Files
fips/testing/root-tests.sh
Johnathan Corgan 26cfb4fad3 Merge maint into master: the lookup forward limiter's per-origin limits, the peer ACL and Nostr traversal docs, the root-only test script, and the narrow-path TCP fixes
The peer ACL changelog entry joins master's existing Links and
transports section under Security rather than opening a second one.

testing/root-tests.sh carries master's filters rather than maint's, in
two groups: the TUN warning, DNS responder and gateway mesh tests (18),
and the Linux TUN namespace tests (13). It needs ip, nft and sysctl
here, and its probe also writes a sysctl inside the namespace, as four
of the TUN tests do, so a container with a read-only /proc/sys is
reported as such.

GitHub's Linux job ran those groups as two steps; they are now one step
that runs the script under sudo. ci-local.sh takes --root-tests or
FIPS_CI_ROOT_TESTS=1 alongside --baremetal and --slow-tests, and --rerun
rejects both forms. The README section sits beside the known-defect
lane, and the comments that said which runners run these tests now name
the script.

The held-SYN re-clamp and ingress fragmentation for paths under 1280
are ported rather than taken as written: on master the IPv6 adapter
hands the node compressed datagrams and the node judges them in
handle_datagram.

- The fragmenter moves to src/ipv6tun/ipv6_frag.rs with its body and
  tests unchanged, listed as one of the adapter's decision modules.
- outbound.rs gains can_split, split and reclamp, pure functions that
  rebuild a held datagram's packet, fragment it or clamp its SYN, and
  compress the result. mss.rs gains flow_ceiling, the per-flow MSS
  ceiling the TUN reader and the release-time re-clamp now share.
- handle_datagram lets a datagram over the node-wide limit or an
  established session's path MTU carry on whole when it can be
  fragmented. The established send and the flush of held datagrams go
  through send_tun_packet, which sends a datagram whole or as fragments
  by the smaller of the node's own limit and the session's path MTU, and
  the flush clamps a held SYN again from the node's own path MTU map.
- A Packet Too Big for a packet over 1280 on a path under 1280 reports
  1280, both for a datagram the node refuses and for one the adapter
  answers itself.
- The new node tests are ported to master's TUN taps and compressed held
  payloads. The two existing Packet Too Big tests that now use a
  Hop-by-Hop header give it a real length, since master's ICMPv6 builder
  does not answer a packet whose extension headers run past its end.
  The host-link tests of refusals send a packet with a Hop-by-Hop header
  where they mean a packet that is refused rather than fragmented, and
  one of them also checks that a fragmentable oversized packet to a key
  the node cannot resolve is told no route.

The IPv6 adapter design document now says that on this structure the
node decides fragmentation at its two refusal sites and splits a
datagram where it sends it, that it clamps a held SYN again from its
own path MTU map when it releases the datagrams it held, and that both
apply to an app-owned TUN because the node holds and releases the
datagrams of either path. It also names the functions an app-owned
TUN's packets pass through on this structure, Handles::outbound and
handle_datagram, and the MTU document says it is the ingress node that
fragments.
2026-10-29 00:03:54 +00:00

