mirror of
https://src.fedoraproject.org/rpms/llvm.git
synced 2024-11-24 09:32:42 +00:00
4573a69a87
Try to simply lit config file generation by adding standalone fedora specific config file. In the future, when we need to customize the lit config for the llvm-test package, we can just update lit.fedora.cfg.py rather than trying to patch the default config from the spec file. This patch also makes the run-lit-tests script identical on all arches, so it will no longer cause multi-lib conflicts.
17 lines
807 B
Python
17 lines
807 B
Python
|
|
config.llvm_tools_dir = '/usr/bin'
|
|
config.llvm_shlib_dir = '%(llvm_shlib_dir)s' % lit_config.params
|
|
|
|
if hasattr(config, 'host_triple'):
|
|
# This means we are running lit regression tests
|
|
|
|
# Regression tests write output to this directory, so we need to be able to specify
|
|
# a temp directory when invoking lit. e.g. lit -Dllvm_obj_root=/tmp/lit
|
|
config.llvm_obj_root = "%(llvm_obj_root)s" % lit_config.params
|
|
lit_config.load_config(config, '%(llvm_test_root)s/lit.cfg.py' % lit_config.params)
|
|
else:
|
|
# This means we are running lit unit tests
|
|
|
|
# For unit tests, llvm_obj_root is used to find the unit test binaries.
|
|
config.llvm_obj_root = '%(llvm_unittest_bindir)s' % lit_config.params
|
|
lit_config.load_config(config, '%(llvm_test_root)s/Unit/lit.cfg.py' % lit_config.params)
|