rocprofiler-systems - add gfx containers to ghcr (#883)

* Initial skeleton code for rocprofiler-systems-continuous-integration.yml

* Add python3-devel to opensuse and rhel ci images

* Update rocprofiler-systems-containers.yml to include TheRock tarballs

* Update pip install command for Dockerfile.ubuntu.ci

* Fix pip install again for Dockerfile.ubuntu.ci

* Remove skeleton workflow for CI

* Add new ci-gfx containers for TheRock installs

* Add set -e and pipefail to ci Dockerfiles to detect errors

* Upgrade pip in Dockerfile.ubuntu.ci

* revert pipefail set -e change

* Replace build-docker-ci.sh script with Docker step for ci-base

* Add support for gfx950, add containers-ci-gfx.yml

* Add working-directory to matrix setup steps

* Try changing containers-ci-gfx.yml

* make more changes to containers-ci-gfx.yml

* Remove build-docker-ci.sh script from gfx step, fix typo in Dockerfile

* Remove gfx110X and gfx120X for now

* Update ci-gfx docker workflow to use ghcr.io

* Temporary change to test one image

* Enable push to test out ghcr package

* Add labels to debug oauth issue

* add pacakages permissions to step

* add rocprofiler-systems-ghcr.yml workflow

* Remove cache from Docker push action step

* Add prefix to tag

* Add back gfx94X and gfx950 support, add back no push on PR

* Remove gfx container creation from rocprofiler-systems-containers.yml

* Add a gfx950 image for now

* Revert change

.github/workflows/rocprofiler-systems-containers.yml

@@ -34,12 +34,12 @@ jobs:
     steps:
       - uses: actions/checkout@v4
         with:
-          sparse-checkout: projects/rocprofiler-systems
+          sparse-checkout: projects/rocprofiler-systems/docker
 
       - name: Output data for containers matrix
+        working-directory: projects/rocprofiler-systems/docker
         id: generate_matrix_ci
         run: |
-          pushd projects/rocprofiler-systems/docker
           MATRIX_CONTENT=$(cat containers-ci.yml | yq '.matrix' -I=0 -o=json)
           echo "matrix_data=$MATRIX_CONTENT" >> $GITHUB_OUTPUT
 
@@ -55,7 +55,7 @@ jobs:
     steps:
       - uses: actions/checkout@v4
         with:
-          sparse-checkout: projects/rocprofiler-systems
+          sparse-checkout: projects/rocprofiler-systems/docker
           submodules: recursive
 
       - name: Set up QEMU
@@ -70,31 +70,39 @@ jobs:
           username: ${{ secrets.ROCPROF_SYS_DOCKER_LOGIN }}
           password: ${{ secrets.ROCPROF_SYS_DOCKER_TOKEN }}
 
