Update cpuinfo and apply thread-safe deinitialization patch

[Vineeth Chelur (crvineeth97)]

Description

• Update pytorch/cpuinfo from 4628dc060ce4e82345dc166bbac875609db4ff69 to 66ee79c038d70dad9f08705b2c9b3e58f6d8f512, the latest commit on cpuinfo main as of August 27, 2026.
• Carry the thread-safe, reference-counted initialization and deinitialization changes from pytorch/cpuinfo#400 as one shared ORT patch used by both FetchContent and vcpkg.
• Reset Windows ARM64 cache-population state on each initialization so repeated DLL load/unload cycles cannot reuse stale cache indices.
• Scope XNNPACK's cpuinfo references to hardware discovery so XNNPACK-enabled ORT builds do not retain unmatched references during DLL unload.

The cpuinfo patch can be removed after pytorch/cpuinfo#400, including the ARM64 reinitialization fix, is merged and ORT updates to a revision containing it. The XNNPACK compatibility patch can be removed after the corresponding lifecycle fix is available upstream.

Motivation and Context

microsoft/onnxruntime#28245 integrated cpuinfo_deinitialize() after pytorch/cpuinfo#387 added it upstream. That implementation was later reverted by pytorch/cpuinfo#411 because initialization and deinitialization were not safe for multiple consumers.

Pinning the latest cpuinfo main without an ORT-side patch would therefore make cpuinfo_deinitialize() a no-op again. Carrying the corrected implementation keeps ORT independent of the pending upstream review while preserving safe cleanup during dynamic DLL unload.

Testing

• onnxruntime_cpuinfo_refcount_test covers sequential consumers, concurrent consumers, and reinitialization after final release.
• Verified this test fails against ORT main's pinned cpuinfo revision (4628dc060ce4e82345dc166bbac875609db4ff69) and passes against the patched revision (66ee79c038d70dad9f08705b2c9b3e58f6d8f512).
• In XNNPACK-enabled builds, the refcount test initializes XNNPACK and verifies hardware discovery does not retain a cpuinfo reference.
• onnxruntime_shared_lib_cpuinfo_dlopen_test loads a small DLL containing ORT's CPUIDInfo and cpuinfo, captures a cpuinfo process-heap allocation, unloads the DLL, and verifies that allocation was released.
• Verified the FetchContent patch sequences for the default, Linux, and Windows ARM64/ARM64EC paths, plus the vcpkg patch sequence, apply with zero rejected hunks.

[Copilot]

Pull request overview

Updates cpuinfo and adds lifecycle patches to support thread-safe initialization, cleanup, and repeated DLL loading.

Changes:

• Updates cpuinfo and applies reference-counted deinitialization.
• Balances XNNPACK cpuinfo usage.
• Adds concurrency and DLL unload regression tests.

Reviewed changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
onnxruntime/test/shared_lib/cpuinfo_refcount_test.cc	Tests cpuinfo lifecycle concurrency.
onnxruntime/test/shared_lib/cpuinfo_dlopen_test.cc	Tests repeated DLL unloading and heap growth.
cmake/vcpkg-ports/xnnpack/portfile.cmake	Applies the XNNPACK lifecycle patch.
cmake/vcpkg-ports/cpuinfo/portfile.cmake	Updates and patches cpuinfo.
cmake/patches/xnnpack/release_cpuinfo_after_hardware_config.patch	Releases XNNPACK’s cpuinfo references.
cmake/patches/cpuinfo/enable_deinit_refcounting.patch	Implements synchronized, reference-counted cleanup.
cmake/patches/.gitattributes	Preserves mixed line endings in the cpuinfo patch.
cmake/onnxruntime_unittests.cmake	Registers the new tests.
cmake/external/xnnpack.cmake	Applies the XNNPACK patch for fetched sources.
cmake/external/onnxruntime_external_deps.cmake	Applies the cpuinfo patch for fetched sources.
cmake/deps.txt	Updates the pinned cpuinfo revision.

---

💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Pull request overview

Copilot reviewed 10 out of 10 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated no new comments.

Suppressed comments (1)

onnxruntime/test/shared_lib/cpuinfo_dlopen_test.cc:14

• The test implementation does not match the PR's stated coverage: it loads this custom test DLL once, rather than repeatedly loading/unloading onnxruntime.dll, exercising XNNPACK, or measuring sustained process-heap growth. Either implement the described end-to-end loop or update the PR's Testing section so it does not claim coverage that this test does not provide.

  HMODULE library = LoadLibraryW(ORT_CPUINFO_DLOPEN_TEST_LIBRARY);

[Copilot]

🔵 Needs a closer look

It carries a large unmerged, cross-platform memory-lifecycle patch requiring final human validation.

Review details

• Files reviewed: 13/13 changed files
• Comments generated: 1
• Review effort level: Balanced