Add ppc64le wheel build support

Author: sandeepgupta12

.ci/docker/manywheel/Dockerfile_ppc64le

@@ -0,0 +1,44 @@
+# Use UBI 9.3 as base image
+FROM registry.access.redhat.com/ubi9/ubi:9.3
+
+# Install necessary dependencies
+RUN dnf install -y \
+    https://dl.fedoraproject.org/pub/epel/epel-release-latest-9.noarch.rpm && \
+    dnf install -y git cmake ninja-build gcc-toolset-13 rust cargo zip \
+    python3 python3-devel && \
+    dnf clean all
+
+ENV PATH="/opt/rh/gcc-toolset-13/root/usr/bin:$PATH"
+ENV MANPATH="/opt/rh/gcc-toolset-13/root/usr/share/man"
+ENV INFOPATH="/opt/rh/gcc-toolset-13/root/usr/share/info"
+ENV PCP_DIR="/opt/rh/gcc-toolset-13/root"
+ENV LD_LIBRARY_PATH="/opt/rh/gcc-toolset-13/root/usr/lib64:/opt/rh/gcc-toolset-13/root/usr/lib"
+
+# Set Python and pip aliases to use Python 3.9
+RUN ln -sf /usr/bin/python3 /usr/bin/python && \
+    ln -sf /usr/bin/pip3 /usr/bin/pip
+
+COPY requirements.txt .
+# Install Python packages via pip
+RUN pip install wheel setuptools pyyaml typing_extensions expecttest
+
+#RUN source /opt/rh/gcc-toolset-13/enable && pip install -r requirements.txt
+RUN pip install -r requirements.txt
+
+# Copy the PyTorch source code into the container
+COPY . /workspace/pytorch
+
+WORKDIR /workspace/pytorch
+
+# Ensure submodules are initialized and updated
+RUN git submodule update --init --recursive
+
+# Copy the build script and make it executable
+COPY .github/scripts/ppc64le-build.sh /ppc64le-build.sh
+RUN chmod +x /ppc64le-build.sh
+
+# Verify permissions and ensure Unix line endings
+RUN dos2unix /ppc64le-build.sh || sed -i 's/\r$//' /ppc64le-build.sh
+RUN chmod +x /ppc64le-build.sh
+
+

.github/scripts/ppc64le-build.sh

@@ -0,0 +1,34 @@
+#!/usr/bin/env bash
+
+# Environment variables
+PACKAGE_NAME=pytorch
+PACKAGE_VERSION=${PACKAGE_VERSION:-v2.4.0}
+
+cd /workspace/$PACKAGE_NAME
+
+# Clean up old artifacts
+rm -rf build/ dist/ torch.egg-info/
+
+# Build and install PyTorch wheel
+if ! (MAX_JOBS=4 python setup.py bdist_wheel && pip install dist/*.whl); then
+    echo "------------------$PACKAGE_NAME:install_fails-------------------------------------"
+    exit 1
+fi
+
+# register PrivateUse1HooksInterface
+python test/test_utils.py TestDeviceUtilsCPU.test_device_mode_ops_sparse_mm_reduce_cpu_bfloat16
+python test/test_utils.py TestDeviceUtilsCPU.test_device_mode_ops_sparse_mm_reduce_cpu_float16
+python test/test_utils.py TestDeviceUtilsCPU.test_device_mode_ops_sparse_mm_reduce_cpu_float32
+python test/test_utils.py TestDeviceUtilsCPU.test_device_mode_ops_sparse_mm_reduce_cpu_float64
+
+cd ..
+pip install pytest pytest-xdist
+
+if ! pytest "$PACKAGE_NAME/test/test_utils.py"; then
+    echo "------------------$PACKAGE_NAME:install_success_but_test_fails---------------------"
+    exit 2
+    
+else
+    echo "------------------$PACKAGE_NAME:install_and_test_both_success-------------------------"
+    exit 0
+fi

