This is a valid request, not sure if just changing the check is enough because it also matters where packets match endpoints.
Begin forwarded message: Date: Mon, 06 May 2019 21:18:30 +0000 From: bugzilla-dae...@bugzilla.kernel.org To: step...@networkplumber.org Subject: [Bug 203533] New: VxLAN: Cannot create multiple vxlan devices for a given VNI and different source addresses. https://bugzilla.kernel.org/show_bug.cgi?id=203533 Bug ID: 203533 Summary: VxLAN: Cannot create multiple vxlan devices for a given VNI and different source addresses. Product: Networking Version: 2.5 Kernel Version: 4.9.110 Hardware: All OS: Linux Tree: Mainline Status: NEW Severity: normal Priority: P1 Component: IPV4 Assignee: step...@networkplumber.org Reporter: hasanrazaonl...@gmail.com Regression: No When trying to create two vxlan terminations with different source VTEPs, only one termination is allowed and second netdevice creation fails with EEXIST error. root@:~# ip link add vxlan-10 type vxlan id 10 local 10.1.1.1 dstport 4789 root@:~# ip link add vxlan2-10 type vxlan id 10 local 20.1.1.1 dstport 4789 RTNETLINK answers: File exists root@:~# It is a valid use-case scenario where a router is terminating multiple VxLAN endpoints. But this doesn't seem to be allowed in the kernel. The EEXIST error is coming from here: vxlan_dev_configure(): ... list_for_each_entry(tmp, &vn->vxlan_list, next) { if (tmp->cfg.vni == conf->vni && | (tmp->default_dst.remote_ip.sa.sa_family == AF_INET6 || | tmp->cfg.saddr.sa.sa_family == AF_INET6) == use_ipv6 && | tmp->cfg.dst_port == vxlan->cfg.dst_port && | (tmp->flags & VXLAN_F_RCV_FLAGS) == | (vxlan->flags & VXLAN_F_RCV_FLAGS)) { pr_info("duplicate VNI %u\n", be32_to_cpu(conf->vni)); return -EEXIST; } } The uniqueness of vxlan device should be checked with <local-endpoint, vni, dstport> instead of just <vni, dstport>. Similarly, vxlan_vs_find_vni() should lookup vxlan device using <src-addr, vni> instead of just <vni>: static struct vxlan_dev *vxlan_vs_find_vni(struct vxlan_sock *vs, __be32 vni) As per the code, I can see this issue exists in v5.1 as well. -- You are receiving this mail because: You are the assignee for the bug.
