Guard GridSample CUDA coordinate conversion

[Gopalakrishnan Nallasamy (GopalakrishnanN)]

Description

Guards the CUDA GridSample coordinate conversion paths for non-finite and extreme grid values, matching the existing CPU behavior for float grids. The CUDA path now sanitizes coordinates before integer conversion, uses wider intermediate indices where needed, and clamps reflected indices before sampling.

Per review feedback, GsReflect performs the isfinite check before computing the reflection range (x_max - x_min), so a non-finite coordinate returns early without the extra subtraction. The same reorder is applied to the CPU GsReflect (core/providers/cpu/tensor/grid_sample.cc) to keep the CPU and CUDA implementations in lockstep; behavior is unchanged.

Tests

• .\.venv\Scripts\python.exe tools\ci_build\build.py --config RelWithDebInfo --build --parallel --target onnxruntime_provider_test --build_dir build\Windows
• .\build\Windows\RelWithDebInfo\RelWithDebInfo\onnxruntime_provider_test.exe --gtest_filter=*Grid*
  • 180 tests passed
• .\.venv\Scripts\clang-format.exe on touched C++ files

Note: the local build is CPU-only (onnxruntime_USE_CUDA=OFF), so CUDA compilation/runtime coverage will come from CUDA CI.

Test coverage

The hardening is applied at shared choke-points: coordinate sanitization runs before interpolation-mode dispatch, and the int64_t index widening and reflected-index clamps are single shared branches. So one representative case per {mode, padding, dimensionality} exercises the hardened path rather than the full cross-product. Regression tests use constant-valued images, so the expected output is well-defined regardless of which (now sanitized/clamped) index each adversarial coordinate resolves to; this also sidesteps a pre-existing CPU-vs-CUDA reflected-index difference (double-reflect + round-half-to-even). The new GridSampleCudaHardeningTest cases run on CPU and CUDA (and CUDA-NHWC when ENABLE_CUDA_NHWC_OPS is enabled); the CUDA kernel is registered for float only, so double is exercised through the shared templated CPU path.

[Copilot]

Pull request overview

This PR hardens the CUDA implementation of the ONNX GridSample operator against non-finite and extreme grid coordinates, aligning CUDA behavior with existing CPU float-grid handling and preventing undefined behavior during float→integer conversions.

Changes:

• Added device-side guards and coordinate sanitization in CUDA GridSample kernels to ensure float→int64 conversions are well-defined for NaN/Inf/extreme values.
• Widened intermediate index types to int64_t where needed and clamped reflected indices before sampling to avoid out-of-range accesses.
• Added regression tests that run on CPU and (when enabled) CUDA EPs to cover these adversarial coordinate cases.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
onnxruntime/test/providers/cpu/tensor/grid_sample_test_custom.cc	Adds CUDA-focused hardening regression tests and a CPU+CUDA EP runner helper (excluding CoreML/WebGPU due to differing semantics).
onnxruntime/core/providers/cuda/tensor/grid_sample_impl.cu	Sanitizes unsafe coordinates, switches key index computations to int64_t, guards reflection math, and clamps reflected indices before sampling.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.