.github/scripts/ppc64le-ci/README.md

@@ -0,0 +1,50 @@
+# Configuring the builder.
+
+## Install prerequisites.
+
+```
+Install Docker 
+```
+## Clone pytorch repository
+
+## Add services.
+
+```
+$ sudo cp self-hosted-builder/*.service /etc/systemd/system/
+$ sudo systemctl daemon-reload
+```
+Next step is to build `actions-runner` image using:
+
+```
+## clone gaplib repo (https://github.com/anup-kodlekere/gaplib.git) and copy runner-sdk-8.ppc64le patch from gaplib/build-files into pytorch/.github\scripts\ppc64le-ci\self-hosted-builder
+
+$ cd self-hosted-builder
+$ sudo docker build \
+      --pull \
+      -f actions-runner.Dockerfile \
+      --build-arg RUNNERPATCH="runner-sdk-8.ppc64le.patch" \
+      -t iiilinuxibmcom/actions-runner.<name> \
+      .
+```
+
+Now prepare all necessary files for runner registration:
+
+```
+$ sudo mkdir -p /etc/actions-runner/<name>
+$ sudo chmod 755 /etc/actions-runner/<name>
+$ sudo /bin/cp <github_app_private_key_file> /etc/actions-runner/<name>/key_private.pem
+$ sudo echo <github_app_id> | sudo tee /etc/actions-runner/<name>/appid.env
+$ sudo echo <github_app_install_id> | sudo tee /etc/actions-runner/<name>/installid.env
+$ sudo echo NAME=<worker_name> | sudo tee    /etc/actions-runner/<name>/env
+$ sudo echo OWNER=<github_owner>   | sudo tee -a /etc/actions-runner/<name>/env
+$ sudo echo REPO=pytorch   | sudo tee -a /etc/actions-runner/<name>/env
+$ cd self-hosted-builder
+$ sudo /bin/cp helpers/*.sh /usr/local/bin/
+$ sudo chmod 755 /usr/local/bin/app_token.sh /usr/local/bin/gh_token_generator.sh
+```
+
+## Autostart the runner.
+
+```
+$ sudo systemctl enable --now actions-runner@$NAME
+```

