amazon linux 2023 support

Signed-off-by: shiva kumar <[email protected]>

.common-ci.yml

@@ -93,6 +93,10 @@ trigger-pipeline:
     DIST: ubuntu22.04
     CVE_UPDATES: "openssl"
 
+.dist-amzn2023:
+  variables:
+    DIST: amzn2023
+
 .dist-rhel8:
   variables:
     DIST: rhel8
@@ -162,6 +166,14 @@ trigger-pipeline:
   rules:
     - if: $CI_PIPELINE_SOURCE != "schedule"
 
+.release-amzn2023:
+  # Perform for each DRIVER_VERSION
+  extends:
+    - .release-generic
+    - .driver-versions
+  rules:
+    - if: $CI_PIPELINE_SOURCE != "schedule"
+
 .release-rhel9:
   # Perform for each DRIVER_VERSION
   extends:
@@ -199,6 +211,15 @@ trigger-pipeline:
     OUT_REGISTRY: "${CI_REGISTRY}"
     OUT_IMAGE_NAME: "${CI_REGISTRY_IMAGE}/staging/driver"
 
+.release:staging-amzn2023:
+  extends:
+    - .release-amzn2023
+  variables:
+    OUT_REGISTRY_USER: "${CI_REGISTRY_USER}"
+    OUT_REGISTRY_TOKEN: "${CI_REGISTRY_PASSWORD}"
+    OUT_REGISTRY: "${CI_REGISTRY}"
+    OUT_IMAGE_NAME: "${CI_REGISTRY_IMAGE}/staging/driver"
+
 .release:staging-rhel9:
   extends:
     - .release-rhel9

.github/workflows/image.yaml

@@ -40,6 +40,7 @@ jobs:
         dist:
           - ubuntu20.04
           - ubuntu22.04
+          - amzn2023
           - rhel8
         ispr:
           - ${{github.event_name == 'pull_request'}}
@@ -50,6 +51,21 @@ jobs:
           - ispr: true
             dist: ubuntu20.04
             driver: 550.127.05
+          - ispr: true
+            dist: ubuntu20.04
+            driver: 560.35.03
+          - ispr: true
+            dist: ubuntu22.04
+            driver: 560.35.03
+          - ispr: true
+            dist: amzn2023
+            driver: 535.216.01
+          - ispr: true
+            dist: amzn2023
+            driver: 550.127.05
+          - ispr: true
+            dist: amzn2023
+            driver: 560.35.03
       fail-fast: false
     steps:
       - uses: actions/checkout@v4

.gitlab-ci.yml

@@ -49,6 +49,15 @@ include:
   rules:
     - if: $CI_PIPELINE_SOURCE != "schedule"
 
+# Define the image build targets
+.image-build-amzn2023:
+  # Perform for each DRIVER_VERSION
+  extends:
+    - .driver-versions
+    - .image-build-generic
+  rules:
+    - if: $CI_PIPELINE_SOURCE != "schedule"
+
 # Define the image build targets
 .image-build-rhel9:
   # Perform for each DRIVER_VERSION
@@ -69,6 +78,11 @@ image-ubuntu22.04:
     - .image-build-ubuntu22.04
     - .dist-ubuntu22.04
 
+image-amzn2023:
+  extends:
+    - .image-build-amzn2023
+    - .dist-amzn2023
+
 image-rhel8:
   extends:
     - .image-build

.nvidia-ci.yml

@@ -184,6 +184,18 @@ image-rhel8:
     - if: $CI_PIPELINE_SOURCE == "merge_request_event"
     - !reference [.pipeline-trigger-rules, rules]
 
+.scan-amzn2023:
+  # Repeat for each DRIVER_VERSION
+  extends:
+    - .driver-versions
+    - .scan-generic
+  rules:
+    - !reference [.scan-rules-common, rules]
+    - if: $CI_PIPELINE_SOURCE == "schedule"
+      when: never
+    - if: $CI_PIPELINE_SOURCE == "merge_request_event"
+    - !reference [.pipeline-trigger-rules, rules]
+
 .scan-precompiled-ubuntu22.04:
   variables:
     DIST: signed_ubuntu22.04
@@ -278,6 +290,12 @@ release:ngc-ubuntu22.04:
     - .dist-ubuntu22.04
     - .driver-versions
 
+release:ngc-amzn2023:
+  extends:
+    - .release:ngc
+    - .dist-amzn2023
+    - .driver-versions
+
 release:ngc-precompiled-ubuntu22.04:
   variables:
     DIST: signed_ubuntu22.04
@@ -439,3 +457,6 @@ sign:ngc-ubuntu-rhel-rhcos:
     - SIGN_JOB_NAME: ["rhcos"]
       VERSION: ["4.12","4.13","4.14","4.15", "4.16", "4.17"]
       DRIVER_VERSION: ["535.216.01", "550.127.05", "565.57.01"]
+    - SIGN_JOB_NAME: ["amzn"]
+      VERSION: ["2023"]
+      DRIVER_VERSION: ["565.57.01"]

