CVE-2016-8863 is a heap-based buffer overflow (CWE-119) in create_url_list() in libupnp, the Portable SDK for UPnP Devices, before 1.6.21: a SUBSCRIBE request whose CALLBACK header holds a valid URI followed by an invalid one writes past the allocated list. NVD scores it 9.8. libupnp 1.6.21 fixes it.
Vulnerability summary
CVSS base score
9.8
Critical severity
EPSS score (FIRST)
8.6%
Probability of exploitation in the wild in the next 30 days
CISA KEV catalog
No
Not listed
Published
Vulnerability ID: CVE-2016-8863 · Source: NVD
Affected components
| Component | CPE | Version range |
|---|---|---|
| Portable SDK for UPnP Devices (libupnp, pupnp) | cpe:2.3:a:libupnp_project:libupnp:*:*:*:*:*:*:*:* | through 1.6.20 (NVD); fixed in 1.6.21 |
| OpenWrt packages feed: libupnp | cpe:2.3:a:libupnp_project:libupnp:*:*:*:*:*:*:*:* | openwrt-18.06 packages 1.6.19 with no patch directory; master, openwrt-24.10 and openwrt-25.12 package 1.14.18 (read 2026-11-18) |
| Debian libupnp | cpe:2.3:o:debian:debian_linux:8.0:*:*:*:*:*:*:* | fixed in 1:1.6.19+git20141001-1+deb8u1 (jessie, DSA-3736-1) and 1:1.6.17-1.2+deb7u2 (wheezy, DLA-747-1) |
What does this mean for affected devices?
NVD's description: "Heap-based buffer overflow in the create_url_list function in gena/gena_device.c in Portable UPnP SDK (aka libupnp) before 1.6.21 allows remote attackers to cause a denial of service (crash) or possibly execute arbitrary code via a valid URI followed by an invalid one in the CALLBACK header of an SUBSCRIBE request." NVD's Primary score is CVSS v3.0 9.8 (CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H), with CVSS v2 7.5 and CWE-119. Its CPE configuration covers libupnp through 1.6.20 and Debian 8.0. The record was published on 2017-03-07 (NVD).
What does the 1.6.21 fix change?
The libupnp ChangeLog entry of 2016-12-08 under Version 1.6.21, "Fix out-of-bound access in create_url_list() (CVE-2016-8863)", explains the bug: "If there is an invalid URL in URLS->buf after a valid one, uri_parse is called with out pointing after the allocated memory. As uri_parse writes to *out before returning an error the loop in create_url_list must be stopped early to prevent an out-of-bound access." gena_device.c is the device side of GENA eventing, so the code runs in products that publish evented UPnP services, such as media servers and gateways, when a control point subscribes. The CALLBACK header is the same field that CallStranger, CVE-2020-12695, abuses at the protocol level.
Which firmware build systems carry it?
On 2026-11-18 the OpenWrt packages feed builds libupnp 1.14.18 on master, openwrt-24.10 and openwrt-25.12, and Buildroot's libupnp package builds 1.14.25, both well past 1.6.21. The openwrt-18.06 branch packaged libupnp 1.6.19 and its package has no patch directory, so images built from that release carry the flaw. Debian fixed its 1.6.19-based jessie package in DSA-3736-1 and wheezy in DLA-747-1 and DLA-748-1, according to the Debian security tracker, so a version string alone over-reports on Debian-derived systems. Vendor SDKs that froze an older libupnp need the same check; the 2012 SSDP overflows such as CVE-2012-5958 sit in the same library and were fixed in 1.6.18. Our MiniUPnPd and UPnP firmware guide shows how to tell libupnp from MiniUPnPd in an image.
How do you fix or mitigate it?
Upgrade libupnp to a maintained 1.14.x release or backport the 1.6.21 change to create_url_list(). If the firmware cannot change, keep the UPnP device port off untrusted networks: in the 1.14 source the miniserver starts from TCP 49152 when the application passes no port, and only the LAN segment that needs discovery should reach it. The later libupnp DNS-rebinding flaw CVE-2021-29462 is fixed only in 1.14.6, so a 1.6.21 or 1.8.x build still needs work. The memory corruption fix guide covers this class of bug in device code.
CVE-2016-8863 is not in the CISA KEV catalog as of 2026-11-18 (catalog version 2026.11.17). FIRST EPSS gives it 0.0856, the 95th percentile, in the 2026-11-17 dataset (CISA KEV, FIRST EPSS).
Evidence
- NVD: heap-based buffer overflow in create_url_list in gena/gena_device.c in libupnp before 1.6.21, valid URI followed by invalid one in SUBSCRIBE CALLBACK; Primary CVSS v3.0 9.8, v2 7.5; CWE-119; CPE libupnp through 1.6.20 and Debian 8.0; published 2017-03-07. — NVD
- libupnp ChangeLog, Version 1.6.21, 2016-12-08: 'Fix out-of-bound access in create_url_list() (CVE-2016-8863)', with the explanation quoted. — pupnp project (GitHub)
- OpenWrt packages libs/libupnp/Makefile: 1.14.18 on master, openwrt-24.10, openwrt-25.12; 1.6.19 on openwrt-18.06 with no patches directory; read 2026-11-18. — OpenWrt packages
- Buildroot package/libupnp/libupnp.mk: LIBUPNP_VERSION = 1.14.25, read 2026-11-18. — Buildroot
- Debian security tracker: libupnp fixed in wheezy 1:1.6.17-1.2+deb7u2 (DLA-747-1), jessie 1:1.6.19+git20141001-1+deb8u1 (DSA-3736-1); libupnp4 wheezy DLA-748-1. — Debian security tracker
- libupnp miniserver.c: #define APPLICATION_LISTENING_PORT 49152. — pupnp project (GitHub)
- CVE-2016-8863 is not in the CISA KEV catalog, version 2026.11.17 (1,782 entries), checked 2026-11-18. — CISA KEV
- EPSS 0.0856, percentile 0.95157, dataset 2026-11-17. — FIRST EPSS

Related advisories
Sources
Frequently asked questions
Which libupnp versions are affected by CVE-2016-8863?
NVD marks libupnp through 1.6.20 as vulnerable, and the libupnp ChangeLog lists the create_url_list() fix under Version 1.6.21. Debian backported it to its 1.6.17 and 1.6.19 packages in DLA-747-1 and DSA-3736-1 (per NVD, the libupnp ChangeLog and the Debian security tracker).
Which part of a UPnP device is exposed to CVE-2016-8863?
The device side of GENA eventing. create_url_list() parses the CALLBACK header of a SUBSCRIBE request in gena_device.c, so products that publish evented UPnP services with libupnp are exposed to anyone who can reach their UPnP HTTP port (per NVD and the libupnp ChangeLog).
Is an OpenWrt image affected by CVE-2016-8863?
Images built from openwrt-18.06 package libupnp 1.6.19 with no patches and carry it. The master, openwrt-24.10 and openwrt-25.12 branches package libupnp 1.14.18, which is not affected (per the OpenWrt packages feed, read 2026-11-18).
Is CVE-2016-8863 in the CISA KEV catalog?
No, not as of 2026-11-18 (catalog version 2026.11.17). FIRST EPSS gives it 0.0856, the 95th percentile, in the 2026-11-17 dataset (per CISA KEV and FIRST EPSS).
Questions about this page?