Testing Pthreads with Visual Studio .Net

Testing Pthreads with Visual Studio .Net

[Pietrobon Marcello]

I tested Pthreads with Visual Studio .Net build 7.0.9372.
I run Windows 2000 on a Pentium 4 processor (1.0 GHz) and using the CVS
code repository Updated the 2002-02-19 12:31pm

The library compiles fine but not all the tests are passed
1)
condvar1_2:
assert (result == 0); failed because result==11 (EAGAIN: Not all CVs
were broadcast)

2)
Mutext8.c, Mutext8n.c, Mutext8e.c, Mutext8r.c do not exist but the
Makefile requires them

3)
cancel6a.c, cancel6d.c  do not exist but the Makefile requires them

4)
spin4.c
assert(result > 1000); failed   because result = 0

5)
exception3.c
assert(caught == NUMTHREADS);  failed   because caught = 0

6)
exit4.c :
assert(counter == init_counter_value);  failed   because counter = 4

Best Regards,
Marcello

P.S. What is the best command to use with csv in order to know if the
code repository has been changed, and not having an output too long ?
(it is the first time I use csv)

Re: Testing Pthreads with Visual Studio .Net

[Scott McCaskill]

> P.S. What is the best command to use with csv in order to know if the
> code repository has been changed, and not having an output too long ?
> (it is the first time I use csv)

cvs -q -n update -d

Re: Testing Pthreads with Visual Studio .Net

[Pietrobon Marcello]

I went again to the cvs and now I could find the files
Mutext8.c, Mutext8n.c, Mutext8e.c, Mutext8r.c and cancel6a.c, cancel6d.c
so 2) and 3) are not an issue.

Marcello

Re: Testing Pthreads with Visual Studio .Net

[Pietrobon Marcello]

This is just for information.
If I test pthreads-2001-07-12 things are a little bit different.
Once I fix barriers.c and spin.c with:
#if defined (__MINGW32__) || (_MSC_VER >= 1300)
# define _LONG long
# define _LPLONG long*
#else
# define _LONG PVOID
# define _LPLONG PVOID*
#endif


these are the tests errors I get:
spin2.c :
assert(pthread_spin_trylock(&lock) == EBUSY);  failed  because
pthread_spin_trylock(&lock) = 22

spin3.c :
assert(pthread_spin_unlock(&spin) == EPERM);  failed  because
pthread_spin_unlock(&spin) = 22

spin4.exe :
 The exception Privileged instruction. - (0xc0000096) occurred in the
application at location 0x77e8699b.

exception3.c :
assert(caught == NUMTHREADS);  failed  because  caught = 0

Regards,
Marcello

Re: Testing Pthreads with Visual Studio .Net

[Ross Johnson]

Pietrobon Marcello wrote:
> 
> I went again to the cvs and now I could find the files
> Mutext8.c, Mutext8n.c, Mutext8e.c, Mutext8r.c and cancel6a.c, cancel6d.c
> so 2) and 3) are not an issue.
> 

Yes, sorry, they were missing and I've just added them.

Hopefully 1) is now fixed.
There was a recently added but misplaced CS lock in
pthread_cond_destroy(), which the condvar1_2.c test was
finding on your SMP machine. This is a new test for
the new pthread_timechange_handler_np() routine.
Good to know that the test actually caught a problem
inside the library.

Also have updated spin4.c to try to fix 4).

5) is a known bug, but no idea what is wrong here.
Any help would be appreciated. Interesting to know
that it's not just MSVC 6.0 related.

6) is also known, but again not sure why.

5) and 6) pass for the GNU g++ compiler versions.

Cheers, and many thanks for the feedback, especially
the SMP testsuite results.

Ross