CVE-2026-55093
CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:L/I:N/A:H
Summary
Tract is a tiny, no-nonsense, self-contained TensorFlow and ONNX inference toolkit. Prior to 0.21.16, 0.22.2, and 0.23.1, tract-nnef uses unchecked usize multiplication in nnef/src/tensors.rs read_tensor for attacker-controlled tensor dimensions, the allocation size, and the reported tensor length. Loading a crafted NNEF archive through model_for_path or model_for_read reaches the default DatLoader and can make the wrapped size check accept a small allocation while data/src/tensor.rs as_slice_unchecked creates a much larger logical slice. Model construction through as_uniform can then read beyond the heap allocation and disclose adjacent data, and later access can terminate the process with a segmentation fault. The affected dense numeric tensor path does not include the independently guarded bool, String, or block-quant paths, and no out-of-bounds write or code execution was demonstrated. This issue is fixed in versions 0.21.16, 0.22.2, and 0.23.1.
Affected Software
| Vendor | Product | Version Range | Status |
|---|---|---|---|
| sonos | tract | < 0.21.16 | affected |
| sonos | tract | >= 0.22.0, < 0.22.2 | affected |
| sonos | tract | >= 0.23.0, < 0.23.1 | affected |
Weaknesses
- CWE-125: CWE-125: Out-of-bounds Read
- CWE-190: CWE-190: Integer Overflow or Wraparound
ADP Enrichment
CISA ADP Vulnrichment
- SSVC:
- Exploitation: poc
- Automatable: no
- Technical Impact: partial
Additional References
References
- https://github.com/sonos/tract/security/advisories/GHSA-x5mv-8wgw-29hg
- https://github.com/sonos/tract/commit/34c7df2c9bd2a36583e09b52f3e6319bf23102e8
- https://github.com/sonos/tract/commit/828ec20f74c14c0923c7d7f0cff062ed222d630a
- https://github.com/sonos/tract/commit/ee5f019d39fb97f7d4431b883c3ce97b74c28a64
- https://github.com/sonos/tract/releases/tag/0.21.16
- https://github.com/sonos/tract/releases/tag/0.22.2
- https://github.com/sonos/tract/releases/tag/v0.23.1
Feedback
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.