Integer overflow in Tensorflow
Description
Tensorflow is an Open Source Machine Learning Framework. Under certain scenarios, Grappler component of TensorFlow is vulnerable to an integer overflow during cost estimation for crop and resize. Since the cropping parameters are user controlled, a malicious person can trigger undefined behavior. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.
Affected packages
Versions sourced from the GitHub Security Advisory.
| Package | Affected versions | Patched versions |
|---|---|---|
tensorflowPyPI | < 2.5.3 | 2.5.3 |
tensorflowPyPI | >= 2.6.0, < 2.6.3 | 2.6.3 |
tensorflowPyPI | >= 2.7.0, < 2.7.1 | 2.7.1 |
tensorflow-cpuPyPI | < 2.5.3 | 2.5.3 |
tensorflow-cpuPyPI | >= 2.6.0, < 2.6.3 | 2.6.3 |
tensorflow-cpuPyPI | >= 2.7.0, < 2.7.1 | 2.7.1 |
tensorflow-gpuPyPI | < 2.5.3 | 2.5.3 |
tensorflow-gpuPyPI | >= 2.6.0, < 2.6.3 | 2.6.3 |
tensorflow-gpuPyPI | >= 2.7.0, < 2.7.1 | 2.7.1 |
Affected products
1- Range: >= 2.7.0, < 2.7.1
Patches
10aaaae6eca5aPrevent overflow in grappler cost estimation of crop&resize op.
1 file changed · +22 −7
tensorflow/core/grappler/costs/op_level_cost_estimator.cc+22 −7 modified@@ -2681,27 +2681,42 @@ Status OpLevelCostEstimator::PredictCropAndResize(const OpContext& op_context, // calculation differs from rough estimate in implementation, as it separates // out cost per box from cost per pixel and cost per element. + // Since crop arguments are user controlled, check for overflow. + int64_t crop_area = MultiplyWithoutOverflow(crop_height, crop_width); + if (crop_area < 0) + return errors::InvalidArgument("Cannot estimate cost, multiplying ", + crop_height, " with ", crop_width, + " would overflow"); + int64_t crop_volume = MultiplyWithoutOverflow(crop_area, num_boxes); + if (crop_volume < 0) + return errors::InvalidArgument("Cannot estimate cost, multiplying ", + crop_area, " with ", num_boxes, + " would overflow"); + int64_t crop_depth = MultiplyWithoutOverflow(crop_height, num_boxes); + if (crop_depth < 0) + return errors::InvalidArgument("Cannot estimate cost, multiplying ", + crop_height, " with ", num_boxes, + " would overflow"); + // Ops for variables height_scale and width_scale. int64_t ops = (sub_cost * 6 + mul_cost * 2 + div_cost * 2) * num_boxes; // Ops for variable in_y. - ops += (mul_cost * 2 + sub_cost + add_cost) * crop_height * num_boxes; + ops += (mul_cost * 2 + sub_cost + add_cost) * crop_depth; // Ops for variable in_x (same computation across both branches). - ops += (mul_cost * 2 + sub_cost + add_cost) * crop_height * crop_width * - num_boxes; + ops += (mul_cost * 2 + sub_cost + add_cost) * crop_volume; // Specify op_cost based on the method. if (use_bilinear_interp) { // Ops for variables top_y_index, bottom_y_index, y_lerp. - ops += (floor_cost + ceil_cost + sub_cost) * crop_height * num_boxes; + ops += (floor_cost + ceil_cost + sub_cost) * crop_depth; // Ops for variables left_x, right_x, x_lerp; - ops += (floor_cost + ceil_cost + sub_cost) * crop_height * crop_width * - num_boxes; + ops += (floor_cost + ceil_cost + sub_cost) * crop_volume; // Ops for innermost loop across depth. ops += (cast_to_float_cost * 4 + add_cost * 3 + sub_cost * 3 + mul_cost * 3) * output_elements; } else /* method == "nearest" */ { // Ops for variables closest_x_index and closest_y_index. - ops += round_cost * 2 * crop_height * crop_width * num_boxes; + ops += round_cost * 2 * crop_volume; // Ops for innermost loop across depth. ops += cast_to_float_cost * output_elements; }
Vulnerability mechanics
Generated by null/stub on May 9, 2026. Inputs: CWE entries + fix-commit diffs from this CVE's patches. Citations validated against bundle.
References
7- github.com/advisories/GHSA-8jj7-5vxc-pg2qghsaADVISORY
- nvd.nist.gov/vuln/detail/CVE-2022-23587ghsaADVISORY
- github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2022-96.yamlghsaWEB
- github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2022-151.yamlghsaWEB
- github.com/tensorflow/tensorflow/blob/a1320ec1eac186da1d03f033109191f715b2b130/tensorflow/core/grappler/costs/op_level_cost_estimator.ccghsax_refsource_MISCWEB
- github.com/tensorflow/tensorflow/commit/0aaaae6eca5a7175a193696383f582f53adab23fghsax_refsource_MISCWEB
- github.com/tensorflow/tensorflow/security/advisories/GHSA-8jj7-5vxc-pg2qghsax_refsource_CONFIRMWEB
News mentions
0No linked articles in our index yet.