Rework lock_core / timers (#378)

- Add recursive_mutex
  - Make all locking primitives and sleep use common overridable wait/notify support to allow RTOS
    implementations to replace WFE/SEV with something more appropriate
  - Add busy_wait_ms
This commit is contained in:
Graham Sanderson
2021-05-05 11:46:25 -05:00
committed by GitHub
parent ec0dc7a88b
commit 6d87da4c59
15 changed files with 434 additions and 152 deletions

View File

@ -10,15 +10,18 @@
static_assert(sizeof(critical_section_t) == 8, "");
#endif
void critical_section_init(critical_section_t *critsec) {
critical_section_init_with_lock_num(critsec, (uint)spin_lock_claim_unused(true));
void critical_section_init(critical_section_t *crit_sec) {
critical_section_init_with_lock_num(crit_sec, (uint)spin_lock_claim_unused(true));
}
void critical_section_init_with_lock_num(critical_section_t *critsec, uint lock_num) {
lock_init(&critsec->core, lock_num);
void critical_section_init_with_lock_num(critical_section_t *crit_sec, uint lock_num) {
crit_sec->spin_lock = spin_lock_instance(lock_num);
__mem_fence_release();
}
void critical_section_deinit(critical_section_t *critsec) {
spin_lock_unclaim(spin_lock_get_num(critsec->core.spin_lock));
void critical_section_deinit(critical_section_t *crit_sec) {
spin_lock_unclaim(spin_lock_get_num(crit_sec->spin_lock));
#ifndef NDEBUG
crit_sec->spin_lock = (spin_lock_t *)-1;
#endif
}

View File

@ -22,11 +22,12 @@ extern "C" {
* from the other core, and from (higher priority) interrupts on the same core. It does the former
* using a spin lock and the latter by disabling interrupts on the calling core.
*
* Because interrupts are disabled by this function, uses of the critical_section should be as short as possible.
* Because interrupts are disabled when a critical_section is owned, uses of the critical_section
* should be as short as possible.
*/
typedef struct __packed_aligned critical_section {
lock_core_t core;
spin_lock_t *spin_lock;
uint32_t save;
} critical_section_t;
@ -38,16 +39,16 @@ typedef struct __packed_aligned critical_section {
* critical sections, however if you do so you *must* use \ref critical_section_init_with_lock_num
* to ensure that the spin lock's used are different.
*
* \param critsec Pointer to critical_section structure
* \param crit_sec Pointer to critical_section structure
*/
void critical_section_init(critical_section_t *critsec);
void critical_section_init(critical_section_t *crit_sec);
/*! \brief Initialise a critical_section structure assigning a specific spin lock number
* \ingroup critical_section
* \param critsec Pointer to critical_section structure
* \param crit_sec Pointer to critical_section structure
* \param lock_num the specific spin lock number to use
*/
void critical_section_init_with_lock_num(critical_section_t *critsec, uint lock_num);
void critical_section_init_with_lock_num(critical_section_t *crit_sec, uint lock_num);
/*! \brief Enter a critical_section
* \ingroup critical_section
@ -55,20 +56,32 @@ void critical_section_init_with_lock_num(critical_section_t *critsec, uint lock_
* If the spin lock associated with this critical section is in use, then this
* method will block until it is released.
*
* \param critsec Pointer to critical_section structure
* \param crit_sec Pointer to critical_section structure
*/
static inline void critical_section_enter_blocking(critical_section_t *critsec) {
critsec->save = spin_lock_blocking(critsec->core.spin_lock);
static inline void critical_section_enter_blocking(critical_section_t *crit_sec) {
crit_sec->save = spin_lock_blocking(crit_sec->spin_lock);
}
/*! \brief Release a critical_section
* \ingroup critical_section
*
* \param critsec Pointer to critical_section structure
* \param crit_sec Pointer to critical_section structure
*/
static inline void critical_section_exit(critical_section_t *critsec) {
spin_unlock(critsec->core.spin_lock, critsec->save);
static inline void critical_section_exit(critical_section_t *crit_sec) {
spin_unlock(crit_sec->spin_lock, crit_sec->save);
}
/*! \brief De-Initialise a critical_section created by the critical_section_init method
* \ingroup critical_section
*
* This method is only used to free the associated spin lock allocated via
* the critical_section_init method (it should not be used to de-initialize a spin lock
* created via critical_section_init_with_lock_num). After this call, the critical section is invalid
*
* \param crit_sec Pointer to critical_section structure
*/
void critical_section_deinit(critical_section_t *crit_sec);
#ifdef __cplusplus
}
#endif

