From 3db7a53ace90e3b5191486cc52e3b9c808eb7b02 Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Wed, 18 Oct 2023 14:25:39 +0200
Subject: [PATCH 01/18] Build CUDA under

---
 .../nvidia/install_cuda_host_injections.sh    | 92 +++++++++++++++++++
 scripts/utils.sh                              | 80 +++++++++++++++-
 2 files changed, 167 insertions(+), 5 deletions(-)
 create mode 100644 gpu_support/nvidia/install_cuda_host_injections.sh

diff --git a/gpu_support/nvidia/install_cuda_host_injections.sh b/gpu_support/nvidia/install_cuda_host_injections.sh
new file mode 100644
index 0000000000..b33efbfedd
--- /dev/null
+++ b/gpu_support/nvidia/install_cuda_host_injections.sh
@@ -0,0 +1,92 @@
+#!/usr/bin/env bash
+
+# Initialise our bash functions
+TOPDIR=$(dirname $(realpath $BASH_SOURCE))
+source "$TOPDIR"/../../scripts/utils.sh
+
+# Make sure EESSI is initialised
+check_eessi_initialised()
+
+if [[ $# -eq 0 ]] ; then
+    fatal_error "You must provide the CUDA version as an argument, e.g.:\n $0 11.3.1"
+fi
+install_cuda_version=$1
+if [[ -z "${EESSI_SOFTWARE_PATH}" ]]; then
+  fatal_error "This script cannot be used without having first defined EESSI_SOFTWARE_PATH"
+else
+  # As an installation location just use $EESSI_SOFTWARE_PATH but replacing `versions` with `host_injections`
+  # (CUDA is a binary installation so no need to worry too much about the EasyBuild setup)
+  cuda_install_parent=${EESSI_SOFTWARE_PATH/versions/host_injections}
+fi
+
+# Only install CUDA if specified version is not found.
+# This is only relevant for users, the shipped CUDA installation will
+# always be in versions instead of host_injections and have symlinks pointing
+# to host_injections for everything we're not allowed to ship
+# (existence of easybuild subdir implies a successful install)
+if [ -d "${cuda_install_parent}"/software/CUDA/"${install_cuda_version}"/easybuild ]; then
+  echo_green "CUDA software found! No need to install CUDA again, proceed with testing."
+else
+  # We need to be able write to the installation space so let's make sure we can
+  if ! create_directory_structure "${cuda_install_parent}"/software/CUDA ; then
+    fatal_error "No write permissions to directory ${cuda_install_parent}/software/CUDA"
+  fi
+
+  # we need a directory we can use for temporary storage
+  if [[ -z "${CUDA_TEMP_DIR}" ]]; then
+    tmpdir=$(mktemp -d)
+  else
+    tmpdir="${CUDA_TEMP_DIR}"/temp
+    if ! mkdir "$tmpdir" ; then
+      fatal_error "Could not create directory ${tmpdir}"
+    fi
+  fi
+
+  required_space_in_tmpdir=50000
+  # Let's see if we have sources and build locations defined if not, we use the temporary space
+  if [[ -z "${EASYBUILD_BUILDPATH}" ]]; then
+    export EASYBUILD_BUILDPATH=${tmpdir}/build
+    required_space_in_tmpdir=$((required_space_in_tmpdir + 5000000))
+  fi
+  if [[ -z "${EASYBUILD_SOURCEPATH}" ]]; then
+    export EASYBUILD_SOURCEPATH=${tmpdir}/sources
+    required_space_in_tmpdir=$((required_space_in_tmpdir + 5000000))
+  fi
+
+  # The install is pretty fat, you need lots of space for download/unpack/install (~3*5GB),
+  # need to do a space check before we proceed
+  avail_space=$(df --output=avail "${cuda_install_parent}"/ | tail -n 1 | awk '{print $1}')
+  if (( avail_space < 5000000 )); then
+    fatal_error "Need at least 5GB disk space to install CUDA under ${cuda_install_parent}, exiting now..."
+  fi
+  avail_space=$(df --output=avail "${tmpdir}"/ | tail -n 1 | awk '{print $1}')
+  if (( avail_space < required_space_in_tmpdir )); then
+    error="Need at least ${required_space_in_tmpdir} disk space under ${tmpdir}.\n"
+    error="${error}Set the environment variable CUDA_TEMP_DIR to a location with adequate space to pass this check."
+    error="${error}You can alternatively set EASYBUILD_BUILDPATH and/or EASYBUILD_SOURCEPATH "
+    error="${error}to reduce this requirement. Exiting now..."
+    fatal_error "${error}"
+  fi
+
+  if [[ -z "${EBROOTEASYBUILD}" ]]; then
+    echo_yellow "Loading EasyBuild module to do actual install"
+    module load EasyBuild
+  fi
+
+  # we need the --rebuild option and a (random) dir for the module since we are
+  # fixing the broken links of the EESSI-shipped installation
+  extra_args="--rebuild --installpath-modules=${tmpdir}"
+
+  # We don't want hooks used in this install, we need a vanilla CUDA installation
+  touch "$tmpdir"/none.py
+  # shellcheck disable=SC2086  # Intended splitting of extra_args
+  eb --prefix="$tmpdir" ${extra_args} --hooks="$tmpdir"/none.py --installpath="${cuda_install_parent}"/ CUDA-"${install_cuda_version}".eb
+  ret=$?
+  if [ $ret -ne 0 ]; then
+    fatal_error  "CUDA installation failed, please check EasyBuild logs..."
+  else
+    echo_green "CUDA installation at ${cuda_install_parent}/software/CUDA/${install_cuda_version} succeeded!"
+  fi
+  # clean up tmpdir
+  rm -rf "${tmpdir}"
+fi
diff --git a/scripts/utils.sh b/scripts/utils.sh
index d0da95e87f..07760f0dd0 100644
--- a/scripts/utils.sh
+++ b/scripts/utils.sh
@@ -14,7 +14,7 @@ ANY_ERROR_EXITCODE=1
 function fatal_error() {
     echo_red "ERROR: $1" >&2
     if [[ $# -gt 1 ]]; then
-      exit $2
+      exit "$2"
     else
       exit "${ANY_ERROR_EXITCODE}"
     fi
@@ -32,11 +32,81 @@ function check_exit_code {
     fi
 }
 
