Le 02/07/2025 à 18:10, Jakub Kicinski a écrit : > On Wed, 2 Jul 2025 17:14:42 +0200 Nicolas Dichtel wrote: >>> Should we invert the polarity? It appears that the condition below only >>> let's this setting _disable_ forwarding. IMO calling it "force" suggests >>> to the user that it will force it to be enabled. >> Not sure to follow you. When force_forwarding is set to 1 the forwarding is >> always enabled. >> >> sysctl | all.forwarding | iface.force_forwarding | packet processing from iface >> | 0 | 0 | no forward >> | 0 | 1 | forward >> | 1 | 0 | forward >> | 1 | 1 | forward > > Ugh, I can't read comparisons to zero. > Let's switch to more sane logic: > > if (idev && !READ_ONCE(idev->cnf.force_forwarding) && > !READ_ONCE(net->ipv6.devconf_all->forwarding)) +1
