[ET][Windows] Relax -Werror for ExecuTorch CPU kernels on the Windows host#20949
Open
SS-JIA wants to merge 1 commit into
Open
[ET][Windows] Relax -Werror for ExecuTorch CPU kernels on the Windows host#20949SS-JIA wants to merge 1 commit into
SS-JIA wants to merge 1 commit into
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20949
Note: Links to docs will display an error until the docs builds have been completed. ❌ 23 New FailuresAs of commit 426b782 with merge base ceecc9e ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This PR needs a
|
meta-cla
Bot
added
the
CLA Signed
SS-JIA
added a commit
that referenced
this pull request
Jul 14, 2026
… host Pull Request resolved: #20949 The arvr Windows clang toolchain enables a stricter `-Werror` warning set than the sanctioned ET build platforms, and vendored third-party headers that ExecuTorch does not own trip it. VMA's `vk_mem_alloc.h` hits `-Werror,-Winconsistent-missing-destructor-override`, and PyTorch's `ATen/cpu/vec/vec_base.h` uses `#if __GNUC__ <= 12` without `defined(__GNUC__)`, hitting `-Werror,-Wundef`. Because the offending code is in vendored headers we cannot patch, this disables warnings-as-errors for the affected compiles, scoped to `ovr_config//os:windows` only, so the clang target compiles are relaxed while the MSVC host tools and all non-Windows platforms are untouched. `backends/vulkan/targets.bzl` adds `-Wno-error` to the Vulkan targets (VMA header). The CPU-kernel build is split across four independent macros/targets, each of which includes the ATen vec headers and so needs the same relaxation: `kernels/portable/op_registration_util.bzl`, `kernels/optimized/op_registration_util.bzl`, the `binary_ops` support library in `kernels/optimized/cpu/targets.bzl`, and the `libblas` support library in `kernels/optimized/lib_defs.bzl`. ghstack-source-id: 402995445 @exported-using-ghexport Differential Revision: [D112012049](https://our.internmc.facebook.com/intern/diff/D112012049/)
Gasoonjia
approved these changes
Jul 14, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stack from ghstack (oldest at bottom):
The arvr Windows clang toolchain enables a stricter
-Werrorwarning set than the sanctioned ET build platforms, and vendored third-party headers that ExecuTorch does not own trip it. VMA'svk_mem_alloc.hhits-Werror,-Winconsistent-missing-destructor-override, and PyTorch'sATen/cpu/vec/vec_base.huses#if __GNUC__ <= 12withoutdefined(__GNUC__), hitting-Werror,-Wundef. Because the offending code is in vendored headers we cannot patch, this disables warnings-as-errors for the affected compiles, scoped toovr_config//os:windowsonly, so the clang target compiles are relaxed while the MSVC host tools and all non-Windows platforms are untouched.backends/vulkan/targets.bzladds-Wno-errorto the Vulkan targets (VMA header). The CPU-kernel build is split across four independent macros/targets, each of which includes the ATen vec headers and so needs the same relaxation:kernels/portable/op_registration_util.bzl,kernels/optimized/op_registration_util.bzl, thebinary_opssupport library inkernels/optimized/cpu/targets.bzl, and thelibblassupport library inkernels/optimized/lib_defs.bzl.Differential Revision: D112012049