+function check_eessi_initialised() {
+  if [[ -z "${EESSI_SOFTWARE_PATH}" ]]; then
+    fatal_error "EESSI has not been initialised!"
+  else
+    return 0
+  fi
+}
+
+function float_greater_than() {
+  # Make sure we have two arguments
+  if [ $# -ne 2 ]; then
+    echo_red "greater_than_float requires two (float) numbers" >&2
+    return $ANY_ERROR_EXITCODE
+  fi
+  # Make sure the arguments are numbers
+  if [[ ! $1 =~ ^[+-]?[0-9]+\.?[0-9]*$ ]]; then
+    echo_yellow "Input to float_greater_than is not a float, ignoring"
+    return $ANY_ERROR_EXITCODE
+  fi
+  if [[ ! $2 =~ ^[+-]?[0-9]+\.?[0-9]*$ ]]; then
+    echo_yellow "Input to float_greater_than is not a float, ignoring"
+    return $ANY_ERROR_EXITCODE
+  fi
+  # Now do the actual evaluation
+  return_code=$ANY_ERROR_EXITCODE
+  result=$(echo "$1" "$2" | awk '{if ($1 > $2) print "true"}')
+  if [ "$result" = true ] ; then
+    return_code=0
+  fi
+  return $return_code
+}
+
+function check_in_prefix_shell() {
+  # Make sure EPREFIX is defined
+  if [[ -z "${EPREFIX}" ]]; then
+    fatal_error "This script cannot be used without having first defined EPREFIX"
+  fi
+  if [[ ! ${SHELL} = ${EPREFIX}/bin/bash ]]; then
+    fatal_error "Not running in Gentoo Prefix environment, run '${EPREFIX}/startprefix' first!"
+  fi
+}
+
+function create_directory_structure() {
+  # Ensure we are given a single path argument
+  if [ $# -ne 1 ]; then
+    echo_red "Function requires a single (relative or absolute) path argument" >&2
+    return $ANY_ERROR_EXITCODE
+  fi
+  dir_structure="$1"
+
+  # Attempt to create the directory structure
+  error_message=$(mkdir -p "$dir_structure" 2>&1)
+  return_code=$?
+  # If it fails be explicit about the error
+  if [ ${return_code} -ne 0 ]; then
+    real_dir=$(realpath -m "$dir_structure")
+    echo_red "Creating ${dir_structure} (real path ${real_dir}) failed with:\n ${error_message}" >&2
+  else
+    # If we're creating it, our use case is that we want to be able to write there
+    # (this is a check in case the directory already existed)
+    if [ ! -w "${dir_structure}" ]; then
+      real_dir=$(realpath -m "$dir_structure")
+      echo_red "You do not have (required) write permissions to ${dir_structure} (real path ${real_dir})!"
+      return_code=$ANY_ERROR_EXITCODE
+    fi
+  fi
+
+  return $return_code
+}
+
 function get_path_for_tool {
     tool_name=$1
     tool_envvar_name=$2
 
-    which_out=$(which ${tool_name} 2>&1)
+    which_out=$(which "${tool_name}" 2>&1)
     exit_code=$?
     if [[ ${exit_code} -eq 0 ]]; then
         echo "INFO: found tool ${tool_name} in PATH (${which_out})" >&2
@@ -68,7 +138,7 @@ function get_host_from_url {
     url=$1
     re="(http|https)://([^/:]+)"
     if [[ $url =~ $re ]]; then
-        echo ${BASH_REMATCH[2]}
+        echo "${BASH_REMATCH[2]}"
         return 0
     else
         echo ""
@@ -80,7 +150,7 @@ function get_port_from_url {
     url=$1
     re="(http|https)://[^:]+:([0-9]+)"
     if [[ $url =~ $re ]]; then
-        echo ${BASH_REMATCH[2]}
+        echo "${BASH_REMATCH[2]}"
         return 0
     else
         echo ""
@@ -90,7 +160,7 @@ function get_port_from_url {
 
 function get_ipv4_address {
     hname=$1
-    hipv4=$(grep ${hname} /etc/hosts | grep -v '^[[:space:]]*#' | cut -d ' ' -f 1)
+    hipv4=$(grep "${hname}" /etc/hosts | grep -v '^[[:space:]]*#' | cut -d ' ' -f 1)
     # TODO try other methods if the one above does not work --> tool that verifies
     #      what method can be used?
     echo "${hipv4}"

From 82893a136bad5c0a3125c91aaa2bf354f70c7450 Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Wed, 18 Oct 2023 14:32:22 +0200
Subject: [PATCH 02/18] Make script executable

---
 gpu_support/nvidia/install_cuda_host_injections.sh | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 mode change 100644 => 100755 gpu_support/nvidia/install_cuda_host_injections.sh

diff --git a/gpu_support/nvidia/install_cuda_host_injections.sh b/gpu_support/nvidia/install_cuda_host_injections.sh
old mode 100644
new mode 100755

From 3797953057ff83e1e82c1be0232ee9aa61e18b1f Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Wed, 18 Oct 2023 14:42:23 +0200
Subject: [PATCH 03/18] Fix calling function

---
 gpu_support/nvidia/install_cuda_host_injections.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gpu_support/nvidia/install_cuda_host_injections.sh b/gpu_support/nvidia/install_cuda_host_injections.sh
index b33efbfedd..d6278ff3d7 100755
--- a/gpu_support/nvidia/install_cuda_host_injections.sh
+++ b/gpu_support/nvidia/install_cuda_host_injections.sh
@@ -5,7 +5,7 @@ TOPDIR=$(dirname $(realpath $BASH_SOURCE))
 source "$TOPDIR"/../../scripts/utils.sh
 
 # Make sure EESSI is initialised
-check_eessi_initialised()
+check_eessi_initialised
 
 if [[ $# -eq 0 ]] ; then
     fatal_error "You must provide the CUDA version as an argument, e.g.:\n $0 11.3.1"

From 445048a248bf61ae82bb0456acf242fcb48428ed Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 19 Oct 2023 15:07:33 +0200
Subject: [PATCH 04/18] Make additional bind mounts for CUDA install to succeed
 in container

---
 build_container.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/build_container.sh b/build_container.sh
index 23a9e665c9..42f7afade9 100755
--- a/build_container.sh
+++ b/build_container.sh
@@ -24,12 +24,12 @@ echo "Using $EESSI_TMPDIR as parent for temporary directories..."
 
 # create temporary directories
 mkdir -p $EESSI_TMPDIR/{home,overlay-upper,overlay-work}
-mkdir -p $EESSI_TMPDIR/{var-lib-cvmfs,var-run-cvmfs}
+mkdir -p $EESSI_TMPDIR/{var-lib-cvmfs,var-run-cvmfs,var-log,opt-eessi,usr-local-cuda}
 # configure Singularity
 export SINGULARITY_CACHEDIR=$EESSI_TMPDIR/singularity_cache
 
 # take into account that $SINGULARITY_BIND may be defined already, to bind additional paths into the build container
-BIND_PATHS="$EESSI_TMPDIR/var-run-cvmfs:/var/run/cvmfs,$EESSI_TMPDIR/var-lib-cvmfs:/var/lib/cvmfs,$EESSI_TMPDIR"
+BIND_PATHS="$EESSI_TMPDIR/var-log:/var/log,$EESSI_TMPDIR/usr-local-cuda:/usr/local/cuda,$EESSI_TMPDIR/var-run-cvmfs:/var/run/cvmfs,$EESSI_TMPDIR/var-lib-cvmfs:/var/lib/cvmfs,$EESSI_TMPDIR/opt-eessi:/opt/eessi,$EESSI_TMPDIR"
 if [ -z $SINGULARITY_BIND ]; then
     export SINGULARITY_BIND="$BIND_PATHS"
 else

From 617c8ada42cf8ca3421e9a8d7f41453d7697309e Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Tue, 31 Oct 2023 16:12:45 +0100
Subject: [PATCH 05/18] Switch GPU support to eessi_container.sh

---
 build_container.sh |  4 +--
 eessi_container.sh | 64 +++++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 62 insertions(+), 6 deletions(-)

diff --git a/build_container.sh b/build_container.sh
index 42f7afade9..23a9e665c9 100755
--- a/build_container.sh
+++ b/build_container.sh
@@ -24,12 +24,12 @@ echo "Using $EESSI_TMPDIR as parent for temporary directories..."
 
