Description

FFmpeg versions 4.4 through 8.1.2 contain an out-of-bounds memory access vulnerability in the ADX audio decoder within libavcodec/adxdec.c that allows attackers to trigger both out-of-bounds reads and writes by supplying a crafted ADX or AAX audio file with a mid-stream channel layout change. When AV_PKT_DATA_NEW_EXTRADATA side data is received mid-stream, the adx_decode_frame function re-parses the stream header but fails to update the internal channel state, causing subsequent decoding operations to access the prev[] state array using a stale channel count.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "8.8",
  "pubDate": "2026-07-22T18:17:05.917Z",
  "pubdate": "2026-07-22T18:17:05.917Z",
  "executiveSummary": "An out-of-bounds memory access vulnerability exists in the ADX audio decoder within FFmpeg versions 4.4 through 8.1.2. The flaw resides in libavcodec/adxdec.c and allows malicious actors to trigger both out-of-bounds reads and writes. This is achieved by supplying a specially crafted ADX or AAX audio file containing a mid-stream channel layout change. Successful exploitation can lead to memory corruption, potential arbitrary code execution, or application crashes, posing severe risks to systems processing untrusted multimedia streams. Attackers require the capability to supply a malformed audio file to a vulnerable application utilizing the FFmpeg decoding pipeline, with no specific authentication or privilege requirements needed to trigger the parsing error.",
  "technicalDetails": "The vulnerability stems from improper state management during mid-stream configuration updates in the ADX audio decoder. The vulnerable component is located in libavcodec/adxdec.c, specifically within the adx_decode_frame function. When processing input streams, the decoder can receive AV_PKT_DATA_NEW_EXTRADATA side data mid-stream, signaling a change in the audio stream configuration or channel layout. Upon receiving this side data, the adx_decode_frame function re-parses the stream header to adapt to the new parameters.\nHowever, the root cause of the vulnerability is that the decoding logic fails to properly update the internal channel state and associated tracking structures following the header re-parsing. Consequently, subsequent decoding operations continue to access the internal prev[] state array using a stale channel count that no longer matches the newly established stream layout. This mismatch between the expected channel dimensions and the stale state array indexing results in out-of-bounds memory reads and writes.\nThe step-by-step attack flow involves an attacker generating or supplying a crafted ADX or AAX audio file engineered with a deliberate mid-stream channel layout alteration paired with AV_PKT_DATA_NEW_EXTRADATA side data. When a vulnerable instance of FFmpeg parses this payload, the decoder processes the initial header correctly, establishes the initial channel state, and then encounters the mid-stream alteration. The header is re-parsed, but the internal state tracking remains desynchronized. As subsequent audio frames are decoded, buffer calculations relying on the stale channel count access memory outside the bounds of the prev[] state array. Depending on heap layout and memory structures, this memory corruption can lead to denial-of-service via segmentation faults or potentially be leveraged for arbitrary code execution within the context of the running application."
}