Boost logo

Boost Testing :

From: Boris Gubenko (Boris.Gubenko_at_[hidden])
Date: 2007-11-29 18:36:48


  This is what I got so far:

  <http://tinyurl.com/32exwz> describes condition_test failure. The
  test fails the same way on Tru64 and HP-UX: it hangs on the call to
  pthread_cond_wait() function in interprocess_mutex::lock() function
  in boost/interprocess/sync/posix/interprocess_mutex.hpp. I verified
  it with the debugger on both systems and, also, by adding some
  diagnostics to the source. Eventually, the test gets killed by bjam:
  the -lx option.

inline void interprocess_mutex::lock()
{
    int res = 0;
    res = pthread_mutex_lock(&m_mut);
    assert(res == 0);

    while (m_locked){
        res = pthread_cond_wait(&m_cond, &m_mut); <<<< hangs
        assert(res == 0);
    }
    ...

  I did not spend any time investigating the hang because I figured
  you may just know what might be causing it and why only these two
  Unix systems are affected (some configuration problem? The compiler
  does not seem to be the factor: it is gcc on HP-UX and cxx on Tru64
  (cxx is EDG-based compiler).

  Please let me know how you want to proceed. If it is unclear to you
  what might be causing the hang, I'll investigate.

  Thanks,
    Boris

----- Original Message -----
From: "Ion Gaztañaga" <igaztanaga_at_[hidden]>
To: "Boost Testing List" <boost-testing_at_[hidden]>
Sent: Tuesday, November 27, 2007 3:34 PM
Subject: [Boost-testing] [interprocess] HP-UX_ia64_gcc failures

> Hi,
>
> After recent changes, HP-UX_ia64_gcc (gcc-4.2.1) failures have been
> reduced to 13 and those are the only remaining failures for
> Interprocess. There are some strange failures. For example:
>
> http://tinyurl.com/32exwz
>
> Run [2007-11-26 19:09:49 UTC]: fail
>
> do_test_condition_notify_one<N5boost12interprocess22interprocess_conditionE,N5boost12interprocess18interprocess_mutexE
> do_test_condition_notify_all<N5boost12interprocess22interprocess_conditionE,N5boost12interprocess18interprocess_mutexE
>
> fails without the return error. Maybe an exception?
>
> This one (http://tinyurl.com/2o7ohr) fails with:
>
> Run [2007-11-26 19:09:52 UTC]: fail
>
> Unhandled exception: Not enough space
> terminate called after throwing an instance of
> 'boost::interprocess::interprocess_exception'
> what(): Not enough space
>
> EXIT STATUS: 134
>
> The test is trying to create a mapped file. Maybe a permission issue
> when creating the file?
>
> The same tests on the same gcc version but on linux (Sandia-gcc-64) seem
> ok, so I need a bit of help here.
>
> Regards,
>
> Ion
> _______________________________________________
> Boost-Testing mailing list
> Boost-Testing_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-testing
>


Boost-testing list run by mbergal at meta-comm.com