 # create temporary directories
 mkdir -p $EESSI_TMPDIR/{home,overlay-upper,overlay-work}
-mkdir -p $EESSI_TMPDIR/{var-lib-cvmfs,var-run-cvmfs,var-log,opt-eessi,usr-local-cuda}
+mkdir -p $EESSI_TMPDIR/{var-lib-cvmfs,var-run-cvmfs}
 # configure Singularity
 export SINGULARITY_CACHEDIR=$EESSI_TMPDIR/singularity_cache
 
 # take into account that $SINGULARITY_BIND may be defined already, to bind additional paths into the build container
-BIND_PATHS="$EESSI_TMPDIR/var-log:/var/log,$EESSI_TMPDIR/usr-local-cuda:/usr/local/cuda,$EESSI_TMPDIR/var-run-cvmfs:/var/run/cvmfs,$EESSI_TMPDIR/var-lib-cvmfs:/var/lib/cvmfs,$EESSI_TMPDIR/opt-eessi:/opt/eessi,$EESSI_TMPDIR"
+BIND_PATHS="$EESSI_TMPDIR/var-run-cvmfs:/var/run/cvmfs,$EESSI_TMPDIR/var-lib-cvmfs:/var/lib/cvmfs,$EESSI_TMPDIR"
 if [ -z $SINGULARITY_BIND ]; then
     export SINGULARITY_BIND="$BIND_PATHS"
 else
diff --git a/eessi_container.sh b/eessi_container.sh
index 48c4653ba9..b58bddf309 100755
--- a/eessi_container.sh
+++ b/eessi_container.sh
@@ -46,6 +46,7 @@ SAVE_ERROR_EXITCODE=$((${ANY_ERROR_EXITCODE} << 8))
 HTTP_PROXY_ERROR_EXITCODE=$((${ANY_ERROR_EXITCODE} << 9))
 HTTPS_PROXY_ERROR_EXITCODE=$((${ANY_ERROR_EXITCODE} << 10))
 RUN_SCRIPT_MISSING_EXITCODE=$((${ANY_ERROR_EXITCODE} << 11))
+NVIDIA_MODE_UNKNOWN_EXITCODE=$((${ANY_ERROR_EXITCODE} << 12))
 
 # CernVM-FS settings
 CVMFS_VAR_LIB="var-lib-cvmfs"
@@ -72,12 +73,17 @@ display_help() {
   echo "  -a | --access {ro,rw}  - ro (read-only), rw (read & write) [default: ro]"
   echo "  -c | --container IMG   - image file or URL defining the container to use"
   echo "                           [default: docker://ghcr.io/eessi/build-node:debian11]"
-  echo "  -h | --help            - display this usage information [default: false]"
   echo "  -g | --storage DIR     - directory space on host machine (used for"
   echo "                           temporary data) [default: 1. TMPDIR, 2. /tmp]"
+  echo "  -h | --help            - display this usage information [default: false]"
+  echo "  -i | --host-injections - directory to link to for host_injections "
+  echo "                           [default: /..storage../opt-eessi]"
   echo "  -l | --list-repos      - list available repository identifiers [default: false]"
   echo "  -m | --mode MODE       - with MODE==shell (launch interactive shell) or"
   echo "                           MODE==run (run a script or command) [default: shell]"
+  echo "  -n | --nvidia MODE     - configure the container to work with NVIDIA GPUs,"
+  echo "                           MODE==install for a CUDA installation, MODE==run to"
+  echo "                           attach a GPU, MODE==all for both [default: false]"
   echo "  -r | --repository CFG  - configuration file or identifier defining the"
   echo "                           repository to use [default: EESSI-pilot via"
   echo "                           default container, see --container]"
@@ -111,6 +117,8 @@ VERBOSE=0
 STORAGE=
 LIST_REPOS=0
 MODE="shell"
+SETUP_NVIDIA=0
+ADDITIONAL_SINGULARITY_FLAGS=
 REPOSITORY="EESSI-pilot"
 RESUME=
 SAVE=
@@ -141,6 +149,10 @@ while [[ $# -gt 0 ]]; do
       display_help
       exit 0
       ;;
+    -i|--host-injections)
+      USER_HOST_INJECTIONS="$2"
+      shift 2
+      ;;
     -l|--list-repos)
       LIST_REPOS=1
       shift 1
@@ -149,6 +161,11 @@ while [[ $# -gt 0 ]]; do
       MODE="$2"
       shift 2
       ;;
+    -n|--nvidia)
+      SETUP_NVIDIA=1
+      NVIDIA_MODE="$2"
+      shift 2
+      ;;
     -r|--repository)
       REPOSITORY="$2"
       shift 2
@@ -224,6 +241,13 @@ if [[ "${MODE}" != "shell" && "${MODE}" != "run" ]]; then
     fatal_error "unknown execution mode '${MODE}'" "${MODE_UNKNOWN_EXITCODE}"
 fi
 
+# Also validate the NVIDIA GPU mode (if present)
+if [[ ${SETUP_NVIDIA} -eq 1 ]]; then
+    if [[ "${NVIDIA_MODE}" != "run" && "${NVIDIA_MODE}" != "install" && "${NVIDIA_MODE}" != "all" ]]; then
+        fatal_error "unknown NVIDIA mode '${NVIDIA_MODE}'" "${NVIDIA_MODE_UNKNOWN_EXITCODE}"
+    fi
+fi
+
 # TODO (arg -r|--repository) check if repository is known
 # REPOSITORY_ERROR_EXITCODE
 if [[ ! -z "${REPOSITORY}" && "${REPOSITORY}" != "EESSI-pilot" && ! -r ${EESSI_REPOS_CFG_FILE} ]]; then
@@ -294,6 +318,7 @@ else
   echo "Using ${EESSI_HOST_STORAGE} as tmp directory (to resume session add '--resume ${EESSI_HOST_STORAGE}')."
 fi
 
+
 # if ${RESUME} is a file (assume a tgz), unpack it into ${EESSI_HOST_STORAGE}
 if [[ ! -z ${RESUME} && -f ${RESUME} ]]; then
   tar xf ${RESUME} -C ${EESSI_HOST_STORAGE}
@@ -310,12 +335,25 @@ fi
 #      |-overlay-work
 #      |-home
 #      |-repos_cfg
+#      |-opt-eessi (unless otherwise specificed for host_injections)
 
 # tmp dir for EESSI
 EESSI_TMPDIR=${EESSI_HOST_STORAGE}
 mkdir -p ${EESSI_TMPDIR}
 [[ ${VERBOSE} -eq 1 ]] && echo "EESSI_TMPDIR=${EESSI_TMPDIR}"
 
+# Set host_injections directory and ensure it is a writable directory (if user provided)
+if [ -z ${USER_HOST_INJECTIONS+x} ]; then
+    # Not set, so use our default
+    HOST_INJECTIONS=${EESSI_TMPDIR}/opt-eessi
+    mkdir -p $HOST_INJECTIONS
+else
+    # Make sure the host_injections directory specified exists and is a folder
+    mkdir -p ${USER_HOST_INJECTIONS} || fatal_error "host_injections directory ${USER_HOST_INJECTIONS} is either not a directory or cannot be created"
+    HOST_INJECTIONS=${USER_HOST_INJECTIONS}
+fi
+[[ ${VERBOSE} -eq 1 ]] && echo "HOST_INJECTIONS=${HOST_INJECTIONS}"
+
 # configure Singularity: if SINGULARITY_CACHEDIR is already defined, use that
 #   a global SINGULARITY_CACHEDIR would ensure that we don't consume
 #   storage space again and again for the container & also speed-up
@@ -394,12 +432,30 @@ fi
 [[ ${VERBOSE} -eq 1 ]] && echo "SINGULARITY_HOME=${SINGULARITY_HOME}"
 
