CVE-2026-121629
HighCVSS 7.5Summary
A flaw was found in GStreamer's adaptive streaming demuxers (HLS, DASH, MSS). When processing a media playlist or manifest, URIs referenced in the manifest content are fetched without restricting which URI schemes are permitted. An attacker who controls a manifest URL, or who can inject content into an HTTP manifest response, can include file:// URIs pointing to arbitrary local files, causing the affected application to read and process those files.
Risk Assessment
In automated or server-side deployments such as media recorders, proxies, and downloaders, this can lead to disclosure of sensitive local files. Existence of files can also be inferred through differing error responses even when content is not directly observable.
Recommendation
Restrict permitted URI schemes in GStreamer demuxers to the necessary ones (e.g., http/https) and update the library to a patched version. Avoid processing untrusted manifests in server-side contexts.
Other vulnerabilities in GStreamer
See all- CVE-2017-5843High
Multiple use-after-free vulnerabilities in the gst_mini_object_unref, gst_tag_list_unref, and gst_mxf_demux_update_essence_tracks functions in GStreamer before 1.10.3 allow remote attackers to cause a denial of service (crash) via vectors involving stream tags.
- CVE-2017-5840High
The qtdemux_parse_samples function in gst/isomp4/qtdemux.c in GStreamer before version 1.10.3 allows remote attackers to cause a denial of service (out-of-bounds heap read) via vectors involving the current stts index.
- CVE-2017-5838High
The gst_date_time_new_from_iso8601_string function in GStreamer before version 1.10.3 allows remote attackers to cause a denial of service via a malformed datetime string, leading to an out-of-bounds heap read.
- CVE-2016-9447High
The ROM mappings in the NSF decoder in gstreamer 0.10.x allow remote attackers to cause a denial of service (out-of-bounds read or write) and possibly execute arbitrary code via a crafted NSF music file.
- CVE-2016-9446High
The vmnc decoder in gstreamer does not initialize the render canvas, allowing remote attackers to obtain sensitive information. This is demonstrated by thumbnailing a simple 1 frame vmnc movie that does not draw to the allocated render canvas.
- CVE-2016-9445High
Integer overflow in the vmnc decoder in gstreamer allows remote attackers to cause a denial of service (crash) via large width and height values, which triggers a buffer overflow.
- CVE-2016-9809High
Off-by-one error in the gst_h264_parse_set_caps function in GStreamer before version 1.10.2 allows remote attackers to have unspecified impact via a crafted file, leading to an out-of-bounds read.
- CVE-2016-9808High
The FLIC decoder in GStreamer before 1.10.2 allows remote attackers to cause a denial of service (out-of-bounds write and crash) via a crafted series of skip and count pairs.
- CVE-2026-111183Critical
The rtpL16depay element in GStreamer has a heap-based buffer overflow due to missing validation of user-supplied data length before copying. This allows a remote attacker to execute arbitrary code in the context of the current process.
- CVE-2024-47606Critical
An integer underflow was detected in GStreamer's qtdemux_parse_theora_extension function in qtdemux.c. This leads to an allocation size overflow, overwriting the GstMapInfo structure and potentially hijacking a function pointer, which may allow arbitrary code execution.
Original NVD description (English source)
A flaw was found in GStreamer's adaptive streaming demuxers (HLS, DASH, MSS). When processing a media playlist or manifest, URIs referenced in the manifest content are fetched without restricting which URI schemes are permitted. An attacker who controls a manifest URL, or who can inject content into an HTTP manifest response, can include file:// URIs pointing to arbitrary local files, causing the affected application to read and process those files. In automated or server-side deployments such as media recorders, proxies, and downloaders, this can lead to disclosure of sensitive local files. Existence of files can also be inferred through differing error responses even when content is not directly observable.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS