GHSA-m87f-9fvv-2mgg
Deserialization of Untrusted Data in parlai
Packages
Details
As written in the advisory (Markdown source, shown as plain text).
### Impact
Due to use of unsafe YAML deserialization logic, an attacker with the ability to modify local YAML configuration files could provide malicious input, resulting in remote code execution or similar risks.
### Patches
The issue can be patched by upgrading to v1.1.0 or later. It can also be patched by replacing YAML deserialization with equivalent safe_load calls.
### References
- https://github.com/facebookresearch/ParlAI/commit/507d066ef432ea27d3e201da08009872a2f37725
- https://github.com/facebookresearch/ParlAI/commit/4374fa2aba383db6526ab36e939eb1cf8ef99879
- https://anon-artist.github.io/blogs/blog3.html
References
External links from the advisory. They lead to third-party sites.
- https://github.com/facebookresearch/ParlAI/security/advisories/GHSA-m87f-9fvv-2mgg
- https://nvd.nist.gov/vuln/detail/CVE-2021-39207
- https://github.com/facebookresearch/ParlAI/commit/4374fa2aba383db6526ab36e939eb1cf8ef99879
- https://github.com/facebookresearch/ParlAI/commit/507d066ef432ea27d3e201da08009872a2f37725
- https://github.com/advisories/GHSA-mwgj-7x7j-6966
- https://github.com/facebookresearch/ParlAI
- https://github.com/facebookresearch/ParlAI/releases/tag/v1.1.0
- https://github.com/pypa/advisory-database/tree/main/vulns/parlai/PYSEC-2021-330.yaml
- https://github.com/pypa/advisory-database/tree/main/vulns/parlai/PYSEC-2021-334.yaml
- http://packetstormsecurity.com/files/164136/Facebook-ParlAI-1.0.0-Code-Execution-Deserialization.html