On Wed, 2016-06-01 at 08:53 -0700, Stephen Hemminger wrote:
> On Wed, 01 Jun 2016 07:23:31 -0700
> Eric Dumazet <eric.duma...@gmail.com> wrote:
> > + } else if (strcmp(*argv, "memory_limit") == 0) {
> > + NEXT_ARG();
> > + if (get_size(&memory, *argv)) {
> > + fprintf(stderr, "Illegal \"memory_limit\"\n");
> > + return -1;
> > + }
>
> Do you really want to allow memory limit of 0?
>
Why not ? If there is a kernel bug we need to fix it anyway ?
What would be the enforced 'minimum' ?
>
> > + if (tb[TCA_FQ_CODEL_MEMORY_LIMIT] &&
> > + RTA_PAYLOAD(tb[TCA_FQ_CODEL_MEMORY_LIMIT]) >= sizeof(__u32)) {
> > + memory_limit = rta_getattr_u32(tb[TCA_FQ_CODEL_MEMORY_LIMIT]);
> > +
> > + fprintf(f, "memory_limit %s ", sprint_size(memory_limit, b1));
> > + }
>
> Why the size check? other parameters don't do it?
Existing code style in tc/q_fq_codel.c
I really think all TLV consumers should never trust producers, even if
it is the kernel ;)
