Re: [PATCH net-next v2] net/sched: act_police: add support for packet-per-second policing

On Fri, Jan 29, 2021 at 09:30:00AM -0500, Jamal Hadi Salim wrote:

On 2021-01-29 5:28 a.m., Simon Horman wrote:

quoted

From: Baowen Zheng <redacted>

Allow a policer action to enforce a rate-limit based on packets-per-second,
configurable using a packet-per-second rate and burst parameters. This may
be used in conjunction with existing byte-per-second rate limiting in the
same policer action.

e.g.
tc filter add dev tap1 parent ffff: u32 match \
               u32 0 0 police pkts_rate 3000 pkts_burst 1000

Testing was unable to uncover a performance impact of this change on
existing features.

Ido's comment is important: Why not make packet rate vs byte rate
mutually exclusive? If someone uses packet rate then you make sure
they dont interleave with attributes for byte rate and vice-versa.

I dont see featurewise impact - but certainly the extra check
in the fastpath will likely have a small performance impact
at high rates. Probably not a big deal (if Eric D. is not looking).
Side note: this policer (with your addition) is now supporting 3 policer
algorithms - i wonder if it makes sense to start spliting the different
policers (which will solve the performance impact).

Sorry, I somehow missed Ido's email until you and he pointed it out
in this thread.

Regarding splitting up the policer action. I think there is some value to
the current setup in terms of code re-use and allowing combinations of
features. But I do agree it would be a conversation worth having at some
point.

On 2021-02-01 7:33 a.m., Simon Horman wrote:

On Fri, Jan 29, 2021 at 09:30:00AM -0500, Jamal Hadi Salim wrote:

quoted

Ido's comment is important: Why not make packet rate vs byte rate
mutually exclusive? If someone uses packet rate then you make sure
they dont interleave with attributes for byte rate and vice-versa.

Sorry, I somehow missed Ido's email until you and he pointed it out
in this thread.

This one i think is still important. Potential for misconfig
exists with both on.
The check for exclusivity is rather simple in init().
Also please see if you can add a test in the policer tests in tdc.

Regarding splitting up the policer action. I think there is some value to
the current setup in terms of code re-use and allowing combinations of
features. But I do agree it would be a conversation worth having at some
point.

Sounds reasonable.

cheers,
jamal