-      - name: Build CI Container (PR - No Push)
-        if: github.event_name == 'pull_request'
-        timeout-minutes: 45
-        uses: nick-fields/retry@v3
+      - name: Set up Docker variables
+        id: setup_vars
+        run: |
+          if [ ${{ matrix.distro }} = "opensuse" ]; then
+            DISTRO_IMAGE="opensuse/leap"
+          elif [ ${{ matrix.distro }} = "rhel" ]; then
+            DISTRO_IMAGE="rockylinux/rockylinux"
+          else
+            DISTRO_IMAGE=${{ matrix.distro }}
+          fi
+          echo "distro_image=${DISTRO_IMAGE}" >> $GITHUB_OUTPUT
+
+          if [ ${{ matrix.distro }} = "debian" ]; then
+            DOCKER_FILE=Dockerfile.ubuntu.ci
+          else
+            DOCKER_FILE=Dockerfile.${{ matrix.distro }}.ci
+          fi
+          echo "docker_file=${DOCKER_FILE}" >> $GITHUB_OUTPUT
+          
+      - name: Build CI Base Container (Does not Push on PR)
+        uses: docker/build-push-action@v6
         with:
-          retry_wait_seconds: 60
-          timeout_minutes: 45
-          max_attempts: 3
-          command: |
-            pushd projects/rocprofiler-systems/docker
-            ./build-docker-ci.sh --distro ${{ matrix.distro }} --versions ${{ matrix.version }} --user ${{ secrets.ROCPROF_SYS_DOCKER_LOGIN }} --jobs 2 --elfutils-version 0.188 --boost-version 1.79.0
-            popd
-      
-      - name: Build CI Container (Push)
-        if: github.event_name != 'pull_request'
-        timeout-minutes: 45
-        uses: nick-fields/retry@v3
-        with:
-          retry_wait_seconds: 60
-          timeout_minutes: 45
-          max_attempts: 3
-          command: |
-            pushd projects/rocprofiler-systems/docker
-            ./build-docker-ci.sh --distro ${{ matrix.distro }} --versions ${{ matrix.version }} --user ${{ secrets.ROCPROF_SYS_DOCKER_LOGIN }} --push --jobs 2 --elfutils-version 0.188 --boost-version 1.79.0
-            popd
+          file: projects/rocprofiler-systems/docker/${{ steps.setup_vars.outputs.docker_file }}
+          platforms: linux/amd64
+          push: ${{ github.event_name != 'pull_request' }}
+          build-args: |
+            DISTRO=${{ steps.setup_vars.outputs.distro_image }}
+            VERSION=${{ matrix.version }}
+            NJOBS=2
+            ELFUTILS_DOWNLOAD_VERSION=0.188
+            BOOST_DOWNLOAD_VERSION=1.79.0
+          tags: |
+            ${{ secrets.ROCPROF_SYS_DOCKER_LOGIN }}/rocprofiler-systems:ci-base-${{ matrix.distro }}-${{ matrix.version }}
 
   prepare_matrix_release:
     if: github.repository == 'ROCm/rocm-systems'
@@ -108,9 +116,9 @@ jobs:
           sparse-checkout: projects/rocprofiler-systems
 
       - name: Output data for containers matrix
+        working-directory: projects/rocprofiler-systems/docker
         id: generate_matrix_release
         run: |
-          pushd projects/rocprofiler-systems/docker
           MATRIX_CONTENT=$(cat containers.yml | yq '.matrix' -I=0 -o=json)
           echo "matrix_data=$MATRIX_CONTENT" >> $GITHUB_OUTPUT
           
@@ -150,7 +158,11 @@ jobs:
           max_attempts: 3
           command: |
             pushd projects/rocprofiler-systems/docker
-            ./build-docker.sh --distro ${{ matrix.os-distro }} --versions ${{ matrix.os-version }} --rocm-versions ${{ matrix.rocm-version }} --user ${{ secrets.ROCPROF_SYS_DOCKER_LOGIN }}
+            ./build-docker.sh \
+              --distro ${{ matrix.os-distro }} \
+              --versions ${{ matrix.os-version }} \
+              --rocm-versions ${{ matrix.rocm-version }} \
+              --user ${{ secrets.ROCPROF_SYS_DOCKER_LOGIN }}
             popd
       
       - name: Build Base Container (Push)
@@ -163,5 +175,10 @@ jobs:
           max_attempts: 3
           command: |
             pushd projects/rocprofiler-systems/docker
-            ./build-docker.sh --distro ${{ matrix.os-distro }} --versions ${{ matrix.os-version }} --rocm-versions ${{ matrix.rocm-version }} --user ${{ secrets.ROCPROF_SYS_DOCKER_LOGIN }} --push
+            ./build-docker.sh \
+              --distro ${{ matrix.os-distro }} \
+              --versions ${{ matrix.os-version }} \
+              --rocm-versions ${{ matrix.rocm-version }} \
+              --user ${{ secrets.ROCPROF_SYS_DOCKER_LOGIN }} \
+              --push
             popd

