ONNX: Out-of-bounds read in ConvTranspose shape inference via undersized weight rank

[ Description ]Official description

Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.2.1 until 1.22.0, ConvTranspose shape inference processes a weight tensor with fewer than three dimensions without a kernel_shape attribute by deriving an empty effective_kernel_shape and indexing it for the input's spatial dimensions. The flaw affects convTransposeShapeInference_opset11() and the current inference path, and an untrusted model that can pass onnx.checker.check_model() can crash onnx.shape_inference.infer_shapes(), causing denial of service without demonstrated confidentiality or integrity impact. This issue is fixed in version 1.22.0.

[ Fixes ]Fixes and patches

• onnx 1.22.0Fixed version · GitHub Advisories · pip package onnx: upgrade to 1.22.0 or later (affected >= 1.2.1, <= 1.21.0)

[ Packages ]Affected open-source packages

• onnx pipAffected >= 1.2.1, <= 1.21.0 · fixed in 1.22.0 · GHSA-6pwm-x34p-fvj2

[ References ]References

• github.com/onnx/onnx/commit/2ec288908952eea9dcf48d4b197915e1f72c6abf
• github.com/onnx/onnx/pull/7997
• github.com/onnx/onnx/releases/tag/v1.22.0
• github.com/onnx/onnx/security/advisories/GHSA-6pwm-x34p-fvj2