reconFTW is a tool designed to perform automated recon on a target domain by running the best set of tools to perform scanning and finding out vulnerabilities
Shell bug-bountybugbountybugbounty-tooldnshackingnucleiosintpenetration-testingpentestpentest-toolpentestingreconreconnaissancescannersecuritysecurity-toolssubdomainsubdomain-enumerationvulnerabilities
dns: detect a metadata service that exists but requires a token (#1050)
six2dez1fe734822119
diff --git a/modules/utils.sh b/modules/utils.shindex c9d710a..29e6047 100644--- a/modules/utils.sh+++ b/modules/utils.sh@@ -1398,7 +1398,27 @@ _ip_is_public_ipv4() {# Returns 0 if running on a cloud VPS, 1 otherwise._is_cloud_vps() {[[ "${DRY_RUN:-false}" == "true" ]] && return 1- curl -sf --max-time 2 -o /dev/null http://169.254.169.254/ 2>/dev/null && return 0++ # Probe for the PRESENCE of a metadata service, not for its willingness to+ # serve an unauthenticated read (issue #1050).+ #+ # The previous `curl -sf` could not tell those apart. IMDSv2 has been the+ # default on new EC2 instances since 2019 and answers an unauthenticated GET+ # with 401; `-f` turns any >=400 status into exit 22, so a perfectly live+ # metadata service was read as "no metadata service at all". Every modern+ # AWS box was therefore misclassified as a home/NAT network — which is not+ # cosmetic: _can_use_puredns() uses this to pick the resolver, so+ # DNS_RESOLVER=auto silently downgraded puredns to dnsx on exactly the hosts+ # puredns exists to be fast on.+ #+ # %{http_code} is '000' only when no HTTP response was received at all+ # (connection refused, timeout, no route) — which is what a non-cloud host+ # gives for link-local 169.254.169.254. Any real status (200, 401, 403, 404)+ # proves something is listening there, and that is the whole question.+ local _code+ _code=$(curl -s --max-time 2 -o /dev/null -w '%{http_code}' \+ http://169.254.169.254/ 2>/dev/null)+ [[ -n "$_code" ]] && [[ "$_code" != "000" ]] && return 0return 1}diff --git a/tests/unit/test_cloud_vps_detection.bats b/tests/unit/test_cloud_vps_detection.batsnew file mode 100644index 0000000..12f3bbb--- /dev/null+++ b/tests/unit/test_cloud_vps_detection.bats@@ -0,0 +1,113 @@+#!/usr/bin/env bats+#+# Regression coverage for issue #1050.+#+# _is_cloud_vps() decides whether a metadata service exists at link-local+# 169.254.169.254. _can_use_puredns() consumes that answer to pick the DNS+# resolver, so a false negative is not cosmetic: DNS_RESOLVER=auto silently+# downgrades puredns to dnsx on precisely the cloud hosts puredns is for.+#+# The old probe was `curl -sf`. IMDSv2 — the default on new EC2 instances since+# 2019 — answers an unauthenticated GET with 401, and `-f` maps any >=400 status+# to exit 22. A live metadata service was therefore read as no metadata service.+#+# These tests drive a stub `curl` so they assert the BRANCHING on HTTP status+# rather than reaching the network, which unit-fast cannot do anyway.+#+# ASSERTION STYLE: `run` captures status, so each case asserts on $status+# directly; the suite restores errexit via tests/helpers/common.bash.++setup() {+ source "$(cd "$(dirname "$BATS_TEST_FILENAME")" && pwd)/../helpers/common.bash"+ setup_recon_env++ TEST_DIR="$(mktemp -d)"+ export TEST_DIR+ cd "$TEST_DIR"++ # Stub curl: emulate a response by echoing $STUB_HTTP_CODE for -w '%{http_code}'.+ # '000' is what real curl reports when no HTTP response arrived at all.+ #+ # It MUST honour -f, or these tests are decoration: the bug under test is+ # precisely that -f maps a >=400 status to exit 22. A stub that ignored -f+ # would keep reporting success for the old `curl -sf` probe, so reinstating+ # the bug would leave every test below green. Verified by mutation, not+ # assumed — see the note on the IMDSv2 case.+ mkdir -p bin+ cat > bin/curl <<'STUB'+#!/usr/bin/env bash+fail_on_error=0+for _a in "$@"; do+ case "$_a" in+ -f | --fail) fail_on_error=1 ;;+ -*f*) [[ "$_a" != --* ]] && fail_on_error=1 ;;+ esac+done+code="${STUB_HTTP_CODE:-000}"+printf '%s' "$code"+if [[ "$fail_on_error" -eq 1 ]] && [[ "$code" =~ ^[45][0-9][0-9]$ ]]; then+ exit 22+fi+[[ "$code" == "000" ]] && exit 7+exit 0+STUB+ chmod +x bin/curl+ export PATH="$TEST_DIR/bin:$PATH"++ export DRY_RUN=false+}++teardown() {+ cd /+ rm -rf "$TEST_DIR"+}++@test "_is_cloud_vps detects AWS IMDSv2, which answers an unauthenticated GET with 401 (issue #1050)" {+ # The exact case the issue reports. `curl -sf` returned exit 22 here and the+ # instance was misclassified as a non-cloud NAT network.+ export STUB_HTTP_CODE="401"+ run _is_cloud_vps+ [ "$status" -eq 0 ]+}++@test "_is_cloud_vps detects a metadata service that answers 403" {+ export STUB_HTTP_CODE="403"+ run _is_cloud_vps+ [ "$status" -eq 0 ]+}++@test "_is_cloud_vps still detects a plain IMDSv1-style 200" {+ export STUB_HTTP_CODE="200"+ run _is_cloud_vps+ [ "$status" -eq 0 ]+}++@test "_is_cloud_vps returns false when nothing answers at 169.254.169.254" {+ # No HTTP response at all: the non-cloud case, which must stay false so a+ # home/office network is never handed to puredns.+ export STUB_HTTP_CODE="000"+ run _is_cloud_vps+ [ "$status" -ne 0 ]+}++@test "_is_cloud_vps returns false when curl produces no output" {+ export STUB_HTTP_CODE=""+ run _is_cloud_vps+ [ "$status" -ne 0 ]+}++@test "_is_cloud_vps short-circuits to false under DRY_RUN" {+ export DRY_RUN=true+ export STUB_HTTP_CODE="200"+ run _is_cloud_vps+ [ "$status" -ne 0 ]+}++@test "_can_use_puredns now says yes for a private-IP EC2 box behind IMDSv2" {+ # The end-to-end consequence of the fix: private IP + IMDSv2 401 is a cloud+ # VPS with 1:1 NAT, which is safe for puredns. Before the fix this returned+ # false and the run silently fell back to dnsx.+ export STUB_HTTP_CODE="401"+ run _can_use_puredns "10.0.0.5"+ [ "$status" -eq 0 ]+}