RE: [PATCH v4 1/9] IB/cma: pass the port number to ib_create_qp

["Steve Wise" <swise-7bPotxP6k4+P2YhJcF5u+vpXobYPEAuW@public.gmane.org>]

> On Tue, Mar 08, 2016 at 09:55:14PM +0200, Leon Romanovsky wrote:
> > > At the expense of replicating this code and forcing all users to
> remember to set this.
> >
> > There are no much such users to update, these users need to set
> > qp_init_attr structure anyway.
> 
> But they need to extract the port_num first, while we already get
> the cm_id that has the right port_id passed to this function.  Not
> setting it in the qp_init_attr changes the interface from one that
> just works to one that is arcane, and prone to generate hard to
> detect errors (passing a 0 port_num will just work for all current
> drivers, but if someone at some point actually introduces different
> capabilities for differnet ports it will break for just that case!)

While I'm for keeping your change as-is, it turns out the port numbers
are 1 relative, so 0 could be treated as an error. From
read_port_immutable():


        /**
         * device->port_immutable is indexed directly by the port number
to make
         * access to this data as efficient as possible.
         *
         * Therefore port_immutable is declared as a 1 based array with
         * potential empty slots at the beginning.
         */
        device->port_immutable = kzalloc(sizeof(*device->port_immutable)
                                         * (end_port + 1),
                                         GFP_KERNEL);

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html