Correct the Windows cross-compiling commands

[ykhrustalev]

Summary

During the cross-compilation using Clang on Ubuntu for Windows, there are several commands that are executed on the host and they are failing.

Add the guarding condition which checks for CMAKE_CROSSCOMPILING and picks the Linux alternative.

Test plan

1. Create a docker image for cross compilation

ARG GROUP_ID=1000
ARG USER_ID=1000
ARG USER_NAME=docker-user
FROM amd64/ubuntu:24.04
# https://bugs.launchpad.net/cloud-images/+bug/2005129
RUN userdel -r ubuntu
ENV DEBIAN_FRONTEND=noninteractive
ARG GROUP_ID
ARG USER_ID
ARG USER_NAME
RUN apt-get -y update \
&& apt-get install --no-install-recommends -y \
ccache \
curl \
lsb-release \
wget \
software-properties-common \
gnupg \
ca-certificates \
build-essential \
git \
make \
ninja-build \
patch \
&& rm -rf /var/lib/apt/lists/*
RUN wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | apt-key add - \
&& wget https://apt.llvm.org/llvm.sh \
&& chmod +x llvm.sh \
&& ./llvm.sh 19 all \
&& apt update && apt install -y \
clang-19 \
lld-19 \
llvm-19 \
&& rm -rf /var/lib/apt/lists/*
RUN update-alternatives --install /usr/bin/clang-cl clang-cl /usr/bin/clang-cl-19 60 \
&& update-alternatives --install /usr/bin/clang clang /usr/bin/clang-19 60 \
&& update-alternatives --install /usr/bin/clang++ clang++ /usr/bin/clang++-19 60 \
&& update-alternatives --install /usr/bin/llvm-rc llvm-rc /usr/bin/llvm-rc-19 60 \
&& update-alternatives --install /usr/bin/llvm-mt llvm-mt /usr/bin/llvm-mt-19 60 \
&& update-alternatives --install /usr/bin/lld-link lld-link /usr/bin/lld-link-19 60 \
&& update-alternatives --install /usr/bin/lldb lldb /usr/bin/lldb-19 60 \
&& update-alternatives --install /usr/bin/llvm-lib llvm-lib /usr/bin/llvm-lib-19 60 \
&& update-alternatives --install /usr/bin/llvm-ar llvm-ar /usr/bin/llvm-ar-19 60 \
&& update-alternatives --install /usr/bin/llvm-ranlib llvm-ranlib /usr/bin/llvm-ranlib-19 60 \
&& update-alternatives --install /usr/bin/llvm-nm llvm-nm /usr/bin/llvm-nm-19 60 \
&& update-alternatives --install /usr/bin/llvm-objdump llvm-objdump /usr/bin/llvm-objdump-19 60 \
&& update-alternatives --install /usr/bin/llvm-objcopy llvm-objcopy /usr/bin/llvm-objcopy-19 60 \
&& update-alternatives --install /usr/bin/llvm-strip llvm-strip /usr/bin/llvm-strip-19 60 \
&& update-alternatives --install /usr/bin/cc cc /usr/bin/clang-19 100 \
&& update-alternatives --install /usr/bin/c++ c++ /usr/bin/clang++-19 100 \
&& update-alternatives --install /usr/bin/ld ld /usr/bin/ld.lld-19 100
# -----------------------------------------------------------------------------
# user
# -----------------------------------------------------------------------------
RUN groupadd --gid $GROUP_ID docker-user \
&& useradd --uid $USER_ID --gid docker-user --create-home $USER_NAME
USER $USER_NAME
# -----------------------------------------------------------------------------
# python uv
# -----------------------------------------------------------------------------
RUN curl -LsSf https://astral.sh/uv/0.7.17/install.sh | sh
# -----------------------------------------------------------------------------
# rust, x86_64-pc-windows-msvc target
# -----------------------------------------------------------------------------
RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- --profile minimal -y
# -----------------------------------------------------------------------------
# xwin
# -----------------------------------------------------------------------------
RUN ~/.cargo/bin/cargo install xwin cargo-cache \
&& ~/.cargo/bin/cargo-cache -a \
&& ~/.cargo/bin/xwin --cache-dir /tmp/xwin-cache --accept-license --variant desktop --arch x86_64 --include-atl \
splat --preserve-ms-arch-notation --include-debug-libs --output ~/.xwin/x86_64 \
&& rm -rf /tmp/xwin-cache

