On Tue, May 20, 2014 at 12:25:50PM -0700, Jarno Rajahalme wrote:
> Acked-by: Jarno Rajahalme <[email protected]>
Thanks.
> > @@ -733,7 +731,7 @@ do_add_port(struct dp_netdev *dp, const char *devname,
> > const char *type,
> > }
> > ovs_refcount_init(&port->ref_cnt);
> >
> > - hmap_insert(&dp->ports, &port->node, hash_int(odp_to_u32(port_no), 0));
> > + cmap_insert(&dp->ports, &port->node, hash_int(odp_to_u32(port_no), 0));
>
> Could use the new hash_port_no() here.
Thanks, I made that change here and in another place.
> > @@ -1783,10 +1784,9 @@ dpif_netdev_run(struct dpif *dpif)
> > {
> > struct dp_netdev_port *port;
> > struct dp_netdev *dp = get_dp_netdev(dpif);
> > + struct cmap_cursor cursor;
> >
> > - ovs_rwlock_rdlock(&dp->port_rwlock);
> > -
>
> Unless it is OK to sometimes skip a port (due a concurrent insert,
> etc.) then we should still exclude writers here.
It's OK to skip a port here. We'll hit it on the next call.
> > @@ -1804,10 +1802,9 @@ dpif_netdev_wait(struct dpif *dpif)
> > {
> > struct dp_netdev_port *port;
> > struct dp_netdev *dp = get_dp_netdev(dpif);
> > + struct cmap_cursor cursor;
> >
> > - ovs_rwlock_rdlock(&dp->port_rwlock);
> > -
>
> Same here.
I thought I had a reason it wasn't needed here, but I can't remember
it. I added the lock.
> > @@ -1831,12 +1827,12 @@ pmd_load_queues(struct pmd_thread *f,
> > struct dp_netdev *dp = f->dp;
> > struct rxq_poll *poll_list = *ppoll_list;
> > struct dp_netdev_port *port;
> > + struct cmap_cursor cursor;
> > int id = f->id;
> > int index;
> > int i;
> >
> > /* Simple scheduler for netdev rx polling. */
> > - ovs_rwlock_rdlock(&dp->port_rwlock);
>
> And here.
This shouldn't need the lock; it gets called again if the port
membership changes, based on port_seq values.
_______________________________________________
dev mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/dev
