unleashed-firmware/furi/core/kernel.h
あく acc39a4bc0
Api Symbols: replace asserts with checks (#3507)
* Api Symbols: replace asserts with checks
* Api Symbols: replace asserts with checks part 2
* Update no args function signatures with void, to help compiler to track incorrect usage
* More unavoidable void
* Update PVS config and code to make it happy
* Format sources
* nfc: fix checks
* dead code cleanup & include fixes

Co-authored-by: gornekich <n.gorbadey@gmail.com>
Co-authored-by: hedger <hedger@users.noreply.github.com>
Co-authored-by: hedger <hedger@nanode.su>
2024-03-19 23:43:52 +09:00

126 lines
3.4 KiB
C

/**
* @file kernel.h
* Furi Kernel primitives
*/
#pragma once
#include <core/base.h>
#ifdef __cplusplus
extern "C" {
#endif
/** Check if CPU is in IRQ or kernel running and IRQ is masked
*
* Originally this primitive was born as a workaround for FreeRTOS kernel primitives shenanigans with PRIMASK.
*
* Meaningful use cases are:
*
* - When kernel is started and you want to ensure that you are not in IRQ or IRQ is not masked(like in critical section)
* - When kernel is not started and you want to make sure that you are not in IRQ mode, ignoring PRIMASK.
*
* As you can see there will be edge case when kernel is not started and PRIMASK is not 0 that may cause some funky behavior.
* Most likely it will happen after kernel primitives being used, but control not yet passed to kernel.
* It's up to you to figure out if it is safe for your code or not.
*
* @return true if CPU is in IRQ or kernel running and IRQ is masked
*/
bool furi_kernel_is_irq_or_masked(void);
/** Check if kernel is running
*
* @return true if running, false otherwise
*/
bool furi_kernel_is_running(void);
/** Lock kernel, pause process scheduling
*
* @warning This should never be called in interrupt request context.
*
* @return previous lock state(0 - unlocked, 1 - locked)
*/
int32_t furi_kernel_lock(void);
/** Unlock kernel, resume process scheduling
*
* @warning This should never be called in interrupt request context.
*
* @return previous lock state(0 - unlocked, 1 - locked)
*/
int32_t furi_kernel_unlock(void);
/** Restore kernel lock state
*
* @warning This should never be called in interrupt request context.
*
* @param[in] lock The lock state
*
* @return new lock state or error
*/
int32_t furi_kernel_restore_lock(int32_t lock);
/** Get kernel systick frequency
*
* @return systick counts per second
*/
uint32_t furi_kernel_get_tick_frequency(void);
/** Delay execution
*
* @warning This should never be called in interrupt request context.
*
* Also keep in mind delay is aliased to scheduler timer intervals.
*
* @param[in] ticks The ticks count to pause
*/
void furi_delay_tick(uint32_t ticks);
/** Delay until tick
*
* @warning This should never be called in interrupt request context.
*
* @param[in] tick The tick until which kerel should delay task execution
*
* @return The furi status.
*/
FuriStatus furi_delay_until_tick(uint32_t tick);
/** Get current tick counter
*
* System uptime, may overflow.
*
* @return Current ticks in milliseconds
*/
uint32_t furi_get_tick(void);
/** Convert milliseconds to ticks
*
* @param[in] milliseconds time in milliseconds
* @return time in ticks
*/
uint32_t furi_ms_to_ticks(uint32_t milliseconds);
/** Delay in milliseconds
*
* This method uses kernel ticks on the inside, which causes delay to be aliased to scheduler timer intervals.
* Real wait time will be between X+ milliseconds.
* Special value: 0, will cause task yield.
* Also if used when kernel is not running will fall back to `furi_delay_us`.
*
* @warning Cannot be used from ISR
*
* @param[in] milliseconds milliseconds to wait
*/
void furi_delay_ms(uint32_t milliseconds);
/** Delay in microseconds
*
* Implemented using Cortex DWT counter. Blocking and non aliased.
*
* @param[in] microseconds microseconds to wait
*/
void furi_delay_us(uint32_t microseconds);
#ifdef __cplusplus
}
#endif