GHSA-4W8F-HJM9-XWGF
Vulnerability from github – Published: 2022-06-06 21:24 – Updated: 2024-09-16 21:49Impact
It was possible to traverse the entire AWS S3 bucket and in most cases to access or delete files. The issue was discovered by the maintainer. There were no reports of the vulnerability being known to or exploited by a third party, before the release of the patch.
If the AWS_LOCATION setting was set, traversal was limited to that location only.
If all your files handling views (like form views) require authentication or special permission, the thread is limited to privileged users.
Patches
The vulnerability has been fixed in version 5.5.1 and above.
Workarounds
There is no feasible workaround. We must urge all users to immediately updated to a patched version.
Detailed attack vector description
An attacker may use a request with malicious form data to traverse the entire AWS S3 bucket and perform destructive operations.
An attack could look as follows:
curl -X POST -F "s3file=file" -F "file=/priviliged/location/secrets.txt" https://www.example.com/any/path/will/work/
This will result in a request with files set and opened:
>>> request.FILES.getlist("file")
[File("/priviliged/location/secrets.txt")]
Since this behavior is injected via a middleware, any view can be called this way and will carry any files defined by the attacker.
Via the s3file form field, any input name can be specified, including multiple inputs. For each input, multiple files can be freely
picked of the S3 bucket.
Scenarios and their practicality
There are four scenarios that would be considered practical in most setups:
- Illegal file injection,
- file deletion,
- file retrieval & tree traversal.
- code injection & remote code execution.
File deletion
An attacker knows the location of a privileged file, like a static asset. Next, the file is injected into a form view. The upload to function will move the file to a new location. This is effectively deleting the file, since the previous references to it are invalid, and will cause S3 to return a 404. Furthermore, the new location is unknown to the site operator.
File retrieval & tree traversal
An attacker knows the URL of a secret file and injects it into a form view. The view will move the file to a public location, making it accessible to the attacker. Since most form views will not be rate limited, this could also be used to guess files and traverse the file tree.
Illegal file injection
An attacker uses any form to upload a file to the temporary upload location. Next, the attacker injects that file into a request, does not validate the contents or is not equipped to handle the mime type. The latter could be used as a potential DOS vector.
In practice, this is not a practical risk in most hardened setup. Files should always be sanitized before processing, since files can be included in a request even without this security issues.
For more information
If you have any questions or comments about this advisory: * Open an issue on GitHub * Email us at johannes@maron.family
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "django-s3file"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "5.5.1"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2022-24840"
],
"database_specific": {
"cwe_ids": [
"CWE-22",
"CWE-96"
],
"github_reviewed": true,
"github_reviewed_at": "2022-06-06T21:24:24Z",
"nvd_published_at": "2022-06-09T04:15:00Z",
"severity": "CRITICAL"
},
"details": "### Impact\n\nIt was possible to traverse the entire AWS S3 bucket and in most cases to access or delete files.\nThe issue was discovered by the maintainer. There were no reports of the vulnerability\nbeing known to or exploited by a third party, before the release of the patch.\n\nIf the `AWS_LOCATION` setting was set, traversal was limited to that location only.\nIf all your files handling views (like form views) require authentication or special permission, the thread is limited to privileged users.\n\n### Patches\n\nThe vulnerability has been fixed in version 5.5.1 and above.\n\n### Workarounds\n\nThere is no feasible workaround. We must urge all users to immediately updated to a patched version.\n\n### Detailed attack vector description\n\nAn attacker may use a request with malicious form data to traverse the entire AWS S3 bucket and perform destructive operations.\n\nAn attack could look as follows:\n```bash\ncurl -X POST -F \"s3file=file\" -F \"file=/priviliged/location/secrets.txt\" https://www.example.com/any/path/will/work/\n```\n\nThis will result in a request with files set and opened:\n\n```python\n\u003e\u003e\u003e request.FILES.getlist(\"file\")\n[File(\"/priviliged/location/secrets.txt\")]\n```\n\nSince this behavior is injected via a middleware, any view can be called this way and will carry any files defined by the attacker.\n\nVia the `s3file` form field, any input name can be specified, including multiple inputs. For each input, multiple files can be freely\npicked of the S3 bucket.\n\n#### Scenarios and their practicality\n\nThere are four scenarios that would be considered practical in most setups:\n\n1. Illegal file injection,\n2. file deletion,\n3. file retrieval \u0026 tree traversal.\n4. code injection \u0026 remote code execution.\n\n##### File deletion\n\nAn attacker knows the location of a privileged file, like a static asset. Next, the file is injected into a form view. The upload to function will move the file to a new location. This is effectively deleting the file, since the previous references to it are invalid, and will cause S3 to return a 404. Furthermore, the new location is unknown to the site operator.\n\n##### File retrieval \u0026 tree traversal\n\nAn attacker knows the URL of a secret file and injects it into a form view. The view will move the file to a public location, making it accessible to the attacker. Since most form views will not be rate limited, this could also be used to guess files and traverse the file tree.\n\n##### Illegal file injection\n\nAn attacker uses any form to upload a file to the temporary upload location. Next, the attacker injects that file into a request, does not validate the contents or is not equipped to handle the mime type. The latter could be used as a potential DOS vector.\n\nIn practice, this is not a practical risk in most hardened setup. Files should always be sanitized before processing, since files can be included in a request even without this security issues.\n\n### For more information\nIf you have any questions or comments about this advisory:\n* Open an issue on [GitHub](https://github.com/codingjoe/django-s3file/issues)\n* Email us at [johannes@maron.family](mailto:johannes@maron.family)\n",
"id": "GHSA-4w8f-hjm9-xwgf",
"modified": "2024-09-16T21:49:30Z",
"published": "2022-06-06T21:24:24Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/codingjoe/django-s3file/security/advisories/GHSA-4w8f-hjm9-xwgf"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2022-24840"
},
{
"type": "WEB",
"url": "https://github.com/codingjoe/django-s3file/commit/68ccd2c621a40eb66fdd6af2be9d5fcc9c373318"
},
{
"type": "PACKAGE",
"url": "https://github.com/codingjoe/django-s3file"
},
{
"type": "WEB",
"url": "https://github.com/codingjoe/django-s3file/releases/tag/5.5.1"
},
{
"type": "WEB",
"url": "https://github.com/pypa/advisory-database/tree/main/vulns/django-s3file/PYSEC-2022-208.yaml"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:N",
"type": "CVSS_V3"
},
{
"score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "Path Traversal in django-s3file"
}
Sightings
| Author | Source | Type | Date |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.