Makefile

@@ -54,7 +54,7 @@ OUT_IMAGE_TAG = $(OUT_IMAGE_VERSION)-$(OUT_DIST)
 OUT_IMAGE = $(OUT_IMAGE_NAME):$(OUT_IMAGE_TAG)
 
 ##### Public rules #####
-DISTRIBUTIONS := ubuntu18.04 ubuntu20.04 ubuntu22.04 signed_ubuntu20.04 signed_ubuntu22.04 rhel8 rhel9 flatcar fedora36 sles15.3 precompiled_rhcos
+DISTRIBUTIONS := amzn2023 flatcar fedora36 precompiled_rhcos rhel8 rhel9 signed_ubuntu20.04 signed_ubuntu22.04 sles15.3 ubuntu18.04 ubuntu20.04 ubuntu22.04
 PUSH_TARGETS := $(patsubst %, push-%, $(DISTRIBUTIONS))
 BASE_FROM := jammy focal
 PUSH_TARGETS := $(patsubst %, push-%, $(DISTRIBUTIONS))

amzn2023/Dockerfile

@@ -0,0 +1,111 @@
+FROM nvcr.io/nvidia/cuda:12.6.2-base-amzn2023 AS build
+
+ARG TARGETARCH
+
+SHELL ["/bin/bash", "-c"]
+
+# Remove cuda repository to avoid GPG errors
+RUN rm -f /etc/yum.repos.d/cuda*
+
+RUN dnf update -y && dnf makecache && \
+    dnf install -y \
+    gcc \
+    gcc-c++ \
+    make \
+    ca-certificates \
+    git \
+    tar && \
+    dnf clean all && rm -rf /var/cache/yum/*
+
+ENV GOLANG_VERSION=1.23.2
+
+# download appropriate binary based on the target architecture for multi-arch builds
+RUN curl https://storage.googleapis.com/golang/go${GOLANG_VERSION}.linux-${TARGETARCH}.tar.gz \
+    | tar -C /usr/local -xz
+
+ENV PATH /usr/local/bin:$PATH
+ENV PATH /usr/local/go/bin:$PATH
+
+WORKDIR /work
+
+RUN git clone https://github.com/NVIDIA/gpu-driver-container driver && \
+    cd driver/vgpu/src && \
+    go build -o vgpu-util && \
+    mv vgpu-util /work
+
+FROM nvcr.io/nvidia/cuda:12.6.2-base-amzn2023
+
+SHELL ["/bin/bash", "-c"]
+
+ARG BASE_URL=https://us.download.nvidia.com/tesla
+ARG TARGETARCH
+ENV TARGETARCH=$TARGETARCH
+ENV DRIVER_ARCH=${TARGETARCH/amd64/x86_64}
+ARG DRIVER_VERSION
+ENV DRIVER_VERSION=$DRIVER_VERSION
+
+# Arg to indicate if driver type is either of passthrough(baremetal) or vgpu
+ARG DRIVER_TYPE=passthrough
+ENV DRIVER_TYPE=$DRIVER_TYPE
+ARG DRIVER_BRANCH=560
+ENV DRIVER_BRANCH=$DRIVER_BRANCH
+ARG VGPU_LICENSE_SERVER_TYPE=NLS
+ENV VGPU_LICENSE_SERVER_TYPE=$VGPU_LICENSE_SERVER_TYPE
+# Enable vGPU version compability check by default
+ARG DISABLE_VGPU_VERSION_CHECK=true
+ENV DISABLE_VGPU_VERSION_CHECK=$DISABLE_VGPU_VERSION_CHECK
+ENV NVIDIA_VISIBLE_DEVICES=void
+
+RUN echo "TARGETARCH=$TARGETARCH"
+
+ADD install.sh /tmp
+
+RUN NVIDIA_GPGKEY_SUM=d0664fbbdb8c32356d45de36c5984617217b2d0bef41b93ccecd326ba3b80c87 && \
+    curl -fsSL https://developer.download.nvidia.com/compute/cuda/repos/amzn2023/${DRIVER_ARCH}/D42D0685.pub | sed '/^Version/d' > /etc/pki/rpm-gpg/RPM-GPG-KEY-NVIDIA && \
+    echo "$NVIDIA_GPGKEY_SUM  /etc/pki/rpm-gpg/RPM-GPG-KEY-NVIDIA" | sha256sum -c --strict - && \
+    curl -fsSL -o /etc/yum.repos.d/cuda.repo https://developer.download.nvidia.com/compute/cuda/repos/amzn2023/${DRIVER_ARCH}/cuda-amzn2023.repo
+
+RUN dnf clean all && dnf makecache && dnf update -y && dnf install -y shadow-utils
+RUN /tmp/install.sh reposetup && /tmp/install.sh depinstall && \
+    curl -fsSL -o /usr/local/bin/donkey https://github.com/3XX0/donkey/releases/download/v1.1.0/donkey && \
+    chmod +x /usr/local/bin/donkey
+
+COPY nvidia-driver /usr/local/bin
+COPY --from=build /work/vgpu-util /usr/local/bin
+
+RUN curl -fsSL -o /usr/local/bin/extract-vmlinux https://raw.githubusercontent.com/torvalds/linux/master/scripts/extract-vmlinux && \
+    chmod +x /usr/local/bin/extract-vmlinux
+
+ADD drivers drivers/
+
+# Fetch the installer automatically for passthrough/baremetal types
+RUN if [ "$DRIVER_TYPE" != "vgpu" ]; then \
+    cd drivers && \
+    /tmp/install.sh download_installer; fi
+
+# Check for nvidia-fabric-manager or nvidia-fabricmanager availability and install
+RUN if [ "$DRIVER_TYPE" != "vgpu" ] && [ "$TARGETARCH" != "arm64" ]; then \
+    # Initialize the fabric manager package variable
+    FABRIC_PACKAGE=""; \
+    if dnf list nvidia-fabric-manager-${DRIVER_VERSION}-1 &>/dev/null; then \
+        FABRIC_PACKAGE="nvidia-fabric-manager-${DRIVER_VERSION}-1"; \
+    elif dnf list nvidia-fabricmanager-${DRIVER_BRANCH}-${DRIVER_VERSION}-1 &>/dev/null; then \
+        FABRIC_PACKAGE="nvidia-fabricmanager-${DRIVER_BRANCH}-${DRIVER_VERSION}-1"; \
+    else \
+        echo "Error: No suitable package found for fabric manager version ${DRIVER_VERSION}"; \
+        exit 1; \
+    fi; \
+    dnf install -y "$FABRIC_PACKAGE" libnvidia-nscq-${DRIVER_BRANCH}-${DRIVER_VERSION}-1; fi
+
+WORKDIR  /drivers
+
+ARG PUBLIC_KEY=empty
+COPY ${PUBLIC_KEY} kernel/pubkey.x509
+
+# Remove cuda repository to avoid GPG errors
+# clean cache
+# Add NGC DL license from the CUDA image
+RUN rm -f /etc/yum.repos.d/cuda* && dnf clean all && \
+    mkdir /licenses && mv /NGC-DL-CONTAINER-LICENSE /licenses/NGC-DL-CONTAINER-LICENSE
+
+ENTRYPOINT ["nvidia-driver", "init"]

