[mpich-discuss] Fwd: gcov question with MPICH2

[Dave Goodell]

I've never hit this personally on my Mac, but I don't run coverage  
tests by hand very often and our nightly coverage tests aren't run on  
Mac.

-Dave

On Sep 21, 2009, at 10:55 PM, Barry Smith wrote:

>
>  When I run the same tests on a Linux system I get a very small  
> number of these messages and it never slows down or hangs.
> I guess the problem is with gcov or file locking on the Mac.
>
>   Barry
>
>
> Begin forwarded message:
>
>> From: Barry Smith <bsmith at mcs.anl.gov>
>> Date: September 21, 2009 3:17:06 PM CDT
>> To: mpich-discuss at mcs.anl.gov
>> Subject: Fwd: gcov question with MPICH2
>>
>>
>>
>>>>
>>>>
>>>>
>>>>
>>>> This isn't a bug report on MPICH2 but a question.
>>>>
>>>> When I compile and build PETSc with gcov (and MPICH2), when I run  
>>>> our PETSc test suite I sometimes get messages like
>>>>
>>>> profiling:/Users/barrysmith/Src/petsc-dev/externalpackages/ 
>>>> mpich2-1.0.8/src/util/mem/argstr.gcda:Invocation mismatch - some  
>>>> data files may have been removed
>>>>
>>>> but most of the tests do not give these messages and sometimes it  
>>>> really slows down or hangs (but not always).
>>>>
>>>> This is on a Mac running 10.5
>>>>
>>>> Have you ever seen these when running MPICH tests with gcov?
>>>>
>>>> Thanks
>>>>
>>>> Barry
>>>>
>>>> Note: the "complaints" are always for MPICH source code, never  
>>>> PETSc source??
>>>>
>>>
>>
>