View File

@ -8,26 +8,183 @@
#define _PICO_LOCK_CORE_H
#include "pico.h"
#include "pico/time.h"
#include "hardware/sync.h"
/** \file lock_core.h
* \defgroup lock_core lock_core
* \ingroup pico_sync
* \brief base synchronization/lock primitive support
*
* Most of the pico_sync locking primitives contain a lock_core_t structure member. This currently just holds a spin
* lock which is used only to protect the contents of the rest of the structure as part of implementing the synchronization
* primitive. As such, the spin_lock member of lock core is never still held on return from any function for the primitive.
*
* \ref critical_section is an exceptional case in that it does not have a lock_core_t and simply wraps a pin lock, providing
* methods to lock and unlock said spin lock.
*
* lock_core based structures work by locking the spin lock, checking state, and then deciding whether they additionally need to block
* or notify when the spin lock is released. In the blocking case, they will wake up again in the future, and try the process again.
*
* By default the SDK just uses the processors' events via SEV and WEV for notification and blocking as these are sufficient for
* cross core, and notification from interrupt handlers. However macros are defined in this file that abstract the wait
* and notify mechanisms to allow the SDK locking functions to effectively be used within an RTOS on other environment.
*
* When implementing an RTOS, it is desirable for the SDK synchronization primitives that wait, to block the calling task (and immediately yield),
* and those that notify, to wake a blocked task which isn't on processor. At least the wait macro implementation needs to be atomic with the protecting
* spin_lock unlock from the callers point of view; i.e. the task should unlock the spin lock when as it starts its wait. Such implementation is
* up to the RTOS integration, however the macros are defined such that such operations are always combined into a single call
* (so they can be perfomed atomically) even though the default implementation does not need this, as a WFE which starts
* following the corresponding SEV is not missed.
*/
// PICO_CONFIG: PARAM_ASSERTIONS_ENABLED_LOCK_CORE, Enable/disable assertions in the lock core, type=bool, default=0, group=pico_sync
#ifndef PARAM_ASSERTIONS_ENABLED_LOCK_CORE
#define PARAM_ASSERTIONS_ENABLED_LOCK_CORE 0
#endif
/** \file lock_core.h
* \ingroup pico_sync
* \ingroup lock_core
*
* Base implementation for locking primitives protected by a spin lock
* Base implementation for locking primitives protected by a spin lock. The spin lock is only used to protect
* access to the remaining lock state (in primitives using lock_core); it is never left locked outside
* of the function implementations
*/
typedef struct lock_core {
struct lock_core {
// spin lock protecting this lock's state
spin_lock_t *spin_lock;
// note any lock members in containing structures need not be volatile;
// they are protected by memory/compiler barriers when gaining and release spin locks
} lock_core_t;
};
typedef struct lock_core lock_core_t;
/*! \brief Initialise a lock structure
* \ingroup lock_core
*
* Inititalize a lock structure, providing the spin lock number to use for protecting internal state.
*
* \param core Pointer to the lock_core to initialize
* \param lock_num Spin lock number to use for the lock. As the spin lock is only used internally to the locking primitive
* method implementations, this does not need to be globally unique, however could suffer contention
*/
void lock_init(lock_core_t *core, uint lock_num);
#ifndef lock_owner_id_t
/*! \brief type to use to store the 'owner' of a lock.
* \ingroup lock_core
* By default this is int8_t as it only needs to store the core number or -1, however it may be
* overridden if a larger type is required (e.g. for an RTOS task id)
*/
#define lock_owner_id_t int8_t
#endif
#ifndef LOCK_INVALID_OWNER_ID
/*! \brief marker value to use for a lock_owner_id_t which does not refer to any valid owner
* \ingroup lock_core
*/
#define LOCK_INVALID_OWNER_ID ((lock_owner_id_t)-1)
#endif
#ifndef lock_get_caller_owner_id
/*! \brief return the owner id for the caller
* \ingroup lock_core
* By default this returns the calling core number, but may be overridden (e.g. to return an RTOS task id)
*/
#define lock_get_caller_owner_id() ((lock_owner_id_t)get_core_num())
#endif
#ifndef lock_internal_spin_unlock_with_wait
/*! \brief Atomically unlock the lock's spin lock, and wait for a notification.
* \ingroup lock_core
*
* _Atomic_ here refers to the fact that it should not be possible for a concurrent lock_internal_spin_unlock_with_notify
* to insert itself between the spin unlock and this wait in a way that the wait does not see the notification (i.e. causing
