2.1
CVSSv2

CVE-2018-6260

Published: 13/11/2018 Updated: 18/04/2019
CVSS v2 Base Score: 2.1 | Impact Score: 2.9 | Exploitability Score: 3.9
CVSS v3 Base Score: 5.5 | Impact Score: 3.6 | Exploitability Score: 1.8
VMScore: 187
Vector: AV:L/AC:L/Au:N/C:P/I:N/A:N

Vulnerability Summary

It exists that the NVIDIA graphics drivers incorrectly handled the GPU performance counters. A local attacker could possibly use this issue to access the application data processed on the GPU.

Vulnerability Trend

Vulnerable Product Search on Vulmon Subscribe to Product

nvidia gpu driver -

Vendor Advisories

Debian Bug report logs - #913467 nvidia-graphics-drivers: CVE‑2018‑6260: access to application data processed on the GPU through a side channel exposed by the GPU performance counters Package: src:nvidia-graphics-drivers; Maintainer for src:nvidia-graphics-drivers is Debian NVIDIA Maintainers <pkg-nvidia-devel@listsaliothdebiano ...
NVIDIA graphics drivers could be made to expose sensitive information ...
NVIDIA graphics driver contains a vulnerability that may allow access to application data processed on the GPU through a side channel exposed by the GPU performance counters Local user access is required This is not a network or remote attack vector (CVE-2018-6260) ...
HP has been notified of potential security vulnerabilities with the GPU Display Driver for certain NVIDIA products These vulnerabilities may lead to denial of service, escalation of privileges, unauthorized code execution, or information disclosure ...
HP has been notified of potential security vulnerabilities with the GPU Display Driver for certain NVIDIA products These vulnerabilities may lead to denial of service, escalation of privileges, unauthorized code execution, or information disclosure ...

Github Repositories

PyProf2: PyTorch Profiling tool

PyProf2 - PyTorch Profiling tool What does this tool do? Analyzing the performance of deep neural networks is hard Getting kernels out of NVProf or NSight Compute provides some generic kernel names and execution times, but not detailed information regarding the following: Which layer launched it: eg the association of ComputeOffsetsKernel with a concrete PyTorch layer or AP