CVE-2026-48484: 
Python Analyse et atténuation des vulnérabilités

CVE-2026-48484 is a Denial of Service (DoS) vulnerability in pyLoad, a free and open-source download manager written in Python, caused by a lack of input size validation during file uploads. The rpc function in api_blueprint.py reads the entire content of a multipart/form-data uploaded file into memory using file.read() without enforcing any size limit, allowing an authenticated attacker to exhaust server memory and terminate the process. All versions of pyload-ng prior to 0.5.0b3.dev101 are affected. The vulnerability was published on October 8–9, 2026, and carries a CVSS v3.1 base score of 6.5 (Medium) (Github Advisory).

The root cause is improper input validation (CWE-20) combined with uncontrolled resource consumption (CWE-400). In src/pyload/webui/app/blueprints/api_blueprint.py at line 73, when the request MIME type is multipart/form-data, the handler calls file.read() unconditionally — loading the entire uploaded file into memory before passing it to the underlying API function — with no MAX_CONTENT_LENGTH or equivalent size cap configured (pyload source). An attacker must be authenticated (via session or API key) and can target any API endpoint that accepts file uploads, such as check_online_status_container, by sending a crafted multipart/form-data POST request to /api/rpc with an arbitrarily large file (Github Advisory). The fix adds MAX_CONTENT_LENGTH = 16 * 1024 * 1024 (16 MB) to the Flask application configuration (pyload commit).

Successful exploitation causes the pyLoad process to be terminated by the operating system's Out-Of-Memory (OOM) killer, or renders the entire system unresponsive due to swap thrashing or memory exhaustion. All active downloads and tasks managed by pyLoad are interrupted upon process termination. There is no confidentiality or integrity impact; the vulnerability is limited to a high availability impact on the affected pyLoad instance (Github Advisory).

No public proof-of-concept exploit code or in-the-wild exploitation has been reported as of the disclosure date. The EPSS score is 0.0, indicating a very low probability of exploitation in the near term, and the vulnerability is not listed in the CISA Known Exploited Vulnerabilities (KEV) catalog (Github Advisory). Exploitation requires low-privilege authenticated access, which somewhat limits the attack surface, though API keys may be obtainable through credential theft or weak configurations.

1. Obtain credentials or API key: Acquire valid pyLoad credentials or an API key for a low-privilege account through credential theft, brute force, or reuse of default credentials.
2. Identify target endpoint: Confirm the pyLoad instance is running a version prior to 0.5.0b3.dev101 and that the /api/rpc endpoint is accessible over the network.
3. Prepare a large file: Create a large sparse file (e.g., 10 GB) on the attacker's system using a command such as truncate -s 10G large_file.bin.
4. Send malicious multipart upload: Submit a multipart/form-data POST request to the /api/rpc endpoint with the large file attached, targeting a file-accepting function:

curl -X POST "http://<target>:8000/api/rpc" \
  -H "X-API-Key: YOUR_API_KEY" \
  -F "func=check_online_status_container" \
  -F "container=@large_file.bin"

5. Memory exhaustion: The server reads the entire file into memory via file.read(), exhausting available RAM and triggering the OOM killer, which terminates the pyLoad process and disrupts all active downloads (Github Advisory).

• Network: Unusually large HTTP POST requests (multi-gigabyte bodies) to /api/rpc with Content-Type: multipart/form-data; sustained high-bandwidth inbound traffic to the pyLoad web port (default 8000).
• Logs: Web server or Flask access logs showing POST requests to /api/rpc with very large Content-Length values; repeated requests from the same source IP targeting file-upload API functions.
• Process/System: Sudden spike in memory consumption by the pyLoad Python process; OOM killer log entries in /var/log/syslog or dmesg referencing the pyLoad process (e.g., Out of memory: Kill process <pid> (python)); unexpected process termination of pyLoad.
• File System: Presence of large temporary files in system or application temp directories associated with the pyLoad process during or after the attack.

Upgrade pyload-ng to version 0.5.0b3.dev101 or later, which introduces a MAX_CONTENT_LENGTH = 16 MB limit in the Flask application configuration, preventing unbounded file reads (pyload commit). As an interim workaround, enforce upload size limits at the reverse proxy or web server layer (e.g., set client_max_body_size in Nginx or LimitRequestBody in Apache) to reject oversized requests before they reach the pyLoad application (Github Advisory). Additionally, restrict API access to trusted networks and rotate or revoke API keys if unauthorized access is suspected.