GHSA-jfp7-4j67-8r3q Heap buffer overflow caused by rounding

• PyPI tensorflow: < 2.1.4 → fixed in 2.1.4 • PyPI tensorflow: >= 2.2.0, < 2.2.3 → fixed in 2.2.3 • PyPI tensorflow: >= 2.3.0, < 2.3.3 → fixed in 2.3.3 • PyPI tensorflow: >= 2.4.0, < 2.4.2 → fixed in 2.4.2 • PyPI tensorflow-cpu: < 2.1.4 → fixed in 2.1.4 • PyPI tensorflow-cpu: >= 2.2.0, < 2.2.3 → fixed in 2.2.3 • PyPI tensorflow-cpu: >= 2.3.0, < 2.3.3 → fixed in 2.3.3 • PyPI tensorflow-cpu: >= 2.4.0, < 2.4.2 → fixed in 2.4.2 • PyPI tensorflow-gpu: < 2.1.4 → fixed in 2.1.4 • PyPI tensorflow-gpu: >= 2.2.0, < 2.2.3 → fixed in 2.2.3 • PyPI tensorflow-gpu: >= 2.3.0, < 2.3.3 → fixed in 2.3.3 • PyPI tensorflow-gpu: >= 2.4.0, < 2.4.2 → fixed in 2.4.2

Impact An attacker can trigger a heap buffer overflow in `tf.raw_ops.QuantizedResizeBilinear` by manipulating input values so that float rounding results in off-by-one error in accessing image elements: ```python import tensorflow as tf l = [256, 328, 361, 17, 361, 361, 361, 361, 361, 361, 361, 361, 361, 361, 384] images = tf.constant(l, shape=[1, 1, 15, 1], dtype=tf.qint32) size = tf.constant([12, 6], shape=[2], dtype=tf.int32) min = tf.constant(80.22522735595703) max = tf.constant(80.39215850830078) tf.raw_ops.QuantizedResizeBilinear(images=images, size=size, min=min, max=max, align_corners=True, half_pixel_centers=True) ``` This is because the implementation computes two integers (representing the upper and lower bounds for interpolation) by ceiling and flooring a floating point value: ```cc const float in_f = std::floor(in); interpolation->lower[i] = std::max(static_cast(in_f), static_cast(0)); interpolation->upper[i] = std::min(static_cast(std::ceil(in)), in_size - 1); ``` For some values of `in`, `interpolation->upper[i]` might be smaller than `interpolation->lower[i]`. This is an issue if `interpolation->upper[i]` is capped at `in_size-1` as it means that `interpolation->lower[i]` points outside of the image. Then, in the interpolation code, this would result in heap buffer overflow: ```cc template inline void OutputLerpForChannels(const InterpolationCache& xs, const int64 x, const T_SCALE ys_ilerp, const int channels, const float min, const float max, const T* ys_input_lower_ptr, const T* ys_input_upper_ptr, T* output_y_ptr) { const int64 xs_lower = xs.lower[x]; ... for (int c = 0; c upper[i]` is smaller than `interpolation->lower[i]`, we can set them to be equal without affecting the output. Patches We have patched the issue in GitHub commit f851613f8f0fb0c838d160ced13c134f778e3ce7. The fix will be included in TensorFlow 2.5.0. We will also cherrypick this commit on TensorFlow 2.4.2, TensorFlow 2.3.3, TensorFlow 2.2.3 and TensorFlow 2.1.4, as these are also affected and still in supported range. For more information Please consult our security guide for more information regarding the security model and how to contact us with issues and questions. Attribution This vulnerability has been reported by Ying Wang and Yakun Zhang of Baidu X-Team.