Arbitrary writes via (tarfile) realpath overflow

The vulnerability allows an attacker to write files to arbitrary locations on the system during tar archive extraction. This can lead to a complete system compromise, including overwriting critical system files, injecting malicious code, or exfiltrating sensitive data. The filter="data" option, intended for data integrity checks, is exploited to bypass extraction path restrictions. Successful exploitation requires an attacker to control the contents of the tar archive being extracted. This vulnerability shares similarities with other archive-related vulnerabilities where improper handling of file paths can lead to arbitrary write access.

Applications and systems that rely on the Python tarfile module to process untrusted tar archives are at risk. This includes build systems, deployment pipelines, and any application that accepts tar archives from external sources. Shared hosting environments where multiple users can upload files are particularly vulnerable.

• python / server:

find / -name '*.tar.gz' -type f -print0 | xargs -0 grep -i 'data filter'

• python / supply-chain: Inspect Python dependencies for vulnerable versions of tarfile. • generic web: Monitor web server access logs for requests containing tar.gz archives, especially those originating from untrusted sources.

1. 2025-06-03Disclosure

   disclosure

1. Reserviert2025-05-09
2. Veröffentlicht2025-06-03
3. Geändert2026-04-21
4. EPSS aktualisiert2026-03-23

The primary mitigation is to upgrade to Python 3.14.0b3 or later, where the default filter value has been changed to "data". If upgrading is not immediately feasible, implement strict validation of the extraction path before calling TarFile.extractall() or TarFile.extract(). This validation should ensure that the extraction path is within a controlled and trusted directory. Consider using a WAF or proxy to inspect incoming tar archives for suspicious patterns or malicious file names. Avoid using the filter parameter with untrusted archives.