CVE-2021-37712: Arbitrary File Creation/Overwrite via insufficient symlink protection due to directory cache poisoning using symbolic links

A flaw was found in the npm package "tar" (aka node-tar). Extracting tar files that contain two directories and a symlink with names containing Unicode values that normalize to the same value on Windows systems made it possible to bypass node-tar symlink checks on directories. This allows an untrusted tar file to extract and overwrite files into an arbitrary location. The highest threat from this vulnerability is to integrity and system availability.

node-tar aims to guarantee that any file whose location would be modified by a symbolic link is not extracted. This is, in part, achieved by ensuring that extracted directories are not symlinks. Additionally, in order to prevent unnecessary stat calls to determine whether a given path is a directory, paths are cached when directories are created.

This logic was insufficient when extracting tar files that contained two directories and a symlink with names containing unicode values that normalized to the same value. Additionally, on Windows systems, long path portions would resolve to the same file system entities as their 8.3 "short path" counterparts. A specially crafted tar archive could thus include directories with two forms of the path that resolve to the same file system entity, followed by a symbolic link with a name in the first form, lastly followed by a file using the second form. It led to bypassing node-tar symlink checks on directories, essentially allowing an untrusted tar file to symlink into an arbitrary location and subsequently extracting arbitrary files into that location, thus allowing arbitrary file creation and overwrite.

Reference: https://github.com/npm/node-tar/security/advisories/GHSA-qq89-hq3f-393p

— Red Hat

Node.js tar module could allow a local attacker to execute arbitrary code on the system, caused by an arbitrary file creation/overwrite vulnerability. By creating a directory, and then replacing that directory with a symlink that had a different apparent name that resolved to the same entry in the filesystem, an attacker could use an untrusted tar file to symlink into an arbitrary location and extract arbitrary files into that location to create or overwrite arbitrary files and execute arbitrary code on the system.

— IBM

The npm package "tar" (aka node-tar) before versions 4.4.18, 5.0.10, and 6.1.9 has an arbitrary file creation/overwrite and arbitrary code execution vulnerability. node-tar aims to guarantee that any file whose location would be modified by a symbolic link is not extracted. This is, in part, achieved by ensuring that extracted directories are not symlinks. Additionally, in order to prevent unnecessary stat calls to determine whether a given path is a directory, paths are cached when directories are created. This logic was insufficient when extracting tar files that contained both a directory and a symlink with names containing unicode values that normalized to the same value. Additionally, on Windows systems, long path portions would resolve to the same file system entities as their 8.3 "short path" counterparts. A specially crafted tar archive could thus include a directory with one form of the path, followed by a symbolic link with a different string that resolves to the same file system entity, followed by a file using the first form. By first creating a directory, and then replacing that directory with a symlink that had a different apparent name that resolved to the same entry in the filesystem, it was thus possible to bypass node-tar symlink checks on directories, essentially allowing an untrusted tar file to symlink into an arbitrary location and subsequently extracting arbitrary files into that location, thus allowing arbitrary file creation and overwrite. These issues were addressed in releases 4.4.18, 5.0.10 and 6.1.9. The v3 branch of node-tar has been deprecated and did not receive patches for these issues. If you are still using a v3 release we recommend you update to a more recent version of node-tar. If this is not possible, a workaround is available in the referenced GHSA-qq89-hq3f-393p.