michaelh/zephyr
1
0
Fork 0
Code Releases Activity

kernel/queue: Clean up scheduler API usage

Browse source 
This was the only spot where the scheduler-internal
_peek_first_pending_thread() API was used.  Given that this kind of
thing is inherently racy (it may not be pending as long as you expect
if a timeout expires, etc...), it would be nice to retire it.

And as it happens all the queue code was using it for was to detect
the case of a non-empty wait_q over which it was looping, which is
trivial to do without API support.

Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This commit is contained in:
Andy Ross 2018-03-09 13:05:15 -08:00 • committed by Anas Nashif
commit 345553b19b
1 changed files with 9 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

18
kernel/queue.c
View file

@ -148,26 +148,21 @@ void k_queue_append_list(struct k_queue *queue, void *head, void *tail)
{
__ASSERT(head && tail, "invalid head or tail");
int need_sched = 0;
unsigned int key = irq_lock();
#if !defined(CONFIG_POLL)
struct k_thread *first_thread, *thread;
struct k_thread *thread;
first_thread = _peek_first_pending_thread(&queue->wait_q);
while (head && ((thread = _unpend_first_thread(&queue->wait_q)))) {
prepare_thread_to_run(thread, head);
head = *(void **)head;
need_sched = 1;
}
if (head) {
sys_slist_append_list(&queue->data_q, head, tail);
}
if (first_thread) {
if (!_is_in_isr() && _must_switch_threads()) {
(void)_Swap(key);
return;
}
}
#else
sys_slist_append_list(&queue->data_q, head, tail);
if (handle_poll_events(queue, K_POLL_STATE_DATA_AVAILABLE)) {
@ -176,7 +171,12 @@ void k_queue_append_list(struct k_queue *queue, void *head, void *tail)
}
#endif /* !CONFIG_POLL */
irq_unlock(key);
if (need_sched) {
_reschedule_threads(key);
return;
} else {
irq_unlock(key);
}
}
void k_queue_merge_slist(struct k_queue *queue, sys_slist_t *list)