proxy/routes: Remove fromEgressProxyRule for cilium downgrade

[ upstream commit: 53133ff ]

[ backporter's note: v1.14 uses bpf/init.sh to install proxy rules so we
have to do a customized backport. ]

Although we don't install fromEgressProxyRule for now, this commit
insists on removing it to make sure further downgrade can go smoothly.

Soon We'll have another PR to install fromEgressProxyRule, and cilium
downgrade from that PR to branch tip (patch downgrade, 1.X.Y ->
1.X.{Y-1}) will be broken if we don't handle the new ip rule carefullly.

Without this patch, downgrade from higher version will leave
fromEgressProxyRule on the lower version cilium, cluster will be in a
wrong status of "having stale ip rule + not having other necessary
settings (iptables)", breaking the connectivity.

Signed-off-by: Zhichuan Liang <gray.liang@isovalent.com>
Signed-off-by: Zhichuan Liang <gray.liang@isovalent.com>

bpf/init.sh

@@ -135,6 +135,9 @@ function setup_proxy_rules()
 					ip -4 rule delete $from_ingress_rulespec 2> /dev/null || true
 				fi
 			fi
+                        if [ ! -z "$(ip -4 rule list $from_egress_rulespec)" ]; then
+                                ip -4 rule delete $from_egress_rulespec 2> /dev/null || true
+                        fi
 		fi
 
 		# Traffic to the host proxy is local
@@ -168,6 +171,9 @@ function setup_proxy_rules()
 					ip -6 rule delete $from_ingress_rulespec 2> /dev/null || true
 				fi
 			fi
+                        if [ ! -z "$(ip -6 rule list $from_egress_rulespec)" ]; then
+                                ip -6 rule delete $from_egress_rulespec 2> /dev/null || true
+                        fi
 		fi
 
 		IP6_LLADDR=$(ip -6 addr show dev $HOST_DEV2 | grep inet6 | head -1 | awk '{print $2}' | awk -F'/' '{print $1}')