Improve search for ROCm-LLVM dep and add tests

Author: zerefwayne

.github/workflows/test-eb-hooks.yml

@@ -116,3 +116,48 @@ jobs:
               eb --hooks=$PWD/eb_hooks.py "$INCOMPATIBLE_EASYCONFIG" --stop fetch 2>&1 1>/dev/null | grep -q "does not contain a valid list of dictionaries"
               echo "Incorrect format for EESSI_SITE_TOP_LEVEL_TOOLCHAINS caught"
 
+  check_inject_gpu_property:
+    runs-on: ubuntu-24.04
+    strategy:
+      matrix:
+        # ROCm-LLVM / rocm-compilers / rompi only exist in the 2025.06 stack
+        EESSI_VERSION:
+          - '2025.06'
+
+    steps:
+        - name: Check out software-layer repository
+          uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
+
+        - name: Mount EESSI CernVM-FS repository
+          uses: eessi/github-action-eessi@v3
+          with:
+            eessi_stack_version: ${{matrix.EESSI_VERSION}}
+            use_eessi_module: true
+
+        - name: Test that inject_gpu_property tags GPU software correctly
+          if: ${{ github.event_name == 'pull_request' }}
+          run: |
+              module load EESSI-extend/${{matrix.EESSI_VERSION}}-easybuild
+
+              # Control: a non-GPU easyconfig must not receive the gpu property
+              eb --hooks=$PWD/eb_hooks.py --extended-dry-run M4-1.4.19-GCCcore-14.2.0.eb 2>&1 | grep -qF 'add_property("arch","gpu")' && exit 1
+              echo "Non-GPU easyconfig untouched"
+
+              # ROCm-LLVM as a direct dependency (rocm-compilers bundle)
+              OUT=$(eb --hooks=$PWD/eb_hooks.py --extended-dry-run rocm-compilers-19.0.0-ROCm-6.4.1.eb 2>&1)
+              echo "$OUT" | grep -qF 'add_property("arch","gpu")'
+              echo "$OUT" | grep -qF 'setenv("EESSIROCMMINLLVMVERSION","19.0.0-ROCm-6.4.1")'
+              echo "Direct ROCm-LLVM dependency detected"
+
+              # rocm-compilers as the toolchain
+              OUT=$(eb --hooks=$PWD/eb_hooks.py --extended-dry-run hwloc-ROCm-2.11.2-rocm-compilers-19.0.0-ROCm-6.4.1.eb 2>&1)
+              echo "$OUT" | grep -qF 'add_property("arch","gpu")'
+              echo "$OUT" | grep -qF 'setenv("EESSIROCMMINLLVMVERSION","19.0.0-ROCm-6.4.1")'
+              echo "rocm-compilers toolchain handled"
+
+              # rompi as the toolchain (ROCm-LLVM nested inside rocm-compilers bundle)
+              OUT=$(eb --hooks=$PWD/eb_hooks.py --extended-dry-run OSU-Micro-Benchmarks-7.5-rompi-2025a.eb 2>&1)
+              echo "$OUT" | grep -qF 'add_property("arch","gpu")'
+              echo "$OUT" | grep -qF 'setenv("EESSIROCMMINLLVMVERSION","19.0.0-ROCm-6.4.1")'
+              echo "rompi toolchain handled"
+

eb_hooks.py

@@ -11,7 +11,11 @@
 import easybuild.tools.environment as env
 from easybuild.easyblocks.generic.configuremake import obtain_config_guess
 from easybuild.framework.easyconfig.constants import EASYCONFIG_CONSTANTS
-from easybuild.framework.easyconfig.easyconfig import get_toolchain_hierarchy
+from easybuild.framework.easyconfig.easyconfig import (
+    get_toolchain_hierarchy,
+    process_easyconfig,
+    robot_find_easyconfig,
+)
 from easybuild.tools import config
 from easybuild.tools.build_log import EasyBuildError, print_msg, print_warning
 from easybuild.tools.config import build_option, install_path, update_build_option
@@ -1934,45 +1938,69 @@ def replace_binary_non_distributable_files_with_symlinks(log, install_dir, pkg_n
                     symlink(host_inj_path, full_path)
 