174 lines
6.7 KiB
Bash
Executable File

#!/bin/bash
# ── Root-only unit tests ────────────────────────────────────────────────────
# Runs the unit tests that need root and network namespaces. They are
# `#[ignore]`d, so the ordinary unit-test runs skip them, and this script is
# the one place that says which `--ignored` filters run and how many tests
# each must pass.
#
# Usage:
# testing/root-tests.sh <bin>
#
# <bin> the built `fips-mesh` library test executable (cargo's `fips` lib
# target), for instance the path printed for it by
# `cargo test -p fips-mesh --lib --no-run --message-format=json`.
# A bare name with no slash is a file in the current directory,
# not a command looked up on PATH.
#
# The script builds and locates nothing, and needs neither cargo nor git, so
# it runs wherever the binary and this file are present. It never calls sudo:
# it must already be running as root, and the caller decides how. GitHub's
# Linux job and testing/ci-local.sh --root-tests run it under sudo; a
# container runs it as uid 0.
#
# A container needs, besides uid 0: the NET_ADMIN and SYS_ADMIN capabilities,
# /dev/net/tun, an AppArmor profile that allows the network-namespace mount
# (Docker's default profile refuses `mount --make-shared /run/netns`; for
# instance --security-opt apparmor=unconfined), and a writable /proc/sys (a
# default Docker container mounts it read-only; for instance
# --security-opt systempaths=unconfined). Before any test runs, the script
# creates and deletes a namespace and checks /dev/net/tun, plus, where the
# tests set sysctls, writes one inside the namespace, so a missing permission
# is reported as such rather than as a set of failing tests.
#
# Each group runs as one `<bin> --ignored <filters> --test-threads=1`, and
# passes only if the binary exits 0 and libtest reports exactly the group's
# count passed. The count is the guard against a filter that quietly matches
# fewer tests: a filter that matches nothing still reports `ok`. Adding or
# removing a root-only test therefore changes the count here. Every group
# runs even after one fails, so one run reports every failure.
#
# Output: the test output, then one line per group, then a final line that is
# always last:
# root-tests: PASS: <g> groups, <n> tests
# root-tests: FAIL: <k> of <g> groups failed
# root-tests: ERROR: <reason>
#
# Exit 0 = every group passed with its exact count. Exit 1 = a test failed,
# the binary failed or a count did not match. Exit 2 = the script could not
# run (usage, the binary, privilege, a missing tool or permission); never a
# pass.
# ─────────────────────────────────────────────────────────────────────────────
set -uo pipefail
# The groups, each "<count>|<filters>", run in order.
TEST_GROUPS=(
"18|node::tests::tun_warning:: ipv6tun::dns:: gateway::dns::mesh_tests::"
"13|ipv6tun::tun::name_tests::"
)
# The tools the tests in the groups above run: ip for the namespaces, nft for
# the DNS tests' rules and packet counters, and sysctl for the TUN tests that
# turn IPv6 off inside a namespace. When sysctl is listed, the probe also
# writes a sysctl inside its namespace, as those tests do.
TEST_TOOLS=(ip nft sysctl)
PROBE_NS="root-tests-probe-$$"
probe_made=false
tmp=""
# Remove the output file and the probe namespace, whichever exist.
# shellcheck disable=SC2317 # run by the EXIT trap
cleanup() {
[[ -n "$tmp" ]] && rm -f "$tmp"
if [[ "$probe_made" == true ]]; then
ip netns del "$PROBE_NS" 2>/dev/null
fi
return 0
}
trap cleanup EXIT
trap 'exit 130' INT
trap 'exit 143' TERM
# Print the could-not-run reason as the last line and exit 2.
die() {
echo "root-tests: ERROR: $1" >&2
exit 2
}
# Print the usage on stderr.
usage() {
echo "usage: $0 <bin>" >&2
echo " <bin> the fips-mesh library test executable; run as root" >&2
}
# The checks that decide whether the tests can run at all.
preflight() {
local t out
[[ ${#TEST_GROUPS[@]} -ge 1 ]] || die "no test groups are listed in $0"
for t in "${TEST_GROUPS[@]}"; do
[[ "$t" =~ ^[1-9][0-9]*\|[^[:space:]] ]] \
|| die "malformed test group in $0: '$t' (want <count>|<filters>)"
done
for t in mktemp tee grep sed "${TEST_TOOLS[@]}"; do
command -v "$t" >/dev/null 2>&1 || die "$t not found on PATH"
done
[[ "$(id -u)" -eq 0 ]] \
|| die "must run as root: under sudo, or as uid 0 in a container with NET_ADMIN and SYS_ADMIN"
out=$(ip netns add "$PROBE_NS" 2>&1) \
|| die "cannot create a network namespace (ip netns add): ${out:-no message}"
probe_made=true
[[ -c /dev/net/tun ]] || die "/dev/net/tun is not a character device"
for t in "${TEST_TOOLS[@]}"; do
[[ "$t" == sysctl ]] || continue
out=$(ip netns exec "$PROBE_NS" sysctl -qw net.ipv6.conf.all.disable_ipv6=1 2>&1) \
|| die "cannot write /proc/sys inside a network namespace (sysctl -w): ${out:-no message}"
done
out=$(ip netns del "$PROBE_NS" 2>&1) \
|| die "cannot delete a network namespace (ip netns del): ${out:-no message}"
probe_made=false
return 0
}
# run_group <count> <filters...>: run one group, print its line, and return
# 0 if it passed.
run_group() {
local n="$1" rc result name
shift
"$bin" --ignored "$@" --test-threads=1 2>&1 | tee "$tmp"
rc=${PIPESTATUS[0]}
if [[ $rc -eq 0 ]] && grep -qE "^test result: ok\. $n passed;" "$tmp"; then
echo "root-tests: ok: $*: $n passed"
return 0
fi
result=$(grep -m 1 -E '^test result: ' "$tmp")
echo "root-tests: FAILED: $*: expected $n passed; exit $rc; ${result:-no test result line}"
while IFS= read -r name; do
echo "root-tests: failed test: $name"
done < <(sed -n 's/^test \(.*\) \.\.\. FAILED$/\1/p' "$tmp")
return 1
}
if [[ $# -ne 1 ]]; then
usage
die "expected one argument, got $#"
fi
bin="$1"
# Without a slash the shell would look the name up on PATH when running it.
[[ "$bin" == */* ]] || bin="./$bin"
[[ -e "$bin" ]] || die "$bin does not exist"
[[ -f "$bin" && -x "$bin" ]] || die "$bin is not an executable file"
preflight
tmp=$(mktemp) || die "mktemp failed"
failed=0
total=0
for entry in "${TEST_GROUPS[@]}"; do
count="${entry%%|*}"
read -r -a filters <<< "${entry#*|}"
if run_group "$count" "${filters[@]}"; then
total=$((total + count))
else
failed=$((failed + 1))
fi
done
if [[ $failed -eq 0 ]]; then
echo "root-tests: PASS: ${#TEST_GROUPS[@]} groups, $total tests"
exit 0
fi
echo "root-tests: FAIL: $failed of ${#TEST_GROUPS[@]} groups failed"
exit 1