[openib-general] [PATCH] [RFC] RDMA generic CMA updates

James Lentini jlentini at netapp.com
Tue Sep 27 07:43:52 PDT 2005



On Mon, 26 Sep 2005, Sean Hefty wrote:

> James Lentini wrote:
> > - move listen declaration closer to accept and reject
> 
> Accepted - will be pushed in with next version.
> 
> > - add private data and new cma_id fields to event structure
> 
> Added private_data_len field to cma_id structure.  Would like to get some
> additional feedback before adding the new_cma_id field.  If there are no
> objections, I'll add this.
> 
> > - record need to address information in the event structure
> 
> The address information is only sent in the CM REQ.  There shouldn't 
> be a need to carry it back in the CM REP.

ok
 
> > - implement private data handling for IB_CM_REQ_RECEIVED and
> > IB_CM_REP_RECEIVED
> 
> See below.
> 
> > @@ -177,8 +177,6 @@
> >  	if (!route->path_rec)
> >  		goto err;
> >  -	ib_event->private_data += sizeof *addr;
> 
> Used to skip address information sent in CM REQ.
> 
> >  	case IB_CM_REQ_RECEIVED:
> > -		cma_id_priv = cma_req_recv(cma_id_priv, ib_event);
> > -		if (!cma_id_priv)
> > +		new_cma_id_priv = cma_req_recv(cma_id_priv, ib_event);
> > +		if (!new_cma_id_priv)
> >  			return -ENOMEM;
> >  		event.event = RDMA_CMA_EVENT_CONNECT_REQUEST;
> > +		event.private_data = ib_event->private_data + +
> > sizeof struct cma_addr;
> private_data pointer is set at the end of this routine.

I see it now.

> 
> > +		event.private_data_len = IB_CM_REQ_PRIVATE_DATA_SIZE - +
> > sizeof struct cma_addr;
> added this.
> 
> > +		event.private_data = ib_event->private_data + +
> > sizeof struct cma_addr;
> > +		event.private_data_len = IB_CM_REQ_PRIVATE_DATA_SIZE - +
> > sizeof struct cma_addr;
> Set private_data_len = IB_CM_ *REP* _PRIVATE_DATA_SIZE.

right

> > -int rdma_cma_reject(struct rdma_cma_id *cma_id,
> > -		    const void *private_data, u8 private_data_len)
> > +int rdma_cma_reject(struct rdma_cma_id *cma_id, const void *private_data, +
> > u8 private_data_len)
> 
> I prefer that the private data variables appear together...

ok



More information about the general mailing list