test amazon linux 2023 support

.common-ci.yml

@@ -81,6 +81,12 @@ trigger-pipeline:
     matrix:
       - DRIVER_VERSION: [535.183.06, 550.90.12, 560.35.03]
 
+# Define the driver versions for jobs that can be run in parallel for amzn2023
+.driver-versions-amzn2023:
+  parallel:
+    matrix:
+      - DRIVER_VERSION: [550.90.12, 560.35.03]
+
 # Define the matrix of precompiled jobs that can be run in parallel for ubuntu22.04
 .driver-versions-precompiled-ubuntu22.04:
   parallel:
@@ -105,6 +111,10 @@ trigger-pipeline:
     DIST: ubuntu22.04
     CVE_UPDATES: "openssl"
 
+.dist-amzn2023:
+  variables:
+    DIST: amzn2023
+
 .dist-rhel8:
   variables:
     DIST: rhel8
@@ -174,6 +184,14 @@ trigger-pipeline:
   rules:
     - if: $CI_PIPELINE_SOURCE != "schedule"
 
+.release-amzn2023:
+  # Perform for each DRIVER_VERSION
+  extends:
+    - .release-generic
+    - .driver-versions-amzn2023
+  rules:
+    - if: $CI_PIPELINE_SOURCE != "schedule"
+
 .release-rhel9:
   # Perform for each DRIVER_VERSION
   extends:
@@ -211,6 +229,15 @@ trigger-pipeline:
     OUT_REGISTRY: "${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/ci.yaml

@@ -21,6 +21,16 @@ on:
       - completed
     branches:
       - main
+  pull_request:
+    types:
+      - opened
+      - synchronize
+    branches:
+      - testshivaku
+  push:
+    branches:
+      - testshivaku
+
 
 jobs:
   e2e-tests-nvidiadriver:
@@ -36,6 +46,8 @@ jobs:
         AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
         AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
         AWS_SSH_KEY: ${{ secrets.AWS_SSH_KEY }}
+        AWS_SESSION_TOKEN: ${{ secrets.AWS_SESSION_TOKEN }}
+        AWS_DEFAULT_REGION: "us-west-1"
       with:
         aws_access_key_id: ${{ secrets.AWS_ACCESS_KEY_ID }}
         aws_secret_access_key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}

.github/workflows/image.yaml

@@ -40,6 +40,7 @@ jobs:
         dist:
           - ubuntu20.04
           - ubuntu22.04
+          - amzn2023
           - rhel8
           - rhel9
         ispr:
@@ -53,6 +54,10 @@ jobs:
           - ispr: true
             dist: ubuntu20.04
             driver: 550.90.12
+          - dist: amzn2023
+            driver: 535.183.06
+          - dist: amzn2023
+            driver: 550.90.12
       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-amzn2023
+    - .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

@@ -121,6 +121,12 @@ image-rhel9:
     - .dist-rhel9
     - .driver-versions-rhel9
 
+image-amzn2023:
+  extends:
+    - .image-pull
+    - .dist-amzn2023
+    - .driver-versions-amzn2023
+
 # The .scan step forms the base of the image scan operation performed before releasing
 # images.
 .scan-generic:
@@ -190,6 +196,18 @@ image-rhel9:
     - if: $CI_PIPELINE_SOURCE == "merge_request_event"
     - !reference [.pipeline-trigger-rules, rules]
 
+.scan-amzn2023:
+  # Repeat for each DRIVER_VERSION
+  extends:
+    - .driver-versions-amzn2023
+    - .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
@@ -235,6 +253,14 @@ scan-ubuntu22.04-arm64:
   needs:
     - image-ubuntu22.04
 
+scan-amzn2023-amd64:
+  extends:
+    - .scan-amzn2023
+    - .dist-amzn2023
+    - .platform-amd64
+  needs:
+    - image-amzn2023
+
 scan-precompiled-ubuntu22.04-amd64:
   variables:
     PLATFORM: linux/amd64
@@ -302,6 +328,12 @@ release:ngc-ubuntu22.04:
     - .dist-ubuntu22.04
     - .driver-versions-ubuntu22.04
 
