Hi Petr,

On Tue, 26 May 2020 at 20:11, Petr Machata [off-list ref] wrote:

The Spectrum hardware allows execution of one of several actions as a
result of queue management events: tail-dropping, early-dropping, marking a
packet, or passing a configured latency threshold or buffer size. Such
packets can be mirrored, trapped, or sampled.

Modeling the action to be taken as simply a TC action is very attractive,
but it is not obvious where to put these actions. At least with ECN marking
one could imagine a tree of qdiscs and classifiers that effectively
accomplishes this task, albeit in an impractically complex manner. But
there is just no way to match on dropped-ness of a packet, let alone
dropped-ness due to a particular reason.

To allow configuring user-defined actions as a result of inner workings of
a qdisc, this patch set introduces a concept of qevents. Those are attach
points for TC blocks, where filters can be put that are executed as the
packet hits well-defined points in the qdisc algorithms. The attached
blocks can be shared, in a manner similar to clsact ingress and egress
blocks, arbitrary classifiers with arbitrary actions can be put on them,
etc.

For example:

# tc qdisc add dev eth0 root handle 1: \
        red limit 500K avpkt 1K qevent early block 10
# tc filter add block 10 \
        matchall action mirred egress mirror dev eth1

Patch #1 of this set introduces several helpers to allow easy and uniform
addition of qevents to qdiscs. The following two patches, #2 and #3, then
add two qevents to the RED qdisc: "early" qevent fires when a packet is
early-dropped; "mark" qevent, when it is ECN-marked.

This patch set does not deal with offloading. The idea there is that a
driver will be able to figure out that a given block is used in qevent
context by looking at binder type. A future patch-set will add a qdisc
pointer to struct flow_block_offload, which a driver will be able to
consult to glean the TC or other relevant attributes.

Petr Machata (3):
  net: sched: Introduce helpers for qevent blocks
  net: sched: sch_red: Split init and change callbacks
  net: sched: sch_red: Add qevents "early" and "mark"

 include/net/flow_offload.h     |   2 +
 include/net/pkt_cls.h          |  48 +++++++++++++++
 include/uapi/linux/pkt_sched.h |   2 +
 net/sched/cls_api.c            | 107 +++++++++++++++++++++++++++++++++
 net/sched/sch_red.c            | 100 ++++++++++++++++++++++++++----
 5 files changed, 247 insertions(+), 12 deletions(-)

--
2.20.1

I only took a cursory glance at your patches. Can these "qevents" be
added to code outside of the packet scheduler, like to the bridge, for
example? Or can the bridge mark the packets somehow, and then any
generic qdisc be able to recognize this mark without specific code?
A very common use case which is currently not possible to implement is
to rate-limit flooded (broadcast, unknown unicast, unknown multicast)
traffic. Can your "qevents" be used to describe this, or must it be
described separately?

Thanks,
-Vladimir