CHECK-fail in LSTM with zero-length input in TensorFlow

Impact Running an LSTM/GRU model where the LSTM/GRU layer receives an input with zero-length results in a CHECK failure when using the CUDA backend. This can result in a query-of-death vulnerability, via denial of service, if users can control the input to the layer. Patches We have patched the...

Write to immutable memory region in TensorFlow

Impact The tf.rawops.ImmutableConst operation returns a constant tensor created from a memory mapped file which is assumed immutable. However, if the type of the tensor is not an integral type, the operation crashes the Python interpreter as it tries to write to the memory area: python import...

GHSA-HHVC-G5HV-48C6 Write to immutable memory region in TensorFlow

Impact The tf.rawops.ImmutableConst operation returns a constant tensor created from a memory mapped file which is assumed immutable. However, if the type of the tensor is not an integral type, the operation crashes the Python interpreter as it tries to write to the memory area: python import...

GHSA-C9F3-9WFR-WGH7 Lack of validation in data format attributes in TensorFlow

Impact The tf.rawops.DataFormatVecPermute API does not validate the srcformat and dstformat attributes. The code assumes that these two arguments define a permutation of NHWC. However, these assumptions are not checked and this can result in uninitialized memory accesses, read outside of bounds a...

Lack of validation in data format attributes in TensorFlow

Impact The tf.rawops.DataFormatVecPermute API does not validate the srcformat and dstformat attributes. The code assumes that these two arguments define a permutation of NHWC. However, these assumptions are not checked and this can result in uninitialized memory accesses, read outside of bounds a...

Uninitialized memory access in TensorFlow

Impact Under certain cases, a saved model can trigger use of uninitialized values during code execution. This is caused by having tensor buffers be filled with the default value of the type but forgetting to default initialize the quantized floating point types in Eigen: cc struct QUInt8 QUInt8 /...

GHSA-QHXX-J73R-QPM2 Uninitialized memory access in TensorFlow

Impact Under certain cases, a saved model can trigger use of uninitialized values during code execution. This is caused by having tensor buffers be filled with the default value of the type but forgetting to default initialize the quantized floating point types in Eigen: cc struct QUInt8 QUInt8 /...

Google TensorFlow 缓冲区错误漏洞

Google TensorFlow is a suite of end-to-end open source platforms for machine learning from Google USA. Google TensorFlow suffers from a buffer overflow vulnerability that stems from the DataFormatVecPermute API not validating the src format and dst format attributes. An attacker could exploit the...

Google TensorFlow 缓冲区错误漏洞

Google TensorFlow is a suite of end-to-end open source platforms for machine learning from Google USA. Google TensorFlow has a security vulnerability that can be exploited by an attacker to access a special directory...

Google TensorFlow 输入验证错误漏洞

Google TensorFlow is a suite of end-to-end open source platforms for machine learning from Google USA. Google TensorFlow suffers from a denial-of-service vulnerability that stems from the LSTM GRU layer receiving a zero-length input when using a CUDA backend, which results in a check failure. An...

PT-2020-16389 · Google · Tensorflow

Name of the Vulnerable Software and Affected Versions: TensorFlow versions prior to 1.15.5 TensorFlow versions prior to 2.0.4 TensorFlow versions prior to 2.1.3 TensorFlow versions prior to 2.2.2 TensorFlow versions prior to 2.3.2 TensorFlow versions prior to 2.4.0 Description: In affected versio...

PT-2020-16393 · Google · Tensorflow

Name of the Vulnerable Software and Affected Versions: TensorFlow versions prior to 1.15.5 TensorFlow versions prior to 2.0.4 TensorFlow versions prior to 2.1.3 TensorFlow versions prior to 2.2.2 TensorFlow versions prior to 2.3.2 TensorFlow versions prior to 2.4.0 Description: In affected versio...

PT-2020-16392 · Google · Tensorflow

Name of the Vulnerable Software and Affected Versions: TensorFlow versions 2.4.0rc Description: The general implementation for matching filesystem paths to globbing pattern in TensorFlow is vulnerable to an access out of bounds of the array holding the directories. This issue arises due to...

Google TensorFlow 缓冲区错误漏洞

Google TensorFlow is a suite of end-to-end open source platforms for machine learning from Google USA. Google TensorFlow suffers from a buffer overflow vulnerability that stems from a tensor buffer being populated with the default value for the type, but forgetting to initialize the quantized...

Google TensorFlow 缓冲区错误漏洞

Google TensorFlow is a suite of end-to-end open source platforms for machine learning from Google USA. TensorFlow suffers from a buffer overflow vulnerability that can be exploited by an attacker to leak addresses from the library...

Google TensorFlow 安全漏洞

Google TensorFlow is a suite of end-to-end open source platforms for machine learning from Google USA. Google TensorFlow suffers from a buffer overflow vulnerability that can be exploited by an attacker to cause a buffer overflow...

PT-2020-16391 · Google · Tensorflow

Name of the Vulnerable Software and Affected Versions: TensorFlow versions prior to 1.15.5 TensorFlow versions prior to 2.0.4 TensorFlow versions prior to 2.1.3 TensorFlow versions prior to 2.2.2 TensorFlow versions prior to 2.3.2 TensorFlow versions prior to 2.4.0 Description: The tf.raw...

Security Bulletin: IBM Watson Discovery for IBM Cloud Pak for Data affected by vulnerability in TensorFlow

Summary IBM Watson Discovery for IBM Cloud Pak for Data contains a vulnerable version of TensorFlow. Vulnerability Details CVEID: CVE-2020-15190 DESCRIPTION: TensorFlow is vulnerable to a denial of service, caused by improper input validation by the tf.rawops.Switch operation in eager mode. By...

GHSA-XWHF-G6J5-J5GC Float cast overflow undefined behavior

Impact When the boxes argument of tf.image.cropandresize has a very large value, the CPU kernel implementation receives it as a C++ nan floating point value. Attempting to operate on this is undefined behavior which later produces a segmentation fault. Patches We have patched the issue in...

Float cast overflow undefined behavior

Impact When the boxes argument of tf.image.cropandresize has a very large value, the CPU kernel implementation receives it as a C++ nan floating point value. Attempting to operate on this is undefined behavior which later produces a segmentation fault. Patches We have patched the issue in...