Merge pull request #579 from ocaisa/override_lmod_gpu_check

Allow overriding the Lmod GPU driver check

Author: trz42

EESSI-install-software.sh

@@ -17,6 +17,11 @@ display_help() {
   echo "  --skip-cuda-install    -  disable installing a full CUDA SDK in the host_injections prefix (e.g. in CI)"
 }
 
+# Function to check if a command exists
+function command_exists() {
+    command -v "$1" >/dev/null 2>&1
+}
+
 function copy_build_log() {
     # copy specified build log to specified directory, with some context added
     build_log=${1}
@@ -238,10 +243,11 @@ else
     echo "Skipping installation of CUDA SDK in host_injections, since the --skip-cuda-install flag was passed OR no EasyBuild module was found"
 fi
 
-# Install drivers in host_injections
-# TODO: this is commented out for now, because the script assumes that nvidia-smi is available and works;
-#       if not, an error is produced, and the bot flags the whole build as failed (even when not installing GPU software)
-# ${EESSI_PREFIX}/scripts/gpu_support/nvidia/link_nvidia_host_libraries.sh
+# Install NVIDIA drivers in host_injections (if they exist)
+if command_exists "nvidia-smi"; then
+    echo "Command 'nvidia-smi' found. Installing NVIDIA drivers for use in prefix shell..."
+    ${EESSI_PREFIX}/scripts/gpu_support/nvidia/link_nvidia_host_libraries.sh
+fi
 
 # use PR patch file to determine in which easystack files stuff was added
 changed_easystacks=$(cat ${pr_diff} | grep '^+++' | cut -f2 -d' ' | sed 's@^[a-z]/@@g' | grep '^easystacks/.*yml$' | egrep -v 'known-issues|missing') 

bot/build.sh

@@ -223,7 +223,15 @@ mkdir -p ${TARBALL_TMP_BUILD_STEP_DIR}
 BUILD_STEP_ARGS+=("--save" "${TARBALL_TMP_BUILD_STEP_DIR}")
 BUILD_STEP_ARGS+=("--storage" "${STORAGE}")
 # add options required to handle NVIDIA support
-BUILD_STEP_ARGS+=("--nvidia" "all")
+if command_exists "nvidia-smi"; then
+    echo "Command 'nvidia-smi' found, using available GPU"
+    BUILD_STEP_ARGS+=("--nvidia" "all")
+else
+    echo "No 'nvidia-smi' found, no available GPU but allowing overriding this check"
+    BUILD_STEP_ARGS+=("--nvidia" "install")
+fi
+# Retain location for host injections so we don't reinstall CUDA
+# (Always need to run the driver installation as available driver may change)
 if [[ ! -z ${SHARED_FS_PATH} ]]; then
     BUILD_STEP_ARGS+=("--host-injections" "${SHARED_FS_PATH}/host-injections")
 fi

create_lmodsitepackage.py

@@ -131,8 +131,9 @@
     end
     -- when loading CUDA enabled modules check if the necessary driver libraries are accessible to the EESSI linker,
     -- otherwise, refuse to load the requested module and print error message
-    local haveGpu = mt:haveProperty(simpleName,"arch","gpu")
-    if haveGpu then
+    local checkGpu = mt:haveProperty(simpleName,"arch","gpu")
+    local overrideGpuCheck = os.getenv("EESSI_OVERRIDE_GPU_CHECK")
+    if checkGpu and (overrideGpuCheck == nil) then
         local arch = os.getenv("EESSI_CPU_FAMILY") or ""
         local cudaVersionFile = "/cvmfs/software.eessi.io/host_injections/nvidia/" .. arch .. "/latest/cuda_version.txt"
         local cudaDriverFile = "/cvmfs/software.eessi.io/host_injections/nvidia/" .. arch .. "/latest/libcuda.so"
@@ -141,7 +142,9 @@
         if not (cudaDriverExists or singularityCudaExists)  then
             local advice = "which relies on the CUDA runtime environment and driver libraries. "
             advice = advice .. "In order to be able to use the module, you will need "
-            advice = advice .. "to make sure EESSI can find the GPU driver libraries on your host system.\\n"
+            advice = advice .. "to make sure EESSI can find the GPU driver libraries on your host system. You can "
+            advice = advice .. "override this check by setting the environment variable EESSI_OVERRIDE_GPU_CHECK but "
+            advice = advice .. "the loaded application will not be able to execute on your system.\\n"
             advice = advice .. refer_to_docs
             LmodError("\\nYou requested to load ", simpleName, " ", advice)
         else

eessi_container.sh

@@ -464,10 +464,9 @@ if [[ ${SETUP_NVIDIA} -eq 1 ]]; then
         BIND_PATHS="${BIND_PATHS},${EESSI_VAR_LOG}:/var/log,${EESSI_USR_LOCAL_CUDA}:/usr/local/cuda"
         [[ ${VERBOSE} -eq 1 ]] && echo "BIND_PATHS=${BIND_PATHS}"
         if [[ "${NVIDIA_MODE}" == "install" ]] ; then
-            # We need to "trick" our LMOD_RC file to allow us to load CUDA modules even without a CUDA driver
-            # (this works because we build within a container and the LMOD_RC recognises that)
-            touch ${EESSI_TMPDIR}/libcuda.so
-            export SINGULARITY_CONTAINLIBS="${EESSI_TMPDIR}/libcuda.so"
+            # No GPU so we need to "trick" Lmod to allow us to load CUDA modules even without a CUDA driver
+            # (this variable means EESSI_OVERRIDE_GPU_CHECK=1 will be set inside the container)
+            export SINGULARITYENV_EESSI_OVERRIDE_GPU_CHECK=1
         fi
     fi
 fi

run_in_compat_layer_env.sh

@@ -26,6 +26,9 @@ fi
 if [ ! -z ${EESSI_VERSION_OVERRIDE} ]; then
     INPUT="export EESSI_VERSION_OVERRIDE=${EESSI_VERSION_OVERRIDE}; ${INPUT}"
 fi
+if [ ! -z ${EESSI_OVERRIDE_GPU_CHECK} ]; then
+    INPUT="export EESSI_OVERRIDE_GPU_CHECK=${EESSI_OVERRIDE_GPU_CHECK}; ${INPUT}"
+fi
 if [ ! -z ${http_proxy} ]; then
     INPUT="export http_proxy=${http_proxy}; ${INPUT}"
 fi

scripts/utils.sh

@@ -78,6 +78,11 @@ function create_directory_structure() {
   return $return_code
 }
 
+# Function to check if a command exists
+function command_exists() {
+    command -v "$1" >/dev/null 2>&1
+}
+
 function get_path_for_tool {
     tool_name=$1
     tool_envvar_name=$2