Hal Finkel via llvm-dev
2015-Oct-17 14:08 UTC
[llvm-dev] [lld] [ELF2/PPC64] test-suite status
Hi everyone, Here's a quick update - I ran the test suite using lld2 on my big-Endian PPC64/Linux machine (ELF v1), which is clean using the system linker, and here are the outstanding failures: SingleSource/Regression/C/2008-01-07-LongDouble | pass 0.0343 0.0737 * 0.0010 0.0080 This is due to our lack of symbol-versioning support (there was a slight ABI change affecting printf and long doubles, and we need to pick up printf@@GLIBC_2.4, not printf, for this to work). SingleSource/UnitTests/initp1 | pass 0.1512 0.2294 * 0.0011 0.0238 Support for __attribute__((init_priority(N))) incomplete I suppose. SingleSource/Benchmarks/Shootout-C++/EH/except | pass 0.4039 0.6252 * 0.0017 0.1093 SingleSource/Regression/C++/EH/class_hierarchy | pass 0.0668 0.2312 * 0.0013 0.0908 SingleSource/Regression/C++/EH/ctor_dtor_count | pass 0.1155 0.1741 * 0.0011 0.0937 SingleSource/Regression/C++/EH/ctor_dtor_count-2 | pass 0.0466 0.1044 * 0.0011 0.0950 SingleSource/Regression/C++/EH/exception_spec_test | pass 0.0990 0.1530 * 0.0013 0.0099 SingleSource/Regression/C++/EH/function_try_block | pass 0.0485 0.0756 * 0.0013 0.0981 SingleSource/Regression/C++/EH/inlined_cleanup | pass 0.0527 0.0866 * 0.0015 0.0937 SingleSource/Regression/C++/EH/recursive-throw | pass 0.0451 0.0777 * 0.0010 0.0902 SingleSource/Regression/C++/EH/simple_rethrow | pass 0.0341 0.0692 * 0.0013 0.0890 SingleSource/Regression/C++/EH/simple_throw | pass 0.0319 0.0632 * 0.0014 0.0792 SingleSource/Regression/C++/EH/throw_rethrow_test | pass 0.0562 0.0977 * 0.0014 0.0995 Something with exception handling does not work (obviously). SingleSource/UnitTests/Threads/2010-12-08-tls | * 0.0406 0.0962 * 0.0001 0.0019 SingleSource/UnitTests/Threads/tls | * 0.0709 0.3802 * 0.0001 0.0015 Expected failures from our lack of TLS support. External/SPEC/CINT2006/471.omnetpp/471.omnetpp | pass 74.8991 90.5139 * 1.5816 1.8107 SingleSource/UnitTests/2009-04-16-BitfieldInitialization | pass 0.0642 0.1969 * 0.0017 0.0098 I've not yet investigated these in detail. The remaining ~500 tests pass as expected. In short, things are coming along quite nicely. Thanks again everyone, Hal -- Hal Finkel Assistant Computational Scientist Leadership Computing Facility Argonne National Laboratory