+release:ngc-amzn2023:
+  extends:
+    - .release:ngc
+    - .dist-amzn2023
+    - .driver-versions-amzn2023
+
 release:ngc-precompiled-ubuntu22.04:
   variables:
     DIST: signed_ubuntu22.04
@@ -478,6 +510,9 @@ sign:ngc-ubuntu-rhel-rhcos:
     - SIGN_JOB_NAME: ["ubuntu"]
       VERSION: ["20.04"]
       DRIVER_VERSION: ["535.183.06", "550.90.12", "560.35.03"]
+    - SIGN_JOB_NAME: ["amzn"]
+      VERSION: ["2023"]
+      DRIVER_VERSION: ["560.35.03"]
     - SIGN_JOB_NAME: ["rhel"]
       VERSION: ["8.8", "8.10"]
       DRIVER_VERSION: ["535.183.06", "550.90.12", "560.35.03"]

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 := ubuntu18.04 ubuntu20.04 ubuntu22.04 amzn2023 signed_ubuntu20.04 signed_ubuntu22.04 rhel8 rhel9 flatcar fedora36 sles15.3 precompiled_rhcos
 PUSH_TARGETS := $(patsubst %, push-%, $(DISTRIBUTIONS))
 BASE_FROM := jammy focal
 PUSH_TARGETS := $(patsubst %, push-%, $(DISTRIBUTIONS))
@@ -92,6 +92,10 @@ pull-signed_ubuntu22.04%: DIST = ubuntu22.04
 pull-signed_ubuntu22.04%: DRIVER_TAG = $(DRIVER_BRANCH)
 pull-signed_ubuntu22.04%: IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
 
+pull-signed_amzn2023%: DIST = amzn2023
+pull-signed_amzn2023%: DRIVER_TAG = $(DRIVER_BRANCH)
+pull-signed_amzn2023%: IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
+
 PLATFORM ?= linux/amd64
 $(DRIVER_PULL_TARGETS): pull-%:
 	$(DOCKER) pull "--platform=$(PLATFORM)" "$(IMAGE)"
@@ -109,6 +113,10 @@ archive-signed_ubuntu22.04%: DIST = ubuntu22.04
 archive-signed_ubuntu22.04%: DRIVER_TAG = $(DRIVER_BRANCH)
 archive-signed_ubuntu22.04%: IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
 
+archive-signed_amzn2023%: DIST = amzn2023
+archive-signed_amzn2023%: DRIVER_TAG = $(DRIVER_BRANCH)
+archive-signed_amzn2023%: IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
+
 $(DRIVER_ARCHIVE_TARGETS): archive-%:
 	$(DOCKER) save "$(IMAGE)" -o "archive.tar"
 
@@ -130,6 +138,11 @@ push-signed_ubuntu22.04%: DRIVER_TAG = $(DRIVER_BRANCH)
 push-signed_ubuntu22.04%: IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
 push-signed_ubuntu22.04%: OUT_IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
 
+push-signed_amzn2023%: DIST = amzn2023
+push-signed_amzn2023%: DRIVER_TAG = $(DRIVER_BRANCH)
+push-signed_amzn2023%: IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
+push-signed_amzn2023%: OUT_IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
+
 # $(DRIVER_BUILD_TARGETS) is in the form of build-$(DIST)-$(DRIVER_VERSION)
 # Parse the target to set the required variables.
 build-%: DIST = $(word 2,$(subst -, ,$@))
@@ -175,6 +188,13 @@ build-signed_ubuntu22.04%: DRIVER_TAG = $(DRIVER_BRANCH)
 build-signed_ubuntu22.04%: IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
 build-signed_ubuntu22.04%: DOCKER_BUILD_ARGS =  --build-arg KERNEL_VERSION="$(KERNEL_VERSION)"
 
