[nvidia] On Ubuntu22, install the NVIDIA driver using the gcc version used to compile the kernel.

This is required because, NVIDIA driver must be compiled with the same gcc version used by the kernel.
If this is not the case, the NVIDIA driver installation would fail a compiler version check.
On newer version of Ubuntu22.04 (kernel 6.8+), the kernel is compiled with gcc-12, however gcc-11 is installed as default version by build-essentials, making this change necessary.

Signed-off-by: Giacomo Marciani <[email protected]>

Author: gmarciani

cookbooks/aws-parallelcluster-platform/resources/nvidia_driver/nvidia_driver_ubuntu20+.rb renamed to cookbooks/aws-parallelcluster-platform/resources/nvidia_driver/nvidia_driver_ubuntu20.rb

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and limitations under the License.
 
 provides :nvidia_driver, platform: 'ubuntu' do |node|
-  node['platform_version'].to_i >= 20
+  node['platform_version'].to_i == 20
 end
 
 use 'partial/_nvidia_driver_common.rb'

cookbooks/aws-parallelcluster-platform/resources/nvidia_driver/nvidia_driver_ubuntu22.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+# Copyright:: 2024 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License").
+# You may not use this file except in compliance with the License.
+# A copy of the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "LICENSE.txt" file accompanying this file.
+# This file is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, express or implied.
+# See the License for the specific language governing permissions and limitations under the License.
+
+provides :nvidia_driver, platform: 'ubuntu' do |node|
+  node['platform_version'].to_i == 22
+end
+
+use 'partial/_nvidia_driver_common.rb'
+
+def rebuild_initramfs?
+  true
+end
+
+def compiler_path
+  gcc_major_version = gcc_major_version_used_by_kernel
+
+  # If the gcc version used to compile the kernel cannot be detected,
+  # empty string is returned, meaning that the NVIDIA driver will be compiled
+  # using the system default compiler.
+  return "" if gcc_major_version.nil?
+
+  "CC=/usr/bin/gcc-#{gcc_major_version}"
+end

cookbooks/aws-parallelcluster-platform/resources/nvidia_driver/partial/_nvidia_driver_common.rb

@@ -73,14 +73,13 @@
   end
 
   # Install driver
-  # TODO remove --no-cc-version-check when we can update ubuntu 22 images
   bash 'nvidia.run advanced' do
     user 'root'
     group 'root'
     cwd '/tmp'
     code <<-NVIDIA
       set -e
-      #{compiler_path} ./nvidia.run --silent --dkms --disable-nouveau --no-cc-version-check -m=#{nvidia_kernel_module}
+      #{compiler_path} ./nvidia.run --silent --dkms --disable-nouveau -m=#{nvidia_kernel_module}
       rm -f /tmp/nvidia.run
     NVIDIA
     creates '/usr/bin/nvidia-smi'

cookbooks/aws-parallelcluster-shared/libraries/helpers.rb

@@ -106,3 +106,16 @@ def wait_sync_file(path)
     timeout 5
   end
 end
+
+def gcc_major_version_used_by_kernel
+  # Detects the gcc major version used to compile the kernel, e.g. 12.
+  # If the version cannot be detected, nil is returned.
+  begin
+    gcc_major_version = shell_out("awk '{print $8}' /proc/version | tr -d ',' | cut -d '.' -f 1").stdout.strip
+  rescue => error
+    Chef::Log.error("Cannot detect gcc version used to compile the kernel: #{error}")
+    return nil
+  end
+  Chef::Log.info("Detected version of gcc used to compile the kernel is: #{gcc_major_version}")
+  gcc_major_version
+end