timespec_add * * Description: * Add timespec ts1 to to2 and return the result in ts3 * * Input Parameters: * ts1 and ts2: The two timespecs to be added * t23: The location to return the result (may be ts1 or ts2) * * Returned Value: * None * ****************************************************************************/ void clock_timespec_add(FAR const struct timespec *ts1, FAR const struct timespec *ts2, FAR struct timespec *ts3); /**************************************************************************** * Name: clock_timespec_subtract * * Description: * Subtract timespec ts2 from to1 and return the result in ts3. * Zero is returned if the time difference is negative. * * Input Parameters: * ts1 and ts2: The two timespecs to be subtracted (ts1 - ts2) * t23: The location to return the result (may be ts1 or ts2) * * Returned Value: * None * ****************************************************************************/ void clock_timespec_subtract(FAR const struct timespec *ts1, FAR const struct timespec *ts2, FAR struct timespec *ts3); /**************************************************************************** * Name: clock_synchronize * * Description: * Synchronize the system timer to a hardware RTC. This operation is * normally performed automatically by the system during clock * initialization. However, the user may also need to explicitly re- * synchronize the system timer to the RTC under certain conditions where * the system timer is known to be in error. For example, in certain low- * power states, the system timer may be stopped but the RTC will continue * keep correct time. After recovering from such low-power state, this * function should be called to restore the correct system time. * * Calling this function could result in system time going "backward" in * time, especially with certain lower resolution RTC implementations. * Time going backward could have bad consequences if there are ongoing * timers and delays. So use this interface with care. * * Input Parameters: * None * * Returned Value: * None * * Assumptions: * ****************************************************************************/ #ifdef CONFIG_RTC void clock_synchronize(void); #endif /**************************************************************************** * Name: clock_resynchronize * * Description: * Resynchronize the system timer to a hardware RTC. The user can * explicitly re-synchronize the system timer to the RTC under certain * conditions where the system timer is known to be in error. For example, * in certain low-power states, the system timer may be stopped but the * RTC will continue keep correct time. After recovering from such * low-power state, this function should be called to restore the correct * system time. Function also keeps monotonic clock at rate of RTC. * * Calling this function will not result in system time going "backward" in * time. If setting system time with RTC would result time going "backward" * then resynchronization is not performed. * * Input Parameters: * diff: amount of time system-time is adjusted forward with RTC * * Returned Value: * None * * Assumptions: * ****************************************************************************/ #if defined(CONFIG_RTC) && !defined(CONFIG_SCHED_TICKLESS) void clock_resynchronize(FAR struct timespec *rtc_diff); #endif /**************************************************************************** * Name: clock_systime_ticks * * Description: * Return the current value of the 32/64-bit system timer counter. * * Indirect access to the system timer counter is required through this * function if the execution environment does not have direct access to * kernel global data. * * Use of this function is also required to assure atomic access to the * 64-bit system timer. * * NOTE: This is an internal OS interface and should not be called from * application code. Rather, the functionally equivalent, standard * interface clock() should be used. * * Input Parameters: * None * * Returned Value: * The current value of the system timer counter * ****************************************************************************/ #if !defined(__HAVE_KERNEL_GLOBALS) || defined(CONFIG_SYSTEM_TIME64) clock_t clock_systime_ticks(void); #endif /**************************************************************************** * Name: clock_systime_timespec * * Description: * Return the current value of the system timer counter as a struct * timespec. * * Input Parameters: * ts - Location to return the time * * Returned Value: * OK (0) on success; a negated errno value on failure. * * Assumptions: * ****************************************************************************/ int clock_systime_timespec(FAR struct timespec *ts); /**************************************************************************** * Name: clock_cpuload * * Description: * Return load measurement data for the select PID. * * Input Parameters: * pid - The task ID of the thread of interest. pid == 0 is IDLE thread. * cpuload - The location to return the CPU load * * Returned Value: * OK (0) on success; a negated errno value on failure. The only reason * that this function can fail is if 'pid' no longer refers to a valid * thread. * * Assumptions: * ****************************************************************************/ #ifdef CONFIG_SCHED_CPULOAD int clock_cpuload(int pid, FAR struct cpuload_s *cpuload); #endif /**************************************************************************** * Name: nxsched_oneshot_extclk * * Description: * Configure to use a oneshot timer as described in * include/nuttx/timers/oneshot.h to provide external clocking to assess * the CPU load. * * Input Parameters: * lower - An instance of the oneshot timer interface as defined in * include/nuttx/timers/oneshot.h * * Returned Value: * None * ****************************************************************************/ #ifdef CONFIG_CPULOAD_ONESHOT struct oneshot_lowerhalf_s; void nxsched_oneshot_extclk(FAR struct oneshot_lowerhalf_s *lower); #endif /**************************************************************************** * Name: nxsched_period_extclk * * Description: * Configure to use a period timer as described in * include/nuttx/timers/timer.h to provide external clocking to assess * the CPU load. * * Input Parameters: * lower - An instance of the period timer interface as defined in * include/nuttx/timers/timer.h * * Returned Value: * None * ****************************************************************************/ #ifdef CONFIG_CPULOAD_PERIOD struct timer_lowerhalf_s; void nxsched_period_extclk(FAR struct timer_lowerhalf_s *lower); #endif #undef EXTERN #ifdef __cplusplus } #endif #endif /* __INCLUDE_NUTTX_CLOCK_H */