blob: 3ea81c752352847fd0981df2fbc7f776f2d708b7 [file] [log] [blame]
/*
* Copyright (c) 2020 Raspberry Pi (Trading) Ltd.
*
* SPDX-License-Identifier: BSD-3-Clause
*/
#include "pico/mutex.h"
#include "pico/time.h"
void mutex_init(mutex_t *mtx) {
lock_init(&mtx->core, next_striped_spin_lock_num());
mtx->owner = LOCK_INVALID_OWNER_ID;
#if PICO_MUTEX_ENABLE_SDK120_COMPATIBILITY
mtx->recursive = false;
#endif
__mem_fence_release();
}
void recursive_mutex_init(recursive_mutex_t *mtx) {
lock_init(&mtx->core, next_striped_spin_lock_num());
mtx->owner = LOCK_INVALID_OWNER_ID;
mtx->enter_count = 0;
#if PICO_MUTEX_ENABLE_SDK120_COMPATIBILITY
mtx->recursive = true;
#endif
__mem_fence_release();
}
void __time_critical_func(mutex_enter_blocking)(mutex_t *mtx) {
#if PICO_MUTEX_ENABLE_SDK120_COMPATIBILITY
if (mtx->recursive) {
recursive_mutex_enter_blocking(mtx);
return;
}
#endif
lock_owner_id_t caller = lock_get_caller_owner_id();
do {
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
if (!lock_is_owner_id_valid(mtx->owner)) {
mtx->owner = caller;
spin_unlock(mtx->core.spin_lock, save);
break;
}
lock_internal_spin_unlock_with_wait(&mtx->core, save);
} while (true);
}
void __time_critical_func(recursive_mutex_enter_blocking)(recursive_mutex_t *mtx) {
lock_owner_id_t caller = lock_get_caller_owner_id();
do {
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
if (mtx->owner == caller || !lock_is_owner_id_valid(mtx->owner)) {
mtx->owner = caller;
uint __unused total = ++mtx->enter_count;
spin_unlock(mtx->core.spin_lock, save);
assert(total); // check for overflow
return;
} else {
lock_internal_spin_unlock_with_wait(&mtx->core, save);
}
} while (true);
}
bool __time_critical_func(mutex_try_enter)(mutex_t *mtx, uint32_t *owner_out) {
#if PICO_MUTEX_ENABLE_SDK120_COMPATIBILITY
if (mtx->recursive) {
return recursive_mutex_try_enter(mtx, owner_out);
}
#endif
bool entered;
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
if (!lock_is_owner_id_valid(mtx->owner)) {
mtx->owner = lock_get_caller_owner_id();
entered = true;
} else {
if (owner_out) *owner_out = (uint32_t) mtx->owner;
entered = false;
}
spin_unlock(mtx->core.spin_lock, save);
return entered;
}
bool __time_critical_func(recursive_mutex_try_enter)(recursive_mutex_t *mtx, uint32_t *owner_out) {
bool entered;
lock_owner_id_t caller = lock_get_caller_owner_id();
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
if (!lock_is_owner_id_valid(mtx->owner) || mtx->owner == caller) {
mtx->owner = caller;
uint __unused total = ++mtx->enter_count;
assert(total); // check for overflow
entered = true;
} else {
if (owner_out) *owner_out = (uint32_t) mtx->owner;
entered = false;
}
spin_unlock(mtx->core.spin_lock, save);
return entered;
}
bool __time_critical_func(mutex_enter_timeout_ms)(mutex_t *mtx, uint32_t timeout_ms) {
return mutex_enter_block_until(mtx, make_timeout_time_ms(timeout_ms));
}
bool __time_critical_func(recursive_mutex_enter_timeout_ms)(recursive_mutex_t *mtx, uint32_t timeout_ms) {
return recursive_mutex_enter_block_until(mtx, make_timeout_time_ms(timeout_ms));
}
bool __time_critical_func(mutex_enter_timeout_us)(mutex_t *mtx, uint32_t timeout_us) {
return mutex_enter_block_until(mtx, make_timeout_time_us(timeout_us));
}
bool __time_critical_func(recursive_mutex_enter_timeout_us)(recursive_mutex_t *mtx, uint32_t timeout_us) {
return recursive_mutex_enter_block_until(mtx, make_timeout_time_us(timeout_us));
}
bool __time_critical_func(mutex_enter_block_until)(mutex_t *mtx, absolute_time_t until) {
#if PICO_MUTEX_ENABLE_SDK120_COMPATIBILITY
if (mtx->recursive) {
return recursive_mutex_enter_block_until(mtx, until);
}
#endif
assert(mtx->core.spin_lock);
lock_owner_id_t caller = lock_get_caller_owner_id();
do {
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
if (!lock_is_owner_id_valid(mtx->owner)) {
mtx->owner = caller;
spin_unlock(mtx->core.spin_lock, save);
return true;
} else {
if (lock_internal_spin_unlock_with_best_effort_wait_or_timeout(&mtx->core, save, until)) {
// timed out
return false;
}
// not timed out; spin lock already unlocked, so loop again
}
} while (true);
}
bool __time_critical_func(recursive_mutex_enter_block_until)(recursive_mutex_t *mtx, absolute_time_t until) {
assert(mtx->core.spin_lock);
lock_owner_id_t caller = lock_get_caller_owner_id();
do {
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
if (!lock_is_owner_id_valid(mtx->owner) || mtx->owner == caller) {
mtx->owner = caller;
uint __unused total = ++mtx->enter_count;
spin_unlock(mtx->core.spin_lock, save);
assert(total); // check for overflow
return true;
} else {
if (lock_internal_spin_unlock_with_best_effort_wait_or_timeout(&mtx->core, save, until)) {
// timed out
return false;
}
// not timed out; spin lock already unlocked, so loop again
}
} while (true);
}
void __time_critical_func(mutex_exit)(mutex_t *mtx) {
#if PICO_MUTEX_ENABLE_SDK120_COMPATIBILITY
if (mtx->recursive) {
recursive_mutex_exit(mtx);
return;
}
#endif
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
assert(lock_is_owner_id_valid(mtx->owner));
mtx->owner = LOCK_INVALID_OWNER_ID;
lock_internal_spin_unlock_with_notify(&mtx->core, save);
}
void __time_critical_func(recursive_mutex_exit)(recursive_mutex_t *mtx) {
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
assert(lock_is_owner_id_valid(mtx->owner));
assert(mtx->enter_count);
if (!--mtx->enter_count) {
mtx->owner = LOCK_INVALID_OWNER_ID;
lock_internal_spin_unlock_with_notify(&mtx->core, save);
} else {
spin_unlock(mtx->core.spin_lock, save);
}
}