.github/scripts/ppc64le-ci/self-hosted-builder/actions-runner.Dockerfile

@@ -0,0 +1,102 @@
+# Self-Hosted IBM Power Github Actions Runner.
+FROM ubuntu:22.04
+
+# Set non-interactive mode for apt
+ENV DEBIAN_FRONTEND=noninteractive
+
+# Fix sources to point to ports.ubuntu.com for ppc64le
+RUN echo "deb [arch=ppc64el] http://ports.ubuntu.com/ubuntu-ports jammy main restricted universe multiverse" > /etc/apt/sources.list && \
+    echo "deb [arch=ppc64el] http://ports.ubuntu.com/ubuntu-ports jammy-updates main restricted universe multiverse" >> /etc/apt/sources.list && \
+    echo "deb [arch=ppc64el] http://ports.ubuntu.com/ubuntu-ports jammy-backports main restricted universe multiverse" >> /etc/apt/sources.list && \
+    echo "deb [arch=ppc64el] http://ports.ubuntu.com/ubuntu-ports jammy-security main restricted universe multiverse" >> /etc/apt/sources.list
+
+# Fix sources for ppc64le and update system
+RUN apt-get update -o Acquire::Retries=5 -o Acquire::http::Timeout="10" && \
+    apt-get -y install --no-install-recommends \
+    build-essential \
+    curl \
+    sudo \
+    jq \
+    gnupg-agent \
+    iptables \
+    ca-certificates \
+    software-properties-common \
+    vim \
+    zip \
+    python3 \
+    python3-pip && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+# Switch to iptables-legacy
+RUN update-alternatives --set iptables /usr/sbin/iptables-legacy && \
+    update-alternatives --set ip6tables /usr/sbin/ip6tables-legacy
+
+
+# Add Docker GPG key and repository
+RUN curl -fsSL https://download.docker.com/linux/ubuntu/gpg | gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg && \
+    echo "deb [arch=ppc64el signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable" > /etc/apt/sources.list.d/docker.list && \
+    apt-get update && apt-get install -y docker-ce docker-ce-cli containerd.io && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+# Install dotnet SDK and other dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    wget \
+    git \
+    dotnet-sdk-8.0 \
+    cmake \
+    make \
+    automake \
+    autoconf \
+    m4 \
+    libtool && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+
+# Setup user and permissions
+RUN useradd -c "Action Runner" -m runner && \
+    usermod -L runner && \
+    echo "runner ALL=(ALL) NOPASSWD: ALL" > /etc/sudoers.d/runner && \
+    groupadd docker || true && \
+    usermod -aG docker runner && \
+    (test -S /var/run/docker.sock && chmod 660 /var/run/docker.sock && chgrp docker /var/run/docker.sock || true)
+
+
+# Add and configure GitHub Actions runner
+ARG RUNNERREPO="https://github.com/actions/runner"
+ARG RUNNERPATCH
+
+ADD ${RUNNERPATCH} /tmp/runner.patch
+
+RUN git clone -q ${RUNNERREPO} /tmp/runner && \
+    cd /tmp/runner && \
+    git checkout main -b build && \
+    git apply /tmp/runner.patch && \
+    sed -i'' -e /version/s/8......\"$/${SDK}.0.100\"/ src/global.json 
+
+RUN  cd /tmp/runner/src && \
+    ./dev.sh layout && \
+    ./dev.sh package && \
+    ./dev.sh test && \
+    rm -rf /root/.dotnet /root/.nuget
+
+RUN mkdir -p /opt/runner && \
+    tar -xf /tmp/runner/_package/*.tar.gz -C /opt/runner && \
+    chown -R  runner:runner /opt/runner && \
+    su - runner -c "/opt/runner/config.sh --version"
+
+RUN     rm -rf /tmp/runner /tmp/runner.patch
+
+# Copy custom scripts and set permissions
+COPY fs/ /
+RUN chmod 777 /usr/bin/actions-runner /usr/bin/entrypoint
+
+# Switch to the runner user
+USER runner
+
+# Set working directory
+WORKDIR /opt/runner
+
+# Define entry point and command
+ENTRYPOINT ["/usr/bin/entrypoint"]
+CMD ["/usr/bin/actions-runner"]
+

.github/scripts/ppc64le-ci/self-hosted-builder/[email protected]

@@ -0,0 +1,32 @@
+[Unit]
+Description=Self-Hosted IBM power Github Actions Runner
+StartLimitIntervalSec=0
+
+[Service]
+Type=simple
+Restart=always
+
+# Cleanup stale containers
+ExecStartPre=-/usr/bin/docker rm --force actions-runner.%i
+ExecStartPre=-/usr/local/bin/gh_token_generator.sh /etc/actions-runner/%i/appid.env /etc/actions-runner/%i/installid.env /etc/actions-runner/%i/key_private.pem /etc/actions-runner/%i/ghtoken.env
+ExecStartPre=-/usr/local/bin/gh_cat_token.sh /etc/actions-runner/%i/ghtoken.env /etc/actions-runner/%i/ghtoken.txt
+
+ExecStart=/usr/bin/docker run \
+              --env-file=/etc/actions-runner/%i/env \
+              --volume /etc/actions-runner/%i/ghtoken.txt:/run/runner_secret \
+              --volume /var/run/docker.sock:/var/run/docker.sock \
+              --init \
+              --interactive \
+              --name=actions-runner.%i \
+              --rm \
+              --privileged \
+              --log-driver=journald \
+              iiilinuxibmcom/actions-runner.%i
+ExecStop=/bin/sh -c "docker exec actions-runner.%i kill -INT -- -1"
+ExecStop=/bin/sh -c "docker wait actions-runner.%i"
+ExecStop=/bin/sh -c "docker rm actions-runner.%i"
+
+ExecStop=/usr/bin/env rm -f /etc/actions-runner/%i/ghtoken.env /etc/actions-runner/%i/ghtoken.txt
+
+[Install]
+WantedBy=multi-user.target

