Re: [PATCH RESEND net-next 0/5] Improve the taprio qdisc's relationship with its children

[Vladimir Oltean]

On Tue, Jun 06, 2023 at 01:42:19PM -0400, Jamal Hadi Salim wrote:
> > > 2) It seems like in mqprio this qdisc can only be root qdisc (like
> > > mqprio)
> >
> > so far so good
> >
> > > and you dont want to replace the children with other types of
> > > qdiscs i.e the children are always pfifo? i.e is it possible or
> > > intended for example to replace 8001:x with bfifo etc? or even change
> > > the pfifo queue size, etc?
> >
> > no, this is not true, why do you say this?
> 
> I am just asking questions trying to understand;-> So if can i
> replace, for example, with a tbf would it make sense even in s/w?
> 
> > The child could have its own complex enqueue() and dequeue() and that is
> > perfectly fine - for example cbs_dequeue_soft() is a valid child dequeue
> > procedure - as long as the process isn't blocked in the sendmsg() call
> > by __qdisc_run() processing packets belonging to unrelated traffic
> > classes.
> 
> Does it matter what type the child enqueue/dequeue? eg can i attach htb, etc?

So in principle, the taprio model is compatible with attaching any child
Qdisc to the per-TXQ child classes - with tc-cbs in particular being of
interest, because that is a TSN shaper, but also, tbf or htb could be
reasonably imagined as children, and taprio doesn't oppose to any Qdisc
as its child.

That being said, a non-offloaded cbs/htb will not work with an offloaded
taprio root anymore after commit 13511704f8d7 ("net: taprio offload:
enforce qdisc to netdev queue mapping"), and IMO what should be done
there is to reject somehow those child Qdiscs which also can't be
offloaded when the root is offloaded.

Offloading a taprio qdisc (potentially with offloaded cbs children) also
affects the autonomous forwarding data path in case of an Ethernet switch.
So it's not just about dev_queue_xmits() from Linux.