2. Create cmake dir.

For some reason I had to disable AVX instructions to get it working, it is not the intention, the intention here is to use a window toolchain

cmake -G Ninja -B build-win \
-DCMAKE_BUILD_TYPE=Release \
-DCMAKE_SYSTEM_NAME=Windows \
-DCMAKE_SYSTEM_VERSION=10.0 \
-DCMAKE_SYSTEM_PROCESSOR=AMD64 \
-DCMAKE_C_COMPILER=clang-cl \
-DCMAKE_CXX_COMPILER=clang-cl \
-DCMAKE_ASM_COMPILER=clang-cl \
-DCMAKE_RC_COMPILER=llvm-rc \
-DCMAKE_LINKER=lld-link \
-DCMAKE_C_COMPILER_TARGET=x86_64-pc-windows-msvc \
-DCMAKE_CXX_COMPILER_TARGET=x86_64-pc-windows-msvc \
-DCMAKE_ASM_COMPILER_TARGET=x86_64-pc-windows-msvc \
-DCMAKE_SYSROOT=/home/docker-user/.xwin/x86_64 \
-DCMAKE_FIND_ROOT_PATH=/home/docker-user/.xwin/x86_64 \
-DCMAKE_FIND_ROOT_PATH_MODE_PROGRAM=NEVER \
-DCMAKE_FIND_ROOT_PATH_MODE_LIBRARY=ONLY \
-DCMAKE_FIND_ROOT_PATH_MODE_INCLUDE=ONLY \
-DCMAKE_CXX_FLAGS="/imsvc /home/docker-user/.xwin/x86_64/crt/include /imsvc /home/docker-user/.xwin/x86_64/sdk/include/ucrt /imsvc /home/docker-user/.xwin/x86_64/sdk/include/um /imsvc /home/docker-user/.xwin/x86_64/sdk/include/shared -Wno-unknown-argument" \
-DCMAKE_C_FLAGS="/imsvc /home/docker-user/.xwin/x86_64/crt/include /imsvc /home/docker-user/.xwin/x86_64/sdk/include/ucrt /imsvc /home/docker-user/.xwin/x86_64/sdk/include/um /imsvc /home/docker-user/.xwin/x86_64/sdk/include/shared -Wno-unknown-argument" \
-DCMAKE_EXE_LINKER_FLAGS="/libpath:/home/docker-user/.xwin/x86_64/crt/lib/x64 /libpath:/home/docker-user/.xwin/x86_64/sdk/lib/um/x64 /libpath:/home/docker-user/.xwin/x86_64/sdk/lib/ucrt/x64" \
-DCMAKE_SHARED_LINKER_FLAGS="/libpath:/home/docker-user/.xwin/x86_64/crt/lib/x64 /libpath:/home/docker-user/.xwin/x86_64/sdk/lib/um/x64 /libpath:/home/docker-user/.xwin/x86_64/sdk/lib/ucrt/x64" \
-DGFLAGS_INTTYPES_FORMAT=VC7 \
-DEXECUTORCH_BUILD_XNNPACK=ON \
-DXNNPACK_ENABLE_ASSEMBLY=OFF \
-DXNNPACK_BUILD_TESTS=OFF \
-DXNNPACK_BUILD_BENCHMARKS=OFF \
-DXNNPACK_ENABLE_AVX512F=OFF \
-DXNNPACK_ENABLE_AVX512SKX=OFF \
-DXNNPACK_ENABLE_AVX512VBMI=OFF \
-DXNNPACK_ENABLE_AVX512VNNI=OFF \
-DXNNPACK_ENABLE_AVX512VNNIGFNI=OFF \
-DXNNPACK_ENABLE_AVX512AMX=OFF \
-DXNNPACK_ENABLE_AVX512FP16=OFF \
--fresh

3. Build

cmake --build build-win --config Release --target xnnpack_schema

The error will have

FAILED: schema/include/executorch/backends/xnnpack/serialization/schema_generated.h /mnt/executorch/build-win/schema/include/executorch/backends/xnnpack/serialization/schema_generated.h
cd /mnt/executorch && /mnt/executorch/build-win/third-party/flatbuffers_external_project/bin/flatc --cpp --cpp-std c++11 --scoped-enums -o /mnt/executorch/build-win/schema/include/executorch/backends/xnnpack/serialization backends/xnnpack/serialization/runtime_schema.fbs && powershell -Command "Move-Item -Path /mnt/executorch/build-win/schema/include/executorch/backends/xnnpack/serialization/runtime_schema_generated.h -Destination /mnt/executorch/build-win/schema/include/executorch/backends/xnnpack/serialization/schema_generated.h"
/bin/sh: 1: powershell: not found
ninja: build stopped: subcommand failed.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/13071

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• ghstack-mergeability-check and Check labels failing with 'Resource not accessible by integration'

✅ No Failures

As of commit 4dc8bc2 with merge base d36e83a ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[kimishpatel]

in the test plan, can you add what steps did you take that failed before this PR and worked after

[ykhrustalev]

@kimishpatel updated