mirror of https://github.com/jmcorgan/fips.git synced 2026-10-30 07:37:36 +00:00

Files

T

Johnathan Corgan b294d6d405 Report a mesh firewall scoped to another interface as not policing the TUN

fips.nft scopes its inbound chain with an iifname return for "fips0".
With tun.name set to anything else the chain filters nothing on the
TUN, yet the listening-sockets report still claimed an active firewall
and showed each listener as dropped, so fipstop hid its "all listeners
exposed" banner.

The classifier now reads the chain's interface scope from the loaded
ruleset and answers whether it polices a given TUN: an unscoped chain
does, a chain scoped to exactly that name does, and one scoped to
another name or in a shape it cannot read does not. The scope is taken
from the first rule that returns on or matches the input interface by
any key, so a chain scoped by iif, iifgroup or an iifname verdict map,
or by a wildcard or named-set iifname, counts as unreadable rather than
unscoped. When the loaded table does not police the node's TUN, the
report says no firewall is active and every listener is exposed. On
Linux the daemon also checks once when it brings the TUN up and warns,
naming both interfaces, when the loaded table is scoped to another one.

Unit tests render the report for a TUN named mesh0 against the ruleset
the shipped fips.nft loads and expect every listener reported
unfiltered, pin the filtered report when the TUN is fips0, and cover
the classifier's reading of each scope shape.

fips.nft now names the interface once, in a fips_if define, and its
header, the mesh firewall how-to, the tun.name reference row and the
install README say it must match tun.name. The firewall suite gains a
case that loads the file scoped to another interface and checks that
mesh traffic gets through, that the report says so, and that reloading
the real file restores the drop.

2026-10-14 10:02:51 +00:00

5.3 KiB

Raw Permalink Blame History

Firewall Baseline Test

End-to-end exercise of the production fips0 nftables baseline at packaging/common/fips.nft. Closes the v0.3.0 audit gap that the default-deny + conntrack + drop-in semantics had no integration coverage.

What this exercises

The fips.nft baseline polices ONLY the fips0 mesh interface and implements default-deny inbound. This suite asserts the four behaviors documented in the file's header are actually true on a live mesh:

(a) Unallowed inbound on fips0 is dropped
(b) Outbound-initiated flows get their reply via the ct state established,related accept rule
(c) ICMPv6 echo-request is accepted (ping6 reachability)
(d) A drop-in .nft file under /etc/fips/fips.d/ adds an allowlisted port and that port is accepted

A drop-counter check after case (a) confirms the connection was actively DROP'd by the fips chain (not silently unrouted).

(e) The file's chain is scoped by iifname != $fips_if return, so a table scoped to an interface that is not the TUN polices nothing on it. The suite loads the production file with its interface name rewritten to notfips0, asserts that node-a then reaches the case-(a) port and that node-b's show_listening_sockets control reply reports firewall_active: false, then reloads the real file and asserts the port is dropped again and the reply reads true. The reply is read once with the real file loaded before the rewrite, so a control socket that does not answer fails the case instead of passing it.

Topology

Two FIPS nodes peered over UDP on a Docker bridge network:

Container	Hostname	Firewall
`fips-fw-container-a`	`host-a`	none (probe)
`fips-fw-container-b`	`host-b`	`fips.nft` + drop-in

The bridge network requests no subnet, so docker assigns one from its own address pool and two concurrent runs never contend for a fixed range. No node's IPv4 address is therefore known before startup, and the generated peer stanzas address each other by docker hostname, resolved through the container's dnsmasq to docker's embedded DNS. The firewall assertions themselves are unaffected: they run over the fips0 overlay, whose addresses are derived from the node npubs.

node-b mounts the production packaging/common/fips.nft read-only at /etc/fips/fips.nft, plus a drop-in at /etc/fips/fips.d/services.nft containing tcp dport 22 accept. node-a is unfirewalled and serves as the probe origin.

Both containers run the unified test image's default mode, which starts dnsmasq + sshd (port 22) + iperf3 + python http.server on port 8000 + the FIPS daemon.

fips-firewall.service activation

The production unit's ExecStart is:

ExecStart=/usr/sbin/nft -f /etc/fips/fips.nft

The unified test image does not run systemd, so test.sh invokes the same nft -f command directly inside node-b after fips0 is up and peering has converged. The deb-install harness covers the systemd unit-enablement path under real systemd separately.

Run

Build the Linux binaries and test image:

./testing/scripts/build.sh --no-docker

Run the suite:

./testing/firewall/test.sh

test.sh regenerates fixtures automatically before starting Docker. Use --skip-build to reuse the existing release binaries and the existing test image: the suite then neither builds nor pulls the image named by FIPS_TEST_IMAGE (default fips-test:latest), so that image must already exist. Use --keep-up to leave the containers running for inspection.

Expected output shape

=== Generating firewall fixtures
=== Starting firewall harness
=== Waiting for fips0 on both nodes
=== Waiting for peer convergence
=== Resolving fips0 addresses
  node-a: fd97:...
  node-b: fd97:...
=== Activating fips-firewall on fips-fw-container-b
PASS: fips-fw-container-b: fips.nft baseline + drop-in loaded
=== Case (c): ICMPv6 echo-request to firewalled node
PASS: (c) ICMPv6 ping node-a → node-b accepted
=== Case (a): unallowed inbound TCP/8000 from node-a → node-b
PASS: (a) inbound TCP/8000 dropped (curl rc=28, timed out as expected)
=== Case (b): node-b initiates outbound TCP, expects reply via conntrack
PASS: (b) outbound from node-b got HTTP 200 via conntrack reply path
=== Case (d): drop-in allowlisted TCP/22 from node-a → node-b
PASS: (d) drop-in allowlisted TCP/22 reachable
=== Drop counter incremented (case a should have ticked it)
PASS: drop counter = N (case a was actually dropped, not just unrouted)
=== Case (e): fips.nft scoped to an interface that is not the TUN
PASS: (e) table scoped to notfips0 does not police fips0: TCP/8000 reachable
PASS: (e) listener report shows the firewall as not policing the TUN
PASS: fips-fw-container-b: fips.nft baseline + drop-in loaded
PASS: (e) real fips.nft reloaded: TCP/8000 dropped and the report shows the firewall active
=== Firewall integration test passed

Inspect the loaded ruleset

docker exec fips-fw-container-b nft list table inet fips

Stop and clean up

docker compose -f testing/firewall/docker-compose.yml down

Generated fixture location

testing/firewall/generated-configs/ (gitignored), or generated-configs<suffix>/ when FIPS_CI_NAME_SUFFIX is set, which is how concurrent runs keep their fixtures apart.

5.3 KiB Raw Permalink Blame History