 # define paths to add to SINGULARITY_BIND (added later when all BIND mounts are defined)
-BIND_PATHS="${EESSI_CVMFS_VAR_LIB}:/var/lib/cvmfs,${EESSI_CVMFS_VAR_RUN}:/var/run/cvmfs"
+BIND_PATHS="${EESSI_CVMFS_VAR_LIB}:/var/lib/cvmfs,${EESSI_CVMFS_VAR_RUN}:/var/run/cvmfs,${HOST_INJECTIONS}:/opt/eessi"
 # provide a '/tmp' inside the container
 BIND_PATHS="${BIND_PATHS},${EESSI_TMPDIR}:${TMP_IN_CONTAINER}"
 
 [[ ${VERBOSE} -eq 1 ]] && echo "BIND_PATHS=${BIND_PATHS}"
 
+# Configure anything we need for NVIDIA GPUs and CUDA installation
+if [[ ${SETUP_NVIDIA} -eq 1 ]]; then
+    if [[ "${NVIDIA_MODE}" == "run" || "${NVIDIA_MODE}" == "all" ]]; then
+        # Give singularity the appropriate flag
+        ADDITIONAL_SINGULARITY_FLAGS="--nv ${ADDITIONAL_SINGULARITY_FLAGS}"
+        [[ ${VERBOSE} -eq 1 ]] && echo "ADDITIONAL_SINGULARITY_FLAGS=${ADDITIONAL_SINGULARITY_FLAGS}"
+    fi
+    if [[ "${NVIDIA_MODE}" == "install" || "${NVIDIA_MODE}" == "all" ]]; then
+        # Add additional bind mounts to allow CUDA to install within a container
+        EESSI_VAR_LOG=${EESSI_TMPDIR}/var-log
+        EESSI_USR_LOCAL_CUDA=${EESSI_TMPDIR}/usr-local-cuda
+        mkdir -p ${EESSI_VAR_LOG}
+        mkdir -p ${EESSI_USR_LOCAL_CUDA}
+        BIND_PATHS="${BIND_PATHS},${EESSI_VAR_LOG}:/var/log,${EESSI_USR_LOCAL_CUDA}:/usr/local/cuda"
+        [[ ${VERBOSE} -eq 1 ]] && echo "BIND_PATHS=${BIND_PATHS}"
+    fi
+fi
+
 # set up repository config (always create directory repos_cfg and populate it with info when
 # arg -r|--repository is used)
 mkdir -p ${EESSI_TMPDIR}/repos_cfg
@@ -558,8 +614,8 @@ if [ ! -z ${EESSI_SOFTWARE_SUBDIR_OVERRIDE} ]; then
 fi
 
 echo "Launching container with command (next line):"
-echo "singularity ${RUN_QUIET} ${MODE} ${EESSI_FUSE_MOUNTS[@]} ${CONTAINER} $@"
-singularity ${RUN_QUIET} ${MODE} "${EESSI_FUSE_MOUNTS[@]}" ${CONTAINER} "$@"
+echo "singularity ${RUN_QUIET} ${MODE} ${ADDITIONAL_SINGULARITY_FLAGS} ${EESSI_FUSE_MOUNTS[@]} ${CONTAINER} $@"
+singularity ${RUN_QUIET} ${MODE} ${ADDITIONAL_SINGULARITY_FLAGS} "${EESSI_FUSE_MOUNTS[@]}" ${CONTAINER} "$@"
 exit_code=$?
 
 # 6. save tmp if requested (arg -s|--save)

From 090351864cbc5b97f35498b278d16041a5b97b1e Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 11:16:55 +0100
Subject: [PATCH 06/18] Address review and improve UI and error reporting

---
 eessi_container.sh                            |   8 +-
 .../nvidia/install_cuda_host_injections.sh    | 132 ++++++++++++++++--
 scripts/utils.sh                              |  24 ----
 3 files changed, 125 insertions(+), 39 deletions(-)

diff --git a/eessi_container.sh b/eessi_container.sh
index b58bddf309..e31808d546 100755
--- a/eessi_container.sh
+++ b/eessi_container.sh
@@ -30,8 +30,8 @@
 # -. initial settings & exit codes
 TOPDIR=$(dirname $(realpath $0))
 
-source ${TOPDIR}/scripts/utils.sh
-source ${TOPDIR}/scripts/cfg_files.sh
+source "${TOPDIR}"/scripts/utils.sh
+source "${TOPDIR}"/scripts/cfg_files.sh
 
 # exit codes: bitwise shift codes to allow for combination of exit codes
 # ANY_ERROR_EXITCODE is sourced from ${TOPDIR}/scripts/utils.sh
@@ -447,6 +447,10 @@ if [[ ${SETUP_NVIDIA} -eq 1 ]]; then
     fi
     if [[ "${NVIDIA_MODE}" == "install" || "${NVIDIA_MODE}" == "all" ]]; then
         # Add additional bind mounts to allow CUDA to install within a container
+        # (Experience tells us that these are necessary, but we don't know _why_
+        # as the CUDA installer is a black box. The suspicion is that the CUDA
+        # installer gets confused by the permissions on these directories when
+        # inside a container)
         EESSI_VAR_LOG=${EESSI_TMPDIR}/var-log
         EESSI_USR_LOCAL_CUDA=${EESSI_TMPDIR}/usr-local-cuda
         mkdir -p ${EESSI_VAR_LOG}
diff --git a/gpu_support/nvidia/install_cuda_host_injections.sh b/gpu_support/nvidia/install_cuda_host_injections.sh
index d6278ff3d7..2c59a891f8 100755
--- a/gpu_support/nvidia/install_cuda_host_injections.sh
+++ b/gpu_support/nvidia/install_cuda_host_injections.sh
@@ -1,16 +1,94 @@
 #!/usr/bin/env bash
 
+# This script can be used to install CUDA under the `.../host_injections` directory.
+# This provides the parts of the CUDA installation that cannot be redistributed as
+# part of EESSI due to license limitations. While GPU-based software from EESSI will
+# _run_ without these, installation of additional CUDA software requires the CUDA
+# installation(s) under `host_injections` to be present.
+#
+# The `host_injections` directory is a variant symlink that by default points to
+# `/opt/eessi`, unless otherwise defined in the local CVMFS configuration (see
+# https://cvmfs.readthedocs.io/en/stable/cpt-repo.html#variant-symlinks). For the
+# installation to be successful, this directory needs to be writeable by the user
+# executing this script.
+
 # Initialise our bash functions
 TOPDIR=$(dirname $(realpath $BASH_SOURCE))
 source "$TOPDIR"/../../scripts/utils.sh
 
+# Function to display help message
+show_help() {
+    echo "Usage: $0 [OPTIONS]"
+    echo "Options:"
+    echo "  --help                           Display this help message"
+    echo "  -c, --cuda-version CUDA_VERSION  Specify a version o CUDA to install (must"
+    echo "                                   have a corresponding easyconfig in the"
+    echo "                                   EasyBuild release)"
+    echo "  -t, --temp-dir /path/to/tmpdir   Specify a location to use for temporary"
+    echo "                                   storage during the CUDA install"
+    echo "                                   (must have >10GB available)"
+}
+
+# Initialize variables
+install_cuda_version=""
+
+# Parse command-line options
+while [[ $# -gt 0 ]]; do
+    case "$1" in
+        --help)
+            show_help
+            exit 0
+            ;;
+        -c|--cuda-version)
+            if [ -n "$2" ]; then
+                install_cuda_version="$2"
+                shift 2
+            else
+                echo "Error: Argument required for $1"
+                show_help
+                exit 1
+            fi
+            ;;
+        -t|--temp-dir)
+            if [ -n "$2" ]; then
+                CUDA_TEMP_DIR="$2"
+                shift 2
+            else
+                echo "Error: Argument required for $1"
+                show_help
+                exit 1
+            fi
+            ;;
+        *)
+            show_help
+            fatal_error "Error: Unknown option: $1"
+            ;;
+    esac
+done
+
+# Make sure the CUDA version supplied is a semantic version
+is_semantic_version() {
+    local version=$1
+    local regex='^[0-9]+\.[0-9]+\.[0-9]+$'
+
+    if [[ $version =~ $regex ]]; then
+        return 0  # Return success (0) if it's a semantic version
+    else
+        return 1  # Return failure (1) if it's not a semantic version
+    fi
+}
+if ! is_semantic_version "$install_cuda_version"; then
+  show_help
+  error="\nYou must provide a semantic version for CUDA (e.g., 12.1.1) via the appropriate\n"
+  error="${error}command line option. This script is intended for use with EESSI so the 'correct'\n"
+  error="${error}version to provide is probably the one that is available under\n"
+  error="${error}$EESSI_SOFTWARE_PATH/software/CUDA\n"
+  fatal_error "${error}"
+fi
+
 # Make sure EESSI is initialised
 check_eessi_initialised
 
