Feature requirement: Skip the conflict detection function of creating a network with other network address segments

[InPRTx]

I have a very useful intranet, which can access the container's IP by using the internal IPv6 address.
However, I have encountered problems in creating a network during use. I hope to add instructions to skip the detection of address segment conflicts.
When I use this command, an error will be reported. I hope to skip the error and bypass the detection.

podman network create --label io.podman.compose.project=app --label com.docker.compose.project=app --ipv6 --subnet fc6a:795e:392:20:2ac1:464c:a2cf::/112 app_in-net2

Error: subnet fc6a:795e:392:20:2ac1:464c:a2cf:0/112 is already used on the host or by another config

But in fact, this route is only included in fc6a:795e::/32 and does not conflict

ip -6 r|grep fc6a
fc6a:795e:392:20:9a11:464c:ea1f:0/112 dev podman1 proto kernel metric 256 pref medium
fc6a:795e:528::/48 via fe80::ff:fe30:334 dev tinc0 proto bird metric 32 pref medium
fc6a:795e:702::/48 via fe80::ff:fe30:334 dev tinc0 proto bird metric 32 pref medium
fc6a:795e::/32 dev tinc0 proto kernel metric 256 pref medium

common/libnetwork/internal/util/validate.go

Lines 31 to 36 in bbb5c4a

	// check that the new subnet does not conflict with existing ones
	if NetworkIntersectsWithNetworks(n, usedNetworks) {
	return fmt.Errorf("subnet %s is already used on the host or by another config", n.String())
	}

common/libnetwork/internal/util/ip.go

Lines 45 to 56 in bbb5c4a

	func NetworkIntersectsWithNetworks(n *net.IPNet, networklist []*net.IPNet) bool {
	for _, nw := range networklist {
	if networkIntersect(n, nw) {
	return true
	}
	}
	return false
	}
	func networkIntersect(n1, n2 *net.IPNet) bool {
	return n2.Contains(n1.IP) || n1.Contains(n2.IP)
	}

[Luap99]

We do not check check routes, we do check the ip addresses and subnets assigned to the interfaces. I guess this just doesn't make sense for ipv6 as this kind of subnetting seems like normal thing there.

I would not like to maintain a option to skip such a check, if it doesn't work on "common" ipv6 networks we should fix it so it does work and not require extra options. I am not sure what the right fix is, should we skip the check entirely for ipv6 or parse the routing rules to figure out if there will be conflicts?

cc @mheon

[InPRTx]

Got it. I solved this problem by reducing the size of the ipv6 subnet.

[Luap99]

Let's keep it open, I think this a valid issue regardless.