|
| 1 | +#!/bin/bash -e |
| 2 | +# vim: set tabstop=8 shiftwidth=4 softtabstop=4 expandtab smarttab colorcolumn=80: |
| 3 | +# |
| 4 | +# Copyright (c) 2025 Red Hat, Inc. |
| 5 | +# Author: Sergio Correia <scorreia@redhat.com> |
| 6 | +# |
| 7 | +# This program is free software: you can redistribute it and/or modify |
| 8 | +# it under the terms of the GNU General Public License as published by |
| 9 | +# the Free Software Foundation, either version 3 of the License, or |
| 10 | +# (at your option) any later version. |
| 11 | +# |
| 12 | +# This program is distributed in the hope that it will be useful, |
| 13 | +# but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 14 | +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 15 | +# GNU General Public License for more details. |
| 16 | +# |
| 17 | +# You should have received a copy of the GNU General Public License |
| 18 | +# along with this program. If not, see <http://www.gnu.org/licenses/>. |
| 19 | +# |
| 20 | +# Test to verify that clevis-decrypt-sss properly kills all child processes |
| 21 | +# and their descendants (grandchildren) after successful decryption. |
| 22 | +# |
| 23 | +# This test addresses issue #460: |
| 24 | +# https://github.com/latchset/clevis/issues/460 |
| 25 | +# |
| 26 | +# The issue: When using SSS with multiple pins (e.g., tang, fido2) and t < n, |
| 27 | +# after successful decryption with t pins, the remaining child processes |
| 28 | +# (and their grandchildren like curl or fido2-assert) were not being killed. |
| 29 | +# |
| 30 | +# The fix uses process groups: each child becomes a process group leader |
| 31 | +# via setpgid(0,0), and cleanup uses kill(-pid, SIGTERM) to kill the |
| 32 | +# entire process group. |
| 33 | +# |
| 34 | +# Test strategy: |
| 35 | +# 1. Start a real tang server that responds quickly |
| 36 | +# 2. Start a "hanging" server that accepts connections but never responds |
| 37 | +# 3. Encrypt with SSS t=1, two tang pins (real + hanging) |
| 38 | +# 4. Decrypt - the real tang succeeds, the hanging one's child should be killed |
| 39 | +# 5. Verify no orphaned curl processes remain after decryption |
| 40 | + |
| 41 | +. tang-common-test-functions |
| 42 | + |
| 43 | +TMP="$(mktemp -d)" |
| 44 | + |
| 45 | +# Force exit - kills background jobs first, then exits |
| 46 | +force_exit() { |
| 47 | + local code="$1" |
| 48 | + trap - EXIT # Disable trap to prevent recursion |
| 49 | + |
| 50 | + # Kill all child processes to prevent bash from waiting |
| 51 | + # Use both pkill and direct kill to be thorough |
| 52 | + pkill -9 -P $$ 2>/dev/null || true |
| 53 | + jobs -p 2>/dev/null | xargs -r kill -9 2>/dev/null || true |
| 54 | + |
| 55 | + if [ "${code}" -eq 0 ]; then |
| 56 | + # Success - normal exit after killing children |
| 57 | + exit 0 |
| 58 | + else |
| 59 | + # Failure - kill entire process group for immediate exit |
| 60 | + kill -9 -$$ 2>/dev/null || kill -9 $$ 2>/dev/null || exit "${code}" |
| 61 | + fi |
| 62 | +} |
| 63 | + |
| 64 | +# No cleanup trap - we'll force kill on exit |
| 65 | + |
| 66 | +# Start a real tang server |
| 67 | +tang_run "${TMP}" sig exc |
| 68 | +port_real=$(tang_get_port "${TMP}") |
| 69 | +url_real="http://localhost:${port_real}" |
| 70 | + |
| 71 | +# Get the advertisement from the real tang server |
| 72 | +adv="${TMP}/adv.jws" |
| 73 | +tang_get_adv "${port_real}" "${adv}" |
| 74 | + |
| 75 | +# Start a "hanging" server - accepts TCP connections but never sends HTTP response. |
| 76 | +# This simulates a tang server that is unreachable or very slow. |
| 77 | +# We use socat to listen and spawn a process that just sleeps forever. |
| 78 | +# Use disown to prevent bash from waiting for this background job on exit. |
| 79 | +"${SOCAT}" TCP4-LISTEN:0,fork EXEC:"sleep 3600" & |
| 80 | +hang_pid=$! |
| 81 | +disown "${hang_pid}" |
| 82 | +echo "${hang_pid}" > "${TMP}/hang.pid" |
| 83 | + |
| 84 | +# Wait for the hanging server to start listening |
| 85 | +sleep 0.5 |
