
Cloud Vulnerability DB
A community-led vulnerabilities database
CVE-2026-48045 is an unbounded TC-deferred queue vulnerability in python-zeroconf (the zeroconf pip package) that allows unauthenticated LAN-adjacent attackers to cause memory exhaustion and quadratic CPU burn via a spoofed-source mDNS flood. It affects all versions of zeroconf prior to 0.149.12. The vulnerability was disclosed by maintainer bdraco on May 21, 2026, with the fix merged the same day and the GitHub Advisory Database entry published on June 11, 2026. It carries a CVSS v3.1 base score of 6.5 (Medium) (Github Advisory, GHSA Security Advisory).
The root cause is uncontrolled resource consumption (CWE-400) and allocation without limits or throttling (CWE-770) in AsyncListener.handle_query_or_defer. When a truncated (TC-bit) mDNS query arrives, the method stores it in self._deferred[addr] and arms a per-address timer in self._timers[addr] per RFC 6762 §18.5, but neither the per-address list length nor the total number of distinct addr keys was capped. Each stored packet can be up to _MAX_MSG_ABSOLUTE = 8966 bytes, and the deduplication check (for incoming in reversed(deferred): if incoming.data == msg.data) runs O(N) over the growing per-address queue on every packet arrival, resulting in quadratic CPU cost. An attacker on the local link can stream byte-distinct TC-flagged UDP/5353 queries to the multicast address (224.0.0.251 / ff02::fb) and trivially spoof source IPs to multiply entries across _deferred and _timers, exhausting both memory and CPU (Github Advisory, Fix PR #1751).
Successful exploitation results in a denial-of-service condition with no confidentiality or integrity impact. On memory-constrained deployments — the canonical example being Home Assistant running on Raspberry Pi-class hardware — sustained flooding OOM-kills the zeroconf process. Under lighter load, event-loop scheduler starvation caused by the quadratic CPU burn disrupts unrelated zeroconf consumers including service discovery, registration, and ServiceBrowser callbacks. There is no lateral movement or data exfiltration risk; the impact is limited to availability of the affected process and any services depending on mDNS (Github Advisory, GHSA Security Advisory).
_deferred and _timers for each fake source, multiplying memory consumption._deferred/_timers, and each packet to an existing source triggers an O(N) dedup scan, causing quadratic CPU burn.oom-kill or Out of memory: Kill process) referencing the zeroconf or Home Assistant process; application-level errors or timeouts in mDNS service discovery, registration, or ServiceBrowser callbacks coinciding with unusual network traffic.Upgrade to zeroconf version 0.149.12 or later, which introduces per-address queue caps (_MAX_DEFERRED_PER_ADDR = 16) and a total-address cap (_MAX_DEFERRED_ADDRS = 512) with FIFO eviction, bounding worst-case memory at approximately 72 MB even under a spoofed-source flood (Release 0.149.12, Fix PR #1751). There is no in-process workaround available for unpatched versions. As a network-level mitigation, restrict mDNS traffic (UDP/5353) to trusted Layer-2 segments using AP client isolation, guest-network separation, or host firewall rules to limit exposure to untrusted LAN peers (Github Advisory).
The fix was authored and merged by the primary maintainer (bdraco) on May 20, 2026, with the security advisory published the following day. The openSUSE security team issued a security announcement for the package update, and Tenable published Nessus detection plugins (IDs 324891 and 327765) for the vulnerability. The fix was also backported into downstream projects such as Music Assistant. No significant broader media coverage or notable researcher commentary beyond the maintainer's own detailed advisory has been identified (openSUSE Announcement, Tenable Plugin).
Source: This report was generated using AI
Free Vulnerability Assessment
Evaluate your cloud security practices across 9 security domains to benchmark your risk level and identify gaps in your defenses.
Get a personalized demo
"Best User Experience I have ever seen, provides full visibility to cloud workloads."
"Wiz provides a single pane of glass to see what is going on in our cloud environments."
"We know that if Wiz identifies something as critical, it actually is."