.github/workflows/rocprofiler-systems-ghcr.yml

@@ -0,0 +1,122 @@
+name: Publish GHCR Packages for rocprofiler-systems CI Images
+
+on:
+  workflow_dispatch:
+  schedule:
+    - cron: 0 5 * * *
+  push:
+    branches:
+      - develop
+    paths:
+      - '.github/workflows/rocprofiler-systems-ghcr.yml'
+      - 'projects/rocprofiler-systems/docker/**'
+  pull_request:
+    paths:
+      - '.github/workflows/rocprofiler-systems-ghcr.yml'
+      - 'projects/rocprofiler-systems/docker/**'
+
+jobs:
+  prepare_matrix_ci:
+    if: github.repository == 'ROCm/rocm-systems'
+    runs-on: ubuntu-latest
+    outputs:
+      matrix_data: ${{ steps.generate_matrix_ci_base.outputs.matrix_data }}
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          sparse-checkout: projects/rocprofiler-systems/docker
+
+      - name: Output data for containers matrix
+        working-directory: projects/rocprofiler-systems/docker
+        id: generate_matrix_ci_base
+        run: |
+          MATRIX_CONTENT=$(cat containers-ci.yml | yq '.matrix' -I=0 -o=json)
+          echo "matrix_data=$MATRIX_CONTENT" >> $GITHUB_OUTPUT
+
+  rocprofiler-systems-ci-gfx:
+    needs: prepare_matrix_ci
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      packages: write
+
+    strategy:
+      fail-fast: false
+      matrix:
+        system: ${{ fromJSON(needs.prepare_matrix_ci.outputs.matrix_data) }}
+        gpu: [ 'gfx94X', 'gfx950' ]
+
+    steps:
+      - name: Checkout Repository
+        uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
+        with:
+          sparse-checkout: projects/rocprofiler-systems
+          submodules: recursive
+
+      - name: Login to GitHub Container Registry
+        uses: docker/login-action@184bdaa0721073962dff0199f1fb9940f07167d1 # v3.5.0
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Set up Docker variables
+        id: setup_vars_gfx
+        run: |
+          if [ ${{ matrix.system.distro }} = "opensuse" ]; then
+            DISTRO_IMAGE="opensuse/leap"
+          elif [ ${{ matrix.system.distro }} = "rhel" ]; then
+            DISTRO_IMAGE="rockylinux/rockylinux"
+          else
+            DISTRO_IMAGE=${{ matrix.system.distro }}
+          fi
+          echo "distro_image=${DISTRO_IMAGE}" >> $GITHUB_OUTPUT
+
+          if [ ${{ matrix.system.distro }} = "debian" ]; then
+            DOCKER_FILE=Dockerfile.ubuntu.ci
+          else
+            DOCKER_FILE=Dockerfile.${{ matrix.system.distro }}.ci
+          fi
+          echo "docker_file=${DOCKER_FILE}" >> $GITHUB_OUTPUT
+      
+      - name: Get the latest build of The Rock tarball
+        id: therock
+        run: |
+          sudo apt-get install -y python3-pip
+          python3 -m pip install -U pip
+          python3 -m pip install -U awscli
+          export PATH=~/.local/bin:$PATH
+          KEY=$(aws s3api list-objects-v2 \
+                --bucket therock-nightly-tarball \
+                --no-sign-request \
+                --output json \
+                --query "sort_by(Contents[?contains(Key, 'linux-${{ matrix.gpu }}')], &LastModified)[-1].Key")
+          KEY=${KEY//\"/}
+          test -n "$KEY" || { echo "No ${{ matrix.gpu }} tarball found"; exit 1; }
+          echo "tarball=${KEY}" >> $GITHUB_OUTPUT
+      
+      - name: Extract metadata (tags, labels) for Docker
+        id: meta
+        uses: docker/metadata-action@c1e51972afc2121e065aed6d45c65596fe445f3f # v5.7.9
+        with:
+          images: ghcr.io/ROCm/rocprofiler-${{ matrix.system.distro }}-${{ matrix.system.version }}
+
+      - name: Build CI GFX Container (Does not Push on PR)
+        uses: docker/build-push-action@263435318d21b8e681c14492fe198d362a7d2c83 # v6.18.0
+        with:
+          file: projects/rocprofiler-systems/docker/${{ steps.setup_vars_gfx.outputs.docker_file }}
+          platforms: linux/amd64
+          push: ${{ github.event_name != 'pull_request' }}
+          build-args: |
+            DISTRO=${{ steps.setup_vars_gfx.outputs.distro_image }}
+            VERSION=${{ matrix.system.version }}
+            TYPE=${{ matrix.gpu }}
+            GPU_TYPE=${{ matrix.gpu }}
+            GPU_TARBALL=${{ steps.therock.outputs.tarball }}
+            NJOBS=2
+            ELFUTILS_DOWNLOAD_VERSION=0.188
+            BOOST_DOWNLOAD_VERSION=1.79.0
+          tags: |
+            ghcr.io/rocm/rocprofiler-${{ matrix.system.distro }}-${{ matrix.system.version }}:systems-ci-${{ matrix.gpu }}
+          labels: ${{ steps.meta.outputs.labels }}

