Re: [PATCH v3 net-next] net/sched: add skbprio scheduler

[Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>]

On Tue, Jul 10, 2018 at 07:25:53PM -0700, Cong Wang wrote:
> On Mon, Jul 9, 2018 at 2:40 PM Marcelo Ricardo Leitner
> <marcelo.leitner@gmail.com> wrote:
> >
> > On Mon, Jul 09, 2018 at 05:03:31PM -0400, Michel Machado wrote:
> > >    Changing TC_PRIO_MAX from 15 to 63 risks breaking backward compatibility
> > > with applications.
> >
> > If done, it needs to be done carefully, indeed. I don't know if it's
> > doable, neither I know how hard is your requirement for 64 different
> > priorities.
> 
> struct tc_prio_qopt {
>         int     bands;                  /* Number of bands */
>         __u8    priomap[TC_PRIO_MAX+1]; /* Map: logical priority -> PRIO band */
> };
> 
> How would you do it carefully?

quick shot, multiplex v1 and v2 formats based on bands and sizeof():

#define TCQ_PRIO_BANDS_V1	16
#define TCQ_PRIO_BANDS_V2	64
#define TC_PRIO_MAX_V2		64

struct tc_prio_qopt_v2 {
        int     bands;                  /* Number of bands */
        __u8    priomap[TC_PRIO_MAX_V2+1]; /* Map: logical priority -> PRIO band */
};

static int prio_tune(struct Qdisc *sch, struct nlattr *opt,
                     struct netlink_ext_ack *extack)
{
        struct prio_sched_data *q = qdisc_priv(sch);
        struct Qdisc *queues[TCQ_PRIO_BANDS_V2];
        int oldbands = q->bands, i;
        struct tc_prio_qopt_v2 *qopt;

        if (nla_len(opt) < sizeof(int))
                return -EINVAL;
        qopt = nla_data(opt);

	if (qopt->bands <= TCQ_PRIO_BANDS_V1 &&
            nla_len(opt) < sizeof(struct tc_prio_qopt))
                return -EINVAL;

	if (qopt->bands <= TCQ_PRIO_BANDS_V2 &&
            nla_len(opt) < sizeof(*qopt))
                return -EINVAL;

	/* By here, if it has up to 3 bands, we can assume it is using the _v1
	 * layout, while if it has up to TCQ_PRIO_BANDS_V2 it is using the _v2
	 * format.
	 */

        if (qopt->bands > TCQ_PRIO_BANDS_V2 || qopt->bands < 2)
                return -EINVAL;
...

With something like this I think it can keep compatibility with old
software while also allowing the new usage.

> Also, it is not only used by prio but also pfifo_fast.

Yes. More is needed, indeed. prio2band would also need to be expanded,
etc. Yet, I still don't see any blocker.