CVE-2026-42369
Published: 04 May 2026
Description
GV-VMS V20 is a Video Monitoring Software used to gather the feeds of many surveillance cameras and manage other security devices. It is a native application accessed locally, but it is also possible to enable remote access via the "WebCam…
more
Server" feature. Once enabled, it is possible to access to the management and monitoring feature via a regular Web interface. This webersever is another native application, compiled without ASLR, which makes exploitation much easier and more likely. Most of the features require authentication before being reachable and leverage a standard login page to grant access. However the `gvapi` endpoint uses its own authentication mechanism via an `HTTP Authorization` header. It supports both `Basic` authentication and the `Digest` modes of authentication. #### Stack-overflow via unbound copy of base64 decoded string The `b64decoder` string is sized dynamically, but it is then copied to the `Buffer` stack variable one character at the time at [0], and there's no bound-check. As such, if the decoded string is bigger than 256 characters (the size of the `Buffer` variable) then a stack overflow occurs. Because the data can be fully controlled by an attacker and lack of ASLR, this vulnerability can easily be exploited to gain full code execution as SYSTEM on the machine running the service.
Mitigating Controls (NIST 800-53 r5)AI
Directly remediates the stack overflow flaw in the WebCam Server's gvapi endpoint by requiring timely patching of known vulnerabilities.
Requires validation of HTTP Authorization header inputs to reject oversized base64-encoded strings that exceed the fixed 256-byte stack buffer.
Enforces memory protections such as ASLR and data execution prevention to mitigate reliable exploitation of the stack overflow despite lack of ASLR in the vulnerable application.
Security SummaryAI
CVE-2026-42369 is a stack overflow vulnerability (CWE-787) affecting GV-VMS V20, a Video Monitoring Software used to manage feeds from surveillance cameras and other security devices. The issue resides in the WebCam Server feature, a native application that enables remote access to management and monitoring via a web interface and is compiled without ASLR. Specifically, the `gvapi` endpoint processes HTTP Authorization headers supporting Basic and Digest authentication; during base64 decoding, the dynamically sized `b64decoder` string is copied character-by-character into a fixed 256-byte stack `Buffer` without bounds checking, triggering an overflow if the decoded string exceeds this size.
A remote network attacker can exploit this unauthenticated (PR:N) by crafting an oversized base64-encoded string in the HTTP Authorization header sent to the `gvapi` endpoint. Full attacker control over the input, combined with the lack of ASLR, enables reliable exploitation for arbitrary code execution as SYSTEM on the host running the WebCam Server service. The vulnerability yields a CVSS v3.1 base score of 10.0 (AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H), indicating critical impact across confidentiality, integrity, and availability.
Advisories are available from Talos Intelligence (https://https://talosintelligence.com/vulnerability_reports/) and GeoVision (https://www.geovision.com.tw/cyber_security.php), which provide additional details on the vulnerability.
Details
- CWE(s)
MITRE ATT&CK Enterprise TechniquesAI
Why these techniques?
Remote unauthenticated stack overflow in public-facing WebCam Server/gvapi web endpoint enables T1190 for initial access; yields arbitrary code execution as SYSTEM, directly facilitating T1068 for privilege escalation.