Re: RFC/PATCH capture qdisc requeue event in stats

[Stephen Hemminger <shemminger@xxxxxxxx>]

On 30 Aug 2004 18:56:32 -0400
jamal <hadi@xxxxxxxxxx> wrote:

> On Mon, 2004-08-30 at 18:44, David S. Miller wrote:
> > On 30 Aug 2004 18:14:48 -0400
> > jamal <hadi@xxxxxxxxxx> wrote:
> 
> > Check the size, if it matches the older tc_stats size sans
> > requeue stat addition, then don't try to reference that struct member.
> > 
> >     if (RTA_PAYLOAD(tb[TCA_STATS]) == sizeof(struct tc_stats_old)) {
> >             /* report everything sans requeue */
> >     } else if (RTA_PAYLOAD(tb[TCA_STATS]) == sizeof(struct tc_stats)) {
> >             /* report all stats, including requeue */
> >     } else {
> >             /* report error, etc. */
> >     }
> 
> Sounds reasonable to me. Some of the BSDs do this to maintain old compat
> - just means keeping old struct around. Steve, agreeable to you? 
> 
> Change to both kernel and user space or just user space?

I have no problem but easier to just do something:
        struct tc_stats mystats;

        memset(&mystats, 0, sizeof(mystats));
        memcpy(&mystats, RTA_DATA(tb[TCA_STATS]), RTA_PAYLOAD(tb[TCA_STATS]));

that way it can grow as much as we want and don't have v1, v2, v3, ... 
size structures.