 
+def find_rocm_llvm_dependency(ec):
+    """
+    Return the ROCm-LLVM dependency for this easyconfig, or None. ROCm-LLVM can
+    be a direct dependency, a direct toolchain component (rocm-compilers as the
+    toolchain), or one level deeper inside the rocm-compilers bundle when the
+    toolchain is rompi/rfbf/rfoss.
+    """
+    for dep in ec.asdict()['dependencies']:
+        if dep['name'] == 'ROCm-LLVM':
+            return dep
+
+    if is_system_toolchain(ec.toolchain.name):
+        return None
+
+    tcdeps = ec.toolchain.tcdeps or []
+    for dep in tcdeps:
+        if dep['name'] == 'ROCm-LLVM':
+            return dep
+    for dep in tcdeps:
+        if dep['name'] == 'rocm-compilers':
+            rocm_compilers_ec = robot_find_easyconfig(dep['name'], dep['version'] + dep['versionsuffix'])
+            if rocm_compilers_ec:
+                rocm_compilers_deps = process_easyconfig(rocm_compilers_ec)[0]['ec'].dependencies(runtime_only=True)
+                for subdep in rocm_compilers_deps:
+                    if subdep['name'] == 'ROCm-LLVM':
+                        return subdep
+
+    return None
+
+
 def inject_gpu_property(ec):
     """
     Add 'gpu' property and EESSI<PACKAGE>VERSION envvars via modluafooter
     easyconfig parameter, and drop dependencies to build dependencies
     """
     ec_dict = ec.asdict()
-    # Check if CUDA, cuDNN, you-name-it is in the dependencies, if so
-    # - drop dependency to build dependency
-    # - add 'gpu' Lmod property
-    # - add envvar with package version
-    pkg_names = ( "CUDA", "cuDNN", "ROCm-LLVM" )
-    # skip dropping step for ROCm-LLVM as it is redistributable
-    drop_to_builddep = ( "CUDA", "cuDNN" )
     pkg_versions = { }
     add_gpu_property = ''
 
-    # If none of the gpu packages are in the easyconfig, do not process further
-    dep_names = {dep[0] for dep in ec_dict['dependencies']}
-    if not any(pkg in dep_names for pkg in pkg_names):
-        return ec
-
-    # Check if pkg_name is in the dependencies, if so drop dependency to build
+    # Check if pkg_name is related to CUDA, if so drop dependency to build
     # dependency and set variable for later adding the 'gpu' Lmod property
     # to '.remove' dependencies from ec_dict['dependencies'] we make a copy,
     # iterate over the copy and can then savely use '.remove' on the original
-    for pkg_name in pkg_names:
+    # ec_dict['dependencies'].
+    for pkg_name in ('CUDA', 'cuDNN'):
         for dep in ec_dict['dependencies'][:]:
             if dep[0] != pkg_name:
                 continue
 
             add_gpu_property = 'add_property("arch","gpu")'
             pkg_versions[pkg_name] = dep[1]
 
-            if pkg_name in drop_to_builddep:
-                ec.log.info("Dropping dependency on %s to build dependency" % pkg_name)
-                ec_dict['dependencies'].remove(dep)
-                # Avoid adding a duplicate
-                if dep not in ec_dict['builddependencies']:
-                    ec_dict['builddependencies'].append(dep)
+            ec.log.info("Dropping dependency on %s to build dependency" % pkg_name)
+            ec_dict['dependencies'].remove(dep)
+            if dep not in ec_dict['builddependencies']:
+                ec_dict['builddependencies'].append(dep)
+
+    # ROCm-LLVM is handled separately: it is redistributable (kept as a runtime dep)
+    # and may be pulled in via a ROCm toolchain rather than as a direct dependency.
+    rocm_llvm_dep = find_rocm_llvm_dependency(ec)
+    if rocm_llvm_dep is not None:
+        add_gpu_property = 'add_property("arch","gpu")'
+        pkg_versions['ROCm-LLVM'] = rocm_llvm_dep['version'] + (rocm_llvm_dep['versionsuffix'] or '')
 
     if add_gpu_property:
         ec.log.info("Injecting gpu as Lmod arch property and envvars for dependencies with their version")