* a missed notification). In other words this method should always wake up in response to a lock_internal_spin_unlock_with_notify
* for the same lock, which completes after this call starts.
*
* In an ideal implementation, this method would return exactly after the corresponding lock_internal_spin_unlock_with_notify
* has subsequently been called on the same lock instance, however this method is free to return at _any_ point before that;
* this macro is _always_ used in a loop which locks the spin lock, checks the internal locking primitive state and then
* waits again if the calling thread should not proceed.
*
* By default this macro simply unlocks the spin lock, and then performs a WFE, but may be overridden
* (e.g. to actually block the RTOS task).
*
* \param lock the lock_core for the primitive which needs to block
* \param save the uint32_t value that should be passed to spin_unlock when the spin lock is unlocked. (i.e. the `PRIMASK`
* state when the spin lock was acquire
*/
#define lock_internal_spin_unlock_with_wait(lock, save) spin_unlock((lock)->spin_lock, save), __wfe()
#endif
#ifndef lock_internal_spin_unlock_with_notify
/*! \brief Atomically unlock the lock's spin lock, and send a notification
* \ingroup lock_core
*
* _Atomic_ here refers to the fact that it should not be possible for this notification to happen during a
* lock_internal_spin_unlock_with_wait in a way that that wait does not see the notification (i.e. causing
* a missed notification). In other words this method should always wake up any lock_internal_spin_unlock_with_wait
* which started before this call completes.
*
* In an ideal implementation, this method would wake up only the corresponding lock_internal_spin_unlock_with_wait
* that has been called on the same lock instance, however it is free to wake up any of them, as they will check
* their condition and then re-wait if necessary/
*
* By default this macro simply unlocks the spin lock, and then performs a SEV, but may be overridden
* (e.g. to actually un-block RTOS task(s)).
*
* \param lock the lock_core for the primitive which needs to block
* \param save the uint32_t value that should be passed to spin_unlock when the spin lock is unlocked. (i.e. the PRIMASK
* state when the spin lock was acquire)
*/
#define lock_internal_spin_unlock_with_notify(lock, save) spin_unlock((lock)->spin_lock, save), __sev()
#endif
#ifndef lock_internal_spin_unlock_with_best_effort_wait_or_timeout
/*! \brief Atomically unlock the lock's spin lock, and wait for a notification or a timeout
* \ingroup lock_core
*
* _Atomic_ here refers to the fact that it should not be possible for a concurrent lock_internal_spin_unlock_with_notify
* to insert itself between the spin unlock and this wait in a way that the wait does not see the notification (i.e. causing
* a missed notification). In other words this method should always wake up in response to a lock_internal_spin_unlock_with_notify
* for the same lock, which completes after this call starts.
*
* In an ideal implementation, this method would return exactly after the corresponding lock_internal_spin_unlock_with_notify
* has subsequently been called on the same lock instance or the timeout has been reached, however this method is free to return
* at _any_ point before that; this macro is _always_ used in a loop which locks the spin lock, checks the internal locking
* primitive state and then waits again if the calling thread should not proceed.
*
* By default this simply unlocks the spin lock, and then calls \ref best_effort_wfe_or_timeout
* but may be overridden (e.g. to actually block the RTOS task with a timeout).
*
* \param lock the lock_core for the primitive which needs to block
* \param save the uint32_t value that should be passed to spin_unlock when the spin lock is unlocked. (i.e. the PRIMASK
* state when the spin lock was acquire)
* \param until the \ref absolute_time_t value
* \return true if the timeout has been reached
*/
#define lock_internal_spin_unlock_with_best_effort_wait_or_timeout(lock, save, until) ({ \
spin_unlock((lock)->spin_lock, save); \
best_effort_wfe_or_timeout(until); \
})
#endif
#ifndef sync_internal_yield_until_before
/*! \brief yield to other processing until some time before the requested time
* \ingroup lock_core
*
* This method is provided for cases where the caller has no useful work to do
* until the specified time.
*
* By default this method does nothing, however if can be overridden (for example by an
* RTOS which is able to block the current task until the scheduler tick before
* the given time)
*
* \param until the \ref absolute_time_t value
*/
#define sync_internal_yield_until_before(until) ((void)0)
#endif
#endif