-if [[ $# -eq 0 ]] ; then
-    fatal_error "You must provide the CUDA version as an argument, e.g.:\n $0 11.3.1"
-fi
-install_cuda_version=$1
 if [[ -z "${EESSI_SOFTWARE_PATH}" ]]; then
   fatal_error "This script cannot be used without having first defined EESSI_SOFTWARE_PATH"
 else
@@ -20,12 +98,9 @@ else
 fi
 
 # Only install CUDA if specified version is not found.
-# This is only relevant for users, the shipped CUDA installation will
-# always be in versions instead of host_injections and have symlinks pointing
-# to host_injections for everything we're not allowed to ship
 # (existence of easybuild subdir implies a successful install)
 if [ -d "${cuda_install_parent}"/software/CUDA/"${install_cuda_version}"/easybuild ]; then
-  echo_green "CUDA software found! No need to install CUDA again, proceed with testing."
+  echo_green "CUDA software found! No need to install CUDA again."
 else
   # We need to be able write to the installation space so let's make sure we can
   if ! create_directory_structure "${cuda_install_parent}"/software/CUDA ; then
@@ -68,19 +143,50 @@ else
     fatal_error "${error}"
   fi
 
-  if [[ -z "${EBROOTEASYBUILD}" ]]; then
-    echo_yellow "Loading EasyBuild module to do actual install"
+  if ! command -v "eb" &>/dev/null; then
+    echo_yellow "Attempting to load an EasyBuild module to do actual install"
     module load EasyBuild
+    # There are some scenarios where this may fail
+    if [ $? -ne 0 ]; then
+      error="'eb' command not found in your environment and\n"
+      error="${error}  module load EasyBuild\n"
+      error="${error}failed for some reason.\n"
+      error="${error}Please re-run this script with the 'eb' command available."
+      fatal_error "${error}"
+    fi
+  fi
+
+  cuda_easyconfig="CUDA-${install_cuda_version}.eb"
+
+  # Check the easyconfig file is available in the release
+  # (eb search always returns 0, so we need a grep to ensure a usable exit code)
+  eb --search ^${cuda_easyconfig}|grep CUDA > /dev/null 2>&1
+  # Check the exit code
+  if [ $? -ne 0 ]; then
+    eb_version=$(eb --version)
+    available_cuda_easyconfigs=$(eb --search ^CUDA-*.eb|grep CUDA)
+
+    error="The easyconfig ${cuda_easyconfig} was not found in EasyBuild version:\n"
+    error="${error}  ${eb_version}\n"
+    error="${error}You either need to give a different version of CUDA to install _or_ \n"
+    error="${error}use a different version of EasyBuild for the installation.\n"
+    error="${error}\nThe versions of available with the current eb command are:\n"
+    error="${error}${available_cuda_easyconfigs}"
+    fatal_error "${error}"
   fi
 
-  # we need the --rebuild option and a (random) dir for the module since we are
-  # fixing the broken links of the EESSI-shipped installation
+  # We need the --rebuild option, as the CUDA module may or may not be on the
+  # `MODULEPATH` yet. Even if it is, we still want to redo this installation
+  # since it will provide the symlinked targets for the parts of the CUDA
+  # installation in the `.../versions/...` prefix
+  # We install the module in our `tmpdir` since we do not need the modulefile,
+  # we only care about providing the targets for the symlinks.
   extra_args="--rebuild --installpath-modules=${tmpdir}"
 
   # We don't want hooks used in this install, we need a vanilla CUDA installation
   touch "$tmpdir"/none.py
   # shellcheck disable=SC2086  # Intended splitting of extra_args
-  eb --prefix="$tmpdir" ${extra_args} --hooks="$tmpdir"/none.py --installpath="${cuda_install_parent}"/ CUDA-"${install_cuda_version}".eb
+  eb --prefix="$tmpdir" ${extra_args} --hooks="$tmpdir"/none.py --installpath="${cuda_install_parent}"/ "${cuda_easyconfig}"
   ret=$?
   if [ $ret -ne 0 ]; then
     fatal_error  "CUDA installation failed, please check EasyBuild logs..."
diff --git a/scripts/utils.sh b/scripts/utils.sh
index 07760f0dd0..b2be3f6221 100644
--- a/scripts/utils.sh
+++ b/scripts/utils.sh
@@ -40,30 +40,6 @@ function check_eessi_initialised() {
   fi
 }
 
-function float_greater_than() {
-  # Make sure we have two arguments
-  if [ $# -ne 2 ]; then
-    echo_red "greater_than_float requires two (float) numbers" >&2
-    return $ANY_ERROR_EXITCODE
-  fi
-  # Make sure the arguments are numbers
-  if [[ ! $1 =~ ^[+-]?[0-9]+\.?[0-9]*$ ]]; then
-    echo_yellow "Input to float_greater_than is not a float, ignoring"
-    return $ANY_ERROR_EXITCODE
-  fi
-  if [[ ! $2 =~ ^[+-]?[0-9]+\.?[0-9]*$ ]]; then
-    echo_yellow "Input to float_greater_than is not a float, ignoring"
-    return $ANY_ERROR_EXITCODE
-  fi
-  # Now do the actual evaluation
-  return_code=$ANY_ERROR_EXITCODE
-  result=$(echo "$1" "$2" | awk '{if ($1 > $2) print "true"}')
-  if [ "$result" = true ] ; then
-    return_code=0
-  fi
-  return $return_code
-}
-
 function check_in_prefix_shell() {
   # Make sure EPREFIX is defined
   if [[ -z "${EPREFIX}" ]]; then

From 12719eccaabf57112332cc4a4958d0d03e382ef4 Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 12:56:11 +0100
Subject: [PATCH 07/18] Make sure users are forced to accept CUDA EULA

---
 .../nvidia/install_cuda_host_injections.sh      | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)

diff --git a/gpu_support/nvidia/install_cuda_host_injections.sh b/gpu_support/nvidia/install_cuda_host_injections.sh
index 2c59a891f8..79af986a0a 100755
--- a/gpu_support/nvidia/install_cuda_host_injections.sh
+++ b/gpu_support/nvidia/install_cuda_host_injections.sh
@@ -21,6 +21,9 @@ show_help() {
     echo "Usage: $0 [OPTIONS]"
     echo "Options:"
     echo "  --help                           Display this help message"
+    echo "  --accept-cuda-eula               You _must_ accept the CUDA EULA to install"
+    echo "                                   CUDA, see the EULA at"
+    echo "                                   https://docs.nvidia.com/cuda/eula/index.html"
     echo "  -c, --cuda-version CUDA_VERSION  Specify a version o CUDA to install (must"
     echo "                                   have a corresponding easyconfig in the"
     echo "                                   EasyBuild release)"
@@ -31,6 +34,7 @@ show_help() {
 
