Hello, Sean!
Thanks for looking at this.
Quoting Sean Hefty <[EMAIL PROTECTED]>:
> + INIT_WORK(work, loopback_addr_handler, id_priv);
> + queue_work(wq, work);
> + return 0;
> +err:
> + kfree(work);
> + return ret;
> +}
I'm not following this: why are we deferring the work?
Cant it be done directly?
> @@ -1508,12 +1602,17 @@ static void cma_remove_one(struct ib_dev
>
> static int cma_init(void)
> {
> + wq = create_singlethread_workqueue("rdma_cm");
> + if (!wq)
> + return -ENOMEM;
> +
> return ib_register_client(&cma_client);
> }
>
> static void cma_cleanup(void)
> {
> ib_unregister_client(&cma_client);
> + destroy_workqueue(wq);
> }
>
> module_init(cma_init);
It would be nice to avoid adding another workqueue.
--
MST
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