.github/scripts/ppc64le-ci/self-hosted-builder/fs/usr/bin/actions-runner

@@ -0,0 +1,75 @@
+#!/usr/bin/env bash
+
+set -e -u
+
+trap cleanup EXIT
+
+token_file=registration-token.json
+
+# Function to clean up and unregister the runner
+cleanup() {
+    echo "Cleaning up temporary files..."
+    [ -f "$token_file" ] && rm -f "$token_file"
+    [ -f "runner-id.json" ] && rm -f "runner-id.json"
+
+    echo "Unregistering the runner from GitHub..."
+    ACCESS_TOKEN="$(cat /run/runner_secret)"
+    runner_id=$(curl -s \
+        -H "Accept: application/vnd.github.v3+json" \
+        -H "Authorization: Bearer ${ACCESS_TOKEN}" \
+        "https://api.github.com/repos/${OWNER}/${REPO}/actions/runners" | \
+        jq --raw-output '.runners[] | select(.name=="'"${NAME}"'") | .id')
+
+    if [ -n "$runner_id" ]; then
+        curl -s \
+            -X DELETE \
+            -H "Accept: application/vnd.github.v3+json" \
+            -H "Authorization: Bearer ${ACCESS_TOKEN}" \
+            "https://api.github.com/repos/${OWNER}/${REPO}/actions/runners/$runner_id"
+        echo "Runner unregistered successfully."
+    else
+        echo "Warning: Runner ID for ${NAME} not found. It may already be removed."
+    fi
+
+    unset ACCESS_TOKEN runner_id
+}
+
+# Fetch GitHub access token
+if [ ! -f /run/runner_secret ]; then
+    echo "Error: Access token file not found at /run/runner_secret."
+    exit 1
+fi
+
+
+ACCESS_TOKEN="$(cat /run/runner_secret)"
+
+# Generate registration token
+curl \
+        -X POST \
+        -H "Accept: application/vnd.github.v3+json" \
+        -H "Authorization: Bearer ${ACCESS_TOKEN}" \
+        "https://api.github.com/repos/${OWNER}/${REPO}/actions/runners/registration-token" \
+        -o "$token_file"
+ 
+unset ACCESS_TOKEN
+
+# register runner as ephemeral runner
+# it does one job, stops and unregisters
+registration_token=$(jq --raw-output .token "$token_file")
+
+./config.sh \
+        --unattended \
+        --ephemeral \
+        --url "https://github.com/${OWNER}/${REPO}" \
+        --token "${registration_token}" \
+        --name "${NAME}" \
+        --no-default-labels \
+        --labels self-hosted,linux.ppc64le
+
+unset registration_token
+rm -f "$token_file"
+
+# Run one job.
+./run.sh
+
+echo "Ephemeral runner workflow completed."

.github/scripts/ppc64le-ci/self-hosted-builder/fs/usr/bin/entrypoint

@@ -0,0 +1,30 @@
+#!/usr/bin/env bash
+
+#
+# Container entrypoint that waits for all spawned processes.
+#
+
+set -e -u
+
+# Create a FIFO and start reading from its read end.
+tempdir=$(mktemp -d "/tmp/done.XXXXXXXXXX")
+trap 'rm -r "$tempdir"' EXIT
+done="$tempdir/pipe"
+mkfifo "$done"
+cat "$done" & waiter=$!
+
+# Start the workload. Its descendants will inherit the FIFO's write end.
+status=0
+if [ "$#" -eq 0 ]; then
+    bash 9>"$done" || status=$?
+else
+    "$@" 9>"$done" || status=$?
+fi
+
+# When the workload and all of its descendants exit, the FIFO's write end will
+# be closed and `cat "$done"` will exit. Wait until it happens. This is needed
+# in order to handle SelfUpdater, which the workload may start in background
+# before exiting.
+wait "$waiter"
+
+exit "$status"