On Mon, Jul 18, 2011 at 18:58, Ethan Jackson <et...@nicira.com> wrote:
>>> +static struct nx_action_bundle *
>>> +parse_bundle_actions(char *actions)
>>> +{
>>> + struct nx_action_bundle *nab;
>>> + struct ofpbuf b;
>>> +
>>> + ofpbuf_init(&b, 0);
>>> + bundle_parse(&b, actions);
>>> + nab = ofpbuf_steal_data(&b);
>>> + ofpbuf_uninit(&b);
>>> +
>>> + if (ntohs(nab->n_slaves) > BD_MAX_SLAVES) {
>>> + ovs_fatal(0, "At most %u slaves are supported", BD_MAX_SLAVES);
>>> + }
>>
>> This implies that bundle_parse() will parse more than BD_MAX_SLAVES
>> slaves. Shouldn't that be checked in bundle_parse() itself?
>
> I'm not sure I fully understand this comment. Hopefully the
> incremental I send out will make it clear that BD_MAX_SLAVES is only
> relevant to the testing code, and that the bundle library has it's own
> maximum (2048 slaves). I'll update bundle_parse to enforce this
> limit.
Oops, now that I look at the code, bundle_parse() does enforce that
only BUNDLE_MAX_SLAVES are used. If there are more it simply
truncates. I suppose I could have it ovs_fatal() out. I'll leave it
how it is pending a comment.
Ethan
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
