Rclone from 1.52.0 before 1.75.2 contains a cross-site request forgery vulnerability in the pCloud backend's localhost OAuth callback, which disables the state check and trusts the hostname parameter. Malicious web pages can send a stateless request to port 53682 during authorization, planting attacker tokens and API hosts so later transfers reach the attacker.
Rclone before 1.75.2 contains a path traversal vulnerability in the local backend's translateLink function, which trims the .rclonelink suffix from a path after the containment check, so a name can resolve outside the destination root. Attackers can plant a directory named ...rclonelink in a source tree synced with --links and --metadata to change ownership, timestamps, and xattrs of the destination's parent directory.
Rclone from v1.68.0 before v1.75.2 contains an authorization bypass vulnerability in serve s3 --auth-proxy that allows authenticated tenants to manipulate other tenants' multipart uploads by referencing their upload IDs. Attackers who learn a victim's bucket, object, and upload ID can inject parts, complete, or abort in-flight uploads and view upload and object metadata across tenants.
rclone serve s3 versions 1.75.0 before 1.75.2 contain an integer overflow vulnerability in multipart reorder-memory accounting that bypasses the configured buffer limit. Attackers can open two concurrent current-part PUT requests declaring huge lengths to wrap the signed int64 counter and retain out-of-order parts in memory, exhausting the host.
vLLM from 0.19.0 before 0.24.0 contains a decompression bomb vulnerability because the AudioMediaIO multimodal path calls load_audio without the max_duration_s decoded-duration limit. Authenticated attackers can send small compressed audio via input_audio or audio_url in /v1/chat/completions requests that expands to huge decoded PCM, exhausting server memory.
vLLM from 0.9.2 before 0.24.0 contains a server-side request forgery vulnerability that allows API callers to force outbound connections by embedding a decode address in the X-Request-Id header. Attackers can make a P2pNcclConnector prefill worker ZMQ-connect to an attacker host, leaking the NCCL unique_id and KV-cache tensors or blocking the worker.
vLLM from 0.10.2 before 0.25.0 contains an uncontrolled resource consumption vulnerability that allows unauthenticated attackers to exhaust memory because the Nemotron-VL processors globally set Image.MAX_IMAGE_PIXELS to None. Attackers can submit a pathological decompression-bomb image to /v1/chat/completions to OOM-kill the worker and deny service to co-tenants.
vLLM from 0.20.0 before 0.30.0 contains a denial of service vulnerability that allows remote attackers to crash the engine by sending inconsistent multimodal metadata to the disaggregated /inference/v1/generate endpoint. Attackers can replay a valid multimodal item with repeated or overlapping mm_placeholders ranges or mismatched kwargs_data and mm_hashes lists, producing a negative text span in Qwen2.5-VL position construction that terminates the engine.
vLLM from 0.18.0 before 0.31.0 contains a resource exhaustion vulnerability that allows API clients to bypass the video/jpeg frame-count cap by overriding num_frames through media_io_kwargs. Attackers can set num_frames to -1 with a large inline video so VideoMediaIO.load_base64 decodes every frame, exhausting memory and crashing the API frontend.
vLLM from 0.8.5 before 0.31.0 has a denial of service flaw. LogprobsTensors.empty_cpu() in vllm/v1/outputs.py allocates an uninitialized int32 tensor, and the V1 model runner leaves prompt positions reported as computed by a KV connector unwritten. A client sharing the engine can send prompt_logprobs (or echo with logprobs) against a connector-cached prefix, making tokenizer decode raise OverflowError, which crashes EngineCore and drops all co-tenant requests.