projects/rocprofiler-systems/docker/Dockerfile.opensuse.ci

@@ -28,8 +28,8 @@ RUN zypper --non-interactive update -y && \
     zypper --non-interactive dist-upgrade -y && \
     zypper --non-interactive install -y -t pattern devel_basis && \
     zypper --non-interactive install -y binutils-gold chrpath cmake curl dpkg-devel \
-        gcc-c++ git iproute2 libnuma-devel ninja openmpi3-devel papi-devel python3-pip \
-        rpm-build sqlite3-devel vim wget && \
+        gcc-c++ git iproute2 libnuma-devel ninja openmpi3-devel papi-devel python3-devel \
+        python3-pip rpm-build sqlite3-devel vim wget && \
     zypper --non-interactive clean --all && \
     python3 -m pip install 'cmake==3.21' perfetto
 
@@ -47,5 +47,14 @@ RUN wget https://github.com/conda-forge/miniforge/releases/latest/download/Minif
     shopt -s dotglob extglob && \
     rm -rf *
 
+# The Rock Tarball
+ARG GPU_TYPE=""
+ARG GPU_TARBALL=""
+RUN if [ -n "$GPU_TYPE" ] && [ -n "$GPU_TARBALL" ]; then \
+        python3 -m pip install -U awscli; \
+        aws s3 cp "s3://therock-nightly-tarball/${GPU_TARBALL}" rocm-${GPU_TYPE}.tar.gz --no-sign-request; \
+        mv rocm-${GPU_TYPE}.tar.gz /opt/rocm-${GPU_TYPE}.tar.gz; \
+    fi
+
 WORKDIR /home
 SHELL [ "/bin/bash", "--login", "-c" ]

projects/rocprofiler-systems/docker/Dockerfile.rhel.ci

@@ -20,7 +20,7 @@ ARG NJOBS="8"
 RUN yum groupinstall -y "Development Tools" && \
     yum install -y epel-release && crb enable && \
     yum install -y --allowerasing chrpath cmake curl dpkg-devel iproute ninja-build numactl-devel \
-        openmpi-devel papi-devel python3-pip sqlite-devel texinfo wget which vim zlib-devel && \
+        openmpi-devel papi-devel python3-devel python3-pip sqlite-devel texinfo wget which vim zlib-devel && \
     yum clean all && \
     python3 -m pip install 'cmake==3.21' perfetto
 
@@ -38,5 +38,14 @@ RUN wget https://github.com/conda-forge/miniforge/releases/latest/download/Minif
     shopt -s dotglob extglob && \
     rm -rf *
 
+# The Rock Tarball
+ARG GPU_TYPE=""
+ARG GPU_TARBALL=""
+RUN if [ -n "$GPU_TYPE" ] && [ -n "$GPU_TARBALL" ]; then \
+        python3 -m pip install -U awscli; \
+        aws s3 cp "s3://therock-nightly-tarball/${GPU_TARBALL}" rocm-${GPU_TYPE}.tar.gz --no-sign-request; \
+        mv rocm-${GPU_TYPE}.tar.gz /opt/rocm-${GPU_TYPE}.tar.gz; \
+    fi
+
 WORKDIR /home
 SHELL [ "/bin/bash", "--login", "-c" ]

projects/rocprofiler-systems/docker/Dockerfile.ubuntu.ci

@@ -1,4 +1,3 @@
-
 ARG DISTRO
 ARG VERSION
 FROM ${DISTRO}:${VERSION}
@@ -17,7 +16,6 @@ ARG ELFUTILS_DOWNLOAD_VERSION="0.188"
 ARG BOOST_DOWNLOAD_VERSION="1.79.0"
 ARG NJOBS="8"
 
