On Thu, 30 Sep 2021 20:00:56 +0200 Pablo Neira Ayuso wrote: > On Thu, Sep 30, 2021 at 09:06:52AM -0700, Jakub Kicinski wrote: > > On Thu, 30 Sep 2021 17:13:37 +0200 Pablo Neira Ayuso wrote: > > > It's just one single bit in this case after all. > > > > ?? > > There are "escape" points such ifb from ingress, where the packets gets > enqueued and then percpu might not help, it might be fragile to use > percpu in this case. You still have to scrub the skb mark at the correct points, otherwise the ignoring egress may propagate beyond the "paired hook". I don't see much difference in fragility TBH. Speaking of ifb, doesn't it have an egress hook? And ingress on the way out? IMHO the "ignore egress" mark should not survive going thru ifb. Anyway, that's just my preference. Whatever you, Daniel and Lukas decide together in the end is fine by me.