On 7 July 2017 at 09:11, Nitesh Jain via lldb-dev <lldb-dev@lists.llvm.org> wrote: > Hi, > > > > The “ninja -j1 check-lldb-single” hangs on MIPS-Linux after running all the > tests successfully (0 failures 0 errors) with latest ToT. It doesn’t print > summary message and does not return to prompt. > > Even with “LLDB_TEST_THREADS=1 check-lldb”, we have observed the same > behaviour. Setting LLDB_TEST_USER_ARGS_NEW to “--test-runner-name serial” > does not help as it does not detect any test cases.
You're using some fairly deep magic here, that I'm not sure if anyone uses on a regular basis, so I'm not surprised that it does not work. :) > With “check-lldb” (without -single), we see few random failures each time we > run the test suite. However these tests pass when run individually (using > dotest.py). > > These random failures occur where tests shares a common source file (for > example tests under packages/Python/lldbsuite/test/tools/lldb-server). Tests from the same directory sharing a source file should not matter, as we run each directory serially. If the sources are in different directories, then that's a different story, but I think we don't have those. What's the error message? Does it have something to do with refused connections? I've noticed that the ipv6 patch made the lldb-server tests unreliable in some environments, but I haven't yet looked into why. pl _______________________________________________ lldb-dev mailing list lldb-dev@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-dev
