Add bounds validation for LinearClassifier coefficients

[vraspar]

Description

Add input validation to the LinearClassifier operator to prevent an out-of-bounds heap read in GEMM when a crafted model provides mismatched coefficients/intercepts sizes.

Fixes https://portal.microsofticm.com/imp/v5/incidents/details/31000000559851/summary

Changes

• Constructor: Validate class_count_ > 0 and coefficients_.size() % class_count_ == 0
• Compute(): Validate coefficients_.size() == class_count * num_features before GEMM call
• Tests: Two regression tests for invalid coefficient sizes

Motivation and Context

MSRC case 109185 (VULN-176698): OOB read via GEMM from crafted model in LinearClassifier operator. Root cause is missing validation that the coefficients vector size matches [class_count, num_features] before passing raw pointers to GEMM.

[Copilot]

Pull request overview

Adds defensive validation to the CPU ML LinearClassifier operator to prevent crafted models from triggering an out-of-bounds read during the GEMM call by ensuring the coefficients vector size matches the expected [class_count, num_features] layout.

Changes:

• Add constructor-time validation for class_count_ > 0 and coefficients_.size() divisibility by class_count_.
• Add runtime validation in Compute() that coefficients_.size() == class_count * num_features before invoking GEMM.
• Add two regression tests covering invalid coefficient sizing scenarios.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
onnxruntime/core/providers/cpu/ml/linearclassifier.cc	Adds attribute and runtime coefficient-size validation to prevent OOB reads in GEMM.
onnxruntime/test/providers/cpu/ml/linearclassifer_test.cc	Adds regression tests asserting failures for malformed coefficient sizes.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.