On Tue, Nov 05, 2024 at 05:33:08PM +0100, Florian Westphal wrote: > Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx> wrote: > > > It will help for SEEN_REPLY. But I don't see how it will avoid this > > > patch. > > > > Not current time from ctnetlink, but use the ecache extension to store > > the timestamp when the conntrack is allocated, ecache is already > > initialized from init_conntrack() path. > > OK, so we do ktime_get_real() twice. > I think its way worse than this proposal, but okay. My proposal is to add more well-known "fixed points" to get numbers. At this stage, there is start= and stop= ktimestamps, where start= represents insertion to hashes (confirmation time). I think this adds more ktimestamp point that can be enabled to collect numbers in an optional fashion, both two timestamps do not need to be turned on necessarily. > I'll work on this. Thanks, I'd rather convince you this is the way to go, if after quickly sketching a patchset you think it is not worth for more reasons, we can revisit.