 # Initialize variables
 install_cuda_version=""
+eula_accepted=0
 
 # Parse command-line options
 while [[ $# -gt 0 ]]; do
@@ -49,6 +53,10 @@ while [[ $# -gt 0 ]]; do
                 exit 1
             fi
             ;;
+        --accept-cuda-eula)
+            eula_accepted=1
+            shift 1
+            ;;
         -t|--temp-dir)
             if [ -n "$2" ]; then
                 CUDA_TEMP_DIR="$2"
@@ -86,6 +94,13 @@ if ! is_semantic_version "$install_cuda_version"; then
   fatal_error "${error}"
 fi
 
+# Make sure they have accepted the CUDA EULA
+if [ "$eula_accepted" -ne 1 ]; then
+  show_help
+  error="\nYou _must_ accept the CUDA EULA via the appropriate command line option.\n"
+  fatal_error "${error}"
+fi
+
 # Make sure EESSI is initialised
 check_eessi_initialised
 
@@ -186,7 +201,7 @@ else
   # We don't want hooks used in this install, we need a vanilla CUDA installation
   touch "$tmpdir"/none.py
   # shellcheck disable=SC2086  # Intended splitting of extra_args
-  eb --prefix="$tmpdir" ${extra_args} --hooks="$tmpdir"/none.py --installpath="${cuda_install_parent}"/ "${cuda_easyconfig}"
+  eb --prefix="$tmpdir" ${extra_args} --accept-eula-for=CUDA --hooks="$tmpdir"/none.py --installpath="${cuda_install_parent}"/ "${cuda_easyconfig}"
   ret=$?
   if [ $ret -ne 0 ]; then
     fatal_error  "CUDA installation failed, please check EasyBuild logs..."

From b005591069f055c4a5c206216279fd428878b887 Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 14:50:16 +0100
Subject: [PATCH 08/18] Add script to link in host NVIDIA drivers

---
 .../nvidia/link_nvidia_host_libraries.sh      | 128 ++++++++++++++++++
 1 file changed, 128 insertions(+)
 create mode 100755 gpu_support/nvidia/link_nvidia_host_libraries.sh

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
new file mode 100755
index 0000000000..eb713fce1b
--- /dev/null
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -0,0 +1,128 @@
+#!/bin/bash
+
+# This script links host libraries related to GPU drivers to a location where
+# they can be found by the EESSI linker
+
+# Initialise our bash functions
+TOPDIR=$(dirname $(realpath $BASH_SOURCE))
+source "$TOPDIR"/../../scripts/utils.sh
+
+# We rely on ldconfig to give us the location of the libraries on the host
+command_name="ldconfig"
+# We cannot use a version of ldconfig that's being shipped under CVMFS
+exclude_prefix="/cvmfs"
+
+found_paths=()
+# Always attempt to use /sbin/ldconfig
+if [ -x "/sbin/$command_name" ]; then
+    found_paths+=("$dir/$command_name")
+fi
+IFS=':' read -ra path_dirs <<< "$PATH"
+for dir in "${path_dirs[@]}"; do
+    if [[ ! "$dir" =~ ^$exclude_prefix ]]; then
+        if [ -x "$dir/$command_name" ]; then
+            found_paths+=("$dir/$command_name")
+        fi
+    fi
+done
+
+if [ ${#found_paths[@]} -gt 0 ]; then
+    echo "Found $command_name in the following locations:"
+    printf -- "- %s\n" "${found_paths[@]}"
+    echo "Using first version"
+    host_ldconfig=${found_paths[0]}
+else
+    error="$command_name not found in PATH or only found in paths starting with $exclude_prefix."
+    fatal_error $error
+fi
+
+# Make sure EESSI is initialised (doesn't matter what version)
+check_eessi_initialised
+
+# Find the CUDA version of the host CUDA drivers
+# (making sure that this can still work inside prefix environment inside a container)
+nvidia_smi_command="LD_LIBRARY_PATH=/.singularity/libs:$LD_LIBRARY_PATH nvidia-smi --query-gpu=driver_version --format=csv,noheader"
+if $nvidia_smi_command; then
+  host_cuda_version=$($nvidia_smi_command | tail -n1)
+else
+  error="Failed to successfully execute\n  $nvidia_smi_command\n"
+  fatal_error $error
+fi
+
+# Let's make sure the driver libraries are not already in place
+link_drivers=1
+
+host_injections_nvidia_dir="/cvmfs/pilot.eessi-hpc.org/host_injections/${EESSI_CPU_FAMILY}/nvidia"
+host_injection_driver_dir="${host_injections_nvidia_dir}/host"
+host_injection_driver_version_file="$host_injection_driver_dir/version.txt"
+if [ -e "$host_injection_driver_version_file" ]; then
+  if grep -q "$host_cuda_version" "$host_injection_driver_version_file"; then
+    echo_green "The host CUDA driver libraries have already been linked!"
+    link_drivers=0
+  else
+    # There's something there but it is out of date
+    echo_yellow "Cleaning out outdated symlinks"
+    rm $host_injection_driver_dir/*
+    if [ $? -ne 0 ]; then
+      error="Unable to remove files under '$host_injection_driver_dir'."
+      fatal_error $error
+    fi
+  fi
+fi
+
+drivers_linked=0
+if [ "$link_drivers" -eq 1 ]; then
+  if ! create_directory_structure "${host_injection_driver_dir}" ; then
+    fatal_error "No write permissions to directory ${host_injection_driver_dir}"
+  fi
+  cd ${host_injection_driver_dir}
+  # Need a small temporary space to hold a couple of files
+  temp_dir=$(mktemp -d)
+
+  # Gather libraries on the host (_must_ be host ldconfig)
+  $host_ldconfig -p | awk '{print $NF}' > "$temp_dir"/libs.txt
+  # Allow for the fact that we may be in a container so the CUDA libs might be in there
+  ls /.singularity.d/libs/* >> "$temp_dir"/libs.txt 2>/dev/null
+
+  # Leverage singularity to find the full list of libraries we should be linking to
+  curl -o "$temp_dir"/nvliblist.conf https://raw.githubusercontent.com/apptainer/apptainer/main/etc/nvliblist.conf
+
+  # Make symlinks to all the interesting libraries
+  grep '.so$' "$temp_dir"/nvliblist.conf | xargs -i grep {} libs.txt | xargs -i ln -s {}
+
+  # Inject CUDA version into dir
+  echo $host_cuda_version > version.txt
+  drivers_linked=1
+
+  # Remove the temporary directory when done
+  rm -r "$temp_dir"
+fi
+
+# Make latest symlink for NVIDIA drivers
+cd $host_injections_nvidia_dir
+symlink="latest"
+if [ -L "$symlink" ]; then
+    # Unless the drivers have been installed, leave the symlink alone
+    if [ "$drivers_linked" -eq 1 ]; then
+      ln -sf host latest
+    fi
+else
+    # No link exists yet
+    ln -s host latest
+fi
+
+# Make sure the libraries can be found by the EESSI linker
+host_injection_linker_dir=${EESSI_EPREFIX/versions/host_injections}
+if [ -L "$host_injection_linker_dir/lib" ]; then
+  target_path=$(readlink -f "$host_injection_linker_dir/lib")
+  if [ "$target_path" != "$$host_injections_nvidia_dir/latest" ]; then
+    cd $host_injection_linker_dir
+    ln -sf $host_injections_nvidia_dir/latest lib
+  fi
+else
+  create_directory_structure $host_injection_linker_dir
+  cd $host_injection_linker_dir
+  ln -s $host_injections_nvidia_dir/latest lib
+fi
+
+echo_green "Host NVIDIA gpu drivers linked successfully for EESSI"
\ No newline at end of file