amzn2023/README.md

@@ -0,0 +1,3 @@
+# AmazonLinux2 [![build status](https://gitlab.com/nvidia/driver/badges/master/build.svg)](https://gitlab.com/nvidia/driver/commits/master)
+
+See https://github.com/NVIDIA/nvidia-docker/wiki/Driver-containers-(Beta)

amzn2023/drivers/README.md

@@ -0,0 +1 @@
+# Folder for downloading vGPU drivers and dependent metadata files

amzn2023/install.sh

@@ -0,0 +1,58 @@
+#!/usr/bin/env bash
+
+set -eu
+
+download_installer () {
+    DRIVER_ARCH=${TARGETARCH/amd64/x86_64} && curl -fSsl -O $BASE_URL/$DRIVER_VERSION/NVIDIA-Linux-$DRIVER_ARCH-$DRIVER_VERSION.run && \
+    chmod +x  NVIDIA-Linux-$DRIVER_ARCH-$DRIVER_VERSION.run;
+}
+
+dep_install () {
+    if [ "$TARGETARCH" = "amd64" ]; then
+        DRIVER_ARCH=${TARGETARCH/amd64/x86_64}
+        dnf update -y && dnf install -y \
+            gcc \
+            make \
+            glibc-devel \
+            ca-certificates \
+            kmod \
+            file \
+            elfutils-libelf-devel \
+            libglvnd-devel \
+            shadow-utils \
+            util-linux \
+            tar \
+            rpm-build \
+            dnf-utils \
+            pkgconfig && \
+            dnf clean all && \
+            rm -rf /var/cache/yum/*
+    fi
+}
+
+repo_setup () {
+    if [ "$TARGETARCH" = "amd64" ]; then
+      echo "[cuda-amzn2023-x86_64]
+name=cuda-amzn2023-x86_64
+baseurl=https://developer.download.nvidia.com/compute/cuda/repos/amzn2023/$DRIVER_ARCH
+enabled=1
+gpgcheck=1
+gpgkey=https://developer.download.nvidia.com/compute/cuda/repos/amzn2023/$DRIVER_ARCH/D42D0685.pub" > /etc/yum.repos.d/cuda.repo && \
+      usermod -o -u 0 -g 0 nobody
+    else
+        echo "TARGETARCH doesn't match a known arch target"
+        exit 1
+    fi
+}
+
+if [ "$1" = "reposetup" ]; then
+  repo_setup
+elif [ "$1" = "depinstall" ]; then
+  dep_install
+elif [ "$1" = "download_installer" ]; then
+  download_installer
+else
+  echo "Unknown function: $1"
+  exit 1
+fi
+