Merge branch '2023.06-software.eessi.io' into ALL_0.9.2_2023a

EESSI-install-software.sh

@@ -189,63 +189,68 @@ pr_diff=$(ls [0-9]*.diff | head -1)
 
 # install any additional required scripts
 # order is important: these are needed to install a full CUDA SDK in host_injections
-install_scripts_changed=$(cat ${pr_diff} | grep '^+++' | cut -f2 -d' ' | sed 's@^[a-z]/@@g' | grep '^install_scripts.sh$' > /dev/null; echo $?)
-if [ ${install_scripts_changed} == '0' ]; then
-    # for now, this just reinstalls all scripts. Note the most elegant, but works
-    ${TOPDIR}/install_scripts.sh --prefix ${EESSI_CVMFS_REPO}
-fi
+# for now, this just reinstalls all scripts. Note the most elegant, but works
+${TOPDIR}/install_scripts.sh --prefix ${EESSI_PREFIX}
 
 # Install full CUDA SDK in host_injections
 # Hardcode this for now, see if it works
 # TODO: We should make a nice yaml and loop over all CUDA versions in that yaml to figure out what to install
-${EESSI_CVMFS_REPO}/gpu_support/nvidia/install_cuda_host_injections.sh 12.1.1
+${EESSI_PREFIX}/scripts/gpu_support/nvidia/install_cuda_host_injections.sh -c 12.1.1 --accept-cuda-eula
 
 # Install drivers in host_injections
-${EESSI_CVMFS_REPO}/gpu_support/nvidia/link_nvidia_host_libraries.sh
+# 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
 
 # use PR patch file to determine in which easystack files stuff was added
-for easystack_file in $(cat ${pr_diff} | grep '^+++' | cut -f2 -d' ' | sed 's@^[a-z]/@@g' | grep '^easystacks/.*yml$' | egrep -v 'known-issues|missing'); do
-
-    echo -e "Processing easystack file ${easystack_file}...\n\n"
-
-    # determine version of EasyBuild module to load based on EasyBuild version included in name of easystack file
-    eb_version=$(echo ${easystack_file} | sed 's/.*eb-\([0-9.]*\).*/\1/g')
-
-    # load EasyBuild module (will be installed if it's not available yet)
-    source ${TOPDIR}/load_easybuild_module.sh ${eb_version}
-
-    ${EB} --show-config
-
-    echo_green "All set, let's start installing some software with EasyBuild v${eb_version} in ${EASYBUILD_INSTALLPATH}..."
-
-    if [ -f ${easystack_file} ]; then
-        echo_green "Feeding easystack file ${easystack_file} to EasyBuild..."
-
-        ${EB} --easystack ${TOPDIR}/${easystack_file} --robot
-        ec=$?
-
-        # copy EasyBuild log file if EasyBuild exited with an error
-        if [ ${ec} -ne 0 ]; then
-            eb_last_log=$(unset EB_VERBOSE; eb --last-log)
-            # copy to current working directory
-            cp -a ${eb_last_log} .
-            echo "Last EasyBuild log file copied from ${eb_last_log} to ${PWD}"
-            # copy to build logs dir (with context added)
-            copy_build_log "${eb_last_log}" "${build_logs_dir}"
+changed_easystacks=$(cat ${pr_diff} | grep '^+++' | cut -f2 -d' ' | sed 's@^[a-z]/@@g' | grep '^easystacks/.*yml$' | egrep -v 'known-issues|missing') 
+if [ -z ${changed_easystacks} ]; then
+    echo "No missing installations, party time!"  # Ensure the bot report success, as there was nothing to be build here
+else
+    for easystack_file in ${changed_easystacks}; do
+
+        echo -e "Processing easystack file ${easystack_file}...\n\n"
+
+        # determine version of EasyBuild module to load based on EasyBuild version included in name of easystack file
+        eb_version=$(echo ${easystack_file} | sed 's/.*eb-\([0-9.]*\).*/\1/g')
+
+        # load EasyBuild module (will be installed if it's not available yet)
+        source ${TOPDIR}/load_easybuild_module.sh ${eb_version}
+
+        ${EB} --show-config
+
+        echo_green "All set, let's start installing some software with EasyBuild v${eb_version} in ${EASYBUILD_INSTALLPATH}..."
+
+        if [ -f ${easystack_file} ]; then
+            echo_green "Feeding easystack file ${easystack_file} to EasyBuild..."
+
+            ${EB} --easystack ${TOPDIR}/${easystack_file} --robot
+            ec=$?
+
+            # copy EasyBuild log file if EasyBuild exited with an error
+            if [ ${ec} -ne 0 ]; then
+                eb_last_log=$(unset EB_VERBOSE; eb --last-log)
+                # copy to current working directory
+                cp -a ${eb_last_log} .
+                echo "Last EasyBuild log file copied from ${eb_last_log} to ${PWD}"
+                # copy to build logs dir (with context added)
+                copy_build_log "${eb_last_log}" "${build_logs_dir}"
+            fi
+
+            $TOPDIR/check_missing_installations.sh ${TOPDIR}/${easystack_file}
+        else
+            fatal_error "Easystack file ${easystack_file} not found!"
         fi
-
-        $TOPDIR/check_missing_installations.sh ${TOPDIR}/${easystack_file}
-    else
-        fatal_error "Easystack file ${easystack_file} not found!"
-    fi
-
-done
+
+    done
+fi
 
