Fix non-ASCII Unicode model path crash across session and provider code

[sagarbhure-msft]

Description

On Windows, std::filesystem::path::string() converts the internal UTF-16 representation to a narrow string using the system's active ANSI code page. When the path contains characters outside that code page (Japanese, Chinese, Korean, etc.), this throws std::system_error with 'No mapping for the Unicode character exists in the target multi-byte code page.'

This affected both the core session telemetry logging (causing InferenceSession::Initialize() to fail) and execution provider code (OpenVINO, TensorRT, TensorRT RTX, QNN, MIGraphX) where model paths are converted for EPContext attributes and profiling.

Motivation and Context

Fix: Replace .filename().string() with PathToUTF8String(.filename().native()) which uses WideCharToMultiByte(CP_UTF8, ...) and handles all Unicode characters correctly. This pattern is already used elsewhere in the codebase for path-to-string conversions.

Note: Two remaining instances in Linux-only code (cann_utils.cc, device_discovery.cc) are left as-is since .string() is safe on Linux where paths are already narrow strings.

Fixes microsoft/WindowsAppSDK#6173

[Copilot]

Pull request overview

Fixes Windows crashes (and related EP telemetry/profiling issues) when model/library paths contain non-ASCII Unicode by converting filenames to UTF-8 safely instead of using std::filesystem::path::string().

Changes:

• Replaced .filename().string() with UTF-8-safe conversions for model filenames in core session telemetry logging.
• Updated multiple execution providers (TensorRT, TensorRT RTX, QNN, OpenVINO, MIGraphX) to use UTF-8-safe filename extraction for EPContext/profiling attributes.
• Updated environment EP library registration telemetry to log UTF-8-safe library filenames.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
onnxruntime/core/session/inference_session.cc	Avoids Windows Unicode-to-ANSI conversion crash when logging model filenames to telemetry.
onnxruntime/core/session/environment.cc	Avoids Windows Unicode-to-ANSI conversion crash when logging EP library filename during registration.
onnxruntime/core/providers/tensorrt/onnx_ctx_model_helper.cc	Uses UTF-8-safe model filename for EP context attributes.
onnxruntime/core/providers/qnn/builder/qnn_profile_serializer.cc	Uses UTF-8-safe output filename for QNN profiling log naming.
onnxruntime/core/providers/qnn/builder/onnx_ctx_model_helper.cc	Uses UTF-8-safe context cache filename for QNN EP context.
onnxruntime/core/providers/openvino/backend_manager.cc	Uses UTF-8-safe blob filename for OpenVINO EP context export.
onnxruntime/core/providers/nv_tensorrt_rtx/onnx_ctx_model_helper.cc	Uses UTF-8-safe engine/model filenames for TensorRT RTX EP context attributes.
onnxruntime/core/providers/migraphx/migraphx_execution_provider_utils.h	Uses UTF-8-safe model filename for MIGraphX graph ID logging.

---

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

[yuslepukhin]

🕐

[yuslepukhin]

/azp run Linux QNN CI Pipeline, Win_TRT_Minimal_CUDA_Test_CI, Windows ARM64 QNN CI Pipeline, Windows GPU Doc Gen CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 4 pipeline(s).

[yuslepukhin]