mpipe_workqueue.h File Reference

Shared P4WQ pool for element-level work parallelism. More...

#include <zephyr/sys/p4wq.h>

Go to the source code of this file.

Variables
struct k_p4wq *const	mpipe_p4wq
	Shared mpipe P4WQ pool instance.

Detailed Description

Shared P4WQ pool for element-level work parallelism.

Elements use the Zephyr P4WQ API directly, k_p4wq_submit() and k_p4wq_wait(), on the shared mpipe_p4wq instance. An element processing tiles in parallel:

#include <zephyr/mpipe/mpipe_workqueue.h>
 
static void tile_handler(struct k_p4wq_work *work) {
    // process one tile...
}
 
static int my_process_fn(struct mpipe_pad *pad, struct net_buf *in,
                      struct net_buf **out) {
    struct k_p4wq_work tile_work[4];
 
    for (int i = 0; i < 4; i++) {
        tile_work[i].handler = tile_handler;
        tile_work[i].priority = 7;
        tile_work[i].deadline = 0;
        tile_work[i].sync = true;
        k_sem_init(&tile_work[i].done_sem, 0, 1);
        k_p4wq_submit(mpipe_p4wq, &tile_work[i]);
    }
    for (int i = 0; i < 4; i++) {
        k_p4wq_wait(&tile_work[i], K_FOREVER);
    }
    // assemble output...
    *out = result;
    return 0;
}