Seems we could shrink the structure by removing the condition variable
itself, and eliminate the signal call since it does appear to be unused.
cq_pop() does not have a mode where it waits for a cq_push().
And while we are at it, you can also optimize cq_pop() on the no connection
case to avoid taking the lock (you only have to lock if something is there
and you are going to modify the memory:
static CQ_ITEM *cq_pop(CQ *cq) {
CQ_ITEM *item;
if (NULL == cq->head)
return NULL;
pthread_mutex_lock(&cq->lock);
item = cq->head;
if (NULL != item) {
cq->head = item->next;
if (NULL == cq->head)
cq->tail = NULL;
}
pthread_mutex_unlock(&cq->lock);
return item;
}
-peter
On 8/13/10 4:08 AM, "ilnarb" <[email protected]> wrote:
> I found in thread.c:109
> pthread_cond_signal(&cq->cond);
> without any waiting for that condition variable.
> I don't know, will it helpful if you eliminate using it.