-
 ENV PATH="/usr/local/bin:${PATH}"
 ENV LIBRARY_PATH="/usr/local/lib:/usr/local/lib64:${LIBRARY_PATH}"
 ENV LD_LIBRARY_PATH="/usr/local/lib:/usr/local/lib64:${LD_LIBRARY_PATH}"
@@ -53,6 +51,15 @@ RUN wget https://github.com/conda-forge/miniforge/releases/latest/download/Minif
     shopt -s dotglob extglob && \
     rm -rf *
 
+# The Rock Tarball
+ARG GPU_TYPE=""
+ARG GPU_TARBALL=""
+RUN if [ -n "$GPU_TYPE" ] && [ -n "$GPU_TARBALL" ]; then \
+        pip install --upgrade pip; \
+        pip install awscli --break-system-packages; \
+        aws s3 cp "s3://therock-nightly-tarball/${GPU_TARBALL}" rocm-${GPU_TYPE}.tar.gz --no-sign-request; \
+        mv rocm-${GPU_TYPE}.tar.gz /opt/rocm-${GPU_TYPE}.tar.gz; \
+    fi
 
 ENV LC_ALL=C.UTF-8
 WORKDIR /home

projects/rocprofiler-systems/docker/build-docker-ci.sh

@@ -3,6 +3,7 @@
 set -e
 
 : ${USER:=$(whoami)}
+: ${TYPE:="base"}
 : ${DISTRO:=ubuntu}
 : ${VERSIONS:=24.04}
 : ${NJOBS=$(nproc)}
@@ -11,6 +12,8 @@ set -e
 : ${PYTHON_VERSIONS:="6 7 8 9 10 11 12 13"}
 : ${PUSH:=0}
 : ${PULL:=--pull}
+: ${GPU_TYPE:=""}
+: ${GPU_TARBALL:=""}
 
 verbose-run()
 {
@@ -45,6 +48,7 @@ usage()
     print_default_option elfutils-version "[0.183..0.188]" "ElfUtils version" "${ELFUTILS_VERSION}"
     print_default_option boost-version "[1.67.0..1.79.0]" "Boost version" "${BOOST_VERSION}"
     print_default_option user "[USERNAME]" "DockerHub username" "${USER}"
+    print_default_option type "[base|gfxXXX]" "Type of image to create" "${TYPE}"
 }
 
 send-error()
@@ -69,6 +73,11 @@ do
             usage
             exit 0
             ;;
+        "--type")
+            shift
+            TYPE=${1}
+            reset-last
+            ;;
         "--distro")
             shift
             DISTRO=${1}
@@ -99,6 +108,16 @@ do
             BOOST_VERSION=${1}
             reset-last
             ;;
+        "--gpu-type")
+            shift
+            GPU_TYPE=${1}
+            reset-last
+            ;;
+        "--gpu-tarball")
+            shift
+            GPU_TARBALL=${1}
+            reset-last
+            ;;
         --user|-u)
             shift
             USER=${1}
@@ -156,19 +175,21 @@ do
     verbose-run docker build . \
         ${PULL} \
         -f ${DOCKER_FILE} \
-        --tag ${USER}/rocprofiler-systems:ci-base-${DISTRO}-${VERSION} \
+        --tag ${USER}/rocprofiler-systems:ci-${TYPE}-${DISTRO}-${VERSION} \
         --build-arg DISTRO=${DISTRO_IMAGE} \
         --build-arg VERSION=${VERSION} \
         --build-arg NJOBS=${NJOBS} \
         --build-arg PYTHON_VERSIONS=\"${PYTHON_VERSIONS}\" \
         --build-arg ELFUTILS_DOWNLOAD_VERSION=${ELFUTILS_VERSION} \
-        --build-arg BOOST_DOWNLOAD_VERSION=${BOOST_VERSION}
+        --build-arg BOOST_DOWNLOAD_VERSION=${BOOST_VERSION} \
+        --build-arg GPU_TYPE=${GPU_TYPE} \
+        --build-arg GPU_TARBALL=${GPU_TARBALL} 
 done
 
 if [ "${PUSH}" -gt 0 ]; then
     for VERSION in ${VERSIONS}
     do
-        verbose-run docker push ${USER}/rocprofiler-systems:ci-base-${DISTRO}-${VERSION}
+        verbose-run docker push ${USER}/rocprofiler-systems:ci-${TYPE}-${DISTRO}-${VERSION}
     done
 fi