 ### add packages here
 
 echo ">> Creating/updating Lmod cache..."
 export LMOD_RC="${EASYBUILD_INSTALLPATH}/.lmod/lmodrc.lua"
-if [ ! -f $LMOD_RC ]; then
+lmodrc_changed=$(cat ${pr_diff} | grep '^+++' | cut -f2 -d' ' | sed 's@^[a-z]/@@g' | grep '^create_lmodrc.py$' > /dev/null; echo $?)
+if [ ! -f $LMOD_RC ] || [ ${lmodrc_changed} == '0' ]; then
     python3 $TOPDIR/create_lmodrc.py ${EASYBUILD_INSTALLPATH}
     check_exit_code $? "$LMOD_RC created" "Failed to create $LMOD_RC"
 fi

bot/build.sh

@@ -176,6 +176,11 @@ mkdir -p ${TARBALL_TMP_BUILD_STEP_DIR}
 declare -a BUILD_STEP_ARGS=()
 BUILD_STEP_ARGS+=("--save" "${TARBALL_TMP_BUILD_STEP_DIR}")
 BUILD_STEP_ARGS+=("--storage" "${STORAGE}")
+# add options required to handle NVIDIA support
+BUILD_STEP_ARGS+=("--nvidia" "all")
+if [[ ! -z ${SHARED_FS_PATH} ]]; then
+    BUILD_STEP_ARGS+=("--host-injections" "${SHARED_FS_PATH}/host-injections")
+fi
 
 # prepare arguments to install_software_layer.sh (specific to build step)
 declare -a INSTALL_SCRIPT_ARGS=()

create_lmodrc.py

@@ -17,6 +17,85 @@
 }
 """
 
+GPU_LMOD_RC ="""require("strict")
+local hook = require("Hook")
+local open = io.open
+
+local function read_file(path)
+    local file = open(path, "rb") -- r read mode and b binary mode
+    if not file then return nil end
+    local content = file:read "*a" -- *a or *all reads the whole file
+    file:close()
+    return content
+end
+
+local function cuda_enabled_load_hook(t)
+    local frameStk  = require("FrameStk"):singleton()
+    local mt        = frameStk:mt()
+    local simpleName = string.match(t.modFullName, "(.-)/")
+    -- If we try to load CUDA itself, check if the full CUDA SDK was installed on the host in host_injections. 
+    -- This is required for end users to build additional CUDA software. If the full SDK isn't present, refuse
+    -- to load the CUDA module and print an informative message on how to set up GPU support for EESSI
+    local refer_to_docs = "For more information on how to do this, see https://www.eessi.io/docs/gpu/.\\n"
+    if simpleName == 'CUDA' then
+        -- get the full host_injections path
+        local hostInjections = string.gsub(os.getenv('EESSI_SOFTWARE_PATH') or "", 'versions', 'host_injections')
+        -- build final path where the CUDA software should be installed
+        local cudaEasyBuildDir = hostInjections .. "/software/" .. t.modFullName .. "/easybuild"
+        local cudaDirExists = isDir(cudaEasyBuildDir)
+        if not cudaDirExists then
+            local advice = "but while the module file exists, the actual software is not entirely shipped with EESSI "
+            advice = advice .. "due to licencing. You will need to install a full copy of the CUDA SDK where EESSI "
+            advice = advice .. "can find it.\\n"
+            advice = advice .. refer_to_docs
+            LmodError("\\nYou requested to load ", simpleName, " ", advice)
+        end
+    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 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"
+        local cudaDriverExists = isFile(cudaDriverFile)
+        local singularityCudaExists = isFile("/.singularity.d/libs/libcuda.so")
+        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 .. refer_to_docs
+            LmodError("\\nYou requested to load ", simpleName, " ", advice)
+        else
+            -- CUDA driver exists, now we check its version to see if an update is needed
+            if cudaDriverExists then
+                local cudaVersion = read_file(cudaVersionFile)
+                local cudaVersion_req = os.getenv("EESSICUDAVERSION")
+                -- driver CUDA versions don't give a patch version for CUDA
+                local major, minor = string.match(cudaVersion, "(%d+)%.(%d+)")
+                local major_req, minor_req, patch_req = string.match(cudaVersion_req, "(%d+)%.(%d+)%.(%d+)")
+                local driver_libs_need_update = false
+                if major < major_req then
+                    driver_libs_need_update = true
+                elseif major == major_req then
+                    if minor < minor_req then
+                        driver_libs_need_update = true
+                    end
+                end
+                if driver_libs_need_update == true then
+                    local advice = "but the module you want to load requires CUDA  " .. cudaVersion_req .. ". "
+                    advice = advice .. "Please update your CUDA driver libraries and then "
+                    advice = advice .. "let EESSI know about the update.\\n"
+                    advice = advice .. refer_to_docs
+                    LmodError("\\nYour driver CUDA version is ", cudaVersion, " ", advice)
+                end
+            end
+        end
+    end
+end
+
+hook.register("load", cuda_enabled_load_hook)
+"""
 
