Setting specific priority queue depth or rate

[edgar-costa]

Add a method to set the depth or rate of priority queues in the simple switch. At the moment the sswitch_CLI.py is using set_egress_queue_depth/rate to set the rate/depth of a given port but there was not way to set if for an individual priority queue.

This patch adds:

• set_egress_priority_queue_depth/rate in the simple_switch implementation
• Updated thrift files
• Modified do_set_queue_depth/rate from sswitch_CLI.py so they call the new functions when a priority is provided. Also added exception catchers when the parameters are not integers.

Update: I realized that this will not build if the multiqueueing is not enabled at compile time. This is because QueueingLogicRL does not know about priorities. What would be the right way to proceed?

#ifdef SSWITCH_PRIORITY_QUEUEING_ON
  bm::QueueingLogicPriRL<std::unique_ptr<Packet>, EgressThreadMapper>
#else
  bm::QueueingLogicRL<std::unique_ptr<Packet>, EgressThreadMapper>
#endif

[antoninbas]

Hi,
Thanks for opening this PR. Are you or your organization members of p4.org? We can only accept contributions from members.
Thanks.

[edgar-costa]

Hi Antonin,

Either me or my organization are members of the p4.org. However I already contacted my University and we started the procedure from: https://p4.org/join/.

What should I do once we are members?

[antoninbas]

Once you are members, I will get notified (although you can also let me know when the procedure is done). I can then review your PR & merge it.

[drobinkent]

Is the suuport for setting rate for priority queus available now?

[jafingerhut]

@sys-ptipu I am hoping by @'ing this Github id, a person respnosible for the (so far) three comments made recently will notice and respond. We would like to know how many PRs you plan to create such comments for in the future. If only three, no problem, but if all open PRs on repositories under https://github.com/p4lang, that would be a lot of "noise" in the PR comments, and it would be great if you might find a different way of recording this information.