On Thu, Jun 9, 2016 at 1:02 PM, William Tu <u9012...@gmail.com> wrote:
>
>> >
>> > /* The only known usage of sample action is having a single
>> > user-space
>> > + * action, or having a truncate action followed by a single
>> > user-space
>> > * action. Treat this usage as a special case.
>> > * The output_userspace() should clone the skb to be sent to the
>> > * user space. This skb will be consumed by its caller.
>> > */
>> > + if (unlikely(nla_type(a) == OVS_ACTION_ATTR_TRUNC)) {
>> > + struct ovs_action_trunc *trunc = nla_data(a);
>> > +
>> > + OVS_CB(skb)->cutlen = skb->len > trunc->max_len ?
>> > + skb->len -
>> > trunc->max_len : 0;
>>
>> There is no need to initialize the cutlen to zero if max-len is
>> greater then skb len. it should be already zero.
>>
>> > + a = nla_next(a, &rem);
>> > + }
>> > +
>> There cutlen is set in skb cb, but next action could be other than
>> userspace, in that case cutlen is leaked to remain action list.
>>
>
> Thanks! I will fix it.
>
> Thinking about cutlen to the remaining action list, should we consider the
> truncate and output in sample action? ex: "sample(truncate(64), output:2)"
>
The patch already covers it. sample calls add_deferred_actions() which
would execute nested set of actions later on.
