PYSEC-2020-302

In affected versions of TensorFlow under certain cases, loading a saved model can result in accessing uninitialized memory while building the computation graph. The MakeEdge function creates an edge between one output tensor of the src node given by outputindex and the input slot of the dst node...

PYSEC-2020-257

In affected versions of TensorFlow under certain cases, loading a saved model can result in accessing uninitialized memory while building the computation graph. The MakeEdge function creates an edge between one output tensor of the src node given by outputindex and the input slot of the dst node...

PYSEC-2020-337

In affected versions of TensorFlow under certain cases, loading a saved model can result in accessing uninitialized memory while building the computation graph. The MakeEdge function creates an edge between one output tensor of the src node given by outputindex and the input slot of the dst node...

CVE-2020-26266

In affected versions of TensorFlow 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...

CVE-2020-26266 Uninitialized memory access in Eigen types in TensorFlow

In affected versions of TensorFlow 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...

CVE-2020-26266

CVE-2020-26266 (TensorFlow) arises from use of uninitialized Eigen quantized floating point types during code execution, triggered by saved-model handling. Affected TensorFlow versions include 1.15.5, 2.0.4, 2.1.3, 2.2.2, 2.3.2, and 2.4.0; fixes are in those same branches as indicated. The issue ...

CVE-2020-26267 Lack of validation in data format attributes in TensorFlow

In affected versions of TensorFlow 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. This can result in uninitialized memory accesses, read outside of bounds and even crashes. Th...

CVE-2020-26267

CVE-2020-26267 affects TensorFlow where tf.raw_ops.DataFormatVecPermute does not validate src_format and dst_format, allowing uninitialized memory accesses, out-of-bounds reads, or crashes by assuming a NHWC permutation. Affected releases include various TensorFlow branches; fixes are published i...

CVE-2020-26268 Write to immutable memory region in TensorFlow

In affected versions of TensorFlow 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...

CVE-2020-26268

CVE-2020-26268 affects TensorFlow: tf.raw_ops.ImmutableConst can crash Python when mapping a file to a non-integral tensor type, due to an allocator not returning an opaque handle. The issue may trigger a segmentation fault if the memory area is large enough; a check prevents the fault if the fil...

CVE-2020-26269 Heap out of bounds read in filesystem glob matching in TensorFlow

In TensorFlow release candidate versions 2.4.0rc, the general implementation for matching filesystem paths to globbing pattern is vulnerable to an access out of bounds of the array holding the directories. There are multiple invariants and preconditions that are assumed by the parallel...

CVE-2020-26269

CVE-2020-26269 affects TensorFlow (release candidate 2.4.0rc* and related master branch changes) and describes an access out-of-bounds read in the general filesystem-globbing path matching implementation. The root cause is that invariants and preconditions assumed by the parallel GetMatchingPaths...

CVE-2020-26270 CHECK-fail in LSTM with zero-length input in TensorFlow

In affected versions of TensorFlow 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...

CVE-2020-26270

CVE-2020-26270 affects TensorFlow: when an LSTM/GRU layer receives a zero-length input, the CUDA backend triggers a CHECK failure leading to a denial-of-service (query-of-death). Public sources consistently describe this as a vulnerability in affected TensorFlow builds, with fixes implemented in ...

CVE-2020-26271 Heap out of bounds access in MakeEdge in TensorFlow

In affected versions of TensorFlow under certain cases, loading a saved model can result in accessing uninitialized memory while building the computation graph. The MakeEdge function creates an edge between one output tensor of the src node given by outputindex and the input slot of the dst node...

CVE-2020-26271

In affected versions of TensorFlow under certain cases, loading a saved model can result in accessing uninitialized memory while building the computation graph. The MakeEdge function creates an edge between one output tensor of the src node given by outputindex and the input slot of the dst node...

CVE-2020-26271

CVE-2020-26271 : TensorFlow contains a heap out-of-bounds access in MakeEdge when wiring graph edges, caused by missing bounds checks on indices into arrays of tensor data. This can lead to uninitialized memory access and, in certain cases, leakage of library addresses. The description notes the ...

GHSA-Q263-FVXM-M5MW Heap out of bounds access in MakeEdge in TensorFlow

Impact Under certain cases, loading a saved model can result in accessing uninitialized memory while building the computation graph. The MakeEdge function creates an edge between one output tensor of the src node given by outputindex and the input slot of the dst node given by inputindex. This is...

Heap out of bounds access in MakeEdge in TensorFlow

Impact Under certain cases, loading a saved model can result in accessing uninitialized memory while building the computation graph. The MakeEdge function creates an edge between one output tensor of the src node given by outputindex and the input slot of the dst node given by inputindex. This is...

GHSA-M648-33QF-V3GP 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...