adafruit-mirror/zephyr
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 9

kernel/mempool: Spinlockify

Browse source 
Really the locking in this file is vestigial.  It only exists because
the scheduler's _unpend_all() call to wake up everyone waiting on a
wait_q is unsynchronized, because it was written to assume
irq_lock-style-locking.  It would be cleaner to put that locking into
the wait_q itself and/or use the scheduler's subsystem lock.  But it's
not clear there's any performance benefit, so let's stick with the
more easily verifiable change first.

Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This commit is contained in:
Andy Ross 2018-07-25 10:24:16 -07:00 committed by Anas Nashif
parent f2b1a4bb64
commit 84b47a9290
1 changed files with 11 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
kernel/mempool.c
View file

@ -16,6 +16,8 @@
extern struct k_mem_pool _k_mem_pool_list_start[];
extern struct k_mem_pool _k_mem_pool_list_end[];
static struct k_spinlock lock;
static struct k_mem_pool *get_pool(int id)
{
return &_k_mem_pool_list_start[id];
@ -106,22 +108,26 @@ int k_mem_pool_alloc(struct k_mem_pool *p, struct k_mem_block *block,
void k_mem_pool_free_id(struct k_mem_block_id *id)
{
int key, need_sched = 0;
int need_sched = 0;
struct k_mem_pool *p = get_pool(id->pool);
_sys_mem_pool_block_free(&p->base, id->level, id->block);
/* Wake up anyone blocked on this pool and let them repeat
* their allocation attempts
*
* (Note that this spinlock only exists because _unpend_all()
* is unsynchronized. Maybe we want to put the lock into the
* wait_q instead and make the API safe?)
*/
key = irq_lock();
k_spinlock_key_t key = k_spin_lock(&lock);
need_sched = _unpend_all(&p->wait_q);
if (need_sched && !_is_in_isr()) {
_reschedule_irqlock(key);
if (need_sched) {
_reschedule(&lock, key);
} else {
irq_unlock(key);
k_spin_unlock(&lock, key);
}
}