View File

@ -28,12 +28,17 @@ extern "C" {
*
* See \ref critical_section.h for protecting access between multiple cores AND IRQ handlers
*/
typedef struct __packed_aligned mutex {
lock_core_t core;
int8_t owner; //! core number or -1 for unowned
lock_owner_id_t owner; //! owner id LOCK_INVALID_OWNER_ID for unowned
uint8_t recursion_state; //! 0 means non recursive (owner or unowned)
//! 1 is a maxed out recursive lock
//! 2-254 is an owned lock
//! 255 is an un-owned lock
} mutex_t;
#define MAX_RECURSION_STATE ((uint8_t)255)
/*! \brief Initialise a mutex structure
* \ingroup mutex
*
@ -41,6 +46,15 @@ typedef struct __packed_aligned mutex {
*/
void mutex_init(mutex_t *mtx);
/*! \brief Initialise a recursive mutex structure
* \ingroup mutex
*
* A recursive mutex may be entered in a nested fashion by the same owner
*
* \param mtx Pointer to mutex structure
*/
void recursive_mutex_init(mutex_t *mtx);
/*! \brief Take ownership of a mutex
* \ingroup mutex
*
@ -129,6 +143,29 @@ static inline bool mutex_is_initialzed(mutex_t *mtx) {
*/
#define auto_init_mutex(name) static __attribute__((section(".mutex_array"))) mutex_t name
/*! \brief Helper macro for static definition of recursive mutexes
* \ingroup mutex
*
* A recursive mutex defined as follows:
*
* ```c
* auto_init_recursive_mutex(my_mutex);
* ```
*
* Is equivalent to doing
*
* ```c
* static mutex_t my_mutex;
*
* void my_init_function() {
* recursive_mutex_init(&my_mutex);
* }
* ```
*
* But the initialization of the mutex is performed automatically during runtime initialization
*/
#define auto_init_recursive_mutex(name) static __attribute__((section(".mutex_array"))) mutex_t name = { .recursion_state = MAX_RECURSION_STATE }
#ifdef __cplusplus
}
#endif

View File

@ -7,37 +7,53 @@
#include "pico/mutex.h"
#include "pico/time.h"
#if !PICO_NO_HARDWARE
static_assert(sizeof(mutex_t) == 8, "");
#endif
static void mutex_init_internal(mutex_t *mtx, uint8_t recursion_state) {
lock_init(&mtx->core, next_striped_spin_lock_num());
mtx->owner = LOCK_INVALID_OWNER_ID;
mtx->recursion_state = recursion_state;
__mem_fence_release();
}
void mutex_init(mutex_t *mtx) {
lock_init(&mtx->core, next_striped_spin_lock_num());
mtx->owner = -1;
__mem_fence_release();
mutex_init_internal(mtx, 0);
}
void recursive_mutex_init(mutex_t *mtx) {
mutex_init_internal(mtx, MAX_RECURSION_STATE);
}
void __time_critical_func(mutex_enter_blocking)(mutex_t *mtx) {
assert(mtx->core.spin_lock);
bool block = true;
do {
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
if (mtx->owner < 0) {
mtx->owner = (int8_t)get_core_num();
block = false;
lock_owner_id_t caller = lock_get_caller_owner_id();
if (mtx->owner == LOCK_INVALID_OWNER_ID) {
mtx->owner = caller;
if (mtx->recursion_state) {
assert(mtx->recursion_state == MAX_RECURSION_STATE);
mtx->recursion_state--;
}
} else if (mtx->owner == caller && mtx->recursion_state > 1) {
mtx->recursion_state--;
} else {
lock_internal_spin_unlock_with_wait(&mtx->core, save);
// spin lock already unlocked, so loop again
continue;
}
spin_unlock(mtx->core.spin_lock, save);
if (block) {
__wfe();
}
} while (block);
break;
} while (true);
}
bool __time_critical_func(mutex_try_enter)(mutex_t *mtx, uint32_t *owner_out) {
bool entered;
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
if (mtx->owner < 0) {
mtx->owner = (int8_t)get_core_num();
lock_owner_id_t caller = lock_get_caller_owner_id();
if (mtx->owner == LOCK_INVALID_OWNER_ID) {
mtx->owner = lock_get_caller_owner_id();
entered = true;
} else if (mtx->owner == caller && mtx->recursion_state > 1) {
mtx->recursion_state--;
entered = true;
} else {
if (owner_out) *owner_out = (uint32_t) mtx->owner;
@ -53,27 +69,41 @@ bool __time_critical_func(mutex_enter_timeout_ms)(mutex_t *mtx, uint32_t timeout
bool __time_critical_func(mutex_enter_block_until)(mutex_t *mtx, absolute_time_t until) {
assert(mtx->core.spin_lock);
bool block = true;
do {
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
if (mtx->owner < 0) {
mtx->owner = (int8_t)get_core_num();
block = false;
}
spin_unlock(mtx->core.spin_lock, save);
if (block) {
if (best_effort_wfe_or_timeout(until)) {
lock_owner_id_t caller = lock_get_caller_owner_id();
if (mtx->owner == LOCK_INVALID_OWNER_ID) {
mtx->owner = caller;
} else if (mtx->owner == caller && mtx->recursion_state > 1) {
mtx->recursion_state--;
} else {
if (lock_internal_spin_unlock_with_best_effort_wait_or_timeout(&mtx->core, save, until)) {
// timed out
return false;
} else {
// not timed out; spin lock already unlocked, so loop again
continue;
}
}
} while (block);
return true;
spin_unlock(mtx->core.spin_lock, save);
return true;
} while (true);
}
void __time_critical_func(mutex_exit)(mutex_t *mtx) {
uint32_t save = spin_lock_blocking(mtx->core.spin_lock);
assert(mtx->owner >= 0);
mtx->owner = -1;
__sev();
spin_unlock(mtx->core.spin_lock, save);
}
assert(mtx->owner != LOCK_INVALID_OWNER_ID);
if (!mtx->recursion_state) {
mtx->owner = LOCK_INVALID_OWNER_ID;
lock_internal_spin_unlock_with_notify(&mtx->core, save);
} else {
mtx->recursion_state++;
assert(mtx->recursion_state);
if (mtx->recursion_state == MAX_RECURSION_STATE) {
mtx->owner = LOCK_INVALID_OWNER_ID;
lock_internal_spin_unlock_with_notify(&mtx->core, save);
} else {
spin_unlock(mtx->core.spin_lock, save);
}
}
}

