CVE-2025-4287
Description
A problematic vulnerability has been identified in PyTorch version 2.6.0 with CUDA 12.4. The affected component is the torch.cuda.nccl.reduce function within the torch/cuda/nccl.py file. Exploitation of this vulnerability can lead to a denial-of-service condition. The attack can be launched from the local host. Public exploit code for this issue is currently available, increasing the potential for exploitation.
Statement
Red Hat Product Security has determined that this vulnerability does not affect any currently supported Red Hat product. This assessment may evolve based on further analysis and discovery. For more information about this vulnerability and the products it affects, please see the linked references.
Mitigation
Mitigation for this issue is either not available or the currently available options do not meet the Red Hat Product Security criteria comprising ease of use and deployment, applicability to widespread installation base or stability.
Understanding the Weakness (CWE)
Availability,Other
Technical Impact: DoS: Resource Consumption (Other); Varies by Context
Most unreleased resource issues result in general software reliability problems, but if an attacker can intentionally trigger a resource leak, the attacker might be able to launch a denial of service attack by depleting the resource pool.
Confidentiality
Technical Impact: Read Application Data
When a resource containing sensitive information is not correctly shutdown, it may expose the sensitive data in a subsequent allocation.
Frequently Asked Questions
Not sure what something means? Check out our Security Glossary.
Want to get errata notifications? Sign up here.