| 86 | +port_hang=$(lsof -nP -a -p "${hang_pid}" -iTCP -sTCP:LISTEN -Fn 2>/dev/null \ |
| 87 | + | grep '^n.*:' | head -1 | sed 's/.*://') |
| 88 | + |
| 89 | +if [ -z "${port_hang}" ]; then |
| 90 | + echo "Failed to start hanging server" >&2 |
| 91 | + force_exit 1 |
| 92 | +fi |
| 93 | + |
| 94 | +url_hang="http://localhost:${port_hang}" |
| 95 | + |
| 96 | +echo "Real tang server on port ${port_real}" |
| 97 | +echo "Hanging server on port ${port_hang}" |
| 98 | + |
| 99 | +# Encrypt with SSS: t=1, two tang pins |
| 100 | +# Both pins use the same advertisement (from the real server) so encryption works. |
| 101 | +# During decryption, one will contact the real server (succeeds) and one will |
| 102 | +# contact the hanging server (curl hangs waiting for response). |
| 103 | +cfg=$(cat <<EOF |
| 104 | +{ |
| 105 | + "t": 1, |
| 106 | + "pins": { |
| 107 | + "tang": [ |
| 108 | + {"url": "${url_real}", "adv": "${adv}"}, |
| 109 | + {"url": "${url_hang}", "adv": "${adv}"} |
| 110 | + ] |
| 111 | + } |
| 112 | +} |
| 113 | +EOF |
| 114 | +) |
| 115 | + |
| 116 | +echo "Encrypting test data..." |
| 117 | +enc="$(echo -n "test-cleanup-data" | clevis encrypt sss "${cfg}" -y)" |
| 118 | +echo "Encryption successful." |
| 119 | + |
| 120 | +# Record curl processes before decryption |
| 121 | +curl_before=$(pgrep -x curl 2>/dev/null | wc -l || echo 0) |
| 122 | + |
| 123 | +echo "Starting decryption... ($(date +%T))" |
| 124 | + |
| 125 | +# Decrypt with a timeout. The real tang should respond quickly, so decryption |
| 126 | +# should complete in a few seconds. The hanging server's child process |
| 127 | +# (and its curl grandchild) should be killed after decryption succeeds. |
| 128 | +# |
| 129 | +# Without the fix: curl connecting to the hanging server would remain running |
| 130 | +# after clevis-decrypt-sss exits (orphaned to init). |
| 131 | +# |
| 132 | +# With the fix: the entire process group is killed, including curl. |
| 133 | + |
| 134 | +# Decrypt the data. Without the fix, decryption succeeds but leaves orphaned |
| 135 | +# grandchild processes (curl) which we detect below. |
| 136 | +dec="" |
| 137 | +if ! dec="$(echo -n "${enc}" | clevis decrypt)"; then |
| 138 | + echo "FAIL: Decryption failed" >&2 |
| 139 | + force_exit 1 |
| 140 | +fi |
| 141 | + |
| 142 | +if [ "${dec}" != "test-cleanup-data" ]; then |
| 143 | + echo "FAIL: Decrypted data mismatch: got '${dec}'" >&2 |
| 144 | + force_exit 1 |
| 145 | +fi |
| 146 | + |
| 147 | +echo "Decryption successful: '${dec}' ($(date +%T))" |
| 148 | + |
| 149 | +# Give a moment for process cleanup |
| 150 | +sleep 0.5 |
| 151 | + |
| 152 | +# Check for orphaned curl processes connecting to the hanging server. |
| 153 | +# Use lsof to find any process with a connection to port_hang. |
| 154 | +# -n: no hostname resolution (faster) |
| 155 | +# -P: no port name resolution (faster) |
| 156 | +# -t: terse output (just PIDs) |
| 157 | +orphaned_connections=$(lsof -nP -i "TCP:${port_hang}" -t 2>/dev/null | grep -v "^${hang_pid}$" || true) |
| 158 | + |
| 159 | +if [ -n "${orphaned_connections}" ]; then |
| 160 | + echo "FAIL: Found orphaned processes connecting to hanging server:" >&2 |
| 161 | + echo "${orphaned_connections}" >&2 |
| 162 | + force_exit 1 |
| 163 | +fi |
| 164 | + |
| 165 | +# Also check that we don't have more curl processes than before |
| 166 | +# (accounting for possible legitimate curl usage) |
| 167 | +curl_after=$(pgrep -x curl 2>/dev/null | wc -l || echo 0) |
| 168 | + |
| 169 | +# We expect curl_after <= curl_before (the curl that was connecting to hang |
| 170 | +# server should have been killed) |
| 171 | +if [ "${curl_after}" -gt "${curl_before}" ]; then |
| 172 | + echo "WARNING: More curl processes after decryption (${curl_after}) than before (${curl_before})" >&2 |
| 173 | + echo "This might indicate orphaned processes, but checking connections is more reliable." >&2 |
| 174 | +fi |
| 175 | + |
| 176 | +echo "SUCCESS: No orphaned processes detected after SSS decryption cleanup" |
| 177 | +force_exit 0 |
0 commit comments