View File

@ -20,64 +20,57 @@ int __time_critical_func(sem_available)(semaphore_t *sem) {
}
void __time_critical_func(sem_acquire_blocking)(semaphore_t *sem) {
bool block = true;
do {
uint32_t save = spin_lock_blocking(sem->core.spin_lock);
if (sem->permits > 0) {
sem->permits--;
__sev();
block = false;
lock_internal_spin_unlock_with_notify(&sem->core, save);
break;
}
spin_unlock(sem->core.spin_lock, save);
if (block) {
__wfe();
}
} while (block);
lock_internal_spin_unlock_with_wait(&sem->core, save);
} while (true);
}
bool __time_critical_func(sem_acquire_timeout_ms)(semaphore_t *sem, uint32_t timeout_ms) {
bool block = true;
absolute_time_t target = nil_time;
do {
uint32_t save = spin_lock_blocking(sem->core.spin_lock);
if (sem->permits > 0) {
sem->permits--;
__sev();
block = false;
lock_internal_spin_unlock_with_notify(&sem->core, save);
return true;
}
spin_unlock(sem->core.spin_lock, save);
if (block) {
if (is_nil_time(target)) {
target = make_timeout_time_ms(timeout_ms);
}
if (best_effort_wfe_or_timeout(target)) {
return false;
}
if (is_nil_time(target)) {
target = make_timeout_time_ms(timeout_ms);
}
} while (block);
return true;
if (lock_internal_spin_unlock_with_best_effort_wait_or_timeout(&sem->core, save, target)) {
return false;
}
} while (true);
}
// todo this should really have a blocking variant for when permits are maxed out
bool __time_critical_func(sem_release)(semaphore_t *sem) {
bool rc;
uint32_t save = spin_lock_blocking(sem->core.spin_lock);
int32_t count = sem->permits;
if (count < sem->max_permits) {
sem->permits = (int16_t)(count + 1);
__sev();
rc = true;
lock_internal_spin_unlock_with_notify(&sem->core, save);
return true;
} else {
rc = false;
spin_unlock(sem->core.spin_lock, save);
return false;
}
spin_unlock(sem->core.spin_lock, save);
return rc;
}
void __time_critical_func(sem_reset)(semaphore_t *sem, int16_t permits) {
assert(permits >= 0 && permits <= sem->max_permits);
uint32_t save = spin_lock_blocking(sem->core.spin_lock);
if (permits > sem->permits) __sev();
sem->permits = permits;
spin_unlock(sem->core.spin_lock, save);
if (permits > sem->permits) {
sem->permits = permits;
lock_internal_spin_unlock_with_notify(&sem->core, save);
} else {
sem->permits = permits;
spin_unlock(sem->core.spin_lock, save);
}
}