> - id_priv->query_id = ib_sa_path_rec_get(&sa_client, id_priv->id.device,
> + if (cma_qos_support) {
> + ib_sa_comp_mask m = IB_SA_PATH_REC_DGID | IB_SA_PATH_REC_SGID |
> + IB_SA_PATH_REC_PKEY | IB_SA_PATH_REC_NUMB_PATH;
> + struct sockaddr *sockaddr = &id_priv->id.route.addr.dst_addr;
> + __be64 svc_id = cma_get_service_id(id_priv->id.ps, sockaddr);
> +
> + path_rec.service_id = svc_id;
> + m |= IB_SA_PATH_REC_SERVICE_ID;
> + if (id_priv->priority) {
> + path_rec.priority = __cpu_to_be16(id_priv->priority);
> + m |= IB_SA_PATH_REC_PRIORITY;
> + }
> + id_priv->query_id = ib_sa_path_rec_get(&sa_client,
> id_priv->id.device,
> + id_priv->id.port_num, &path_rec, m,
> + timeout_ms, GFP_KERNEL,
> + cma_query_handler, work, &id_priv->query);
> + } else
> + id_priv->query_id = ib_sa_path_rec_get(&sa_client,
> id_priv->id.device,
> id_priv->id.port_num, &path_rec,
> IB_SA_PATH_REC_DGID | IB_SA_PATH_REC_SGID |
> IB_SA_PATH_REC_PKEY | IB_SA_PATH_REC_NUMB_PATH,
Minor comment that we might be able to restructure this slightly to have a
single call to ib_sa_path_rec_get().
What I'm not as sure about is separating the call to set the priority from
resolving the route. My first impression was that this was a good idea, but it
may give the impression that the priority can change after route resolution.
Does anyone else have any thoughts on this?
- Sean
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
