PT-2020-6980 · Libsvm+2 · Libsvm+2

Pabloec20

·

Published

2020-11-21

·

Updated

2025-02-04

·

CVE-2020-28975

CVSS v2.0

7.8

High

VectorAV:N/AC:L/Au:N/C:N/I:N/A:C
Name of the Vulnerable Software and Affected Versions Libsvm version v324 scikit-learn version 0.23.2
Description The issue is related to the svm predict values function in svm.cpp of Libsvm, which can cause a denial of service (segmentation fault) when a crafted model SVM with a large value in the n support array is introduced. This can occur via pickle, json, or any other model permanence standard. The scikit-learn vendor notes that this behavior can only happen if the library's API is violated by an application that changes a private attribute.
Recommendations For Libsvm version v324, consider disabling the svm predict values function until a patch is available. For scikit-learn version 0.23.2, avoid using the n support array in the affected svm predict values function until the issue is resolved. At the moment, there is no information about a newer version that contains a fix for this vulnerability.

Exploit

Buffer Overflow

Improper Resource Release

Weakness Enumeration

Related Identifiers

ALT-PU-2023-6666
BDU:2024-07246
CVE-2020-28975
GHSA-JXFP-4RVQ-9H9M
OPENSUSE-SU-2024:11805-1
OPENSUSE-SU-2025:14729-1
PYSEC-2020-108

Affected Products

Alt Linux
Libsvm
Scikit-Learn