Thank you for the feedback. I will re-implement the test by not checking
port-flow-queue combination and generally clean-up the code based on your
comments.
On Tue, 2024-12-10 at 11:37 +0100, Mattias Rönnblom wrote:
>
> > +{
> > + struct rte_event_dev_info dev_info;
> > +
> > + rte_event_dev_info_get(dev_id, &dev_info);
> > + return (dev_info.event_dev_cap &
> > RTE_EVENT_DEV_CAP_MAINTENANCE_FREE) ?
> > + true : false;
>
> return dev_info.event_dev_cap & RTE_EVENT_DEV_CAP_MAINTENANCE_FREE;
>
> will work fine.
>
>
I decided against it in order to maintain consistent styling with similar
functions in the file.
>
> > +static int
> > +worker_wrapper(void *arg)
>
> Delete "wrapper".
All other eventdev-tests name their equivalent functions "worker_wrapper", so I
picked it to be consistent with the other tests.
>
> > +
> > + /* setup one port per worker, linking to all queues */
> > + ret = order_event_dev_port_setup(test, opt, nb_workers, NB_QUEUES);
>
> "order"?
This function is declared in test_order_common.h and is used in all tests. It
is not specific for "ordered" ports, so I thought it was OK to use.
> > +
> > +static void
> > +atomic_queue_opt_dump(struct evt_options *opt)
> > +{
> > + order_opt_dump(opt);
>
> "order"?
Same thing here.
