Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] totalview with OMPI 1.3 on rh5 linux
From: Terry Dontje (Terry.Dontje_at_[hidden])
Date: 2009-05-20 13:52:58


Well, I think George's previous statement of my issue was the compiler
is correct. I can compile OMPI 1.3 and trunk with gcc and things work
(except for the mca_topo issue Ashley brought up). However when I use
Sun's compilers the undef symbols occur in the message queue plugin for
both branches.

doh,

--td

Ashley Pittman wrote:
> On Tue, 2009-05-19 at 13:21 -0400, Terry Dontje wrote:
>
>> Actually playing with Ashley´s program shown that RTLD_NOW does error
>> out in the exact same way. So could the problem be that totalview
>> dlopen´s the plugin with RTLD_NOW passed into dlopen.
>>
>
> Passing the RTLD_NOW flag is probably the right thing to do, the problem
> lies with the library, not TotalView.
>
>
>> I would have
>> thought if this was the case we would have seen this problem sooner.
>>
>
> Indeed. It would certainly be something worth adding a test for.
>
> I can confirm the svn trunk doesn't suffer from this problem, I've been
> using it all week.
>
>

> Ashley,
>
>
>> George Bosilca wrote:
>>
>>> No, this is us trying to outsmart compilers. It turned out that at one
>>> point they outsmart us ...
>>>
>>> Anyway, I blame the compiler on this one. As we need the definition of
>>> the ompi_free_list_t structure we have to include the header file. We
>>> never use anything from this header file, especially not the
>>> ompi_free_list functions. However, as one of the functions in the
>>> header file make a reference to the ompi_free_list_grow function
>>> (defined in the .c file), the compiler include a reference to this in
>>> the library.
>>>
>>> george.
>>>
>>>
>>>
>>> On Fri, 15 May 2009, Jeff Squyres wrote:
>>>
>>>
>>>> Could well be our visibility settings, too... Are those symbols
>>>> OMPI|ORTE|OPAL_DECLSPEC'ed?
>>>>
>>>>
>>>> On May 15, 2009, at 10:26 AM, Ashley Pittman wrote:
>>>>
>>>>
>>>>> On Fri, 2009-05-15 at 07:43 -0600, Ralph Castain wrote:
>>>>>
>>>>>> We are running it with 1.3.2, last I heard - haven't tried the current
>>>>>> 1.3 branch. Ashley reported a problem with some other symbol that
>>>>>> couldn't be loaded that blocked him on message queue debugging, but
>>>>>> that was on the trunk.
>>>>>>
>>>>> My problem is that the library opens but is expecting to find a symbol
>>>>> in the application that isn't there, it sounds like terry problem is
>>>>> the
>>>>> debugger isn't loading the so.
>>>>>
>>>>>
>>>>>> Sounds like we are missing at least a couple of header files.
>>>>>>
>>>>> My first thought for this would be to write a simple C program to
>>>>> dlopen() the so and see if that works, both with RTLD_NOW and RTLD_LAZY
>>>>> as I don't know which TotalView uses. Try something like the
>>>>> following.
>>>>>
>>>>> #include <dlfcn.h>
>>>>>
>>>>> int main (argc, argv)
>>>>> int argc;
>>>>> char *argv[];
>>>>> {
>>>>> void *h;
>>>>> h = dlopen(argv[1],RTLD_NOW);
>>>>> printf("Result is %p\n",h);
>>>>> return 0;
>>>>> }
>>>>>
>>>>> Ashley,
>>>>>
>>>>> _______________________________________________
>>>>> devel mailing list
>>>>> devel_at_[hidden]
>>>>> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>>>>>
>>>>
>>>>
>>> "We must accept finite disappointment, but we must never lose infinite
>>> hope."
>>> Martin Luther King
>>>
>>> _______________________________________________
>>> devel mailing list
>>> devel_at_[hidden]
>>> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>>>
>> _______________________________________________
>> devel mailing list
>> devel_at_[hidden]
>> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>>
>
>
> _______________________________________________
> devel mailing list
> devel_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>