---
id: "CVE-2026-125186"
page: "https://infrastack.ai/live/CVE-2026-125186"
kind: "vulnerability"
aliases: ["CVE-2026-125186"]
title: "FFmpeg: crafted H.264 packets exhaust CPU through dts2pts filter"
severity: "medium"
cvss: 5.7
epss: null
exploited: false
status: "open"
category: "package"
products: []
ecosystems: []
tags: ["ffmpeg", "h264", "media-processing", "package"]
not_affected_if: ["no-untrusted-input"]
published_at: "2026-11-09T19:16:54+00:00"
first_seen_at: "2026-11-09T19:28:17+00:00"
updated_at: "2026-11-09T19:32:05+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-125186", "first_seen_at": "2026-11-09T19:28:17+00:00"}]
exploit_sources: []
---

# FFmpeg: crafted H.264 packets exhaust CPU through dts2pts filter

Link for people: https://infrastack.ai/live/CVE-2026-125186 (this .md is for agents)

FFmpeg through 9.0.2 has an uncontrolled resource consumption flaw in its dts2pts bitstream filter. A crafted H.264 packet with many slices forces unbounded per-slice memory and CPU allocation, letting an attacker cause a denial of service.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Run `ffmpeg -version` and compare against the fixed release named in the sources (unknown here)
- Check whether the `dts2pts` bitstream filter is used in your pipelines (unknown if the sources do not say)

## Timeline

- 2026-11-09T19:28:17+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-125186

## Documents

- [advisory] Third-party advisory: www.vulncheck.com https://www.vulncheck.com/advisories/ffmpeg-through-9.0.2-cpu-exhaustion-via-dts2pts-bitstream-filter
- [fix] Patch: github.com https://github.com/FFmpeg/FFmpeg/commit/d8013c061725631261d7326c07d2af3a67fa4bc1

## From the sources

FFmpeg through 9.0.2 contains an uncontrolled resource consumption vulnerability in the dts2pts bitstream filter that allows attackers to exhaust CPU through unbounded per-packet POC node allocation. Attackers can supply a crafted H.264 packet with many slices of alternating picture order counts, forcing h264_queue_frame() to allocate tree nodes per slice and deny service.

---

_Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai_