CVE-2026-1005: 
wolfSSL vulnerability analysis and mitigation

CVE-2026-1005 is an integer underflow vulnerability in the wolfSSL packet sniffer (versions ≤ 5.8.4) that leads to a heap buffer overflow in the AEAD decryption path. When a TLS record shorter than the explicit IV plus authentication tag is processed by ssl_DecodePacket, a 16-bit length value underflows and wraps to a large value, which is then passed to AEAD decryption routines (AES-GCM, AES-CCM, or ARIA-GCM), causing a heap buffer overflow and crash. An unauthenticated remote attacker can trigger this by injecting malformed TLS Application Data records into traffic being inspected by the sniffer. It carries a CVSS v3.1 base score of 5.3 (Medium) (Feedly, GitHub PR).

The root cause is an integer underflow (CWE-191) in the wolfSSL packet sniffer's ssl_DecodePacket function, which lacks a length validation check before subtracting AESGCM_EXP_IV_SZ and ssl->specs.aead_mac_size from the incoming record size. When the record length is smaller than these combined values, the subtraction on an unsigned 16-bit integer wraps around to a very large number, which is subsequently passed to AEAD decryption routines (AES-GCM, AES-CCM, ARIA-GCM), resulting in an out-of-bounds write (CWE-787) on the heap. The fix, merged on December 23, 2025, adds the missing length check prior to the subtraction operation (GitHub PR). The vulnerability is only exploitable when the wolfSSL packet sniffer feature is enabled and actively inspecting TLS traffic.

Successful exploitation causes a heap buffer overflow that crashes the application or service using the wolfSSL packet sniffer, resulting in a denial-of-service condition. The impact is limited to availability — there is no evidence of confidentiality or integrity compromise, as the overflow leads to a crash rather than controlled code execution in known scenarios. Systems relying on the wolfSSL sniffer for network traffic inspection would lose that monitoring capability during an attack (Feedly).

1. Identify target: Locate a system running wolfSSL version 5.8.4 or earlier with the packet sniffer feature enabled (--enable-sniffer), actively inspecting TLS traffic on a network segment accessible to the attacker.
2. Position for traffic injection: Gain a network position capable of injecting packets into the TLS session being monitored — this could be via ARP spoofing, a rogue access point, or another man-in-the-middle technique on the monitored network segment.
3. Craft malformed TLS record: Construct a TLS Application Data record whose length field is less than AESGCM_EXP_IV_SZ (8 bytes) plus the AEAD MAC size (typically 16 bytes for AES-GCM), i.e., a record body shorter than 24 bytes, while still appearing as a valid TLS record header.
4. Inject the record: Send or replay the malformed TLS Application Data record into the traffic stream being inspected by ssl_DecodePacket.
5. Trigger crash: The sniffer processes the record, the unsigned integer underflow occurs during length calculation, a large value is passed to the AEAD decryption routine, the heap buffer overflow is triggered, and the sniffer process crashes (GitHub PR, Feedly).

• Logs: Unexpected crash or segmentation fault logs from the process using the wolfSSL sniffer; core dump files generated by the wolfSSL sniffer process.
• Network: Anomalous TLS Application Data records with unusually short payload lengths (less than 24 bytes) observed in captured traffic on monitored segments.
• Process: Sudden termination or restart of the packet sniffer process; increased frequency of process restarts if a watchdog is in place.

The fix was merged into the wolfSSL master branch on December 23, 2025 (PR #9571), adding the missing length check in the sniffer for AES-GCM, AES-CCM, and ARIA-GCM decryption paths. Users should upgrade to a wolfSSL release incorporating this fix (versions after 5.8.4). As a workaround, organizations can disable the packet sniffer feature if it is not operationally required, or implement network segmentation to restrict who can inject traffic into monitored segments. A backport patch for the OpenEmbedded meta-openembedded layer was also published in May 2026 (GitHub PR, Feedly).

Coverage has been limited to standard vulnerability database aggregators and automated feeds. The Tenable Nessus plugin 303269 was published to detect this vulnerability, and the FreeBSD Ports collection (freshports.org) noted the patch. No significant vendor statements, researcher commentary, or notable media coverage beyond routine CVE tracking has been observed (Feedly).