From 9d6e91de4c0536b3552fdee9d1bd9d631a9a8be7 Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 14:52:13 +0100
Subject: [PATCH 09/18] Typo

---
 gpu_support/nvidia/link_nvidia_host_libraries.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index eb713fce1b..eb68e8dd69 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -15,7 +15,7 @@ exclude_prefix="/cvmfs"
 found_paths=()
 # Always attempt to use /sbin/ldconfig
 if [ -x "/sbin/$command_name" ]; then
-    found_paths+=("$dir/$command_name")
+    found_paths+=("/sbin/$command_name")
 fi
 IFS=':' read -ra path_dirs <<< "$PATH"
 for dir in "${path_dirs[@]}"; do

From cb126492b53e7c77ce896883dd98c47cbf0cd61d Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 14:54:46 +0100
Subject: [PATCH 10/18] Wrap the command exection because of the envvars

---
 gpu_support/nvidia/link_nvidia_host_libraries.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index eb68e8dd69..77796ac9c9 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -42,7 +42,7 @@ check_eessi_initialised
 # Find the CUDA version of the host CUDA drivers
 # (making sure that this can still work inside prefix environment inside a container)
 nvidia_smi_command="LD_LIBRARY_PATH=/.singularity/libs:$LD_LIBRARY_PATH nvidia-smi --query-gpu=driver_version --format=csv,noheader"
-if $nvidia_smi_command; then
+if $($nvidia_smi_command); then
   host_cuda_version=$($nvidia_smi_command | tail -n1)
 else
   error="Failed to successfully execute\n  $nvidia_smi_command\n"

From f9268e0fd80f726ba6ddf37727d821d1dc231d9c Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 14:57:07 +0100
Subject: [PATCH 11/18] Move envvar setting outside of command

---
 gpu_support/nvidia/link_nvidia_host_libraries.sh | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index 77796ac9c9..56c0b7a2ae 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -41,8 +41,9 @@ check_eessi_initialised
 
 # Find the CUDA version of the host CUDA drivers
 # (making sure that this can still work inside prefix environment inside a container)
-nvidia_smi_command="LD_LIBRARY_PATH=/.singularity/libs:$LD_LIBRARY_PATH nvidia-smi --query-gpu=driver_version --format=csv,noheader"
-if $($nvidia_smi_command); then
+LD_LIBRARY_PATH=/.singularity/libs:$LD_LIBRARY_PATH
+nvidia_smi_command="nvidia-smi --query-gpu=driver_version --format=csv,noheader"
+if $nvidia_smi_command; then
   host_cuda_version=$($nvidia_smi_command | tail -n1)
 else
   error="Failed to successfully execute\n  $nvidia_smi_command\n"

From a06f541445b5f9d5a76d659fc07c03d17003e930 Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 14:58:38 +0100
Subject: [PATCH 12/18] Forgot to add temp location

---
 gpu_support/nvidia/link_nvidia_host_libraries.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index 56c0b7a2ae..a15c1e98d4 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -89,7 +89,7 @@ if [ "$link_drivers" -eq 1 ]; then
   curl -o "$temp_dir"/nvliblist.conf https://raw.githubusercontent.com/apptainer/apptainer/main/etc/nvliblist.conf
 
   # Make symlinks to all the interesting libraries
-  grep '.so$' "$temp_dir"/nvliblist.conf | xargs -i grep {} libs.txt | xargs -i ln -s {}
+  grep '.so$' "$temp_dir"/nvliblist.conf | xargs -i grep {} "$temp_dir"/libs.txt | xargs -i ln -s {}
 
   # Inject CUDA version into dir
   echo $host_cuda_version > version.txt

From 600d4b46880d853e62428f94b7ce8912604b29ba Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 15:01:12 +0100
Subject: [PATCH 13/18] Wrong location under host_injections

---
 gpu_support/nvidia/link_nvidia_host_libraries.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index a15c1e98d4..b4b8699628 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -53,7 +53,7 @@ fi
 # Let's make sure the driver libraries are not already in place
 link_drivers=1
 
-host_injections_nvidia_dir="/cvmfs/pilot.eessi-hpc.org/host_injections/${EESSI_CPU_FAMILY}/nvidia"
+host_injections_nvidia_dir="/cvmfs/pilot.eessi-hpc.org/host_injections/nvidia/${EESSI_CPU_FAMILY}"
 host_injection_driver_dir="${host_injections_nvidia_dir}/host"
 host_injection_driver_version_file="$host_injection_driver_dir/version.txt"
 if [ -e "$host_injection_driver_version_file" ]; then

From b2664a3943115c75a85099936111e5ab231ebd2c Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 15:17:13 +0100
Subject: [PATCH 14/18] Export LD_LIBRARY_PATH in script

---
 gpu_support/nvidia/link_nvidia_host_libraries.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index b4b8699628..c971e4e65e 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -41,7 +41,7 @@ check_eessi_initialised
 
 # Find the CUDA version of the host CUDA drivers
 # (making sure that this can still work inside prefix environment inside a container)
-LD_LIBRARY_PATH=/.singularity/libs:$LD_LIBRARY_PATH
+export LD_LIBRARY_PATH=/.singularity.d/libs:$LD_LIBRARY_PATH
 nvidia_smi_command="nvidia-smi --query-gpu=driver_version --format=csv,noheader"
 if $nvidia_smi_command; then
   host_cuda_version=$($nvidia_smi_command | tail -n1)

From 9854e79edbc5b7a3450aa1ee1484f8e58ad4575d Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 15:29:19 +0100
Subject: [PATCH 15/18] Also export CUDA version to the links directory

---
 gpu_support/nvidia/link_nvidia_host_libraries.sh | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index c971e4e65e..ac826f9572 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -43,8 +43,9 @@ check_eessi_initialised
 # (making sure that this can still work inside prefix environment inside a container)
 export LD_LIBRARY_PATH=/.singularity.d/libs:$LD_LIBRARY_PATH
 nvidia_smi_command="nvidia-smi --query-gpu=driver_version --format=csv,noheader"
-if $nvidia_smi_command; then
-  host_cuda_version=$($nvidia_smi_command | tail -n1)
+if $nvidia_smi_command > /dev/null; then
+  host_driver_version=$($nvidia_smi_command | tail -n1)
+  host_cuda_version=$(nvidia-smi  -q --display=COMPUTE | grep CUDA | awk 'NF>1{print $NF}')
 else
   error="Failed to successfully execute\n  $nvidia_smi_command\n"
   fatal_error $error
@@ -57,7 +58,7 @@ host_injections_nvidia_dir="/cvmfs/pilot.eessi-hpc.org/host_injections/nvidia/${
 host_injection_driver_dir="${host_injections_nvidia_dir}/host"
 host_injection_driver_version_file="$host_injection_driver_dir/version.txt"
 if [ -e "$host_injection_driver_version_file" ]; then
-  if grep -q "$host_cuda_version" "$host_injection_driver_version_file"; then
+  if grep -q "$host_driver_version" "$host_injection_driver_version_file"; then
     echo_green "The host CUDA driver libraries have already been linked!"
     link_drivers=0
   else
@@ -91,8 +92,9 @@ if [ "$link_drivers" -eq 1 ]; then
   # Make symlinks to all the interesting libraries
   grep '.so$' "$temp_dir"/nvliblist.conf | xargs -i grep {} "$temp_dir"/libs.txt | xargs -i ln -s {}
 