+# amzn2023 Precompiled Driver
+build-signed_amzn2023%: DIST = amzn2023
+build-signed_amzn2023%: DRIVER_TAG = $(DRIVER_BRANCH)
+build-signed_amzn2023%: IMAGE_TAG = $(DRIVER_BRANCH)-$(KERNEL_VERSION)-$(DIST)
+build-signed_amzn2023%: DOCKER_BUILD_ARGS =  --build-arg KERNEL_VERSION="$(KERNEL_VERSION)"
+# build-signed_amzn2023%: DOCKER_BUILD_PLATFORM_OPTIONS = --platform=linux/amd64
+
 # base is an image used to poll Canonical for the latest kernel version
 build-base-%: DOCKERFILE = $(CURDIR)/base/Dockerfile
 build-base-%: TARGET = $(word 3,$(subst -, ,$@))

amzn2023/Dockerfile

@@ -0,0 +1,106 @@
+ARG CUDA_VERSION=latest
+
+FROM nvcr.io/nvidia/cuda:${CUDA_VERSION}-base-amzn2023 AS build
+FROM amazonlinux:2
+
+ARG TARGETARCH
+ARG CUDA_VERSION
+
+SHELL ["/bin/bash", "-c"]
+
+# Remove cuda repository to avoid GPG errors
+RUN rm -f /etc/yum.repos.d/cuda*
+
+RUN yum update -y && yum install -y yum-utils && \
+    yum-config-manager --setopt=skip_missing_names_on_install=False && \
+    yum install -y \
+    gcc \
+    gcc-c++ \
+    make \
+    ca-certificates \
+    git -y && \
+    yum clean all
+
+ENV GOLANG_VERSION=1.23.1
+
+# download appropriate binary based on the target architecture for multi-arch builds
+RUN OS_ARCH=$(echo ${TARGETARCH} | sed 's/x86_64/amd64/' ) && \
+    curl -s https://storage.googleapis.com/golang/go${GOLANG_VERSION}.linux-${OS_ARCH}.tar.gz \
+    | tar -C /usr/local -xz
+
+ENV PATH /usr/local/go/bin:$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
+COPY --from=build /work/vgpu-util /usr/local/bin
+
+FROM nvcr.io/nvidia/cuda:${CUDA_VERSION}-base-amzn2023
+
+SHELL ["/bin/bash", "-c"]
+
+ARG BASE_URL=https://us.download.nvidia.com/tesla
+ARG TARGETARCH
+ENV TARGETARCH=$TARGETARCH
+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=550
+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 /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
+
+ADD drivers drivers/
+
+# Fetch the installer automatically for passthrough/baremetal types
+RUN if [ "$DRIVER_TYPE" != "vgpu" ]; then \
+    cd drivers && \
+    /tmp/install.sh download_installer; fi
+
+RUN if [ "$DRIVER_TYPE" != "vgpu" ] && [ "$TARGETARCH" != "arm64" ]; then \
+    yum update -y && \
+    yum install -y \
+    nvidia-fabric-manager-${DRIVER_VERSION}-1 \
+    libnvidia-nscq-${DRIVER_BRANCH}-${DRIVER_VERSION}-1; fi
+
+WORKDIR  /drivers
+
+ARG PUBLIC_KEY=empty
+COPY ${PUBLIC_KEY} kernel/pubkey.x509
+
+# Install / upgrade packages here that are required to resolve CVEs
+ARG CVE_UPDATES
+RUN if [ -n "${CVE_UPDATES}" ]; then \
+        yum update -y && yum install -y yum-utils && \
+        yum-config-manager --setopt=skip_missing_names_on_install=False && \
+        yum install -y \
+        ${CVE_UPDATES} && \
+        yum clean all; fi
+
+# Remove cuda repository to avoid GPG errors
+RUN rm -f /etc/yum.repos.d/cuda*
+
+# Add NGC DL license from the CUDA image
+RUN 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