Neophytos Christou

TensorFlow is an open source platform for machine learning. When `tf.raw ops.FusedResizeAndPadConv2D` is given a large tensor shape, it overflows.

The issue occurs when `tf.raw ops.ResizeNearestNeighborGrad` is given a large `size` input, causing an overflow. This can be exploited with a specific input, such as `size = tf.constant([1879048192,1879048192], shape=[2], dtype=tf.int32)`.

TensorFlow is an open source platform for machine learning. When `tf.raw ops.ImageProjectiveTransformV2` is given a large output shape, it overflows. The issue has been patched in GitHub commit 8faa6ea692985dbe6ce10e1a3168e0bd60a723ba.

The issue occurs when `QuantizedAvgPool` is given `min input` or `max input` tensors of a nonzero rank, resulting in a segfault that can be used to trigger a denial of service attack.

The implementation of `Conv2DBackpropInput` requires `input sizes` to be 4-dimensional. Otherwise, it gives a `CHECK` failure which can be used to trigger a denial of service attack.

The issue arises when `QuantizedInstanceNorm` is given `x min` or `x max` tensors of a nonzero rank, resulting in a segfault that can be used to trigger a denial of service attack. There are no known workarounds for this issue.

The issue arises when `FakeQuantWithMinMaxVars` is given `min` or `max` tensors of a nonzero rank, resulting in a `CHECK` fail that can be used to trigger a denial of service attack. There are no known workarounds for this issue.

The issue arises when `QuantizedBiasAdd` is given `min input`, `max input`, `min bias`, `max bias` tensors of a nonzero rank, resulting in a segfault that can be used to trigger a denial of service attack.

The issue arises when `QuantizedMatMul` is given nonscalar input for: `min a`, `max a`, `min b`, or `max b`. This can trigger a segfault, leading to a denial of service attack.

TensorFlow is an open source platform for machine learning. If `QuantizeDownAndShrinkRange` is given nonscalar inputs for `input min` or `input max`, it results in a segfault that can be used to trigger a denial of service attack.

The issue results in a segfault that can be used to trigger a denial of service attack when `QuantizedRelu` or `QuantizedRelu6` are given nonscalar inputs for `min features` or `max features`.

The issue is related to the `FractionalMaxPoolGrad` function, which validates its inputs with `CHECK` failures instead of returning errors. If it receives incorrectly sized inputs, the `CHECK` failure can be used to trigger a denial of service attack. This can be achieved by providing specific input values to the `tf.raw ops.FractionalMaxPoolGrad` function, such as `orig input`, `orig output`, `out backprop`, `row pooling sequence`, and `col pooling sequence`. The estimated number of potentially affected devices is not provided.

The issue occurs when `tf.quantization.fake quant with min max vars per channel gradient` receives input `min` or `max` of rank other than 1, resulting in a `CHECK` fail that can trigger a denial of service attack.

The issue occurs when `tf.quantization.fake quant with min max vars gradient` receives input `min` or `max` that is nonscalar, resulting in a `CHECK` fail that can trigger a denial of service attack. There are no known workarounds for this issue.

The issue results in a segfault that can be used to trigger a denial of service attack when the `Requantize` function is given `input min`, `input max`, `requested output min`, `requested output max` tensors of a nonzero rank.

The issue arises when `RaggedTensorToVariant` is given a `rt nested splits` list containing tensors of ranks other than one, resulting in a `CHECK` fail that can be used to trigger a denial of service attack. There are no known workarounds for this issue.

The issue arises when `FakeQuantWithMinMaxVarsPerChannel` is given `min` or `max` tensors of a rank other than one, resulting in a `CHECK` fail that can be used to trigger a denial of service attack.

The issue arises when the `QuantizeAndDequantizeV3` function is given a nonscalar `num bits` input tensor, resulting in a `CHECK` fail that can be used to trigger a denial of service attack. There are no known workarounds for this issue.

The implementation of SobolSampleOp is vulnerable to a denial of service via CHECK-failure (assertion failure) caused by assuming `input(0)`, `input(1)`, and `input(2)` to be scalar. This issue can be triggered by providing non-scalar inputs to the SobolSampleOp function.

The implementation of `AvgPool3DGradOp` does not fully validate the input `orig input shape`, resulting in an overflow that causes a `CHECK` failure, which can be used to trigger a denial of service attack.