On 24 Oct 2024, at 17:46, Felix Huettner via dev wrote: > Sometimes it is not easily possible to convert the interface index to a > name. Especially if we run in network namespaces. > To still be able to act on existing information we also return the > ifindex now.
Why are we not changing the namespace and getting the interface name? Is this too much of an overhead? In addition, if the interface is not found, we set change->relevant = false, how does this affect the overall implementation? //Eelco > > Signed-off-by: Felix Huettner <[email protected]> > --- > lib/route-table.c | 1 + > lib/route-table.h | 1 + > 2 files changed, 2 insertions(+) > > diff --git a/lib/route-table.c b/lib/route-table.c > index 80325f6f9..ce70382a4 100644 > --- a/lib/route-table.c > +++ b/lib/route-table.c > @@ -228,6 +228,7 @@ route_table_add_nexthop(struct route_table_msg *change, > } > } > > + nh->ifindex = ifindex; > if (ifindex && !if_indextoname(ifindex, nh->ifname)) { > int error = errno; > > diff --git a/lib/route-table.h b/lib/route-table.h > index f96dbebb3..bcc3205e8 100644 > --- a/lib/route-table.h > +++ b/lib/route-table.h > @@ -30,6 +30,7 @@ > > struct route_data_nexthop { > struct in6_addr rta_gw; > + uint32_t ifindex; > char ifname[IFNAMSIZ]; /* Interface name. */ > }; > > -- _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
