PYSEC-2026-985 TensorFlow has double free in Fractional(Max/Avg)Pool

Impact nnops.fractionalavgpoolv2 and nnops.fractionalmaxpoolv2 require the first and fourth elements of their parameter poolingratio to be equal to 1.0, as pooling on batch and channel dimensions is not supported. python import tensorflow as tf import os import numpy as np from...

Double Free

tensorflow is vulnerable to a Double Free attack. The vulnerability occurs when when the first and the fourth elements of the poolingratio function has parameters not equal to 1.0 in nnops.fractionalavgpoolv2 and nnops.fractionalmaxpoolv2 because pooling on batch and channel dimensions is not...

CVE-2023-25801

CVE-2023-25801 — TensorFlow double-free in fractional pooling : Connected sources confirm a vulnerability in nn_ops.fractional_avg_pool_v2 and nn_ops.fractional_max_pool_v2 where the first and fourth elements of pooling_ratio must be 1.0 for batch/channel pooling to be supported. The issue is tri...