All reports
Medium

tensorflow: Conv2D with 'valid' padding causes XLA compile crash leading to denial of service

Affected versions

Affected versions
Unknown
Fixed version
Not reported
Affected versions
Unknown
Fixed version
Not reported

Exploitation

CVE-2025-55559

Probability of exploitation in the next 30 days: 0.21%

FIRST EPSS · 2026-11-20

Description

An issue was discovered TensorFlow v2.18.0. A Denial of Service (DoS) occurs when padding is set to 'valid' in tf.keras.layers.Conv2D.

A flaw was found in TensorFlow. When a model uses tf.keras.layers.Conv2D with padding='valid' and is compiled using XLA, the compiler miscalculates the output shape and ends up with a negative dimension. This causes the process to crash during compilation, leading to a denial of service. The same operation works correctly in eager mode because it checks dimensions dynamically, but under XLA the static shape calculation fails and stops execution.