CVE-2026-103042

Description

LightLLM through 1.2.0 contains a memory exhaustion vulnerability in the NCCL control channel when started with --pd_trans_mode nccl, allowing unauthenticated attackers to exhaust KV-transfer worker memory. Attackers can call the exposed_set_value method to store unbounded key-value pairs without size limits, causing the worker process to crash and triggering node failure.

References

• https://github.com/ModelTC/LightLLM
• https://github.com/ModelTC/LightLLM/issues/1595
• https://github.com/ModelTC/lightllm/blob/v1.2.0/lightllm/server/router/model_infer/mode_backend/pd/base_kv_move_manager.py#L121-L122
• https://github.com/ModelTC/lightllm/blob/v1.2.0/lightllm/server/router/model_infer/mode_backend/pd/nccl_kv_transporter.py#L416-L420
• https://github.com/ModelTC/lightllm/blob/v1.2.0/lightllm/server/router/model_infer/mode_backend/pd/nccl_kv_transporter.py#L474-L485