Skip to main content
Vulnerability Database/CVE-2026-96405

CVE-2026-96405: NLnet Labs Unbound DNS DoS Vulnerability

CVE-2026-96405 is a denial of service vulnerability in NLnet Labs Unbound DNS resolver that allows attackers to degrade performance through crafted DNS responses. This post explains its impact, affected versions, and mitigation steps.

Published:

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

CVE-2026-96405 Overview

CVE-2026-96405 affects NLnet Labs Unbound through version 1.26.1. The recursive DNS resolver enforces a per-operation limit on compression-pointer traversal but lacks an aggregate inbound decompression-work budget. An attacker with recursive-query access who controls a delegated authority can return crafted replies containing many chained compression pointers. Following those chains consumes substantial CPU, degrading overall resolver performance. The weakness is tracked under [CWE-770] (Allocation of Resources Without Limits or Throttling).

Critical Impact

Remote attackers with recursive-query access can degrade Unbound performance through crafted DNS responses that exhaust CPU during name decompression.

Affected Products

  • NLnet Labs Unbound versions up to and including 1.26.1
  • Deployments exposing recursive-query access to untrusted clients
  • Environments resolving names under attacker-controlled delegated authorities

Discovery Timeline

  • 2026-10-21 - CVE-2026-96405 published to NVD
  • 2026-10-22 - Last updated in NVD database

Technical Details for CVE-2026-96405

Vulnerability Analysis

Unbound processes DNS messages that use name compression as defined in RFC 1035. Compression pointers allow a domain name to reference an earlier offset in the message instead of repeating labels. Unbound caps the number of pointers followed during a single name-decompression operation, which prevents loops and bounded blowup for one name.

The resolver does not maintain an aggregate budget across all names processed in a single message or across chained operations within one query flow. A response can therefore contain many names, each individually within the per-name limit, while the total decompression work scales to exhaust CPU. Each pointer dereference performs memory access and bounds checking, and the aggregate cost dominates when the message is densely packed with chained references.

The impact is availability degradation on the resolver process. Legitimate queries experience increased latency and timeouts while the worker is busy decompressing malicious responses.

Root Cause

The root cause is missing resource accounting at the message or query level. Per-operation bounds are necessary but insufficient when an attacker controls the number of operations triggered by a single response. The vulnerability is categorized under [CWE-770] because the decompression subsystem allocates CPU work without a global throttle.

Attack Vector

Exploitation requires two conditions. The attacker must have recursive-query access to the target Unbound instance, either directly or by inducing a downstream client to query a specific name. The attacker must also control an authoritative server for a delegated zone that Unbound will query. When Unbound resolves a name in that zone, the attacker returns a response packed with chained compression pointers across many record names. Unbound follows the pointers during parsing and consumes CPU proportional to the aggregate pointer count.

The vulnerability mechanism is described in the NLnet Labs CVE-2026-96405 Advisory. No public proof-of-concept code is available at the time of writing.

Detection Methods for CVE-2026-96405

Indicators of Compromise

  • Sustained CPU usage spikes in unbound worker threads without a corresponding increase in client query volume
  • DNS responses from upstream authorities containing abnormally large numbers of compression pointers relative to message size
  • Increased query latency and client-side timeouts for recursion through the affected resolver

Detection Strategies

  • Monitor Unbound statistics via unbound-control stats for rising values in mem.cache.rrset, num.query.tcp, and per-thread CPU time
  • Capture pcap samples of responses from suspect authoritative servers and inspect name fields for excessive pointer chaining
  • Correlate recursive-query logs with CPU utilization to identify specific query names or delegated zones that trigger the condition

Monitoring Recommendations

  • Alert on sustained Unbound CPU usage above baseline for intervals longer than normal query bursts
  • Track response sizes and parse times per upstream authority to detect anomalous servers
  • Log and review NXDOMAIN and SERVFAIL rate changes that may accompany worker starvation

How to Mitigate CVE-2026-96405

Immediate Actions Required

  • Upgrade Unbound to a version later than 1.26.1 that includes the aggregate decompression-work budget fix once released by NLnet Labs
  • Restrict recursive-query access using access-control directives so that only trusted networks can submit queries
  • Review upstream forwarders and delegated zones for untrusted authoritative servers

Patch Information

Consult the NLnet Labs CVE-2026-96405 Advisory for the fixed version and patch details. Apply the vendor-supplied update to all recursive resolver instances.

Workarounds

  • Limit recursion to authenticated or internal clients only via access-control: <cidr> refuse for untrusted ranges
  • Place rate limits on recursive queries using ratelimit: and ip-ratelimit: configuration options to cap per-source query bursts
  • Deploy multiple resolver instances behind a load balancer so that a single degraded worker does not block all client traffic
bash
# Configuration example - restrict recursion and apply rate limits in unbound.conf
server:
    access-control: 0.0.0.0/0 refuse
    access-control: 10.0.0.0/8 allow
    access-control: 192.168.0.0/16 allow
    ratelimit: 1000
    ip-ratelimit: 200

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.