[mpich-discuss] make fails with undefined reference VALGRIND_CHECK_MEM_IS_ADDRESSABLE

[Wei-keng Liao]

When using mpd, strace did not find a problem of file opening.
But hydra works fine. Thanks.

Wei-keng

On May 18, 2010, at 2:08 PM, Dave Goodell wrote:

> Darius made a good point out-of-band.  Are you using mpd?  If so,  
> can you try using hydra instead?  MPD's process mapping support has  
> been flaky in the past, and could possibly cause this sort of error.
>
> http://wiki.mcs.anl.gov/mpich2/index.php/Using_the_Hydra_Process_Manager
>
> -Dave
>
> On May 18, 2010, at 1:46 PM, Dave Goodell wrote:
>
>> That sounds like an unrelated issue.  Can you slap strace on there  
>> and see what file open is failing to open?
>>
>> -Dave
>>
>> On May 18, 2010, at 1:40 PM, Wei-keng Liao wrote:
>>
>>> Hi, Dave,
>>>
>>> I tried your approach by removing "meminit" from "--enable-g" and  
>>> deleting lines 25-35 of "src/include/mpiu_valgrind.h", but got the  
>>> following errors when ran a simple hello program.
>>>
>>> I am using "--enable-g=dbg".
>>>
>>> % mpiexec -l -n 4 hello
>>> 3: Fatal error in MPI_Init: Other MPI error, error stack:
>>> 3: MPIR_Init_thread(394).................: Initialization failed
>>> 3: MPID_Init(135)........................: channel initialization  
>>> failed
>>> 3: MPIDI_CH3_Init(43)....................:
>>> 3: MPID_nem_init(202)....................:
>>> 3: MPIDI_CH3I_Seg_commit(366)............:
>>> 3: MPIU_SHMW_Hnd_deserialize(358)........:
>>> 3: MPIU_SHMW_Seg_open(897)...............:
>>> 3: MPIU_SHMW_Seg_create_attach_templ(671): open failed - No such  
>>> file or directory
>>> rank 3 in job 9  aum0.cse.psu.edu_33436   caused collective abort  
>>> of all ranks
>>> exit status of rank 3: return code 1
>>>
>>>
>>>
>>> Wei-keng
>>>
>>> On May 18, 2010, at 12:31 PM, Dave Goodell wrote:
>>>
>>>> Yeah, that's the reason.  I'll make some adjustments to the build  
>>>> system to handle this.
>>>>
>>>> -Dave
>>>>
>>>> On May 18, 2010, at 11:50 AM, Wei-keng Liao wrote:
>>>>
>>>>> It is valgrind-2.4.0
>>>>>
>>>>> Wei-keng
>>>>>
>>>>> On May 18, 2010, at 11:45 AM, Anthony Chan wrote:
>>>>>
>>>>>>
>>>>>> How about "valgrind --version" ?
>>>>>>
>>>>>> ----- "Wei-keng Liao" <wkliao at ece.northwestern.edu> wrote:
>>>>>>
>>>>>>> I checked valgrind.h on my system, but do not know how to tell  
>>>>>>> the
>>>>>>> valgrind version. Let me know if you have an idea.
>>>>>>>
>>>>>>>
>>>>>>> Wei-keng
>>>>>>>
>>>>>>> On May 18, 2010, at 10:51 AM, Dave Goodell wrote:
>>>>>>>
>>>>>>>> What version of valgrind do you have installed?
>>>>>>>>
>>>>>>>> I suspect that your version of valgrind is older than  
>>>>>>>> anything that
>>>>>>>
>>>>>>>> we've tested with and is slipping between the gap of "is  
>>>>>>>> valgrind.h
>>>>>>>
>>>>>>>> available" and "does valgrind.h work".  I can strengthen the
>>>>>>>> configure test if I have an idea of what version of valgrind  
>>>>>>>> you
>>>>>>> have.
>>>>>>>>
>>>>>>>> In the interim, the best workaround for you is probably to  
>>>>>>>> either
>>>>>>>> remove "meminit" from your "--enable-g" arg, hack up the  
>>>>>>>> configure
>>>>>>>
>>>>>>>> tests for valgrind so that they always fail, or delete lines  
>>>>>>>> 25-35
>>>>>>>
>>>>>>>> of "src/include/mpiu_valgrind.h".
>>>>>>>>
>>>>>>>> -Dave
>>>>>>>>
>>>>>>>> On May 18, 2010, at 10:38 AM, Wei-keng Liao wrote:
>>>>>>>>
>>>>>>>>>
>>>>>>>>> When running make of mpich2-1.2.1p1, I got an error  
>>>>>>>>> complaining
>>>>>>>>> undefined reference to VALGRIND_CHECK_MEM_IS_ADDRESSABLE.  
>>>>>>>>> Please
>>>>>>>>> see the attached standard outputs from configure and make.
>>>>>>>>> The configure command and environment variables used are  
>>>>>>>>> also given
>>>>>>>
>>>>>>>>> in out.configure file.
>>>>>>>>>
>>>>>>>>> I am using GNU compilers, gcc and gfortran, to build MPICH2.
>>>>>>>>>
>>>>>>>>> gcc version 4.0.2 20051125 (Red Hat 4.0.2-8)
>>>>>>>>>
>>>>>>>>> % rpm -q glibc
>>>>>>>>> glibc-2.3.6-3
>>>>>>>>>
>>>>>>>>> % uname -a
>>>>>>>>> Linux 2.6.17-1.2142_FC4smp #1 SMP Tue Jul 11 22:57:02 EDT  
>>>>>>>>> 2006 i686
>>>>>>>
>>>>>>>>> athlon i386 GNU/Linux
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Wei-keng
>>>>>>>>>
>>>>>>>>> <out.configure><out.make>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> mpich-discuss mailing list
>>>>>>>>> mpich-discuss at mcs.anl.gov
>>>>>>>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> mpich-discuss mailing list
>>>>>>>> mpich-discuss at mcs.anl.gov
>>>>>>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> mpich-discuss mailing list
>>>>>>> mpich-discuss at mcs.anl.gov
>>>>>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>>>>> _______________________________________________
>>>>>> mpich-discuss mailing list
>>>>>> mpich-discuss at mcs.anl.gov
>>>>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> mpich-discuss mailing list
>>>>> mpich-discuss at mcs.anl.gov
>>>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>>>
>>>> _______________________________________________
>>>> mpich-discuss mailing list
>>>> mpich-discuss at mcs.anl.gov
>>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>>>
>>>
>>> _______________________________________________
>>> mpich-discuss mailing list
>>> mpich-discuss at mcs.anl.gov
>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>
>> _______________________________________________
>> mpich-discuss mailing list
>> mpich-discuss at mcs.anl.gov
>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>
> _______________________________________________
> mpich-discuss mailing list
> mpich-discuss at mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>