Hello Anatoly,
19/03/2026 17:07, Anatoly Burakov:
> Use a numeric request ID for alarm callback lookup so stale callbacks
> from rolled-back requests become harmless no-ops.
It seems you forgot to convert 2 calls to rte_eal_alarm_set().
[...]
> struct pending_request {
> TAILQ_ENTRY(pending_request) next;
> + unsigned long id;
[...]
> +static struct pending_request *
> +find_pending_request_by_id(unsigned long id)
> +{
> + struct pending_request *r;
> +
> + TAILQ_FOREACH(r, &pending_requests.requests, next) {
> + if (r->id == id)
> + return r;
> + }
> +
> + return NULL;
> +}
[...]
> async_reply_handle(void *arg)
> {
> struct pending_request *req;
> + /* alarm arg carries the request ID packed into a void * via uintptr_t
> */
No, that's wrong.
The pointer passed in some rte_eal_alarm_set() from patch 2
is a struct pending_request.
> + unsigned long id = (uintptr_t)arg;
Either you get the id field from arg,
or you pass dummy->id to rte_eal_alarm_set().
> pthread_mutex_lock(&pending_requests.lock);
> - req = async_reply_handle_thread_unsafe(arg);
> + req = find_pending_request_by_id(id);
> + if (req != NULL)
> + req = async_reply_handle_thread_unsafe(req);
> pthread_mutex_unlock(&pending_requests.lock);
