If not specified, Delegated Prefixes will get an automatic SLA of the

interface index. If the biggest SLA and the assigned prefix fits into a /64
then dhcpcd creates a /64 prefix so that SLAAC works. If bigger than /64
is needed then dhcpcd creates one rounded upto the nearest multiple of 8.

Unless a configured SLA of 0 is assigned, a reject route for the Delegated
Prefix is installed to stop unassigned addresses trying to be resolved upstream.

Addresses added from Delegated Prefixes now have a default address suffix of 1
instead of using a SLAAC style address.

1 files changed, 6 insertions, 2 deletions

diff --git a/ipv6.c b/ipv6.c
index 0edb200d..b20bcd2a 100644
--- a/ipv6.c
+++ b/ipv6.c
@@ -328,8 +328,8 @@ ipv6_userprefix(
 {
 	uint64_t vh, vl, user_low, user_high;
 
-	if (prefix_len < 0 || prefix_len > 64 ||
-	    result_len < 0 || result_len > 64)
+	if (prefix_len < 0 || prefix_len > 120 ||
+	    result_len < 0 || result_len > 120)
 	{
 		errno = EINVAL;
 		return -1;
@@ -869,6 +869,10 @@ make_prefix(const struct interface * ifp, const struct ra *rap,
 	    !(addr->flags & (IPV6_AF_ONLINK | IPV6_AF_DELEGATEDPFX)))
 		return NULL;
 
+	/* Don't install a blackhole route when not creating bigger prefixes */
+	if (addr->flags & IPV6_AF_DELEGATEDZERO)
+		return NULL;
+
 	r = make_route(ifp, rap);
 	if (r == NULL)
 		return NULL;