 def error(msg):
     sys.stderr.write("ERROR: %s\n" % msg)
@@ -36,6 +115,7 @@ def error(msg):
     'dot_lmod': DOT_LMOD,
     'prefix': prefix,
 }
+lmodrc_txt += '\n' + GPU_LMOD_RC
 try:
     os.makedirs(os.path.dirname(lmodrc_path), exist_ok=True)
     with open(lmodrc_path, 'w') as fp:

create_tarball.sh

@@ -40,11 +40,23 @@ echo ">> Collecting list of files/directories to include in tarball via ${PWD}..
 files_list=${tmpdir}/files.list.txt
 module_files_list=${tmpdir}/module_files.list.txt
 
+# include Lmod cache and configuration file (lmodrc.lua),
+if [ -d ${eessi_version}/software/${os}/${cpu_arch_subdir}/.lmod ]; then
+    # skip whiteout files and backup copies of Lmod cache (spiderT.old.*)
+    find ${eessi_version}/software/${os}/${cpu_arch_subdir}/.lmod -type f | egrep -v '/\.wh\.|spiderT.old' >> ${files_list}
+fi
+
 if [ -d ${eessi_version}/software/${os}/${cpu_arch_subdir}/.lmod ]; then
     # include Lmod cache and configuration file (lmodrc.lua),
     # skip whiteout files and backup copies of Lmod cache (spiderT.old.*)
-    find ${eessi_version}/software/${os}/${cpu_arch_subdir}/.lmod -type f | egrep -v '/\.wh\.|spiderT.old' > ${files_list}
+    find ${eessi_version}/software/${os}/${cpu_arch_subdir}/.lmod -type f | egrep -v '/\.wh\.|spiderT.old' >> ${files_list}
 fi
+
+# include scripts that were copied by install_scripts.sh, which we want to ship in EESSI repository
+if [ -d ${eessi_version}/scripts ]; then
+    find ${eessi_version}/scripts -type f | grep -v '/\.wh\.' >> ${files_list}
+fi
+
 if [ -d ${eessi_version}/software/${os}/${cpu_arch_subdir}/modules ]; then
     # module files
     find ${eessi_version}/software/${os}/${cpu_arch_subdir}/modules -type f | grep -v '/\.wh\.' >> ${files_list}
@@ -55,6 +67,7 @@ if [ -d ${eessi_version}/software/${os}/${cpu_arch_subdir}/modules ]; then
         | grep -v '/\.wh\.' | grep -v '/\.modulerc\.lua' | sed -e 's/.lua$//' | sed -e 's@.*/modules/all/@@g' | sort -u \
         >> ${module_files_list}
 fi
+
 if [ -d ${eessi_version}/software/${os}/${cpu_arch_subdir}/software -a -r ${module_files_list} ]; then
     # installation directories but only those for which module files were created
     # Note, we assume that module names (as defined by 'PACKAGE_NAME/VERSION.lua'

easystacks/software.eessi.io/2023.06/README.md

@@ -0,0 +1,7 @@
+File naming matters, since it determines the order in which easystack files are processed.
+
+Software installed with system toolchain should be installed first,
+this includes EasyBuild itself, see `eessi-2023.06-eb-4.8.2-001-system.yml` .
+
+CUDA installations must be done before CUDA is required as dependency for something
+built with a non-system toolchain, see `eessi-2023.06-eb-4.8.2-010-CUDA.yml` .

easystacks/software.eessi.io/2023.06/eessi-2023.06-eb-4.8.2-010-CUDA.yml

@@ -0,0 +1,5 @@
+easyconfigs:
+  - CUDA-12.1.1.eb:
+      options:
+        include-easyblocks-from-pr: 3045
+        accept-eula-for: CUDA

easystacks/software.eessi.io/2023.06/eessi-2023.06-eb-4.8.2-2023a.yml

@@ -35,6 +35,14 @@ easyconfigs:
   - Boost-1.82.0-GCC-12.3.0.eb
   - netCDF-4.9.2-gompi-2023a.eb
   - FFmpeg-6.0-GCCcore-12.3.0.eb
+  - CUDA-Samples-12.1-GCC-12.3.0-CUDA-12.1.1.eb:
+      # use easyconfig that only install subset of CUDA samples,
+      # to circumvent problem with nvcc linking to glibc of host OS,
+      # see https://github.com/easybuilders/easybuild-easyconfigs/pull/19189;
+      # and where additional samples are excluded because they fail to build on aarch64,
+      # see https://github.com/easybuilders/easybuild-easyconfigs/pull/19451;
+      options:
+        from-pr: 19451
   - ALL-0.9.2-foss-2023a.eb:
       options:
         from-pr: 19455