-  # Inject CUDA version into dir
-  echo $host_cuda_version > version.txt
+  # Inject driver and CUDA versions into dir
+  echo $host_driver_version > version.txt
+  echo $host_cuda_version > cuda_version.txt
   drivers_linked=1
 
   # Remove the temporary directory when done

From 59fee787c0b8a709c05fc8bba379f2cb5e2abcd7 Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 15:34:49 +0100
Subject: [PATCH 16/18] Also export CUDA version to the links directory

---
 gpu_support/nvidia/link_nvidia_host_libraries.sh | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index ac826f9572..e71697c8d7 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -45,6 +45,7 @@ export LD_LIBRARY_PATH=/.singularity.d/libs:$LD_LIBRARY_PATH
 nvidia_smi_command="nvidia-smi --query-gpu=driver_version --format=csv,noheader"
 if $nvidia_smi_command > /dev/null; then
   host_driver_version=$($nvidia_smi_command | tail -n1)
+  # If the first worked, this should work too
   host_cuda_version=$(nvidia-smi  -q --display=COMPUTE | grep CUDA | awk 'NF>1{print $NF}')
 else
   error="Failed to successfully execute\n  $nvidia_smi_command\n"
@@ -56,7 +57,7 @@ link_drivers=1
 
 host_injections_nvidia_dir="/cvmfs/pilot.eessi-hpc.org/host_injections/nvidia/${EESSI_CPU_FAMILY}"
 host_injection_driver_dir="${host_injections_nvidia_dir}/host"
-host_injection_driver_version_file="$host_injection_driver_dir/version.txt"
+host_injection_driver_version_file="$host_injection_driver_dir/driver_version.txt"
 if [ -e "$host_injection_driver_version_file" ]; then
   if grep -q "$host_driver_version" "$host_injection_driver_version_file"; then
     echo_green "The host CUDA driver libraries have already been linked!"
@@ -93,7 +94,7 @@ if [ "$link_drivers" -eq 1 ]; then
   grep '.so$' "$temp_dir"/nvliblist.conf | xargs -i grep {} "$temp_dir"/libs.txt | xargs -i ln -s {}
 
   # Inject driver and CUDA versions into dir
-  echo $host_driver_version > version.txt
+  echo $host_driver_version > driver_version.txt
   echo $host_cuda_version > cuda_version.txt
   drivers_linked=1
 

From 3d0c3dd3eb6fd0e9a80652b477379325a040cdba Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Thu, 30 Nov 2023 15:46:46 +0100
Subject: [PATCH 17/18] Add a comment to explain the download

---
 gpu_support/nvidia/link_nvidia_host_libraries.sh | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index e71697c8d7..d714c0ec8b 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -88,6 +88,7 @@ if [ "$link_drivers" -eq 1 ]; then
   ls /.singularity.d/libs/* >> "$temp_dir"/libs.txt 2>/dev/null
 
   # Leverage singularity to find the full list of libraries we should be linking to
+  echo_yellow "Downloading latest version of nvliblist.conf from Apptainer"
   curl -o "$temp_dir"/nvliblist.conf https://raw.githubusercontent.com/apptainer/apptainer/main/etc/nvliblist.conf
 
   # Make symlinks to all the interesting libraries
@@ -129,4 +130,4 @@ else
   ln -s $host_injections_nvidia_dir/latest lib
 fi
 
-echo_green "Host NVIDIA gpu drivers linked successfully for EESSI"
\ No newline at end of file
+echo_green "Host NVIDIA gpu drivers linked successfully for EESSI"

From 480d35643c7740c48f5508303e1cccadf04aae76 Mon Sep 17 00:00:00 2001
From: Alan O'Cais <alan.ocais@cecam.org>
Date: Fri, 1 Dec 2023 15:07:34 +0100
Subject: [PATCH 18/18] Address review comments

---
 .../nvidia/install_cuda_host_injections.sh     | 18 +++++++-----------
 .../nvidia/link_nvidia_host_libraries.sh       | 13 ++++++++-----
 2 files changed, 15 insertions(+), 16 deletions(-)

diff --git a/gpu_support/nvidia/install_cuda_host_injections.sh b/gpu_support/nvidia/install_cuda_host_injections.sh
index 79af986a0a..f02f0da02e 100755
--- a/gpu_support/nvidia/install_cuda_host_injections.sh
+++ b/gpu_support/nvidia/install_cuda_host_injections.sh
@@ -74,6 +74,9 @@ while [[ $# -gt 0 ]]; do
     esac
 done
 
+# Make sure EESSI is initialised
+check_eessi_initialised
+
 # Make sure the CUDA version supplied is a semantic version
 is_semantic_version() {
     local version=$1
@@ -89,7 +92,7 @@ if ! is_semantic_version "$install_cuda_version"; then
   show_help
   error="\nYou must provide a semantic version for CUDA (e.g., 12.1.1) via the appropriate\n"
   error="${error}command line option. This script is intended for use with EESSI so the 'correct'\n"
-  error="${error}version to provide is probably the one that is available under\n"
+  error="${error}version to provide is probably one of those available under\n"
   error="${error}$EESSI_SOFTWARE_PATH/software/CUDA\n"
   fatal_error "${error}"
 fi
@@ -101,16 +104,9 @@ if [ "$eula_accepted" -ne 1 ]; then
   fatal_error "${error}"
 fi
 
-# Make sure EESSI is initialised
-check_eessi_initialised
-
-if [[ -z "${EESSI_SOFTWARE_PATH}" ]]; then
-  fatal_error "This script cannot be used without having first defined EESSI_SOFTWARE_PATH"
-else
-  # As an installation location just use $EESSI_SOFTWARE_PATH but replacing `versions` with `host_injections`
-  # (CUDA is a binary installation so no need to worry too much about the EasyBuild setup)
-  cuda_install_parent=${EESSI_SOFTWARE_PATH/versions/host_injections}
-fi
+# As an installation location just use $EESSI_SOFTWARE_PATH but replacing `versions` with `host_injections`
+# (CUDA is a binary installation so no need to worry too much about the EasyBuild setup)
+cuda_install_parent=${EESSI_SOFTWARE_PATH/versions/host_injections}
 
 # Only install CUDA if specified version is not found.
 # (existence of easybuild subdir implies a successful install)
diff --git a/gpu_support/nvidia/link_nvidia_host_libraries.sh b/gpu_support/nvidia/link_nvidia_host_libraries.sh
index d714c0ec8b..26760f0b82 100755
--- a/gpu_support/nvidia/link_nvidia_host_libraries.sh
+++ b/gpu_support/nvidia/link_nvidia_host_libraries.sh
@@ -19,11 +19,14 @@ if [ -x "/sbin/$command_name" ]; then
 fi
 IFS=':' read -ra path_dirs <<< "$PATH"
 for dir in "${path_dirs[@]}"; do
-    if [[ ! "$dir" =~ ^$exclude_prefix ]]; then
-        if [ -x "$dir/$command_name" ]; then
-            found_paths+=("$dir/$command_name")
-        fi
-    fi
+  if [ "$dir" = "/sbin" ]; then
+    continue  # we've already checked for $command_name in /sbin, don't need to do it twice
+  fi
+  if [[ ! "$dir" =~ ^$exclude_prefix ]]; then
+      if [ -x "$dir/$command_name" ]; then
+          found_paths+=("$dir/$command_name")
+      fi
+  fi
 done
 
 if [ ${#found_paths[@]} -gt 0 ]; then