CVE-2021-37663: Incomplete validation in `QuantizeV2`
7.8
CVSS Score
3.1
Basic Information
CVE ID
GHSA ID
EPSS Score
0.00931%
CWE
Published
8/25/2021
Updated
11/13/2024
KEV Status
No
Technology
Python
Technical Details
CVSS Vector
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
Package Name | Ecosystem | Vulnerable Versions | First Patched Version |
---|---|---|---|
tensorflow | pip | < 2.3.4 | 2.3.4 |
tensorflow | pip | >= 2.4.0, < 2.4.3 | 2.4.3 |
tensorflow | pip | = 2.5.0 | 2.5.1 |
tensorflow-cpu | pip | < 2.3.4 | 2.3.4 |
tensorflow-cpu | pip | >= 2.4.0, < 2.4.3 | 2.4.3 |
tensorflow-cpu | pip | = 2.5.0 | 2.5.1 |
tensorflow-gpu | pip | < 2.3.4 | 2.3.4 |
tensorflow-gpu | pip | >= 2.4.0, < 2.4.3 | 2.4.3 |
tensorflow-gpu | pip | = 2.5.0 | 2.5.1 |
Vulnerability Intelligence
Miggo AI
Root Cause Analysis
The vulnerability stems from missing validation logic in QuantizeV2Op's Compute function as shown in the patch commit. The original code (before 6da6620) did not properly validate that: 1) min_range and max_range have the same number of elements, 2) axis is within valid range for input tensor rank, and 3) min_range/max_range lengths match the input's axis dimension. The patch adds multiple OP_REQUIRES checks for these conditions, confirming these were the missing validations. The function's central role in parameter processing and the direct correlation between the vulnerability description and added validation checks make this identification high confidence.